phd presentation
TRANSCRIPT
Ayman Elnashar
Supervisors:
Prof. Dr. Hamdy El-Mikati Prof. Dr. Said El-Noubi
ECMS (MobiNil)
Mansoura University Alexandria University
30 April 2005
Agenda
Introduction Numerically Robust Multiuser Receivers Quadratically Constrained Robust MUD Robust Adaptive Beamforming Thesis Contributions & Publications
MAP GSM Core
384kpbs-2Mbps
UMTS TD-SCDMA 3GPP CWTS
Up to 14 Mbps/cell
HSDPA
136+
ANSI-136
136 HS IS-95B 64 kbps
CDMA200 1x 307 kbps
EDGE
GPRS i-mode DoCoMo
115 kbps
38
4 kb
ps
GSM TDMA IS-54
CdmaOne IS-95A
PDC
Cellular Standards Evolution Introduction
1980
1995
1999
2001
2002
2002 2003 UWC-136
Japan Europe Americas
TACS NMT/TACS/Other AMPS Traffic is Almost Voice (1st G)
Data 9.6-14.4 Kbps (2nd G)
2004
2005
2006
(2.5 G)
CDMA20001x-EV-DO CDMA20001x-EV-DV
CDMA20003x
ANSI-41 Core
3GPP2 1.4Mpbs 2.4Mbps
3G &
Bey
ond
Multiple Access Techniques
Traffic channels: different users are assigned unique code and transmitted over the entire frequency band, for example, WCDMA and CDMA2000
Traffic channels: different frequency bands are allocated to different users,for example, AMPS and TACS
Traffic channels: different time slots are allocated to different users, for example, DAMPS and GSM
Power
Power
Power
FDMA
TDMA
FDD-CDMA Introduction
TDD-CDMA
Traffic channels: different users are assigned unique code and time slot, for example, TD-SCDMA
Codes
DS/CDMA Systems
In CDMA, users are multiplexed by distinct codes rather than by orthogonal frequency bands, as in FDMA, or by orthogonal time slots, as in TDMA
Introduction
Motivations Limitations
Multiple access interference (MAI) Capacity is interference-limited instead of BW-limited Near/Far Effect: Received power from users near to BS is higher than that of far away users.
We Need tight power control
Admitting asynchronous multiple access
Robustness to frequency selective fading
Multipath combining
Efficient bandwidth utilization
DS-CDMA System Model Introduction
, ,0
( ) ( )jm
j j j m j j mm
t a tα ϕ δ=
= −∑g
( ) ( ) ( )j j jm
n m n m∞
=−∞
= −∑h c .g
( ) ( ). ( )j j j jl
n l n lL τ∞
=−∞
= − −∑u S h
1( ) ( ) ( )
K
jj
n n n=
= +∑x u w Chip rate sampling synchronized to user j
1C
( )tϕ
Channel 1
Channel k Channel noise
Chip Pulse Shaping Filter
Chip Pulse Shaping Filter
Chip Matched Filter
1( )nS
( )j nS
jC
( )cT tϕ −
( )w n
( )ny FIR Linear Filter f
1u
ju
( )nx
Multipath Channel User 1
Data
User k
Data
Signature Sequence
Receiver Filter
Tx
Rx
0 1 2 3 4 5 6 7 8 9 10-0.2
0
0.2
0.4
0.6
0.8
1
1.2
time t (channel length = 10chips)
root
-rai
sed
cosi
ne c
hip
puls
e
Raised Cosine Pulse Shaping Filter
Single user Detection
MF user 1
MF user k
Sync 1 11
Sync k
Sync j 11
MF Bank
Hard
Decision MF user j
( )x n
1y
ˆ jy
ˆky
( )j tc
0
1 (.)bT
bT ∫
1y
jy
ky
Received Signal
Introduction
Multiuser Detection
Multiuser detection considers signals from all users which lead us to joint detection Reduces multiple access interference and hence
leads to capacity increase Alleviates the near/far problem Power Control can be used but not necessary
MUD can be implemented in the base station (BS) or
mobile station (MS), or both
Transmission for the Downlink MS is synchronous and equal-power MUD algorithm is simpler for synchronous CDMA
In case of Uplink the Transmission is Asynchronous
which is more complex and need robust algorithms
Introduction
MUD Techniques
Multiuser Receivers
Optimal MLSE Suboptimal
Linear
Zero- Forcing MMSE
Direct MMSE
Adaptive MMSE
LMS Algorithms
RLS Algorithms DD-MMSE
Blind MMSE
MOE Approach
CMA Approach
Subspace Approach
Polynomial Expansion
Non-linear
Multistage Decision -feedback
Successive interference cancellation
Neural Network
Introduction
Linear Multiuser Receivers
The linear detector output is a linear combination of the received chip sampled signals:
( ) ( ) ( )Hn n n=y f x
{ }( )1 1ˆ ( ) sgn Re ( )n n=s y
{ } { }22( ) ( )H HE n E n= = xxy f x f R f
{ }( ) ( ) ( )Hxx n E n n=R x x
Introduction
Linear receivers are of great significance due to ease of practical implementation
In BPSK the bit decision is made according to:
The detector output energy is given by:
The received signal autocorrelation matrix is given by:
Agenda
Introduction Numerically Robust Multiuser Detection Quadratically Constraint Robust MUD Robust Adaptive Beamforming Thesis Contributions & Publications
IQRD-RLS Algorithm Numerically Robust MUD
• The QR decomposition transforms the RLS problem into a problem that uses only transformed data values by Cholesky factorization of the least-squares data matrix
• This algorithm exhibits a high degree of parallelism, and can be mapped to triangular systolic arrays for efficient parallel implementation.
