research article a fair resource allocation algorithm for ... · called deins (data and energy...
TRANSCRIPT
Research ArticleA Fair Resource Allocation Algorithm for Data andEnergy Integrated Communication Networks
Qin Yu1 Yizhe Zhao1 Lanxin Zhang1 Kun Yang2 and Supeng Leng1
1School of Communication and Information Engineering University of Electronic Science and Technology of ChinaChengdu 611731 China2School of Computer Science and Electronic Engineering University of Essex Colchester UK
Correspondence should be addressed to Qin Yu yuqinuestceducn
Received 3 December 2015 Accepted 18 January 2016
Academic Editor Mianxiong Dong
Copyright copy 2016 Qin Yu et al This is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
With the rapid advancement of wireless network technologies and the rapid increase in the number of mobile devices mobile users(MUs) have an increasing high demand to access the Internet with guaranteed quality-of-service (QoS) Data and energy integratedcommunication networks (DEINs) are emerging as a new type of wireless networks that have the potential to simultaneouslytransfer wireless energy and information via the same base station (BS) This means that a physical BS is virtualized into two partsone is transferring energy and the other is transferring information The former is called virtual energy base station (eBS) and thelatter is named as data base station (dBS) One important issue in such setting is dynamic resource allocation Here the resourceconcerned includes both power and time In this paper we propose a fair data-and-energy resource allocation algorithm for DEINsby jointly designing the downlink energy beamforming and a power-and-time allocation scheme with the consideration of finitecapacity batteries at MUs and power sensitivity of radio frequency (RF) to direct current (DC) conversion circuits Simulationresults demonstrate that our proposed algorithm outperforms the existing algorithms in terms of fairness beamforming designsensitivity and average throughput
1 Introduction
The strength of network virtualization is upsurge such assoftware defined networking (SDN) and collaborative radioaccess networks (C-RAN) For instance SDN is expected totransform the way services are created sourced deployedand supported [1ndash3] This paper discusses another type ofvirtualization namely a base station being virtualized intoproviding not only information transferring but also energytransferring to charge mobile devices This is largely drivenby the fact that mobile devices while getting more powerfulin processing and networking exhaust their battery morequickly
To address this challenge this paper utilizes energyharvesting into wireless communications namely the so-called DEINs (Data and energy integrated communicationnetworks) With the development of energy harvesting (EH)technologies and wireless energy transfer (WET) techniquesthe DEIN becomes an emerging trend focusing on the study
of wireless power and information cooperation communi-cations [4ndash7] Compared with simultaneous wireless infor-mation and power transfer (SWIPT) which mainly focuseson the physical layer DEINs focus on the whole networksystem resource allocation and protocols design in differentlayersThe typical architecture of aDEIN is shown in Figure 1which has three major components that is a virtual energybase station (eBS) a data base station (dBS) and massivemobile usersmobile devices The wireless communicationis divided into two parts The virtual eBS first transfersenergy to multiple mobile users (MUs) that do not haveembedded energy sources via downlink (DL) and then MUsuse the harvested energy to perform uplink (UL) wirelessinformation transmission (WIT) to the dBS
No existent works have taken into account the power sen-sitivity of RF-DC circuits when DL transfer energy in DEINswhich can lead to a falsely higher data rate when receivedRF signals cannot be converted into DC (ie energy transfer)if their power level is lower than the power sensitivity of an
Hindawi Publishing CorporationMobile Information SystemsVolume 2016 Article ID 4806452 10 pageshttpdxdoiorg10115520164806452
2 Mobile Information Systems
BS
Virtual data BS
Virtual energy BS
Information transfer
Energy transfer
Energy transfer
U1
U2
U3
Figure 1 A typical architecture of a DEIN based on virtualization
RF-DC circuit [8] Besides none of these works have consid-ered the possibility of energy overflow or the opportunitiesfor users to optimize the use of harvested energy across ULWIT slots It has been shown that a user using all availableenergy for WIT in each slot achieves a lower data rate thanuniformly distributing energy between energy arrivals [9ndash11]Therefore new dynamic time allocation schemes are neededsince not all MUs can harvest energy in every slot which takeinto consideration the policy that the energy harvesting ofevery user should not overflow in every DL WET phase andthe energy harvested in theDLWETphase of former slotmaybe used in the ULWIT of the next
In this paper we devise a fair resource allocation algo-rithm for a DEIN based on dynamical time-slot division toguarantee the received power fairness among all MUs bymaximizing the minimum average DL WET power via opti-mizing the DL energy beamformer In particular consideringthe power sensitivity of the RF-DC conversion circuits thevirtual eBS transfers energy to these MUs whose estimatedreceived power is larger than the corresponding certainthreshold 120572 The proposed algorithm benefits the followingfeatures
(1) Fairness Since there are more than one user in ourscenario fairness among all MUs should be taken intoaccount naturally While each MU is allocated the sameslot for UL transmission by space-division-multiple-access(SDMA) leading to the fact that the increase or decreaseof slot length would lead to similar trend of each MUrsquosthroughput the fairness of the model is mainly reflected bythe WET energy beamforming To achieve the fairness weaim to allocate the optimal beamformer to maximize theminimum received power among MUs which can furtherlead to the fairness of throughput on the whole
(2) Sensitivity We study the UL transmission powered byWET in a DEIN and model the WET of every user asa Bernoulli process with different probability while takinginto account the different sensitivity of RF-DC circuits andcalculate the accurate WET probability of different users
(3) Throughput Performance Considering the sensitivity ofRF-DC circuits that is the WET probability we propose anew low complexity resource allocation algorithm to achievenear optimization throughput with finite capacity batterieswhich employs the zero-forcing (ZF) based receive beam-forming in the UL information transmission [12]
The rest of this paper is organized as follows Section 2illustrates some related works about DEINs in recent yearsSection 3 presents a multiantenna DEIN model based onvirtualization and formulates problems based on fairnessSection 4 presents a dynamic resource allocation to makea near throughput optimization for this problem Section 5provides simulation results to compare the performances ofproposed solutions with exiting algorithms Finally Section 6concludes the paper
Notations All lowercased and uppercased boldface let-ters represent vectors and matrices respectively Let tr(119883)det(119883) 119883minus1 and 119883
119867 denote the trace determinant inverseand Hermitian of a symmetric matrix 119883 respectively C andR denote the set of complex and real matrices of size 119909 times 119910and C and R denote the set of complex and real vectors ofsize 119909times1 respectively All the log(sdot) functions are of base 2 bydefault and ln(sdot) stands for the natural logarithm All lettersat the right bottom of different variables can be explained bythe following 119897 shows the 119897th slot and 119894 is the different users
2 Related Works
In this section we introduce the related energy harvesting-based data-and-energy transmission techniques
There are so many works focusing on designing resourceallocation schemes for different networks In [13] anenergy-efficient context-aware resource allocation problemin caching-enabled ultradense small cells is investigatedIn [14] a new analytical performance model is developedto evaluate the QoS of multihop cognitive radio networksMoreover we introduce some current researches on DEINsin the following
SWIPT has been recently studied in the literature (seeeg [15ndash22]) where the achievable information versus energytransmission tradeoffs was characterized under differentchannel setups SWIPT proposed in [15] has been exten-sively studied which can offer great convenience to mobileusers with concurrent data and energy supplies SWIPT hasbeen studied in orthogonal frequency division multiplexing(OFDM) systems [16ndash18] and multiuser channel setups suchas relay channels [19 20] and interference channels [21 22]
Moreover another problem addressing the joint designof DL energy transfer and UL information transmission isworth investigating [12 23ndash25] UL WIT powered by DLWET inDEINswas studied in [23] Assuming perfect channelstate information (CSI) the single-user DEIN scenario wasstudied in [23] The authors maximized the energy efficiencyof uplink WIT by jointly optimizing the time duration andtransmitted power for downlinkWET Besides some work isfocusing on resource allocation for optimization throughputin DEINs In [24] a DEIN with single-antenna access point
Mobile Information Systems 3
M
BS
processingSignal
Energy transfer
Information transfer
BatteryRF to DC converter
Transceiver
processingSignal
BatteryRF to DC converter
Transceiver
Mobile user U1
Mobile user UK
(1 minus 120591)T
120591T
(1 minus 120591)T
120591T
Qmax
Qmax
Figure 2 A DEIN system model
(AP) and users has been studied for joint DL energy transferand UL information transmission A harvest-then-transmitprotocol was proposed and orthogonal time allocations forthe DL energy transfer and UL information transmissionsof all users are jointly optimized to maximize the networkthroughput In [12] a DEIN where one multiantenna APcoordinates energy transfer and information transfer tofroma set of single-antenna users was studied and the authormaximizes the minimum throughput among all users by ajoint design of the DL-UL time allocation the DL energybeamforming and the UL transmit power allocation as wellas receive beamforming A WET enabled massive MIMOsystem with imperfect CSI was studied in [25] where it isbased on time-division-duplexing (TDD) protocol and eachframe is divided into three phases the UL channel estimationphase the DLWET phase and the ULWIT phase
But all these works take the idea that the MUs transfer allthe received power in DL ignoring the power sensitivity ofRF-DC circuits [9ndash11] Besides no works take the advantageof the battery storage capacity to make a dynamic powerallocation which can have a higher throughput Even thoughsome works use the parameter (ie the battery storagecapacity) they also overlook the fact that the capacitycannot overflow when making the resource allocation In thefollowing we are going to discuss these questions
3 System Model and Problem Formulation
In this section we present a dynamical time slotted transmis-sion scheme and analyze the DL WET phase and UL WITphase Finally we formulate problems based on fairness
We provide a DEIN model consisting of a BS with119872 antennas and 119870 single-antenna MUs with finite battery
DL WIT
DL WET UL WIT
Premable FCHDL
mapUL
map
Control frame
(1 minus 120591)T120591T
Figure 3 Frame structure
capacity denoted by 119880119894
(119894 = 1 119870) as shown in Figure 2It is assumed that 119872 ge 119870 It should be noted that theBS is virtualized into two parts that is a virtual eBS anda dBS as shown in Figure 1 Each MU uses the harvestedenergy from DL WET phase via beamforming of the virtualeBS to power its UL information transmission via the dBSunder the assumption that two BSs and all MUs are perfectlysynchronized and there is no other energy source ofMUsThetotal capacity of battery storage in every MU is 119876max
A time slotted transmission scheme is considered asshown in Figure 3 We assume that the period of DL DIT inevery slot can be neglected since there is little informationto transmit in the DL phase So each slot has a constantperiod 119879 consisting of two phases namely the DL WETphase of duration 120591 sdot 119879 and the UL WIT phase of duration(1 minus 120591) sdot 119879 where 0 le 120591 le 1 The WET phase starts withseveral control frames including the preamble frame controlheader (FCH) DL map and UL map These frames definetransmission parameters such as coding schemes available
4 Mobile Information Systems
resources the duration of DL and UL transmission and theWET probability (which will be defined in the following)Then virtual energy BS transmits energy to 119880
119894
throughwireless energy beamforming The received power level andthe energy harvested at 119880
119894
in slot 119897 (119897 = 1 119873) aredenoted by 119875
119897119894
and 119864119897119894
It is worth noting that due to thepower sensitivity of RF energy harvesting circuits 119880
119894
cannotharvest any RF energy if the received signal power 119875
119897119894
isbelow a certain level Thus the received power at every MUcan be used if the received power level 119875
119897119894
is larger thanthe corresponding certain threshold (eg minus10 dBm) Thisindicates that the WET of every MU follows a Bernoulliprocess with different probability 119901
119894
where 119901119894
stands for theprobability of delivering energy from virtual eBS Next allMUs do UL WIT phase simultaneously via SDMA which ispowered by the energy stored in the batteries For simplicitywe assume that 119879 = 1 s and that the harvested energy isstored in the battery first and then used for UL informationtransmission Note that since the length of control frames ismuch smaller than that of DL WET and UL WIT we ignorethe time duration of control frames in the following analysis
We consider frame-based transmissions over flat-fadingchannels on a single frequency band [26] (ie which meansthe channel remains constant in each slot) Denote ℎ
119897119894
isin
C119872lowast1 as the UL channel of 119880119894
in the 119897th slot and we have
ℎ119897119894
= (1205720
1003816100381610038161003816119863119894
1003816100381610038161003816