• Unfortunately, the QRD-RLS algorithm suffers from major drawback, namely, back-substitution which is a costly operation to be performed in array structure
The IQRD method is the promising one due to:
1. Pipelined implementation on VLSI
2. Good numerical stability 3. No back-substitution.
RLS Algorithm QRD-RLS Algorithm
IQR
D-R
LS
In the conventional RLS algorithm, the calculation of the Kalman gain requires matrix inversion of the autocovariance matrix of the received signal. If the data matrix is in ill-conditioned, the conventional RLS algorithm will rapidly become impossible.
IQRD-RLS Algorithm ( ) ( )H
xx n n=R R R ( 1) ( )( )H n nn
λ
− −=
R xaQR Decomposition ( 1)
( )( )
( )0
HH
HT
nn
nn
λ
−− −
=
RR
Pj
( ) 0( )
1 ( )n
nb n
=
aPIQRD Updating
1
Systolic Array Implementation Internal Cell
( 1)iij−
ix
ix ( )iij
( )ia n ( )ia n
( ) ( )iP n ( ) ( )iP n
Boundary Cell
( 1) ( )ib n−
( ) ( )ib n
( )ia n( ) ( )iP n
A rotation matrix , which successively annihilates the elements of intermediate vector against into a related Kalman gain value using a sequence of Givens rotations.
( )nP( )na ( 1)H n λ− −R ( )b n
Numerically Robust MUD
Received vector
Detector Parameters
0T
• The MOE linear detector can be obtained by minimizing the output energy of the receiver subject to certain number of constraints.
• The Closed-form solution of the above constrained optimization problem can be obtained using Lagrange method as follows:
Minimum Output Energy
( ) 11 11 1 1
Hopt xx xx
−− −=f R C C R C g
1 =C f gUnder constraints min Hxxf
f R fDetector vector
Covariance matrix
Channel vector
Signature vector matrix
Numerically Robust MUD
MOE Implementation Using IQRD-RLS
max/ min max/ min1max ( ) ( )H H n n
=gf R R f max/ min 1 1( )nβ=fΔ υ
( ) ( 1) ( ) ( )Hn n n nλ= − +Ψ Ψ d d1
1
( 1) ( )( )1 ( ) ( 1 ( )H
n nnn n n
λ
λ
−
−
−=
− −
Π πdπ Π )π
11 max/ min 1
max ( )H n−
== =
gυ g g Π g
{ } 11 1
1 1 1( ) ( ) ( ) ( )H H Hf n n n n−− −
= R R C C R R C g1
1( ) ( ) ( )Hn n n−
= Δ R R C
1( ) ( )Hn n=Π C Δ 1( ) ( ) ( )n n n−=fΔ Π g 1( ) ( 1) ( ) ( )Hn n n nλ−= − −Δ Δ j π 1( ) ( )Hn n=π C j
1( ) ( 1) ( ) ( )Hn n n nλ−= − −Π Π π π2 1 1
1 11
( 1) ( ) ( ) ( 1)( ) ( 1)1 ( ) ( 1) ( )
H
Hn n n nn n
n n nλλ
λ
− −− −
−
− −= − +
− −Π π π ΠΠ Π
π Π π
Detector Estimation
Channel Estimation
Any orthogonal subspace tracking algorithm can be employed for tracking the principle component of the.
• orthogonal projection approximation subspace tracking (OPASTd)
• normalized orthogonal OJA (NOOJA).