minus120573
119862119894
)
12
119892119897119894
119894 = 1 119870 (1)
where 1205720
denotes a constant determined by the RF propa-gation environment 119863
119894
is the propagation distance 120573119894
is thepath loss119862
119894
is Shadow fading and119892119897119894
isin C119872lowast1 is thematrix ofRayleigh fading coefficients and 119892
119896119894
sim 119862119873(0 1) By exploit-ing the channel reciprocity the DL transmission channel canbe obtained as ℎ119867
119897119894
For simplicity we assume that 119862119894
= 1 andthat CSI is available at both two BSs and 119880
119894
31 DL WET Phase Assume that just one energy beam isto transmit energy from virtual energy BS to those userssatisfying the probability 119901
119894
since we just transmitted energysignals in the DL [27] Besides we can allocate the optimalbeamformer on the constraint that the power transferred atall of the antennas is equal to the eBSrsquos transmitted powerHence we can assume the beamformer as a ratio of 119875 whichsatisfies the fact that its norm is a unit value where 119875 isthe transmitted power of eBS Then the practical powertransferred can be denoted as the beamformer multiplied byeBS transmitted power 119875 Moreover ambient channel noiseenergy cannot be harvested Thus the DL received signalreceived power and harvested energy of 119880
119894
in the 119897th slot areexpressed as
119910119897119894
= ℎ119867
119897119894
120596119897
1199091198970
+ 119899119897119894
119894 = 1 119870 (2)
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (3)
119864119897119894
= 120598119897
120591119897
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (4)
where 119899119897119894
sim 119862119873(0 1205902
119894
) is the receiver noise 120596119897
is the 119872 times 1
beamforming and satisfies 120596119897
2
= 1 1199091198970
is the transmission
signal and satisfies 11990921198970
le 119875max where 119875max is the transmitpower constraint and 120598
119894
denotes the energy harvestingefficiency at119880
119894
which should satisfy 0 lt 120598119894
le 1 For simplicitywe assume 120598
119894
= 1
32 UL WIT Phase In the UL WIT phase of each slotMUs use the harvested energy to power UL informationtransmission to the virtual dBS For convenience we assumethe circuit energy consumption at119880
119894
is 0 The received signalat the normal dBS in the 119897th slot is given by
119910119897
=
119870
sum
119894=1
ℎ119897119894
119909119897119894
+ 119899119897
119894 = 1 119870 (5)
where 119899119897
isin C119872times1 denotes the receiver additivewhiteGaussiannoise (AWGN) It is assumed that 119899
119897
sim 119862119873(0 1205902
119897
Γ) Besideswe assume that the normal information BS employs linearreceivers to decode 119909
119897119894
in the UL 119909119897119894
denotes the transmitsignal of 119880
119894
and satisfies 1199092
119897119894
= 1198751015840
119897119894
where 1198751015840
119897119894
is thetransmit power of 119880
119894
Specifically let V119897119894
isin C119872times1 denotethe receive beamforming vector for decoding 119909
119897119894
and define119881 = V
1198971
V119897119870
In order to reduce complicity we employthe ZF based receive beamforming in the normal informationBS proposed by [12] which is not related to 119908
119897
and 120591119897
Define 119867
minus119897119894
= [ℎ1198971
ℎ119897119894minus1
ℎ119897119894+1
ℎ119897119894
]119867
119894 = 1 119870including all the UL channels except ℎ
119897119894
Then the singularvalue decomposition (SVD) of 119867
minus119897119894
is given as 119867minus119897119894
=
119883119897119894
Λ119897119894
119884119867
119897119894
= 119883119897119894
Λ119897119894
[119884119897119894
119897119894
]119867 Thus the beamforming can
be expressed as VZF119897119894
= 119897119894
119867
119897119894
ℎ119897119894
119867
119897119894
ℎ119897119894
Then throughputof 119880119894
in bitssecondHz (bpsHz) can be expressed as
119877ZF119897119894
= (1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
) 119894 = 1 119870 (6)
where ℎ119897119894
= 119897119894
119867ℎ119897119894
2
The energy consumption of UL WIT of 119880119894
in the 119897th slotis given by
119902119897119894
= (1 minus 120591119897
) 1198751015840
119897119894
(7)
Let 119876119897119894
represent the amount of energy available in thebattery of 119880
119894
at slot 119897 and its updating function is as follows
119876119897119894
= min (119876119897minus1119894
+ 119864119897119894
minus 119902119897minus1119894
119876max) (8)
There are two constraints in the UL WIT phase theenergy causality constraint and the battery storage constraint[9 11] Specifically the energy causality constraint requiresthat the UL WIT can only use the energy harvested at thecurrent and previous slots and the battery storage constraintindicates that the energy available of 119880
119894
cannot exceed themaximum battery capacity at any time that is
119873
sum
119897=0
[119864119897119894
minus 119902119897119894
] ge 0
119873+1
sum
119897=0
119864119897119894
minus
119873
sum
119897=0
119902119897119894
le 119876max
(9)
Mobile Information Systems 5
33 Problem Formulation Considering fairness we optimizemaximum-minimum (max-min) average ULWIT through-put of all MUs by jointly optimizing the time allocation 120591
119897
theDL energy beams 119908
119897
and the UL transmit power allocation1198751015840
119897119894
Let 119903ZF(119902119897119894)
denote theULdata rate of119880119894
in slot 119897 as a functionof the allocated energy 119902
119897119894
forUL transmission in slot 119897 Noticethat 119902
119897119894
is a feasible energy allocation policy which satisfies
0 le 119902119897119894
le 119876max
119876119897+1119894
= min (119876119897119894
+ 119864119897+1119894
minus 119902119897119894
119876max)
119902119897119894
= 120601 (119897 119864119894
119897
119898=1
)
(10)
The expression (10) shows energy causality constraint thebattery storage constraint and casualty CSI We should notethat (4) should satisfy the constraint 1199092
1198970
le 119875maxThus the optimization problem satisfying constraints (10)
can be defined as
max min lim119873rarrinfin
1
119873
119873
sum
119897=1
(1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
)
0 le 120591119897
le 120578119897
(11)
where 120578119897
= min(min119897
((119876119897maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery overflow
4 Dynamic Resource Allocation Algorithm
In this section we present a dynamic resource allocationto make a near throughput optimization for the problemformulated in the former section including DL beamformerdesign and allocation of WIT energy and duration
As the optimization problem is formulated we can divideit into two parts one of which is the optimal beamformerdesigning of DLWET for fairness and the other is the optimaltime and power allocation of UL WIT for throughput sincewe employ the ZF based receive beamforming in the normalinformation BS proposed by [12] which is not related to 119908
119897
and 120591119897
41 Design of WET Beamforming and Computation of EHProbability Considering fairness of the received power wedetermine an optimal beamformer design to maximize theminimum received power for differentMUs Tomaximize thereceived power at 119880
119894
we can set 11990921198970
= 119875max Generally thereceived power for 119880
119897
is denoted by
119875119897119894
(120596119897
) = 119875max120596119867
119897
119867119897119894
120596119897
(12)
where 119867119897119894
= ℎ119897119894
ℎ119867
119897119894
Then problem on fairness could beformulated as
max120596119897
min1le119897le119870
119875119897119894
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(13)
It is easy to relax the max-min process by introducing aslack variable 119875
119897
and then problem (13) can be transformedinto the following
max120596119897
119875119897
st 119875119897119894
(120596119897
) ge 119875119897
forall1 le 119894 le 119870
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(14)
Since the problem is convex we can follow the approachof convex theory in [28] and consider its Lagrangian functionis given by
119871 (120582 120596119897
) = minus
119870
sum
119894=1
120582119894
(119875119897119894
(120596119897
) minus 119875119897
) (15)
where 120582 = [1205821
1205822
120582119870
] ge 0 consists of Lagrangemultipliers associated with received power constraints ofMUs in problem (14)
The dual function of problem (14) is then given by
119866 (120582) = min120596119897
119871 (120582 120596119897
) (16)
The following rule can be used to determine whetherproblem (14) is feasible For given 119875
119897
gt 0 problem (14) isinfeasible in the following interpretation if and only if thereexists any 120582 ge 0 such that 119866(120582) gt 0
Next for given 120582 ge 0 we would obtain 119866(120582) by theproblem
max120596119897
119870
sum
119894=1
120582119894
119875119897119894
(120596119897
)
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(17)
With (12) the problem could be transformed as
max120596119897
119875max120596119867
119897
119860119897
120596119897
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(18)
where 119860119894
= sum119870
119894=1
120582119894
119867119897119894
Since 119860 is a symmetric matrix the optimal beamformer
120596lowast
119897
could be obtained by the result of quadratic problemAfter obtaining 120596
lowast
119897
for given 119875119897
and 120582 we can computethe corresponding 119875
119897119894
(120596lowast
119897
) and 119866(120582) in (15) If 119866(120582lowast) le 0 itindicates that problem (14) is infeasibleTherefore we shoulddecrease 119875
119897
and solve the feasibility problem in (14) againOn the other hand if 119866(120582lowast) le 0 we can update 120582 usingellipsoid method until 120582 converges to 120582
lowast which denotes themaximizer of 119866(120582) or the optimal dual solution for problem(14) If 119866(120582lowast) le 0 it indicates that the problem is feasibleThen we should increase 119875
119897
and solve the feasibility problemagain Finally119875
119897
lowast could be obtained numerically by iterativelyupdating 119875
119897
by a simple bisection search Meanwhile thecorresponding120596lowast
119897
should be the optimal beamformer Finally
6 Mobile Information Systems
(1) set a large number119873 and suc time = 0 where suc time isin 119877119870
(2) loop(3) set 119875down = 0 119875up gt 119875
119897
lowast generate a stochastic Rayleigh channel119867119897
= [ℎ1198971
ℎ119897119870
] 119899 = 0(4) loop
(5) 119875119897
=
1
2
(119875down + 119875up)
(6) set 1205822 ge 0(7) Given 120582 obtain the optimal 120596
119897
by quadratic problem(8) Computing 119866(120582) using (15)(9) if 119866(120582) gt 0 119875 is infeasible then(10) set 119875up larr 119875
119897
go to step (5)(11) else(12) update 120582 using ellipsoid method(13) if the stopping criteria of the ellipsoid method is not met then(14) go to step (7)(15) end if(16) end if(17) set 119875down larr 119875
119897
(18) if 119875up minus 119875down lt 120575 where 120575 gt 0 is a given error tolerance then(19) 119899 = 119899 + 1(20) break(21) end if(22) end loop(23) if 119899 = 119873 then(24) obtain energy harvesting probability 119901
119894
= suc time(119894)119873(25) break(26) end if(27) end loop
Algorithm 1 Optimal beamformer and EH probability obtaining algorithm
we can compute the received power for each MU anddetermine whether they could harvest the energy which canbe successful only if 119875
119897119894
ge 120572119894
To obtain the EH probability we can iterate the above
steps for 119873 times where 119873 should be a large number and adifferent transmitting channel is generated under the rule ofRayleigh distribution for each timeThe probability of119880
119894
canbe obtained by the ratio of successful harvested times and119873which is denoted by
119901119894
= lim119873rarrinfin
suc time119894
119873
(19)
where suc time119894
is the number of successful iterate timesTo summarize the algorithm to solve the problem is given
in Algorithm 1 Steps 5 to 20 give the solution to obtain anoptimal beamformer in case that CSI is available
42 Allocation of WIT Energy and Duration After obtainingthe best WET beamformer we have to determine an optimalallocation ofWIT energy and duration Consideringmultipleslots the optimization of the problem would be approachedif we keep the average transmitting energy in each slotThis is because uniformly distributing the energy betweenenergy arrivals maximizes the data rate As proposed in[29] according to the EH probability 119901
119894
we take a fraction119901119894
of available energy of 119880119894
to transmit information inUL duration With this energy allocation we only need to
compute the optimal UL duration in each slot and can obtaina global optimization with multiple slots
In the case of the system model the total throughput of119880119894
in each slot is denoted by
119877119897119894
(120591) = (1 minus 120591119897
) log(1 +ℎ119897119894
119901119894
(119876119897119894
+ 120591119875119897119894
)
(1 minus 120591119897
) 1205902
) (20)
where ℎ119897119894
119901119894
119876119897119894
and 119875119897119894
are considered as constant sincethey are calculated out before duration allocation
Like themethod proposed in Section 41 the optimizationproblem can also be formulated as
max 119877
st 119877119897119894
(120591) ge 119877 forall1 le 119894 le 119870
0 le 120591119897
le 120578119897
(21)
where 120578119897
= min(min119894
((119876119894maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery from overflowSince the problem is convex we can also solve the
problem by the Lagrangian method which is proposed inSection 41 What is different between two problems is howto obtain max
120591119897sum119870
119894=1
120582119894
119877119897119894
(120591119897
)
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
2 Mobile Information Systems
BS
Virtual data BS
Virtual energy BS
Information transfer
Energy transfer
Energy transfer
U1
U2
U3
Figure 1 A typical architecture of a DEIN based on virtualization
RF-DC circuit [8] Besides none of these works have consid-ered the possibility of energy overflow or the opportunitiesfor users to optimize the use of harvested energy across ULWIT slots It has been shown that a user using all availableenergy for WIT in each slot achieves a lower data rate thanuniformly distributing energy between energy arrivals [9ndash11]Therefore new dynamic time allocation schemes are neededsince not all MUs can harvest energy in every slot which takeinto consideration the policy that the energy harvesting ofevery user should not overflow in every DL WET phase andthe energy harvested in theDLWETphase of former slotmaybe used in the ULWIT of the next
In this paper we devise a fair resource allocation algo-rithm for a DEIN based on dynamical time-slot division toguarantee the received power fairness among all MUs bymaximizing the minimum average DL WET power via opti-mizing the DL energy beamformer In particular consideringthe power sensitivity of the RF-DC conversion circuits thevirtual eBS transfers energy to these MUs whose estimatedreceived power is larger than the corresponding certainthreshold 120572 The proposed algorithm benefits the followingfeatures
(1) Fairness Since there are more than one user in ourscenario fairness among all MUs should be taken intoaccount naturally While each MU is allocated the sameslot for UL transmission by space-division-multiple-access(SDMA) leading to the fact that the increase or decreaseof slot length would lead to similar trend of each MUrsquosthroughput the fairness of the model is mainly reflected bythe WET energy beamforming To achieve the fairness weaim to allocate the optimal beamformer to maximize theminimum received power among MUs which can furtherlead to the fairness of throughput on the whole
(2) Sensitivity We study the UL transmission powered byWET in a DEIN and model the WET of every user asa Bernoulli process with different probability while takinginto account the different sensitivity of RF-DC circuits andcalculate the accurate WET probability of different users
(3) Throughput Performance Considering the sensitivity ofRF-DC circuits that is the WET probability we propose anew low complexity resource allocation algorithm to achievenear optimization throughput with finite capacity batterieswhich employs the zero-forcing (ZF) based receive beam-forming in the UL information transmission [12]
The rest of this paper is organized as follows Section 2illustrates some related works about DEINs in recent yearsSection 3 presents a multiantenna DEIN model based onvirtualization and formulates problems based on fairnessSection 4 presents a dynamic resource allocation to makea near throughput optimization for this problem Section 5provides simulation results to compare the performances ofproposed solutions with exiting algorithms Finally Section 6concludes the paper
Notations All lowercased and uppercased boldface let-ters represent vectors and matrices respectively Let tr(119883)det(119883) 119883minus1 and 119883
119867 denote the trace determinant inverseand Hermitian of a symmetric matrix 119883 respectively C andR denote the set of complex and real matrices of size 119909 times 119910and C and R denote the set of complex and real vectors ofsize 119909times1 respectively All the log(sdot) functions are of base 2 bydefault and ln(sdot) stands for the natural logarithm All lettersat the right bottom of different variables can be explained bythe following 119897 shows the 119897th slot and 119894 is the different users
2 Related Works
In this section we introduce the related energy harvesting-based data-and-energy transmission techniques
There are so many works focusing on designing resourceallocation schemes for different networks In [13] anenergy-efficient context-aware resource allocation problemin caching-enabled ultradense small cells is investigatedIn [14] a new analytical performance model is developedto evaluate the QoS of multihop cognitive radio networksMoreover we introduce some current researches on DEINsin the following
SWIPT has been recently studied in the literature (seeeg [15ndash22]) where the achievable information versus energytransmission tradeoffs was characterized under differentchannel setups SWIPT proposed in [15] has been exten-sively studied which can offer great convenience to mobileusers with concurrent data and energy supplies SWIPT hasbeen studied in orthogonal frequency division multiplexing(OFDM) systems [16ndash18] and multiuser channel setups suchas relay channels [19 20] and interference channels [21 22]
Moreover another problem addressing the joint designof DL energy transfer and UL information transmission isworth investigating [12 23ndash25] UL WIT powered by DLWET inDEINswas studied in [23] Assuming perfect channelstate information (CSI) the single-user DEIN scenario wasstudied in [23] The authors maximized the energy efficiencyof uplink WIT by jointly optimizing the time duration andtransmitted power for downlinkWET Besides some work isfocusing on resource allocation for optimization throughputin DEINs In [24] a DEIN with single-antenna access point
Mobile Information Systems 3
M
BS
processingSignal
Energy transfer
Information transfer
BatteryRF to DC converter
Transceiver
processingSignal
BatteryRF to DC converter
Transceiver
Mobile user U1
Mobile user UK
(1 minus 120591)T
120591T
(1 minus 120591)T
120591T
Qmax
Qmax
Figure 2 A DEIN system model
(AP) and users has been studied for joint DL energy transferand UL information transmission A harvest-then-transmitprotocol was proposed and orthogonal time allocations forthe DL energy transfer and UL information transmissionsof all users are jointly optimized to maximize the networkthroughput In [12] a DEIN where one multiantenna APcoordinates energy transfer and information transfer tofroma set of single-antenna users was studied and the authormaximizes the minimum throughput among all users by ajoint design of the DL-UL time allocation the DL energybeamforming and the UL transmit power allocation as wellas receive beamforming A WET enabled massive MIMOsystem with imperfect CSI was studied in [25] where it isbased on time-division-duplexing (TDD) protocol and eachframe is divided into three phases the UL channel estimationphase the DLWET phase and the ULWIT phase
But all these works take the idea that the MUs transfer allthe received power in DL ignoring the power sensitivity ofRF-DC circuits [9ndash11] Besides no works take the advantageof the battery storage capacity to make a dynamic powerallocation which can have a higher throughput Even thoughsome works use the parameter (ie the battery storagecapacity) they also overlook the fact that the capacitycannot overflow when making the resource allocation In thefollowing we are going to discuss these questions
3 System Model and Problem Formulation
In this section we present a dynamical time slotted transmis-sion scheme and analyze the DL WET phase and UL WITphase Finally we formulate problems based on fairness
We provide a DEIN model consisting of a BS with119872 antennas and 119870 single-antenna MUs with finite battery
DL WIT
DL WET UL WIT
Premable FCHDL
mapUL
map
Control frame
(1 minus 120591)T120591T
Figure 3 Frame structure
capacity denoted by 119880119894
(119894 = 1 119870) as shown in Figure 2It is assumed that 119872 ge 119870 It should be noted that theBS is virtualized into two parts that is a virtual eBS anda dBS as shown in Figure 1 Each MU uses the harvestedenergy from DL WET phase via beamforming of the virtualeBS to power its UL information transmission via the dBSunder the assumption that two BSs and all MUs are perfectlysynchronized and there is no other energy source ofMUsThetotal capacity of battery storage in every MU is 119876max
A time slotted transmission scheme is considered asshown in Figure 3 We assume that the period of DL DIT inevery slot can be neglected since there is little informationto transmit in the DL phase So each slot has a constantperiod 119879 consisting of two phases namely the DL WETphase of duration 120591 sdot 119879 and the UL WIT phase of duration(1 minus 120591) sdot 119879 where 0 le 120591 le 1 The WET phase starts withseveral control frames including the preamble frame controlheader (FCH) DL map and UL map These frames definetransmission parameters such as coding schemes available
4 Mobile Information Systems
resources the duration of DL and UL transmission and theWET probability (which will be defined in the following)Then virtual energy BS transmits energy to 119880
119894
throughwireless energy beamforming The received power level andthe energy harvested at 119880
119894
in slot 119897 (119897 = 1 119873) aredenoted by 119875
119897119894
and 119864119897119894
It is worth noting that due to thepower sensitivity of RF energy harvesting circuits 119880
119894
cannotharvest any RF energy if the received signal power 119875
119897119894
isbelow a certain level Thus the received power at every MUcan be used if the received power level 119875
119897119894
is larger thanthe corresponding certain threshold (eg minus10 dBm) Thisindicates that the WET of every MU follows a Bernoulliprocess with different probability 119901
119894
where 119901119894
stands for theprobability of delivering energy from virtual eBS Next allMUs do UL WIT phase simultaneously via SDMA which ispowered by the energy stored in the batteries For simplicitywe assume that 119879 = 1 s and that the harvested energy isstored in the battery first and then used for UL informationtransmission Note that since the length of control frames ismuch smaller than that of DL WET and UL WIT we ignorethe time duration of control frames in the following analysis
We consider frame-based transmissions over flat-fadingchannels on a single frequency band [26] (ie which meansthe channel remains constant in each slot) Denote ℎ
119897119894
isin
C119872lowast1 as the UL channel of 119880119894
in the 119897th slot and we have
ℎ119897119894
= (1205720
1003816100381610038161003816119863119894
1003816100381610038161003816
minus120573
119862119894
)
12
119892119897119894
119894 = 1 119870 (1)
where 1205720
denotes a constant determined by the RF propa-gation environment 119863
119894
is the propagation distance 120573119894
is thepath loss119862
119894
is Shadow fading and119892119897119894
isin C119872lowast1 is thematrix ofRayleigh fading coefficients and 119892
119896119894
sim 119862119873(0 1) By exploit-ing the channel reciprocity the DL transmission channel canbe obtained as ℎ119867
119897119894
For simplicity we assume that 119862119894
= 1 andthat CSI is available at both two BSs and 119880
119894
31 DL WET Phase Assume that just one energy beam isto transmit energy from virtual energy BS to those userssatisfying the probability 119901
119894
since we just transmitted energysignals in the DL [27] Besides we can allocate the optimalbeamformer on the constraint that the power transferred atall of the antennas is equal to the eBSrsquos transmitted powerHence we can assume the beamformer as a ratio of 119875 whichsatisfies the fact that its norm is a unit value where 119875 isthe transmitted power of eBS Then the practical powertransferred can be denoted as the beamformer multiplied byeBS transmitted power 119875 Moreover ambient channel noiseenergy cannot be harvested Thus the DL received signalreceived power and harvested energy of 119880
119894
in the 119897th slot areexpressed as
119910119897119894
= ℎ119867
119897119894
120596119897
1199091198970
+ 119899119897119894
119894 = 1 119870 (2)
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (3)
119864119897119894
= 120598119897
120591119897
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (4)
where 119899119897119894
sim 119862119873(0 1205902
119894
) is the receiver noise 120596119897
is the 119872 times 1
beamforming and satisfies 120596119897
2
= 1 1199091198970
is the transmission
signal and satisfies 11990921198970
le 119875max where 119875max is the transmitpower constraint and 120598
119894
denotes the energy harvestingefficiency at119880
119894
which should satisfy 0 lt 120598119894
le 1 For simplicitywe assume 120598
119894
= 1
32 UL WIT Phase In the UL WIT phase of each slotMUs use the harvested energy to power UL informationtransmission to the virtual dBS For convenience we assumethe circuit energy consumption at119880
119894
is 0 The received signalat the normal dBS in the 119897th slot is given by
119910119897
=
119870
sum
119894=1
ℎ119897119894
119909119897119894
+ 119899119897
119894 = 1 119870 (5)
where 119899119897
isin C119872times1 denotes the receiver additivewhiteGaussiannoise (AWGN) It is assumed that 119899
119897
sim 119862119873(0 1205902
119897
Γ) Besideswe assume that the normal information BS employs linearreceivers to decode 119909
119897119894
in the UL 119909119897119894
denotes the transmitsignal of 119880
119894
and satisfies 1199092
119897119894
= 1198751015840
119897119894
where 1198751015840
119897119894
is thetransmit power of 119880
119894
Specifically let V119897119894
isin C119872times1 denotethe receive beamforming vector for decoding 119909
119897119894
and define119881 = V
1198971
V119897119870
In order to reduce complicity we employthe ZF based receive beamforming in the normal informationBS proposed by [12] which is not related to 119908
119897
and 120591119897
Define 119867
minus119897119894
= [ℎ1198971
ℎ119897119894minus1
ℎ119897119894+1
ℎ119897119894
]119867
119894 = 1 119870including all the UL channels except ℎ
119897119894
Then the singularvalue decomposition (SVD) of 119867
minus119897119894
is given as 119867minus119897119894
=
119883119897119894
Λ119897119894
119884119867
119897119894
= 119883119897119894
Λ119897119894
[119884119897119894
119897119894
]119867 Thus the beamforming can
be expressed as VZF119897119894
= 119897119894
119867
119897119894
ℎ119897119894
119867
119897119894
ℎ119897119894
Then throughputof 119880119894
in bitssecondHz (bpsHz) can be expressed as
119877ZF119897119894
= (1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
) 119894 = 1 119870 (6)
where ℎ119897119894
= 119897119894
119867ℎ119897119894
2
The energy consumption of UL WIT of 119880119894
in the 119897th slotis given by
119902119897119894
= (1 minus 120591119897
) 1198751015840
119897119894
(7)
Let 119876119897119894
represent the amount of energy available in thebattery of 119880
119894
at slot 119897 and its updating function is as follows
119876119897119894
= min (119876119897minus1119894
+ 119864119897119894
minus 119902119897minus1119894
119876max) (8)