Subspace Tracking
Numerically Robust MUD
Subspace Tracking (new)
( ) 111 11
max H Hxx
−−
=gg C R C g
1( , ) ( 1) ( ) ( 1) ( 1)(1 ( 1) ( 1))2
H Hn n n n n n nζ ζ= − − + − − − −Ψ g g Π g g g
Cost Function
( ) ( 1) ( , )n n µ ζ= − − ∇gg gΨ g
Channel Update
21( , ) ( , ) 2 ( 1) ( 1) ( ) ( ) ( 1) ( ) ( ) ( )H H H
n n n n n n n n n nζ ζ µ µ−= + − − ∑ − − ∑ ∑g g gΨ g Ψ g g Π Π
Step-Size Estimation
( 1) ( ) ( )( )
( ) ( ) ( )
H
opt H
n n nn
n n nα
µη
− ∑=∑ ∑ +
g
g g
gΠΠ
Optimum Step-Size
2 ( ) 2 ( )a n b n cζ ζ− + ( 1)opta nµ= −
1 ( 1) ( 1) ( ) ( 1)Hoptb n n n nµ= + − − −gΠ g
2( 1) ( 1) ( ) ( 1) 2 ( 1) ( ) ( 1)H Hoptc n n n n n n nµ= − − − + − −gΠ g g Π g
2
( ) b b acna
ζ − ± −=
Lagrange Multiplier
( ) ( ) ( 1) ( ) ( 1)n n n n nζ∑ = − − −g Π g gGradient Vector
Numerically Robust MUD
Channel Vector Estimation Techniques
max/ min 1 1( )nβ=fΔ υ( ) 111 11 1
max maxH H Hxx xx
−−
= ==
g gf R f g C R C gMax/min Approach
1
1 1 1 1( ) ( ) .H Hxxn n γ−= −φ C R C C C ( ) ( ) ( )IMOE n n n=fΔ g Improved Cost
2( ) ( )fxx xx Nn n ασ= −R R I 1
1 11min ( )H H
xx−
==
gg g C R C g ( ) ( ) ( )MMOE n n n=fΔ gModified Cost
11 1
ˆ1 1
ˆ ˆˆ minˆ ˆ
H Hxx
H H
−
=g
g C R C ggg C C g
ˆ( ) ( ) ( )Capon n n n=fΔ gCapon Method
21 11
min ( )H Hxx−
==
gg g C R C g ( ) ( ) ( )POR n n n=fΔ g Power Method (POR)
{ }{ }11max min . . . .H H H
xxf f s t f s t f f ρ=
= ≤fg
R C g ( ) 1max/ min 1 max/ min
ˆxx Iν −= +f R C g
New Robust Multiuser detection technique
Numerically Robust MUD
Simulation Results (1)
0 100 200 300 400 500 600 700 800 900 10001
2
3
4
5
6
7
8
9
Iteration (n)
Out
put S
INR
(dB
)
MOE-IQRD w. Optimal channelMOE-IQRD w. Lagrange (MC)MOE-IQRD w. NOOja (PC)MOE-IQRD w. Lagrange (PC)MOE-IQRD w. OPASTd (PC)
SINR Comparison of Subspace Tracking Algorithms
Numerically Robust MUD
Simulation Results (2)
0 100 200 300 400 500 600 700 800 900 10000
1
2
3
4
5
6
7
8
9
10
snapshot index
Out
put S
INR
(dB
)
MOE-RLSMOE-RLS w. VLMOE-IQRD w. max/min methodMOE-IQRD w. Improved cost functionMOE-IQRD w. Modified cost functionMOE-IQRD w. Capon methodMOE-IQRD w. POR methodMOE-IQRD w. max/min and VL
Comparison between Output SINR for MOE-IQRD based detectors
Numerically Robust MUD
Complexity Analysis
3371 MOE-IQRD w. max/min and VL
2556 MOE-IQRD w. Capon
4046 MOE-IQRD w. POR
4356 MOE-IQRD w. modified
1356 MOE-IQRD w. Improved
1356 MOE-IQRD w. max/min
2079 - MOE-RLS w. VL
1596 - - MOE-RLS
Special case
Total complexity
Weight vector
Channel vector /VL technique
Intermediate matrix update
Kalman gain
Detector
2a aN N+
2a aN N+
6 fN
6 fN
6 fN
6 fN
6 fN
6 fN
2 2a a f aN N N N+ +
2 2a a f aN N N N+ +
22 f g gN N N+
22 f g gN N N+
2 2 2f g f f gN N N N N+ +
2f g f gN N N N+
22 2 2g g f gN N N N+ +
22 f g gN N N+2 4g gN N+
22 3f fN N+
3 22 2g g gN N N+ +
2 4g gN N+
2 4g gN N+
2 4g gN N+
2 4g gN N+
2 2a aN N+
f gN N
f gN N
f gN N
f gN N
f gN N
f gN N
22 3a a f aN N N N+ +
23 5a a f aN N N N+ +
23 2
4 6f g g
g f
N N NN N
+
+ +23 2
4 6f g g
g f
N N NN N
+
+ +2 2
2
3
4 6f g f f g
g g f
N N N N N
N N N
+ +
+ + +2
2
2
4 6f g f g
g g f
N N N N
N N N
+
+ + +3 23 4
4 6g f g g
g f
N N N NN N+ +
+ +
2 23 2 2