There are two constraints in the UL WIT phase theenergy causality constraint and the battery storage constraint[9 11] Specifically the energy causality constraint requiresthat the UL WIT can only use the energy harvested at thecurrent and previous slots and the battery storage constraintindicates that the energy available of 119880
119894
cannot exceed themaximum battery capacity at any time that is
119873
sum
119897=0
[119864119897119894
minus 119902119897119894
] ge 0
119873+1
sum
119897=0
119864119897119894
minus
119873
sum
119897=0
119902119897119894
le 119876max
(9)
Mobile Information Systems 5
33 Problem Formulation Considering fairness we optimizemaximum-minimum (max-min) average ULWIT through-put of all MUs by jointly optimizing the time allocation 120591
119897
theDL energy beams 119908
119897
and the UL transmit power allocation1198751015840
119897119894
Let 119903ZF(119902119897119894)
denote theULdata rate of119880119894
in slot 119897 as a functionof the allocated energy 119902
119897119894
forUL transmission in slot 119897 Noticethat 119902
119897119894
is a feasible energy allocation policy which satisfies
0 le 119902119897119894
le 119876max
119876119897+1119894
= min (119876119897119894
+ 119864119897+1119894
minus 119902119897119894
119876max)
119902119897119894
= 120601 (119897 119864119894
119897
119898=1
)
(10)
The expression (10) shows energy causality constraint thebattery storage constraint and casualty CSI We should notethat (4) should satisfy the constraint 1199092
1198970
le 119875maxThus the optimization problem satisfying constraints (10)
can be defined as
max min lim119873rarrinfin
1
119873
119873
sum
119897=1
(1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
)
0 le 120591119897
le 120578119897
(11)
where 120578119897
= min(min119897
((119876119897maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery overflow
4 Dynamic Resource Allocation Algorithm
In this section we present a dynamic resource allocationto make a near throughput optimization for the problemformulated in the former section including DL beamformerdesign and allocation of WIT energy and duration
As the optimization problem is formulated we can divideit into two parts one of which is the optimal beamformerdesigning of DLWET for fairness and the other is the optimaltime and power allocation of UL WIT for throughput sincewe employ the ZF based receive beamforming in the normalinformation BS proposed by [12] which is not related to 119908
119897
and 120591119897
41 Design of WET Beamforming and Computation of EHProbability Considering fairness of the received power wedetermine an optimal beamformer design to maximize theminimum received power for differentMUs Tomaximize thereceived power at 119880
119894
we can set 11990921198970
= 119875max Generally thereceived power for 119880
119897
is denoted by
119875119897119894
(120596119897
) = 119875max120596119867
119897
119867119897119894
120596119897
(12)
where 119867119897119894
= ℎ119897119894
ℎ119867
119897119894
Then problem on fairness could beformulated as
max120596119897
min1le119897le119870
119875119897119894
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(13)
It is easy to relax the max-min process by introducing aslack variable 119875
119897
and then problem (13) can be transformedinto the following
max120596119897
119875119897
st 119875119897119894
(120596119897
) ge 119875119897
forall1 le 119894 le 119870
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(14)
Since the problem is convex we can follow the approachof convex theory in [28] and consider its Lagrangian functionis given by
119871 (120582 120596119897
) = minus
119870
sum
119894=1
120582119894
(119875119897119894
(120596119897
) minus 119875119897
) (15)
where 120582 = [1205821
1205822
120582119870
] ge 0 consists of Lagrangemultipliers associated with received power constraints ofMUs in problem (14)
The dual function of problem (14) is then given by
119866 (120582) = min120596119897
119871 (120582 120596119897
) (16)
The following rule can be used to determine whetherproblem (14) is feasible For given 119875
119897
gt 0 problem (14) isinfeasible in the following interpretation if and only if thereexists any 120582 ge 0 such that 119866(120582) gt 0
Next for given 120582 ge 0 we would obtain 119866(120582) by theproblem
max120596119897
119870
sum
119894=1
120582119894
119875119897119894
(120596119897
)
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(17)
With (12) the problem could be transformed as
max120596119897
119875max120596119867
119897
119860119897
120596119897
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(18)
where 119860119894
= sum119870
119894=1
120582119894
119867119897119894
Since 119860 is a symmetric matrix the optimal beamformer
120596lowast
119897
could be obtained by the result of quadratic problemAfter obtaining 120596
lowast
119897
for given 119875119897
and 120582 we can computethe corresponding 119875
119897119894
(120596lowast
119897
) and 119866(120582) in (15) If 119866(120582lowast) le 0 itindicates that problem (14) is infeasibleTherefore we shoulddecrease 119875
119897
and solve the feasibility problem in (14) againOn the other hand if 119866(120582lowast) le 0 we can update 120582 usingellipsoid method until 120582 converges to 120582
lowast which denotes themaximizer of 119866(120582) or the optimal dual solution for problem(14) If 119866(120582lowast) le 0 it indicates that the problem is feasibleThen we should increase 119875
119897
and solve the feasibility problemagain Finally119875
119897
lowast could be obtained numerically by iterativelyupdating 119875
119897
by a simple bisection search Meanwhile thecorresponding120596lowast
119897
should be the optimal beamformer Finally
6 Mobile Information Systems
(1) set a large number119873 and suc time = 0 where suc time isin 119877119870
(2) loop(3) set 119875down = 0 119875up gt 119875
119897
lowast generate a stochastic Rayleigh channel119867119897
= [ℎ1198971
ℎ119897119870
] 119899 = 0(4) loop
(5) 119875119897
=
1
2
(119875down + 119875up)
(6) set 1205822 ge 0(7) Given 120582 obtain the optimal 120596
119897
by quadratic problem(8) Computing 119866(120582) using (15)(9) if 119866(120582) gt 0 119875 is infeasible then(10) set 119875up larr 119875
119897
go to step (5)(11) else(12) update 120582 using ellipsoid method(13) if the stopping criteria of the ellipsoid method is not met then(14) go to step (7)(15) end if(16) end if(17) set 119875down larr 119875
119897
(18) if 119875up minus 119875down lt 120575 where 120575 gt 0 is a given error tolerance then(19) 119899 = 119899 + 1(20) break(21) end if(22) end loop(23) if 119899 = 119873 then(24) obtain energy harvesting probability 119901
119894
= suc time(119894)119873(25) break(26) end if(27) end loop
Algorithm 1 Optimal beamformer and EH probability obtaining algorithm
we can compute the received power for each MU anddetermine whether they could harvest the energy which canbe successful only if 119875
119897119894
ge 120572119894
To obtain the EH probability we can iterate the above
steps for 119873 times where 119873 should be a large number and adifferent transmitting channel is generated under the rule ofRayleigh distribution for each timeThe probability of119880
119894
canbe obtained by the ratio of successful harvested times and119873which is denoted by
119901119894
= lim119873rarrinfin
suc time119894
119873
(19)
where suc time119894
is the number of successful iterate timesTo summarize the algorithm to solve the problem is given
in Algorithm 1 Steps 5 to 20 give the solution to obtain anoptimal beamformer in case that CSI is available
42 Allocation of WIT Energy and Duration After obtainingthe best WET beamformer we have to determine an optimalallocation ofWIT energy and duration Consideringmultipleslots the optimization of the problem would be approachedif we keep the average transmitting energy in each slotThis is because uniformly distributing the energy betweenenergy arrivals maximizes the data rate As proposed in[29] according to the EH probability 119901
119894
we take a fraction119901119894
of available energy of 119880119894
to transmit information inUL duration With this energy allocation we only need to
compute the optimal UL duration in each slot and can obtaina global optimization with multiple slots
In the case of the system model the total throughput of119880119894
in each slot is denoted by
119877119897119894
(120591) = (1 minus 120591119897
) log(1 +ℎ119897119894
119901119894
(119876119897119894
+ 120591119875119897119894
)
(1 minus 120591119897
) 1205902
) (20)
where ℎ119897119894
119901119894
119876119897119894
and 119875119897119894
are considered as constant sincethey are calculated out before duration allocation
Like themethod proposed in Section 41 the optimizationproblem can also be formulated as
max 119877
st 119877119897119894
(120591) ge 119877 forall1 le 119894 le 119870
0 le 120591119897
le 120578119897
(21)
where 120578119897
= min(min119894
((119876119894maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery from overflowSince the problem is convex we can also solve the
problem by the Lagrangian method which is proposed inSection 41 What is different between two problems is howto obtain max
120591119897sum119870
119894=1
120582119894
119877119897119894
(120591119897
)
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mobile Information Systems 3
M
BS
processingSignal
Energy transfer
Information transfer
BatteryRF to DC converter
Transceiver
processingSignal
BatteryRF to DC converter
Transceiver
Mobile user U1
Mobile user UK
(1 minus 120591)T
120591T
(1 minus 120591)T
120591T
Qmax
Qmax
Figure 2 A DEIN system model
(AP) and users has been studied for joint DL energy transferand UL information transmission A harvest-then-transmitprotocol was proposed and orthogonal time allocations forthe DL energy transfer and UL information transmissionsof all users are jointly optimized to maximize the networkthroughput In [12] a DEIN where one multiantenna APcoordinates energy transfer and information transfer tofroma set of single-antenna users was studied and the authormaximizes the minimum throughput among all users by ajoint design of the DL-UL time allocation the DL energybeamforming and the UL transmit power allocation as wellas receive beamforming A WET enabled massive MIMOsystem with imperfect CSI was studied in [25] where it isbased on time-division-duplexing (TDD) protocol and eachframe is divided into three phases the UL channel estimationphase the DLWET phase and the ULWIT phase
But all these works take the idea that the MUs transfer allthe received power in DL ignoring the power sensitivity ofRF-DC circuits [9ndash11] Besides no works take the advantageof the battery storage capacity to make a dynamic powerallocation which can have a higher throughput Even thoughsome works use the parameter (ie the battery storagecapacity) they also overlook the fact that the capacitycannot overflow when making the resource allocation In thefollowing we are going to discuss these questions
3 System Model and Problem Formulation
In this section we present a dynamical time slotted transmis-sion scheme and analyze the DL WET phase and UL WITphase Finally we formulate problems based on fairness
We provide a DEIN model consisting of a BS with119872 antennas and 119870 single-antenna MUs with finite battery
DL WIT
DL WET UL WIT
Premable FCHDL
mapUL
map
Control frame
(1 minus 120591)T120591T
Figure 3 Frame structure
capacity denoted by 119880119894
(119894 = 1 119870) as shown in Figure 2It is assumed that 119872 ge 119870 It should be noted that theBS is virtualized into two parts that is a virtual eBS anda dBS as shown in Figure 1 Each MU uses the harvestedenergy from DL WET phase via beamforming of the virtualeBS to power its UL information transmission via the dBSunder the assumption that two BSs and all MUs are perfectlysynchronized and there is no other energy source ofMUsThetotal capacity of battery storage in every MU is 119876max
A time slotted transmission scheme is considered asshown in Figure 3 We assume that the period of DL DIT inevery slot can be neglected since there is little informationto transmit in the DL phase So each slot has a constantperiod 119879 consisting of two phases namely the DL WETphase of duration 120591 sdot 119879 and the UL WIT phase of duration(1 minus 120591) sdot 119879 where 0 le 120591 le 1 The WET phase starts withseveral control frames including the preamble frame controlheader (FCH) DL map and UL map These frames definetransmission parameters such as coding schemes available
4 Mobile Information Systems
resources the duration of DL and UL transmission and theWET probability (which will be defined in the following)Then virtual energy BS transmits energy to 119880
119894
throughwireless energy beamforming The received power level andthe energy harvested at 119880
119894
in slot 119897 (119897 = 1 119873) aredenoted by 119875
119897119894
and 119864119897119894
It is worth noting that due to thepower sensitivity of RF energy harvesting circuits 119880
119894
cannotharvest any RF energy if the received signal power 119875
119897119894
isbelow a certain level Thus the received power at every MUcan be used if the received power level 119875
119897119894
is larger thanthe corresponding certain threshold (eg minus10 dBm) Thisindicates that the WET of every MU follows a Bernoulliprocess with different probability 119901
119894
where 119901119894
stands for theprobability of delivering energy from virtual eBS Next allMUs do UL WIT phase simultaneously via SDMA which ispowered by the energy stored in the batteries For simplicitywe assume that 119879 = 1 s and that the harvested energy isstored in the battery first and then used for UL informationtransmission Note that since the length of control frames ismuch smaller than that of DL WET and UL WIT we ignorethe time duration of control frames in the following analysis
We consider frame-based transmissions over flat-fadingchannels on a single frequency band [26] (ie which meansthe channel remains constant in each slot) Denote ℎ
119897119894
isin
C119872lowast1 as the UL channel of 119880119894
in the 119897th slot and we have
ℎ119897119894
= (1205720
1003816100381610038161003816119863119894
1003816100381610038161003816
minus120573
119862119894
)
12
119892119897119894
119894 = 1 119870 (1)
where 1205720
denotes a constant determined by the RF propa-gation environment 119863
119894
is the propagation distance 120573119894
is thepath loss119862
119894
is Shadow fading and119892119897119894
isin C119872lowast1 is thematrix ofRayleigh fading coefficients and 119892
119896119894
sim 119862119873(0 1) By exploit-ing the channel reciprocity the DL transmission channel canbe obtained as ℎ119867
119897119894
For simplicity we assume that 119862119894
= 1 andthat CSI is available at both two BSs and 119880
119894
31 DL WET Phase Assume that just one energy beam isto transmit energy from virtual energy BS to those userssatisfying the probability 119901
119894
since we just transmitted energysignals in the DL [27] Besides we can allocate the optimalbeamformer on the constraint that the power transferred atall of the antennas is equal to the eBSrsquos transmitted powerHence we can assume the beamformer as a ratio of 119875 whichsatisfies the fact that its norm is a unit value where 119875 isthe transmitted power of eBS Then the practical powertransferred can be denoted as the beamformer multiplied byeBS transmitted power 119875 Moreover ambient channel noiseenergy cannot be harvested Thus the DL received signalreceived power and harvested energy of 119880
119894
in the 119897th slot areexpressed as
119910119897119894
= ℎ119867
119897119894
120596119897
1199091198970
+ 119899119897119894
119894 = 1 119870 (2)
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (3)
119864119897119894
= 120598119897
120591119897
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (4)
where 119899119897119894
sim 119862119873(0 1205902
119894
) is the receiver noise 120596119897
is the 119872 times 1
beamforming and satisfies 120596119897
2
= 1 1199091198970
is the transmission
signal and satisfies 11990921198970
le 119875max where 119875max is the transmitpower constraint and 120598
119894
denotes the energy harvestingefficiency at119880
119894
which should satisfy 0 lt 120598119894
le 1 For simplicitywe assume 120598
119894
= 1
32 UL WIT Phase In the UL WIT phase of each slotMUs use the harvested energy to power UL informationtransmission to the virtual dBS For convenience we assumethe circuit energy consumption at119880
119894
is 0 The received signalat the normal dBS in the 119897th slot is given by
119910119897
=
119870
sum
119894=1
ℎ119897119894
119909119897119894
+ 119899119897
119894 = 1 119870 (5)
where 119899119897
isin C119872times1 denotes the receiver additivewhiteGaussiannoise (AWGN) It is assumed that 119899
119897
sim 119862119873(0 1205902
119897
Γ) Besideswe assume that the normal information BS employs linearreceivers to decode 119909
119897119894
in the UL 119909119897119894
denotes the transmitsignal of 119880
119894
and satisfies 1199092
119897119894
= 1198751015840
119897119894
where 1198751015840
119897119894
is thetransmit power of 119880
119894
Specifically let V119897119894
isin C119872times1 denotethe receive beamforming vector for decoding 119909
119897119894
and define119881 = V
1198971
V119897119870
In order to reduce complicity we employthe ZF based receive beamforming in the normal informationBS proposed by [12] which is not related to 119908
119897
and 120591119897
Define 119867
minus119897119894
= [ℎ1198971
ℎ119897119894minus1
ℎ119897119894+1
ℎ119897119894
]119867
119894 = 1 119870including all the UL channels except ℎ
119897119894
Then the singularvalue decomposition (SVD) of 119867
minus119897119894
is given as 119867minus119897119894
=
119883119897119894
Λ119897119894
119884119867
119897119894
= 119883119897119894
Λ119897119894
[119884119897119894
119897119894
]119867 Thus the beamforming can
be expressed as VZF119897119894
= 119897119894
119867
119897119894
ℎ119897119894
119867
119897119894
ℎ119897119894
Then throughputof 119880119894
in bitssecondHz (bpsHz) can be expressed as
119877ZF119897119894
= (1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
) 119894 = 1 119870 (6)
where ℎ119897119894
= 119897119894
119867ℎ119897119894
2
The energy consumption of UL WIT of 119880119894
in the 119897th slotis given by
119902119897119894
= (1 minus 120591119897
) 1198751015840
119897119894
(7)
Let 119876119897119894
represent the amount of energy available in thebattery of 119880
119894
at slot 119897 and its updating function is as follows
119876119897119894
= min (119876119897minus1119894
+ 119864119897119894
minus 119902119897minus1119894
119876max) (8)
There are two constraints in the UL WIT phase theenergy causality constraint and the battery storage constraint[9 11] Specifically the energy causality constraint requiresthat the UL WIT can only use the energy harvested at thecurrent and previous slots and the battery storage constraintindicates that the energy available of 119880
119894
cannot exceed themaximum battery capacity at any time that is
119873
sum
119897=0
[119864119897119894
minus 119902119897119894
] ge 0
119873+1
sum
119897=0
119864119897119894
minus
119873
sum
119897=0
119902119897119894
le 119876max
(9)
Mobile Information Systems 5
33 Problem Formulation Considering fairness we optimizemaximum-minimum (max-min) average ULWIT through-put of all MUs by jointly optimizing the time allocation 120591
119897
theDL energy beams 119908
119897
and the UL transmit power allocation1198751015840
119897119894
Let 119903ZF(119902119897119894)
denote theULdata rate of119880119894
in slot 119897 as a functionof the allocated energy 119902
119897119894
forUL transmission in slot 119897 Noticethat 119902
119897119894
is a feasible energy allocation policy which satisfies
0 le 119902119897119894
le 119876max
119876119897+1119894
= min (119876119897119894
+ 119864119897+1119894
minus 119902119897119894
119876max)
119902119897119894
= 120601 (119897 119864119894
119897
119898=1
)
(10)
The expression (10) shows energy causality constraint thebattery storage constraint and casualty CSI We should notethat (4) should satisfy the constraint 1199092
1198970
le 119875maxThus the optimization problem satisfying constraints (10)
can be defined as
max min lim119873rarrinfin
1
119873
119873
sum
119897=1
(1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
)
0 le 120591119897
le 120578119897
(11)
where 120578119897
= min(min119897
((119876119897maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery overflow
4 Dynamic Resource Allocation Algorithm
In this section we present a dynamic resource allocationto make a near throughput optimization for the problemformulated in the former section including DL beamformerdesign and allocation of WIT energy and duration
As the optimization problem is formulated we can divideit into two parts one of which is the optimal beamformerdesigning of DLWET for fairness and the other is the optimaltime and power allocation of UL WIT for throughput sincewe employ the ZF based receive beamforming in the normalinformation BS proposed by [12] which is not related to 119908
119897
and 120591119897
41 Design of WET Beamforming and Computation of EHProbability Considering fairness of the received power wedetermine an optimal beamformer design to maximize theminimum received power for differentMUs Tomaximize thereceived power at 119880
119894
we can set 11990921198970
= 119875max Generally thereceived power for 119880
119897
is denoted by
119875119897119894
(120596119897
) = 119875max120596119867
119897
119867119897119894
120596119897
(12)
where 119867119897119894
= ℎ119897119894
ℎ119867
119897119894
Then problem on fairness could beformulated as
max120596119897
min1le119897le119870
119875119897119894
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(13)
It is easy to relax the max-min process by introducing aslack variable 119875
119897
and then problem (13) can be transformedinto the following
max120596119897
119875119897
st 119875119897119894
(120596119897
) ge 119875119897
forall1 le 119894 le 119870
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(14)
Since the problem is convex we can follow the approachof convex theory in [28] and consider its Lagrangian functionis given by
119871 (120582 120596119897
) = minus
119870
sum
119894=1
120582119894
(119875119897119894
(120596119897
) minus 119875119897
) (15)
where 120582 = [1205821
1205822
120582119870
] ge 0 consists of Lagrangemultipliers associated with received power constraints ofMUs in problem (14)
The dual function of problem (14) is then given by
119866 (120582) = min120596119897
119871 (120582 120596119897
) (16)
The following rule can be used to determine whetherproblem (14) is feasible For given 119875
119897
gt 0 problem (14) isinfeasible in the following interpretation if and only if thereexists any 120582 ge 0 such that 119866(120582) gt 0
Next for given 120582 ge 0 we would obtain 119866(120582) by theproblem
max120596119897
119870
sum
119894=1
120582119894
119875119897119894
(120596119897
)
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(17)
With (12) the problem could be transformed as
max120596119897
119875max120596119867
119897
119860119897
120596119897
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(18)
where 119860119894
= sum119870
119894=1
120582119894
119867119897119894
Since 119860 is a symmetric matrix the optimal beamformer
120596lowast
119897
could be obtained by the result of quadratic problemAfter obtaining 120596
lowast
119897
for given 119875119897
and 120582 we can computethe corresponding 119875
119897119894
(120596lowast
119897
) and 119866(120582) in (15) If 119866(120582lowast) le 0 itindicates that problem (14) is infeasibleTherefore we shoulddecrease 119875
119897
and solve the feasibility problem in (14) againOn the other hand if 119866(120582lowast) le 0 we can update 120582 usingellipsoid method until 120582 converges to 120582
lowast which denotes themaximizer of 119866(120582) or the optimal dual solution for problem(14) If 119866(120582lowast) le 0 it indicates that the problem is feasibleThen we should increase 119875
119897
and solve the feasibility problemagain Finally119875
119897
lowast could be obtained numerically by iterativelyupdating 119875
119897
by a simple bisection search Meanwhile thecorresponding120596lowast
119897
should be the optimal beamformer Finally
6 Mobile Information Systems
(1) set a large number119873 and suc time = 0 where suc time isin 119877119870
(2) loop(3) set 119875down = 0 119875up gt 119875
119897
lowast generate a stochastic Rayleigh channel119867119897
= [ℎ1198971
ℎ119897119870
] 119899 = 0(4) loop
(5) 119875119897
=
1
2
(119875down + 119875up)
(6) set 1205822 ge 0(7) Given 120582 obtain the optimal 120596
119897
by quadratic problem(8) Computing 119866(120582) using (15)(9) if 119866(120582) gt 0 119875 is infeasible then(10) set 119875up larr 119875
119897
go to step (5)(11) else(12) update 120582 using ellipsoid method(13) if the stopping criteria of the ellipsoid method is not met then(14) go to step (7)(15) end if(16) end if(17) set 119875down larr 119875
119897
(18) if 119875up minus 119875down lt 120575 where 120575 gt 0 is a given error tolerance then(19) 119899 = 119899 + 1(20) break(21) end if(22) end loop(23) if 119899 = 119873 then(24) obtain energy harvesting probability 119901
119894
= suc time(119894)119873(25) break(26) end if(27) end loop
Algorithm 1 Optimal beamformer and EH probability obtaining algorithm
we can compute the received power for each MU anddetermine whether they could harvest the energy which canbe successful only if 119875
119897119894
ge 120572119894
To obtain the EH probability we can iterate the above
steps for 119873 times where 119873 should be a large number and adifferent transmitting channel is generated under the rule ofRayleigh distribution for each timeThe probability of119880
119894
canbe obtained by the ratio of successful harvested times and119873which is denoted by
119901119894
= lim119873rarrinfin
suc time119894
119873
(19)
where suc time119894
is the number of successful iterate timesTo summarize the algorithm to solve the problem is given
in Algorithm 1 Steps 5 to 20 give the solution to obtain anoptimal beamformer in case that CSI is available
42 Allocation of WIT Energy and Duration After obtainingthe best WET beamformer we have to determine an optimalallocation ofWIT energy and duration Consideringmultipleslots the optimization of the problem would be approachedif we keep the average transmitting energy in each slotThis is because uniformly distributing the energy betweenenergy arrivals maximizes the data rate As proposed in[29] according to the EH probability 119901
119894
we take a fraction119901119894
of available energy of 119880119894
to transmit information inUL duration With this energy allocation we only need to
compute the optimal UL duration in each slot and can obtaina global optimization with multiple slots
In the case of the system model the total throughput of119880119894
in each slot is denoted by
119877119897119894
(120591) = (1 minus 120591119897
) log(1 +ℎ119897119894
119901119894
(119876119897119894
+ 120591119875119897119894
)
(1 minus 120591119897
) 1205902
) (20)
where ℎ119897119894
119901119894
119876119897119894
and 119875119897119894
are considered as constant sincethey are calculated out before duration allocation
Like themethod proposed in Section 41 the optimizationproblem can also be formulated as
max 119877
st 119877119897119894
(120591) ge 119877 forall1 le 119894 le 119870
0 le 120591119897
le 120578119897
(21)
where 120578119897
= min(min119894
((119876119894maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery from overflowSince the problem is convex we can also solve the
problem by the Lagrangian method which is proposed inSection 41 What is different between two problems is howto obtain max
120591119897sum119870
119894=1
120582119894
119877119897119894
(120591119897
)
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
4 Mobile Information Systems
resources the duration of DL and UL transmission and theWET probability (which will be defined in the following)Then virtual energy BS transmits energy to 119880
119894
throughwireless energy beamforming The received power level andthe energy harvested at 119880
119894
in slot 119897 (119897 = 1 119873) aredenoted by 119875
119897119894
and 119864119897119894
It is worth noting that due to thepower sensitivity of RF energy harvesting circuits 119880
119894
cannotharvest any RF energy if the received signal power 119875
119897119894
isbelow a certain level Thus the received power at every MUcan be used if the received power level 119875
119897119894