4 9f g g f
g f
N N N NN N
+ +
+ +
31, 10f gN N= =
Numerically Robust MUD
VL Techniques Comparison
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
100.7
100.8
QI Constrained Value
Out
put S
INR
Ave
rage
(dB
)
0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24
100.4
100.5
100.6
QI Constrained Value
Out
put S
INR
Ave
rage
(dB
)
MOE-IQRD w. max/min and VL
MOE-RLS w. VL
MOE-IQRD w. max/min
MOE-RLS
Variable Loading Technique Comparison
Numerically Robust MUD
Agenda
Introduction Numerically Robust Multiuser Receivers Quadratically Constrained Robust MUD Robust Adaptive Beamforming Thesis Contributions & Publications
MOE Implemented using PLIC structure
Adaptive Algorithm
+
-
Hqf
Haf
( )cy n( )y n
( )ay n
HB
( )x n
( ) 1
( )H H
a opt xx xx c
−=f B R B B R f
c a= −f f Bf
( ) 1
1 1 1H
c
−=f C C C g
( ) ( )mina
Hc a xx c a− −
ff Bf R f Bf
QC Robust MUD
Received vector
min Hxxf
f R f
Blocking Matrix
Reduced Rank Filter
Non-Adaptive Part
Optimal Detector
Robust MOE with QI constraint
( ) ( )mina
Ha a c a= − −c xxf
f f Bf R f Bf Under constraints 2Ha a β≤f f
( ) 1( ) 0a opt B Bλ −= +f R I p
HB = xxR B R BB B c=p P f H
B xx=P B R
ˆ( ) ( ) ( )a a an n nγ≈ −f f f
( ) ( )1 11 1 10 0( ) ( ) ( ) ( ) ( ) ( )a B B B B an I n n n I n nλ λ
− −− − −= + = +f R R p R f1( ) ( ) ( )a B Bn n n−=f R p
1ˆ ( ) ( ) ( )a B an n n−=f R f
{ }2Re 4
2
b b ac
aγ
− ± −=
Optimal Detector
Taylor Series
Lagrange Multiplier ˆ ˆ( ) ( )Ha aa n n= f f
{ }ˆ2Re ( ) ( )Ha ab n n= − f f
2( ) ( )Ha ac n n β= −f f
2Ha a β≤f f
RLS-based VL
QC Robust MUD
Robust MOE with QI constraint (RSD-VL)
( ) ( ) ( )20
12
H Hc a c a a asλ β= − − + −
af xxΨ f Bf R f Bf f fCost Function
( ) ( 1) ( )a n n nµ= − − ∇aa ff fDetector Update
0( ) ( ) ( ) ( 1) ( 1)H Hxx c xx an n n n nλ∇ = − + − + −
af aB R f B R Bf fGradient Vector
0( ) ( 1) ( ( ) ( 1) ( )) ( 1)a a an n n n n nµ µλ= − − − − − −B a Bf f R f p fRobust Detector
[ ]( ) ( 1) ( ) ( 1) ( )an n n n nµ= − − − −a B a Bf f R f pNon-Robust
( ) ( ) 20 0( ) ( 1) ( ) ( 1)
H
a an n n nµλ µλ β− − − − ≤a af f f fQI Constraint
{ }
2 2 20 0( 1) ( 1) 2 Re ( ) ( 1) ( ) ( ) 0
H HHa a a an n n n n nµ λ µ λ β− − − − + − =a af f f f f f
Quadratic Equation
Lagrange Multiplier 2
04
2b b ac
aλ − ± −
=
22 ( 1)a nµ= −af
{ }2 Re ( ) ( 1)H
aab n nµ= − −f f
2 2( )ac n β= −f
QC Robust MUD
Optimum Step-size of MOE-RSD w. VL
( ) ( ) ( )20
12
H Hc a c a a asλ β= − − + −
af xxΨ f Bf R f Bf f fCost Function
Updated Cost Function ( ) ( )( ) ( 1) ( ) ( ) ( 1) ( )a a a
Hn n n n n nµ µ= − + ∇ − + ∇xxf f fΨ f B R f B
[ ]( ) ( 1) ( ) ( 1) ( )an n n n nµ= − − − −a B a Bf f R f pNon-Robust Detector
2( ) ( 1) 2 ( ) ( ) ( 1) ( ) ( ) ( ) ( )a a a a a
HBn n n n n n n n nµ µ= − + ∇ − + ∇ ∇H
Bf f f f fΨ Ψ P f R
Quadratic Equation
( )2 ( ) ( 1) 2 ( ) ( ) ( ) ( )
( )a
a a a
H HB
nn n n n n n
nµ
µ
∂= ∇ − + ∇ ∇
∂f
Bf f f
ΨP f R
Differentiate zero
2( )
( )( ) ( ) ( )
a
a a
opt H
nn
n n n
αµ
σ
∇=∇ ∇ +
f
Bf fR
Optimum Step-Size
QC Robust MUD
Geometric Approach
O
The RLS-based VL technique The RSD-based VL technique
( )a nfE(SP)
ˆ ( )a nf
ˆ ( )a nf
( )a nf
DAF
ˆ ( )a nγ− f
ˆRe( ) ( )a nγ− f( ) ( )