is larger thanthe corresponding certain threshold (eg minus10 dBm) Thisindicates that the WET of every MU follows a Bernoulliprocess with different probability 119901
119894
where 119901119894
stands for theprobability of delivering energy from virtual eBS Next allMUs do UL WIT phase simultaneously via SDMA which ispowered by the energy stored in the batteries For simplicitywe assume that 119879 = 1 s and that the harvested energy isstored in the battery first and then used for UL informationtransmission Note that since the length of control frames ismuch smaller than that of DL WET and UL WIT we ignorethe time duration of control frames in the following analysis
We consider frame-based transmissions over flat-fadingchannels on a single frequency band [26] (ie which meansthe channel remains constant in each slot) Denote ℎ
119897119894
isin
C119872lowast1 as the UL channel of 119880119894
in the 119897th slot and we have
ℎ119897119894
= (1205720
1003816100381610038161003816119863119894
1003816100381610038161003816
minus120573
119862119894
)
12
119892119897119894
119894 = 1 119870 (1)
where 1205720
denotes a constant determined by the RF propa-gation environment 119863
119894
is the propagation distance 120573119894
is thepath loss119862
119894
is Shadow fading and119892119897119894
isin C119872lowast1 is thematrix ofRayleigh fading coefficients and 119892
119896119894
sim 119862119873(0 1) By exploit-ing the channel reciprocity the DL transmission channel canbe obtained as ℎ119867
119897119894
For simplicity we assume that 119862119894
= 1 andthat CSI is available at both two BSs and 119880
119894
31 DL WET Phase Assume that just one energy beam isto transmit energy from virtual energy BS to those userssatisfying the probability 119901
119894
since we just transmitted energysignals in the DL [27] Besides we can allocate the optimalbeamformer on the constraint that the power transferred atall of the antennas is equal to the eBSrsquos transmitted powerHence we can assume the beamformer as a ratio of 119875 whichsatisfies the fact that its norm is a unit value where 119875 isthe transmitted power of eBS Then the practical powertransferred can be denoted as the beamformer multiplied byeBS transmitted power 119875 Moreover ambient channel noiseenergy cannot be harvested Thus the DL received signalreceived power and harvested energy of 119880
119894
in the 119897th slot areexpressed as
119910119897119894
= ℎ119867
119897119894
120596119897
1199091198970
+ 119899119897119894
119894 = 1 119870 (2)
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (3)
119864119897119894
= 120598119897
120591119897
119875119897119894
= 1199092
1198970
ℎ119867
119897119894
120596119897
120596119867
119897
ℎ119897119894
119894 = 1 119870 (4)
where 119899119897119894
sim 119862119873(0 1205902
119894
) is the receiver noise 120596119897
is the 119872 times 1
beamforming and satisfies 120596119897
2
= 1 1199091198970
is the transmission
signal and satisfies 11990921198970
le 119875max where 119875max is the transmitpower constraint and 120598
119894
denotes the energy harvestingefficiency at119880
119894
which should satisfy 0 lt 120598119894
le 1 For simplicitywe assume 120598
119894
= 1
32 UL WIT Phase In the UL WIT phase of each slotMUs use the harvested energy to power UL informationtransmission to the virtual dBS For convenience we assumethe circuit energy consumption at119880
119894
is 0 The received signalat the normal dBS in the 119897th slot is given by
119910119897
=
119870
sum
119894=1
ℎ119897119894
119909119897119894
+ 119899119897
119894 = 1 119870 (5)
where 119899119897
isin C119872times1 denotes the receiver additivewhiteGaussiannoise (AWGN) It is assumed that 119899
119897
sim 119862119873(0 1205902
119897
Γ) Besideswe assume that the normal information BS employs linearreceivers to decode 119909
119897119894
in the UL 119909119897119894
denotes the transmitsignal of 119880
119894
and satisfies 1199092
119897119894
= 1198751015840
119897119894
where 1198751015840
119897119894
is thetransmit power of 119880
119894
Specifically let V119897119894
isin C119872times1 denotethe receive beamforming vector for decoding 119909
119897119894
and define119881 = V
1198971
V119897119870
In order to reduce complicity we employthe ZF based receive beamforming in the normal informationBS proposed by [12] which is not related to 119908
119897
and 120591119897
Define 119867
minus119897119894
= [ℎ1198971
ℎ119897119894minus1
ℎ119897119894+1
ℎ119897119894
]119867
119894 = 1 119870including all the UL channels except ℎ
119897119894
Then the singularvalue decomposition (SVD) of 119867
minus119897119894
is given as 119867minus119897119894
=
119883119897119894
Λ119897119894
119884119867
119897119894
= 119883119897119894
Λ119897119894
[119884119897119894
119897119894
]119867 Thus the beamforming can
be expressed as VZF119897119894
= 119897119894
119867
119897119894
ℎ119897119894
119867
119897119894
ℎ119897119894
Then throughputof 119880119894
in bitssecondHz (bpsHz) can be expressed as
119877ZF119897119894
= (1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
) 119894 = 1 119870 (6)
where ℎ119897119894
= 119897119894
119867ℎ119897119894
2
The energy consumption of UL WIT of 119880119894
in the 119897th slotis given by
119902119897119894
= (1 minus 120591119897
) 1198751015840
119897119894
(7)
Let 119876119897119894
represent the amount of energy available in thebattery of 119880
119894
at slot 119897 and its updating function is as follows
119876119897119894
= min (119876119897minus1119894
+ 119864119897119894
minus 119902119897minus1119894
119876max) (8)
There are two constraints in the UL WIT phase theenergy causality constraint and the battery storage constraint[9 11] Specifically the energy causality constraint requiresthat the UL WIT can only use the energy harvested at thecurrent and previous slots and the battery storage constraintindicates that the energy available of 119880
119894
cannot exceed themaximum battery capacity at any time that is
119873
sum
119897=0
[119864119897119894
minus 119902119897119894
] ge 0
119873+1
sum
119897=0
119864119897119894
minus
119873
sum
119897=0
119902119897119894
le 119876max
(9)
Mobile Information Systems 5
33 Problem Formulation Considering fairness we optimizemaximum-minimum (max-min) average ULWIT through-put of all MUs by jointly optimizing the time allocation 120591
119897
theDL energy beams 119908
119897
and the UL transmit power allocation1198751015840
119897119894
Let 119903ZF(119902119897119894)
denote theULdata rate of119880119894
in slot 119897 as a functionof the allocated energy 119902
119897119894
forUL transmission in slot 119897 Noticethat 119902
119897119894
is a feasible energy allocation policy which satisfies
0 le 119902119897119894
le 119876max
119876119897+1119894
= min (119876119897119894
+ 119864119897+1119894
minus 119902119897119894
119876max)
119902119897119894
= 120601 (119897 119864119894
119897
119898=1
)
(10)
The expression (10) shows energy causality constraint thebattery storage constraint and casualty CSI We should notethat (4) should satisfy the constraint 1199092
1198970
le 119875maxThus the optimization problem satisfying constraints (10)
can be defined as
max min lim119873rarrinfin
1
119873
119873
sum
119897=1
(1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
)
0 le 120591119897
le 120578119897
(11)
where 120578119897
= min(min119897
((119876119897maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery overflow
4 Dynamic Resource Allocation Algorithm
In this section we present a dynamic resource allocationto make a near throughput optimization for the problemformulated in the former section including DL beamformerdesign and allocation of WIT energy and duration
As the optimization problem is formulated we can divideit into two parts one of which is the optimal beamformerdesigning of DLWET for fairness and the other is the optimaltime and power allocation of UL WIT for throughput sincewe employ the ZF based receive beamforming in the normalinformation BS proposed by [12] which is not related to 119908
119897
and 120591119897
41 Design of WET Beamforming and Computation of EHProbability Considering fairness of the received power wedetermine an optimal beamformer design to maximize theminimum received power for differentMUs Tomaximize thereceived power at 119880
119894
we can set 11990921198970
= 119875max Generally thereceived power for 119880
119897
is denoted by
119875119897119894
(120596119897
) = 119875max120596119867
119897
119867119897119894
120596119897
(12)
where 119867119897119894
= ℎ119897119894
ℎ119867
119897119894
Then problem on fairness could beformulated as
max120596119897
min1le119897le119870
119875119897119894
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(13)
It is easy to relax the max-min process by introducing aslack variable 119875
119897
and then problem (13) can be transformedinto the following
max120596119897
119875119897
st 119875119897119894
(120596119897
) ge 119875119897
forall1 le 119894 le 119870
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(14)
Since the problem is convex we can follow the approachof convex theory in [28] and consider its Lagrangian functionis given by
119871 (120582 120596119897
) = minus
119870
sum
119894=1
120582119894
(119875119897119894
(120596119897
) minus 119875119897
) (15)
where 120582 = [1205821
1205822
120582119870
] ge 0 consists of Lagrangemultipliers associated with received power constraints ofMUs in problem (14)
The dual function of problem (14) is then given by
119866 (120582) = min120596119897
119871 (120582 120596119897
) (16)
The following rule can be used to determine whetherproblem (14) is feasible For given 119875
119897
gt 0 problem (14) isinfeasible in the following interpretation if and only if thereexists any 120582 ge 0 such that 119866(120582) gt 0
Next for given 120582 ge 0 we would obtain 119866(120582) by theproblem
max120596119897
119870
sum
119894=1
120582119894
119875119897119894
(120596119897
)
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(17)
With (12) the problem could be transformed as
max120596119897
119875max120596119867
119897
119860119897
120596119897
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(18)
where 119860119894
= sum119870
119894=1
120582119894
119867119897119894
Since 119860 is a symmetric matrix the optimal beamformer
120596lowast
119897
could be obtained by the result of quadratic problemAfter obtaining 120596
lowast
119897
for given 119875119897
and 120582 we can computethe corresponding 119875
119897119894
(120596lowast
119897
) and 119866(120582) in (15) If 119866(120582lowast) le 0 itindicates that problem (14) is infeasibleTherefore we shoulddecrease 119875
119897
and solve the feasibility problem in (14) againOn the other hand if 119866(120582lowast) le 0 we can update 120582 usingellipsoid method until 120582 converges to 120582
lowast which denotes themaximizer of 119866(120582) or the optimal dual solution for problem(14) If 119866(120582lowast) le 0 it indicates that the problem is feasibleThen we should increase 119875
119897
and solve the feasibility problemagain Finally119875
119897
lowast could be obtained numerically by iterativelyupdating 119875
119897
by a simple bisection search Meanwhile thecorresponding120596lowast
119897
should be the optimal beamformer Finally
6 Mobile Information Systems
(1) set a large number119873 and suc time = 0 where suc time isin 119877119870
(2) loop(3) set 119875down = 0 119875up gt 119875
119897
lowast generate a stochastic Rayleigh channel119867119897
= [ℎ1198971
ℎ119897119870
] 119899 = 0(4) loop
(5) 119875119897
=
1
2
(119875down + 119875up)
(6) set 1205822 ge 0(7) Given 120582 obtain the optimal 120596
119897
by quadratic problem(8) Computing 119866(120582) using (15)(9) if 119866(120582) gt 0 119875 is infeasible then(10) set 119875up larr 119875
119897
go to step (5)(11) else(12) update 120582 using ellipsoid method(13) if the stopping criteria of the ellipsoid method is not met then(14) go to step (7)(15) end if(16) end if(17) set 119875down larr 119875
119897
(18) if 119875up minus 119875down lt 120575 where 120575 gt 0 is a given error tolerance then(19) 119899 = 119899 + 1(20) break(21) end if(22) end loop(23) if 119899 = 119873 then(24) obtain energy harvesting probability 119901
119894
= suc time(119894)119873(25) break(26) end if(27) end loop
Algorithm 1 Optimal beamformer and EH probability obtaining algorithm
we can compute the received power for each MU anddetermine whether they could harvest the energy which canbe successful only if 119875
119897119894
ge 120572119894
To obtain the EH probability we can iterate the above
steps for 119873 times where 119873 should be a large number and adifferent transmitting channel is generated under the rule ofRayleigh distribution for each timeThe probability of119880
119894
canbe obtained by the ratio of successful harvested times and119873which is denoted by
119901119894
= lim119873rarrinfin
suc time119894
119873
(19)
where suc time119894
is the number of successful iterate timesTo summarize the algorithm to solve the problem is given
in Algorithm 1 Steps 5 to 20 give the solution to obtain anoptimal beamformer in case that CSI is available
42 Allocation of WIT Energy and Duration After obtainingthe best WET beamformer we have to determine an optimalallocation ofWIT energy and duration Consideringmultipleslots the optimization of the problem would be approachedif we keep the average transmitting energy in each slotThis is because uniformly distributing the energy betweenenergy arrivals maximizes the data rate As proposed in[29] according to the EH probability 119901
119894
we take a fraction119901119894
of available energy of 119880119894
to transmit information inUL duration With this energy allocation we only need to
compute the optimal UL duration in each slot and can obtaina global optimization with multiple slots
In the case of the system model the total throughput of119880119894
in each slot is denoted by
119877119897119894
(120591) = (1 minus 120591119897
) log(1 +ℎ119897119894
119901119894
(119876119897119894
+ 120591119875119897119894
)
(1 minus 120591119897
) 1205902
) (20)
where ℎ119897119894
119901119894
119876119897119894
and 119875119897119894
are considered as constant sincethey are calculated out before duration allocation
Like themethod proposed in Section 41 the optimizationproblem can also be formulated as
max 119877
st 119877119897119894
(120591) ge 119877 forall1 le 119894 le 119870
0 le 120591119897
le 120578119897
(21)
where 120578119897
= min(min119894
((119876119894maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery from overflowSince the problem is convex we can also solve the
problem by the Lagrangian method which is proposed inSection 41 What is different between two problems is howto obtain max
120591119897sum119870
119894=1
120582119894
119877119897119894
(120591119897
)
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mobile Information Systems 5
33 Problem Formulation Considering fairness we optimizemaximum-minimum (max-min) average ULWIT through-put of all MUs by jointly optimizing the time allocation 120591
119897
theDL energy beams 119908
119897
and the UL transmit power allocation1198751015840
119897119894
Let 119903ZF(119902119897119894)
denote theULdata rate of119880119894
in slot 119897 as a functionof the allocated energy 119902
119897119894
forUL transmission in slot 119897 Noticethat 119902
119897119894
is a feasible energy allocation policy which satisfies
0 le 119902119897119894
le 119876max
119876119897+1119894
= min (119876119897119894
+ 119864119897+1119894
minus 119902119897119894
119876max)
119902119897119894
= 120601 (119897 119864119894
119897
119898=1
)
(10)
The expression (10) shows energy causality constraint thebattery storage constraint and casualty CSI We should notethat (4) should satisfy the constraint 1199092
1198970
le 119875maxThus the optimization problem satisfying constraints (10)
can be defined as
max min lim119873rarrinfin
1
119873
119873
sum
119897=1
(1 minus 120591119897
) log(1 +1198751015840
119897119894
ℎ119897119894
1205902
119894
)
0 le 120591119897
le 120578119897
(11)
where 120578119897
= min(min119897
((119876119897maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery overflow
4 Dynamic Resource Allocation Algorithm
In this section we present a dynamic resource allocationto make a near throughput optimization for the problemformulated in the former section including DL beamformerdesign and allocation of WIT energy and duration
As the optimization problem is formulated we can divideit into two parts one of which is the optimal beamformerdesigning of DLWET for fairness and the other is the optimaltime and power allocation of UL WIT for throughput sincewe employ the ZF based receive beamforming in the normalinformation BS proposed by [12] which is not related to 119908
119897
and 120591119897
41 Design of WET Beamforming and Computation of EHProbability Considering fairness of the received power wedetermine an optimal beamformer design to maximize theminimum received power for differentMUs Tomaximize thereceived power at 119880
119894
we can set 11990921198970
= 119875max Generally thereceived power for 119880
119897
is denoted by
119875119897119894
(120596119897
) = 119875max120596119867
119897
119867119897119894
120596119897
(12)
where 119867119897119894
= ℎ119897119894
ℎ119867
119897119894
Then problem on fairness could beformulated as
max120596119897
min1le119897le119870
119875119897119894
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(13)
It is easy to relax the max-min process by introducing aslack variable 119875
119897
and then problem (13) can be transformedinto the following
max120596119897
119875119897
st 119875119897119894
(120596119897
) ge 119875119897
forall1 le 119894 le 119870
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(14)
Since the problem is convex we can follow the approachof convex theory in [28] and consider its Lagrangian functionis given by
119871 (120582 120596119897
) = minus
119870
sum
119894=1
120582119894
(119875119897119894
(120596119897
) minus 119875119897
) (15)
where 120582 = [1205821
1205822
120582119870
] ge 0 consists of Lagrangemultipliers associated with received power constraints ofMUs in problem (14)
The dual function of problem (14) is then given by
119866 (120582) = min120596119897
119871 (120582 120596119897
) (16)
The following rule can be used to determine whetherproblem (14) is feasible For given 119875
119897
gt 0 problem (14) isinfeasible in the following interpretation if and only if thereexists any 120582 ge 0 such that 119866(120582) gt 0
Next for given 120582 ge 0 we would obtain 119866(120582) by theproblem
max120596119897
119870
sum
119894=1
120582119894
119875119897119894
(120596119897
)
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(17)
With (12) the problem could be transformed as
max120596119897
119875max120596119867
119897
119860119897
120596119897
1003817100381710038171003817120596119897
1003817100381710038171003817
2
= 1
(18)
where 119860119894
= sum119870
119894=1
120582119894
119867119897119894
Since 119860 is a symmetric matrix the optimal beamformer
120596lowast
119897
could be obtained by the result of quadratic problemAfter obtaining 120596
lowast
119897
for given 119875119897
and 120582 we can computethe corresponding 119875
119897119894
(120596lowast
119897
) and 119866(120582) in (15) If 119866(120582lowast) le 0 itindicates that problem (14) is infeasibleTherefore we shoulddecrease 119875
119897
and solve the feasibility problem in (14) againOn the other hand if 119866(120582lowast) le 0 we can update 120582 usingellipsoid method until 120582 converges to 120582
lowast which denotes themaximizer of 119866(120582) or the optimal dual solution for problem(14) If 119866(120582lowast) le 0 it indicates that the problem is feasibleThen we should increase 119875
119897
and solve the feasibility problemagain Finally119875
119897
lowast could be obtained numerically by iterativelyupdating 119875
119897
by a simple bisection search Meanwhile thecorresponding120596lowast
119897
should be the optimal beamformer Finally
6 Mobile Information Systems
(1) set a large number119873 and suc time = 0 where suc time isin 119877119870
(2) loop(3) set 119875down = 0 119875up gt 119875
119897
lowast generate a stochastic Rayleigh channel119867119897
= [ℎ1198971
ℎ119897119870
] 119899 = 0(4) loop
(5) 119875119897
=
1
2
(119875down + 119875up)
(6) set 1205822 ge 0(7) Given 120582 obtain the optimal 120596
119897
by quadratic problem(8) Computing 119866(120582) using (15)(9) if 119866(120582) gt 0 119875 is infeasible then(10) set 119875up larr 119875
119897
go to step (5)(11) else(12) update 120582 using ellipsoid method(13) if the stopping criteria of the ellipsoid method is not met then(14) go to step (7)(15) end if(16) end if(17) set 119875down larr 119875
119897
(18) if 119875up minus 119875down lt 120575 where 120575 gt 0 is a given error tolerance then(19) 119899 = 119899 + 1(20) break(21) end if(22) end loop(23) if 119899 = 119873 then(24) obtain energy harvesting probability 119901
119894
= suc time(119894)119873(25) break(26) end if(27) end loop
Algorithm 1 Optimal beamformer and EH probability obtaining algorithm
we can compute the received power for each MU anddetermine whether they could harvest the energy which canbe successful only if 119875
119897119894
ge 120572119894
To obtain the EH probability we can iterate the above
steps for 119873 times where 119873 should be a large number and adifferent transmitting channel is generated under the rule ofRayleigh distribution for each timeThe probability of119880
119894
canbe obtained by the ratio of successful harvested times and119873which is denoted by
119901119894
= lim119873rarrinfin
suc time119894
119873
(19)
where suc time119894
is the number of successful iterate timesTo summarize the algorithm to solve the problem is given
in Algorithm 1 Steps 5 to 20 give the solution to obtain anoptimal beamformer in case that CSI is available
42 Allocation of WIT Energy and Duration After obtainingthe best WET beamformer we have to determine an optimalallocation ofWIT energy and duration Consideringmultipleslots the optimization of the problem would be approachedif we keep the average transmitting energy in each slotThis is because uniformly distributing the energy betweenenergy arrivals maximizes the data rate As proposed in[29] according to the EH probability 119901
119894
we take a fraction119901119894
of available energy of 119880119894
to transmit information inUL duration With this energy allocation we only need to
compute the optimal UL duration in each slot and can obtaina global optimization with multiple slots
In the case of the system model the total throughput of119880119894
in each slot is denoted by
119877119897119894
(120591) = (1 minus 120591119897
) log(1 +ℎ119897119894
119901119894
(119876119897119894
+ 120591119875119897119894
)
(1 minus 120591119897
) 1205902
) (20)
where ℎ119897119894
119901119894
119876119897119894
and 119875119897119894
are considered as constant sincethey are calculated out before duration allocation
Like themethod proposed in Section 41 the optimizationproblem can also be formulated as
max 119877
st 119877119897119894
(120591) ge 119877 forall1 le 119894 le 119870
0 le 120591119897
le 120578119897
(21)
where 120578119897
= min(min119894
((119876119894maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery from overflowSince the problem is convex we can also solve the
problem by the Lagrangian method which is proposed inSection 41 What is different between two problems is howto obtain max
120591119897sum119870
119894=1
120582119894
119877119897119894
(120591119897
)
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
6 Mobile Information Systems
(1) set a large number119873 and suc time = 0 where suc time isin 119877119870
(2) loop(3) set 119875down = 0 119875up gt 119875
119897
lowast generate a stochastic Rayleigh channel119867119897
= [ℎ1198971
ℎ119897119870
] 119899 = 0(4) loop
(5) 119875119897
=
1
2
(119875down + 119875up)
(6) set 1205822 ge 0(7) Given 120582 obtain the optimal 120596
119897
by quadratic problem(8) Computing 119866(120582) using (15)(9) if 119866(120582) gt 0 119875 is infeasible then(10) set 119875up larr 119875
119897
go to step (5)(11) else(12) update 120582 using ellipsoid method(13) if the stopping criteria of the ellipsoid method is not met then(14) go to step (7)(15) end if(16) end if(17) set 119875down larr 119875
119897
(18) if 119875up minus 119875down lt 120575 where 120575 gt 0 is a given error tolerance then(19) 119899 = 119899 + 1(20) break(21) end if(22) end loop(23) if 119899 = 119873 then(24) obtain energy harvesting probability 119901
119894
= suc time(119894)119873(25) break(26) end if(27) end loop
Algorithm 1 Optimal beamformer and EH probability obtaining algorithm
we can compute the received power for each MU anddetermine whether they could harvest the energy which canbe successful only if 119875
119897119894
ge 120572119894
To obtain the EH probability we can iterate the above
steps for 119873 times where 119873 should be a large number and adifferent transmitting channel is generated under the rule ofRayleigh distribution for each timeThe probability of119880
119894
canbe obtained by the ratio of successful harvested times and119873which is denoted by
119901119894
= lim119873rarrinfin
suc time119894
119873
(19)
where suc time119894
is the number of successful iterate timesTo summarize the algorithm to solve the problem is given
in Algorithm 1 Steps 5 to 20 give the solution to obtain anoptimal beamformer in case that CSI is available
42 Allocation of WIT Energy and Duration After obtainingthe best WET beamformer we have to determine an optimalallocation ofWIT energy and duration Consideringmultipleslots the optimization of the problem would be approachedif we keep the average transmitting energy in each slotThis is because uniformly distributing the energy betweenenergy arrivals maximizes the data rate As proposed in[29] according to the EH probability 119901
119894
we take a fraction119901119894
of available energy of 119880119894
to transmit information inUL duration With this energy allocation we only need to
compute the optimal UL duration in each slot and can obtaina global optimization with multiple slots
In the case of the system model the total throughput of119880119894
in each slot is denoted by
119877119897119894
(120591) = (1 minus 120591119897
) log(1 +ℎ119897119894
119901119894
(119876119897119894
+ 120591119875119897119894
)
(1 minus 120591119897
) 1205902
) (20)
where ℎ119897119894
119901119894
119876119897119894
and 119875119897119894
are considered as constant sincethey are calculated out before duration allocation
Like themethod proposed in Section 41 the optimizationproblem can also be formulated as
max 119877
st 119877119897119894
(120591) ge 119877 forall1 le 119894 le 119870
0 le 120591119897
le 120578119897
(21)
where 120578119897
= min(min119894
((119876119894maxminus119876119897119894)119875119897119894) 1) denotes the upper
bound of 120591119897
preventing the battery from overflowSince the problem is convex we can also solve the
problem by the Lagrangian method which is proposed inSection 41 What is different between two problems is howto obtain max
120591119897sum119870
119894=1
120582119894
119877119897119894
(120591119897
)
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mobile Information Systems 7
(1) set 119877min = 0 119877max gt 119877
lowast(2) loop
(3) 119877 =
1
2
(119877min + 119877max)
(4) set 1205822 ge 0(5) Given 120582 obtain the optimal 120591
119897
according to the golden section search method(6) Computing 119866(120582)(7) if 119866(120582) gt 0 then(8) 119877 is infeasible set 119877max larr 119877 go to step (3)(9) else(10) update 120582 using ellipsoid method(11) if the stopping criteria of the ellipsoid method is not met then(12) go to step (5)(13) end if(14) end if(15) set 119877min larr 119877(16) if 119877max minus 119877min lt 120575 where 120575 gt 0 is a given error tolerance then(17) The corresponding 120591
119897
is the optimal duration(18) break(19) end if(20) end loop
Algorithm 2 Optimal time allocation algorithm
Since (20) is concave with 120582 gt 0 we can easily find thatthe following function is also concave
119891 (120591119897
) =
119870
sum
119894=1
120582119894
119877119897119894
(120591119897
) (22)
Consequently one-dimensional search methods are con-sidered to help us obtain the optimal 120591
119897
The algorithm to solve the problem is given in Algo-
rithm 2
5 Performance Results and Analysis