afn nµ− ∇
( 1)n −af
( )a nf
20( ) ( ) ( 1)an n nµ λ− −f
( )a nf
10( ) ( ) ( 1)an n nµ λ− −f
C1A
C2
O
B
C
QC Robust MUD
Simulation Results (SINR)
0 100 200 300 400 500 600 700 800 900 10004
5
6
7
8
9
10
11
12
13
Iterations
Out
put S
INR
(dB
)
MOE-RLSMOE-RLS w. QCProposed
Output SINR with SNR = 30dB, 5 synchronous users, 31 Gold
Codes, and -10dB weaken user
0 100 200 300 400 500 600 700 800 900 10003
3.5
4
4.5
5
5.5
6
6.5
7
Iterations
Out
put S
INR
(dB
)
MOE-RLSMOE-RSDMOE-RLS w. QCProposed
Output SINR with SNR = 20dB, 5 synchronous users, 31 Gold
Codes, and -10dB weaken user
QC Robust MUD
Simulation Results (3)
0 100 200 300 400 500 600 700 800 900 10003.5
4
4.5
5
5.5
6
6.5
7
Iterations
Out
put S
INR
(dB
)
MOE-RSD, alpha = 0.01MOE-RSD, alpha = 0.1MOE-RSD, alpha = 0.9MOE-RSD w. QC, alpha = 0.01MOE-RSD w. QC, alpha = 0.1MOE-RSD w. QC, alpha = 0.9
MOE-RSD
MOE-RSD w. QC
Output SINR with SNR = 20dB, 5 synchronous users, 31 Gold Codes, and -10dB weaken user and variable step-size
QC Robust MUD
Robust CMA with QI Constraint
LCCMA1
LCCMA2
BSCMA
( )22
1min ( ) HJ E r − f
f f x
1H =C f g
( )22min ( ) ( )
a
Ha c aJ E r − −
ff f Bf x
2Ha a β≤f f
2Ha a β≤f fS.T. &
S.T.
( )1 2
0
1( ) ( ) ( ) 14
NH
a an
f j n jM
−
=
Ψ = − ∑ f Z f2( ) ( )H
a aj j β≤f f
1
( 1)( ) ( ) ( )
iMT
n i Mi n n
−
= −
= ∑Z z z
S.T.
( ){ }2
2min ( ) HJ E r−f
f f x 1H =C f g
{ } { }2 ( ) H H HJ E r E− +f f x f x x f 2 ( ) ( )H HJ n− +f f x f R f
min ( ) ( ) ( ) ( )( )a
H Ha c a c a c aJ R n− − + − −
ff f Bf x f Bf f Bf
2Ha a β≤f f
2Ha a β≤f fS.T. &
S.T.
QC Robust MUD
Simulation Results of Robust CMA
0 100 200 300 400 500 600 700 800 900 1000-10
-5
0
5
10
15
Iterations (n)
Out
put
SIN
R (
dB)
LCCMA1 w/t W.LCCMA1 w. W.LCCMA1 w. VLLCCMA2 w/t QILCCMA2 w. SPLCCMA2 w. VLLCCMA2 w. CG
Output SINR for Different LCCMA receivers with SNR = 30dB, 5 synchronous users, 31
Gold Codes, and -10dB weaken user
0 50 100 150 200 250-4
-3
-2
-1
0
1
2
3
4
5
Block Iteration (j)
Out
put
SIN
R (
dB)
BSCMA w. VLBCGCMA w. VLBGDCMA w. VLBSCMA BCGCMA BGDCMA
Output SINR for BSCMA receivers with SNR = 30dB, 5 synchronous users, 31 Gold Codes,
and -10dB weaken user
QC Robust MUD
Agenda
Introduction Numerically Robust Multiuser Receivers Quadratically Constrained Robust MUD Robust Adaptive Beamforming Thesis Contributions & Publications
LCMV Beamforming
Adaptive beamforming has been exploited in wireless communications, radar, sonar, speech processing, and other areas.
Recently, there has been a great effort to design robust adaptive beamforming techniques which improve robustness against mismatch and modeling errors and enhancing interference cancellation capability.
The mismatch may be caused by uncertainty in direction-of-arrival (DOA), imperfect array calibration, near-far effect, and other mismatch and modeling errors.
The so-called linearly constrained minimum variance (LCMV) beamformer, also known as Capon’s method, has bean a popular beamforming technique.
In LCMV beamforming method, the weights are chosen to minimize the array output power subject to side constraint (s) in the desired look direction (s).
This method assumes that the array manifold is accurately known, unfortunately, even small discrepancy between the presumed and the actual array manifold can substantially degrade its performance.
min Hxxw
w R w 0 0( ) 1H θ =w aS. T. 1
0 00 1
0 0 0 0
( )( ) ( )
xxH
xx
θθ θ
−
−=R aw
a R a
Robust Beamforming
Diagonal Loading Technique
Diagonal loading is a technique where the diagonal of the covariance matrix is augmented with a positive or negative constant prior to inversion
Diagonal loading technique has been a widespread approach to improve robustness against mismatch errors and random perturbations
Moreover, the performance of the signal detectors, which utilize the inverse of the data covariance matrix, experiences serious degradation when the sample support available for estimating the matrix is limited.
This problem can be overcome also by diagonally loading the data covariance matrix
Furthermore, it is well known that antenna sidelobes can be made small if the sample data correlation matrix is diagonally loaded before inversion is performed
Robust Beamforming
Robust Beamforming Design
The SOCP approach can be interpreted as a diagonal loading technique in which the optimal value of diagonal loading is computed based on the known upper bound on the norm of the signal steering vector mismatch
The SeDuMe optimization Matlab toolbox has been used to compute the weight vector of SCOP approach.
Unfortunately, the computational burden of this software seems to be cumbersome which limits the practical implementation of this technique.
The SOCP-based method does not provide any closed-from solution, and does not have simple on-line implementations
In addition, this technique can be regarded as batch algorithm rather than adaptive scheme.
SOCP Approach min H
xxww R w 1 ( )H ε≥ ∀ ∈w c c A { }0( ) | ,e eε ε= = + ≤A c c a
min Hxxw
w R w 0 1H ε≥ +w a w { }0Im 0H =w a
S.T.
S.T. &
Robust Beamforming
0ˆmax min H
xxwaw R w 0 0ˆ ( ) 1H θ =w a 1
0 0 0 0ˆ ˆ( ( ) ) ( ( ) ) 1Hk k−− − ≤a a C a a
10 0ˆ
ˆ ˆmin ( ) ( ) ( )Hxxa
k k k−a R a 20 0ˆ ( )k ε− ≤a a 1 1
ε− =C I
11
0 0( )ˆ xx kλ
−− = +
Ra I a1
0 00 1
0 0 0 0
ˆ ( )ˆˆ ˆ( ) ( )
xxH
xx
θθ θ
−
−=R aw
a R a ( )
2
21
( )1
Mm
j m
zg λ ε
λγ=
=+
∑ Hxx =R UΓU0
H=z U a
S.T. &
S.T. where
Ellipsoidal
Constraint
Robust Beamforming Design (2)
3( )O M Eigendecomposition requires high computational burden of order The adaptive implementation updates both the covariance matrix and its
inverse to compute the diagonal loading value and the robust detector This technique is based on batch algorithm The rank of signal and noise may be uncertain or not exactly known and
need to be estimated in advance. The covariance matrix will be always diagonally loaded even without
mismatch.
Robust Beamforming
Proposed Formulation
ˆ ˆ( ) ( 1) ( ) ( )SDk k k kµ= − −0 0a a g1
( ) ( )( )( ) ( ) ( )
H
SD Hxx
k kkk k kαµ
σ−=+
g gg R g ( )1
0ˆ ˆ( ) ( ) ( 1) ( 1)xxk k k kλ−= − + − −0 0g R a a a
2 2 20 0 0 0( 1) ( 1) ( ) ( ) ( 1) ( 1) ( )H H
SD k k k k k k kµ ε ≤ − − − − − d d g g d d g
{ } { }0 0 0( 1) Re ( 1) , Im ( 1)T Tk k k − − − d d d
{ } { }( ) Re ( ) , Im ( )T Tk k k = g g g
( )21ˆ 0 0 0 0ˆ ˆ ˆ( ) ( ) ( ) ( ) ( )
2H
xxk k k k t kλ ε−Ψ = + − −a a R a a aCost Function
21 1 1 0b a c− ≥Step-Size Constraint
221 0 0ˆ( ) ( 1) 0SDa k kµ= − − >a a ( ) ( ){ }1 0 0 0 0ˆ( ) Re ( ) ( 1)H
SDb k k kµ= − − −a a a a
21 0 0( ) 0c k ε= − − >a a
21 1 1 1
1
( )b b a c
ka
λ± −
=Diagonal Loading Term
( )( )( ) ( )( )( )0 0 0 0 0 0 0 0ˆ ˆ( ) ( ) ( ) ( 1) ( ) ( ) ( ) ( 1)H
SD SDk k k k k k k kµ λ µ λ ε− − − − − − − − ≤a a a a a a a a
Spherical Constraint
( )( )10ˆ ˆ ˆ ˆ( ) ( 1) ( ) ( ) ( 1) ( ) ( 1)SD xxk k k k k k kµ λ−= − − − + − −0 0 0 0a a R a a a 1
0 0 0ˆ ˆ( ) ( 1) ( ) ( ) ( 1)SD xxk k k k kµ −= − − −a a R a
Steering Vector Update
Step-Size
Gradient Vector
Robust Beamforming
Geometric Approach
Q
A
B
C
O
D
( )0ˆ( ) ( ) ( 1)k k kµ λ − −a a
0 ( )ka
0ˆ ( )ka0ˆ ( 1)k +a
a
0 ( 1)k −d
0 ( )kd
0 ( )kd
1
2
ε
Geometric Representation for Robust Capon Beamforming with ellipsoidal constraint
Array direction Ar
ray
broa
dsid
e
Robust Beamforming
000 0ˆˆ
ˆ ˆmax min Hxxwa
w R w 0 0 0ˆ ˆ ( ) 1H θ =w a 20 0ˆ ( )k ε− ≤a a& & S.T.