In this section we focus on algorithm simulations andcomparative analyses among algorithms by using MATLABsimulation tool Moreover we analyze simulation resultsbased on the main problems which we have discussed in pre-vious sections including fairness sensitivity and throughputperformance
51 Parameter Settings In this section several simulationsare executed to testify the performance of the proposedalgorithm in three aspects of fairness sensitivity and max-min throughput We make some comparison between thefair allocation algorithmof using fraction119901 of energy (UFPE)and the algorithm in [12] whereMUswould use all the energystored in their battery (UAE) Before the simulations thefollowing settings are used unless stated otherwise as shownin Table 1
52 Simulation Result and Analysis
521 Fairness While each MU is allocated the same slotto transmit information uplink leading to the fact that the
Table 1 Simulation parameters
Parameters ValueBS antennas number119872 3MU number 119870 2Propagation constant 120572
0
1Path loss exponent 120573 2Shadow fading 119862
119894
1Propagation distance119863
1
10mPropagation distance119863
2
5m119875max 1WChannel noise power 1205902 10
minus7WBattery max capacity 119876max 0005 JPower sensitivity of EH circuits 120572
1
003WPower sensitivity of EH circuits 120572
2
005W
increase or decrease of slot length would lead to the sametrend of throughput of each MU the fairness of the modelis mainly reflected on the WET energy beamforming Wecompare our algorithm of allocating beamformer by channelin UFPE with the algorithm of allocating beamformer bydistance weight in UAE Since the fairness will be moreobvious if received power difference among MUs is smallerwe can obtain the standard deviation of received poweramong all MUs For simplicity assuming that there are onlytwoMUs we can compute the results difference between twoMUs rather than the standard deviation of them
Figure 4 shows the received power difference between twoMUs versus the distance between119880
2
and the virtual eBS It iseasy to see that both curves achieve the minimum when thedistance is around 10m and increase as the distance increasesor decreases from this value It is because of the fact that
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
8 Mobile Information Systems
UFPEUAE
8 9 10 11 12 13 14 157Distance (m)
0
0005
001
0015
002
0025
003
0035
004
Rece
ived
pow
er d
iffer
ence
(W)
Figure 4 Received power difference between two MUs versus thedistance between 119880
2
and virtual energy BS
MU1MU2
005 01 015 02 025 03 035 04 045 050120572 (W)
0
01
02
03
04
05
06
07
08
09
1
EH p
roba
bilit
y
Figure 5 EH probability of two MUs versus circuit sensitivity
the distance between 1198801
and BS is fixed to 10m in thisscenario the factors influencing both channels would bemuch similar when the distance of 119880
2
fluctuates around thisvalueWe can also see that our algorithmUFPE shows a lowerdifference comparedwithUAE indicating that fairness of ourmodel is more highlighted
522 Sensitivity Since the sensitivity of circuit of receiverexits MUs may not always harvest the energy they receivedThe energy harvesting probability of twoMUs versus differentcircuit sensitivity is shown in Figure 5 It is easy to see thatthe energy harvesting probabilities ofMU1 andMU2decreasewith the circuit sensitivity since it will become more difficultfor the receiver to harvest energy when the circuit sensitivity
UFPEUAE
002 003 004 005 006 007 008 01009001120572 (W)
05
15
1
2
25
3
35
4
45
5
Max
-min
thr
ough
put (
bps
Hz)
Figure 6 Maximum-minimum throughput between two MUsversus circuit sensitivity
ismuch too highWhat ismore1198802
shows a higher probabilitythan 119880
1
at the same 120572 since the distance between 1198802
andvirtual eBS is shorter causingmore power to be received than1198801
Since 1198801
rsquos channel is worse its EH probability decreasesrapidly when 120572 is not too big
523Throughput Performance On theWITphase we aim tomaximize the minimum throughput among MUs by obtain-ing an optimal power and time allocationThis subsection canbe divided into two parts one is for power allocation and theother is for time allocation
(a) Power Allocation In this part we compare our algorithmUFPE in which the power allocation is using fraction 119901 ofenergy with UAE in which the power allocation is usingup all the battery energy The max-min throughput amongMUs versus circuit sensitivity can be seen in Figure 6 whichindicates that the max-min throughput of both algorithmsdecreases as the sensitivity grows Moreover UFPE shows ahigher throughput than UAE when 120572 is larger than 002Wand the difference will enlarge as 120572 grows The throughputof UFPE is about 33 times of that of UAE when 120572 achieves009WThis means that our algorithm performs better in thepractical scenario where circuit sensitivity exits
Figure 7 depicts the max-min throughput among MUsversus 119876max It can be seen that the max-min throughput ofboth algorithms increases as 119876max grows It is because of thefact that as 119876max grows the constraints of battery would bemore relaxed which can help MUs allocate a more adaptiveslot to transmit information On the other hand both curvesincreasemore smoothlywhen119876max is higher since the energyharvested at MUs will be more difficult to make the batteryfully charged causing the battery constraint to work nomoreWhat is more UFPE also shows a better performance thanUAE with each specific 119876max
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mobile Information Systems 9
UFPEUAE
Max
-min
thr
ough
put (
bps
Hz)
2
25
3
35
4
45
5
001
000
3
000
4
000
5
000
6
000
7
000
8
000
9
000
20
000
1
Qmax (J)
Figure 7 Maximum-minimum throughput between two MUsversus 119876max
Max
-min
thr
ough
put (
bps
Hz)
002 003 004 005 006 007 008 009001120572 (W)
0
1
2
3
4
5
6
7
8
Dynamic 120591120591 = 01
120591 = 03
120591 = 05
120591 = 07
120591 = 09
Figure 8 Maximum-minimum throughput versus circuit sensitiv-ity with different 120591
(b) Slot Allocation An optimal slot allocation is also necessaryto obtain better throughput performance Figure 8 showsthe comparison of our algorithm (dynamically allocating 120591
in each slot) and other models where the slot 120591 is fixedto several values during every period Generally we choosefive values with 120591 = 01 03 05 07 09 respectively It canbe seen that the algorithm of dynamic allocating 120591 showsbetter performance than any other algorithm with fixed 120591The model of 120591 = 05 performs better than any other modelof fixed 120591 On the contrary the model of 120591 = 01 shows theworst performance
6 Conclusion
In this paper we have studied a DEIN model based onvirtualization with a multiantenna virtual eBS a multi-antenna dBS and 119870 single-antenna MUs with finite batterycapacity We proposed a fair resource allocation algorithmby joint optimization of the DL-UL time allocation DLenergy beamforming andUL transmit power allocation withZF based receive beamforming Considering fairness wefirstly design the DL energy beamforming to maximize theminimum received power After calculating the probabilityof energy being transmitted from the virtual eBS to MUswe propose a simple power allocation with a fraction 119901 ofavailable energy allocated for UL WIT and adaptive timeallocation in each slotThe simulation results have shown thatthe proposed UFPE algorithm achieves good performance
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Thiswork is partly supported by the Experts Recruitment andTraining Program of 985 Project (no A1098531023601064)
References
[1] M Chiosi D Clarke P Willis et al ldquoNetwork functionsvirtualization an introduction benefits enablers challengesand call for actionrdquo in Proceedings of the SDN and OpenFlowWorld Congress pp 22ndash24 Darmstadt Germany October 2012
[2] K Doyle Is Network Function Virtualization (NFV) MovingCloser to Reality Global Knowledge Training LLC 2014
[3] E Hernandez-Valencia S Izzo and B Polonsky ldquoHow willNFVSDN transform service provider opexrdquo IEEE Networkvol 29 no 3 pp 60ndash67 2015
[4] J A Paradiso and T Starner ldquoEnergy scavenging formobile andwireless electronicsrdquo IEEE Pervasive Computing vol 4 no 1 pp18ndash27 2005
[5] S Ulukus A Yener E Erkip et al ldquoEnergy harvesting wirelesscommunications a review of recent advancesrdquo IEEE Journal onSelected Areas in Communications vol 33 no 3 pp 360ndash3812015
[6] A Dolgov R Zane and Z Popovic ldquoPower management sys-tem for online low power RF energy harvesting optimizationrdquoIEEE Transactions on Circuits and Systems I Regular Papersvol 57 no 7 pp 1802ndash1811 2010
[7] X Lu P Wang D Niyato D I Kim and Z Han ldquoWirelessnetworks with RF energy harvesting a contemporary surveyrdquoIEEE Communications Surveys amp Tutorials vol 17 no 2 pp757ndash789 2015
[8] H Liu X Li R Vaddi K Ma S Datta and V NarayananldquoTunnel FET RF rectifier design for energy harvesting applica-tionsrdquo IEEE Journal on Emerging and Selected Topics in Circuitsamp Systems vol 4 no 4 pp 400ndash411 2014
[9] J Yang and S Ulukus ldquoOptimal packet scheduling in anenergy harvesting communication systemrdquo IEEE Transactionson Communications vol 60 no 1 pp 220ndash230 2012
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
10 Mobile Information Systems
[10] O Ozel K Tutuncuoglu J Yang S Ulukus and A YenerldquoTransmission with energy harvesting nodes in fading wirelesschannels optimal policiesrdquo IEEE Journal on Selected Areas inCommunications vol 29 no 8 pp 1732ndash1743 2011
[11] K Tutuncuoglu and A Yener ldquoOptimum transmission policiesfor battery limited energy harvesting nodesrdquo IEEE Transactionson Wireless Communications vol 11 no 3 pp 1180ndash1189 2012
[12] L Liu R Zhang and K C Chua ldquoMulti-antenna wirelesspowered communication with energy beamformingrdquo IEEETransactions on Communications vol 62 no 12 pp 4349ndash43612014
[13] Z Zhou M Dong K Ota and Z Chang ldquoEnergy-efficientcontext-aware matching for resource allocation in ultra-densesmall cellsrdquo IEEE Access vol 3 pp 1849ndash1860 2015
[14] Y L Wu G Min and A Y Al-Dubai ldquoA new analytical modelfor multi-hop cognitive radio networksrdquo IEEE Transactions onWireless Communications vol 11 no 5 pp 1643ndash1648 2012
[15] L R Varshney ldquoTransporting information and energy simulta-neouslyrdquo in Proceedings of the IEEE International Symposium onInformation Theory (ISIT rsquo08) pp 1612ndash1616 Toronto CanadaJuly 2008
[16] P Grover and A Sahai ldquoShannonmeets tesla wireless informa-tion andpower transferrdquo inProceedings of the IEEE InternationalSymposium on Information Theory (ISIT rsquo10) pp 2363ndash2367IEEE Austin Tex USA June 2010
[17] K Huang and E Larsson ldquoSimultaneous information andpower transfer for broadband wireless systemsrdquo IEEE Transac-tions on Signal Processing vol 61 no 23 pp 5972ndash5986 2013
[18] D W K Ng E S Lo and R Schober ldquoWireless informationand power transfer energy efficiency optimization in OFDMAsystemsrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6352ndash6370 2013
[19] A M Fouladgar and O Simeone ldquoOn the transfer of informa-tion and energy in multi-user systemsrdquo IEEE CommunicationsLetters vol 16 no 11 pp 1733ndash1736 2012
[20] A A Nasir X Zhou S Durrani and R A Kennedy ldquoRelayingprotocols for wireless energy harvesting and information pro-cessingrdquo IEEETransactions onWireless Communications vol 12no 7 pp 3622ndash3636 2013
[21] C ShenW-C Li and T-H Chang ldquoSimultaneous informationand energy transfer a two-user MISO interference channelcaserdquo in Proceedings of the IEEE Global Communications Con-ference (GLOBECOM rsquo12) pp 3862ndash3867 Anaheim Calif USADecember 2012
[22] J Park and B Clerckx ldquoJoint wireless information and energytransfer in a two-user MIMO interference channelrdquo IEEETransactions on Wireless Communications vol 12 no 8 pp4210ndash4221 2013
[23] X M Chen X M Wang and X F Chen ldquoEnergy-efficientoptimization for wireless information and power transfer inlarge-scale MIMO systems employing energy beamformingrdquoIEEE Wireless Communications Letters vol 2 no 6 pp 667ndash670 2013
[24] H Ju and R Zhang ldquoThroughput maximization in wire-less powered communication networksrdquo IEEE Transactions onWireless Communications vol 13 no 1 pp 418ndash428 2014
[25] G Yang C K Ho R Zhang and Y L Guan ldquoThroughputoptimization for massive MIMO systems powered by wirelessenergy transferrdquo IEEE Journal on Selected Areas in Communica-tions vol 33 no 8 pp 1640ndash1650 2015
[26] I Ahmed A Ikhlef D W K Ng and R Schober ldquoPower allo-cation for an energy harvesting transmitter with hybrid energysourcesrdquo IEEE Transactions on Wireless Communications vol12 no 12 pp 6255ndash6267 2013
[27] J Xu L Liu and R Zhang ldquoMultiuser MISO beamformingfor simultaneous wireless information and power transferrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo13) pp 4754ndash4758Vancouver Canada May 2013
[28] S Boyd andLVandenbergheConvexOptimization CambridgeUniversity Press Cambridge UK 2004
[29] K Liang L Q Zhao K Yang G Zheng and W Ding ldquoAfair power splitting algorithm for simultaneous wireless infor-mation and energy transfer in comp downlink transmissionrdquoWireless Personal Communications vol 85 no 4 pp 2687ndash27102015
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Submit your manuscripts athttpwwwhindawicom
Computer Games Technology
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Distributed Sensor Networks
International Journal of
Advances in
FuzzySystems
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014
International Journal of
ReconfigurableComputing
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Applied Computational Intelligence and Soft Computing
thinspAdvancesthinspinthinsp
Artificial Intelligence
HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014
Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Journal of
Computer Networks and Communications
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation
httpwwwhindawicom Volume 2014
Advances in
Multimedia
International Journal of
Biomedical Imaging
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ArtificialNeural Systems
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Computational Intelligence and Neuroscience
Industrial EngineeringJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Human-ComputerInteraction
Advances in
Computer EngineeringAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014