0 0ˆ ˆH τ≤w w
Joint Constraint Approach
00 0ˆ
ˆ ˆmax Hxxa
w R w1
LH
xx i i i ni=
= + ∑Rρ s s R
0
20 0 0 0ˆ
1
ˆ ˆ ˆ ˆmaxL
H H Hi i i
iσ
=
+
∑awρ s s w w w
( )( )
10
0 10 0
ˆ ( )ˆ ˆ( ) ( )
xxH
xx
kk k
υ
υ
−
−
+=
+
R I aw
a R I a
( ) ( )( ) ( )
1 10
0 1 10 0( )
xx xx xxH
xx xx xxkυ λ
λ λ
− −
− −
+ +=
+ +
R I R R I aw
a R I R R I a
( ) 10
0 10 0
ˆ ( )ˆ ˆ( ) ( )
xxH
xx
kk kυ −
−
+=
R I aw
a R a
( ) 11 10
0 10 0
ˆ ( )ˆ ˆ( ) ( )
xx xxH
xx
kk k
υ−− −
−
+=
I R R aw
a R a
( )1 10
0 10 0
ˆ ( )ˆ ˆ( ) ( )
xx xxH
xx
kk k
υ − −
−
−≈
I R R aw
a R a
10 0ˆxx
−=w R w0 0 0ˆ υ≈ −w w w
11
0 0( )ˆ xx kλ
−− = +
Ra I a
Robust Beamforming
Simulation Scenario
Presumed DOA
Actual DOA Jammer 1 Direction
2λ
0.03π
2ϕ
1ϕMismatch angle
Jammer 2 direction
Robust Beamforming
1w
1w1w1w
1w
∑
Array Output
Control algorithm
Adaptive processor
Signal processor
Beamformer
Simulation Results (SINR)
0 100 200 300 400 500 600 700 800 900 10000
5
10
15
20
25
Iterations (n)
SIN
R (d
B)
Standared CaponRobust Capon (Batch)Robust Capon (SS)Robust (SOCP)Proposed1Proposed2
Output SINR versus snapshot for SNR =40 dB, two 10dB interference, 0.3pi mismatch angle
0 100 200 300 400 500 600 700 800 900 1000-20
-15
-10
-5
0
5
10
15
Iterations (n)
SIN
R (d
B)
Standared CaponRobust Capon (Batch)Robust Capon (SS)Robust (SOCP)Proposed1Proposed2
Output SINR versus snapshot for SNR =20 dB, two 10dB interference, 0.3pi mismatch angle
Robust Beamforming
Simulation Results (Beampatterns)
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2-60
-50
-40
-30
-20
-10
0
Standared CaponRobust Capon (Batch)Robust Capon (SS)Robust (SOCP)Proposed
Angle (radian)steady state beampatterns for versus snapshot for SNR
=40 dB, two 10dB interference, 0.3pi mismatch angle
Robust Beamforming
Simulation Results (Moving Interference)
0 100 200 300 400 500 600 700 800 900 10000
5
10
15
20
25
Iterations (n)
SIN
R (d
B)
Standared CaponRobust Capon (Batch)Robust Capon (SS)Robust (SOCP)Proposed
Output SINR versus snapshot for SNR =20 dB, two moving 10dB interference, 0.3pi
mismatch angle
0 100 200 300 400 500 600 700 800 900 10000
2
4
6
8
10
12
14
16
18
20
Iterations (n)
SINR
(dB)
Standared CaponRobust Capon (Batch)Robust Capon (SS)Robust (SOCP)Proposed
Output SINR versus snapshot for SNR =20 dB, two coherent moving 10dB interference, 0.3pi mismatch angle
Robust Beamforming
Agenda
Introduction Numerically Robust Multiuser Receivers Quadratically Constraint Robust MUD Robust Adaptive Beamforming Thesis Contributions & Publications
Contributions Summary (1)
• A general DS/CDMA system model which account for asynchronism, multipath propagation, near-far effect, signature mismatch, and inter-symbol-interference (ISI) is developed.
• MUD survey and performance comparison for existing techniques is performed anchored in the proposed model.
• A fast subspace tracking algorithm is Developed and deployed for channel estimation with MOE detector.
• A generalized frame work for building IQRD-based multiuser receivers is offered.
• Based on the above proposed frame work, comparative analyses between the recently proposed channel estimation techniques, subspace tracking and the proposed techniques is conducted.
• A combined subspace approach and a quadratic constraint is proposed to produce robust and optimum multiuser receiver.
• The systolic array implementation is exploited to facilitate real-time implementation of the proposed IQRD-based receivers.
Contributions &Publications
Contributions Summery (2)
• A new VL technique is devised in this thesis and integrated into a recursive steepest descent (RSD) algorithm rather than the RLS algorithms to produce robust MOE detector with low-computational complexity. This VL is exploited to fulfill the quadratic constraint on the detector norm to improve the performance of the multiuser receiver against modeling and mismatch errors.
• Additionally, an optimum step-size closed-form expression for the proposed RSD algorithm is derived.
• The proposed VL technique has been integrated also into the LCCMA algorithms and the BSCMA algorithm to produce robust constant modulus based receivers for sample-by-sample and block-adaptive, respectively.
• We have proposed a low-complexity recursive implementation for the robust Capon beamforming algorithm which incorporating ellipsoidal constraint on the steering vector using the proposed RSD algorithm and the recursive conjugate gradient (RCG) algorithm.
• Additionally, a joint constraint approach is proposed to produce robust beamforming algorithm which is capable of providing robustness against steering vector mismatch and noise enhancement at low SNR.
• A comparative analysis is conducted between most recent beamforming algorithms as well as the proposed approaches in the presence of moving and coherent jamming.
Contributions &Publications
Publications List (updated on Oct. 2011) • Major Publications in Refereed Journals: • A. Elnashar, “On efficient implementation of robust adaptive beamforming based on worst-case
performance optimization” IET Signal Processing, Vol. 2, No. 4, pp. 381-393, Dec. 2008. • A. Elnashar, S. Elnoubi, and H. Elmikati “Performance Analysis of Blind Adaptive MOE Multiuser
Receivers using Inverse QRD-RLS Algorithm,” IEEE Trans. On Circuits and systems I, Vol. 55, No. 1, pp. 398-411, Feb. 2008.
• A. Elnashar, S. Elnoubi, and H. Elmikati “Further study on Robust Adaptive Beamforming with optimum diagonal loading,” IEEE Trans. on Antennas and Propagation, Vol. 54, No 12, pp. 3647-3658, Dec. 2006.
• A. Elnashar, S. Elnoubi, and H. Elmikati, “Low-Complexity Robust Adaptive Generalized Sidelobe Canceller Detector for DS/CDMA Systems,” International Journal of Adaptive Control and Signal Processing, vol. 23, no. 3, pp. 293-310,March 2008, John Wiley & Sons, Ltd.
• T. Samir, S. Elnoubi, and A. Elnashar, “Block-Shanno Minimum BER Beamforming,” IEEE transactions on Vehicular Technology, Vol. 57, No. 5, pp. 2981-2990, Sept. 2008.
• International Conferences: • T. Samir, S. Elnoubi, and A. Elnashar “Block-Shanno MBER algorithm in a spatial multiuser
MIMO/OFDM” in Proc. 14th European Wireless Conference EW2008, 22-25 June 2008. • T. Samir, S. Elnoubi, and A. Elnashar “Class of Minimum Bit Error Rate Algorithms,” in Proc. ICACT
2007, Korea, Feb. 2007, pp. 168-173. • T. Samir, S. Elnoubi, and A. Elnashar, “Block-Shanno Minimum BER Beamforming” in Proc. ISSPA
2007, UAE, Feb. 2007. • A. Elnashar, “Robust Adaptive Beamforming,” ACE2 Network of Excellence Workshop on Smart
Antennas, MIMO Systems and Related Technologies, Myconos, Greece, 8 June 2006.
Contributions &Publications
Publications List (Cont.)
• A. Elnashar, S. Elnoubi, and H. Elmikati, “Performance analysis of robust MOE detectors at low SNR based on the IQRD-RLS algorithm,” In Proc. IST Mobile and Wireless Communications Summit, Myconos, Greece, 4-8 June, 2006.
• A. Elnashar, S. Elnoubi, and H. Elmikati “Robust Adaptive Beamforming with Variable Diagonal Loading,” In Proc. Sixth International Conference on 3G and Beyond - 3G 2005, 07-09 November 2005, The IEE, Savoy Place, London, UK, pp. 489-493.
• A. Elnashar, S. Elnoubi, and H. Elmikati “A Robust Block-Shanno Adaptive Blind Multiuser Receiver for DS-CDMA Systems,” In Proc. IST Mobile & Wireless Communications Summit 2005, Dresden, Germany, 19-23 June, 2005.
• A. Elnashar, S. Elnoubi, and H. Elmikati “A Robust Linearly Constrained CMA for Adaptive Blind Multiuser Detection,” In Proc. IEEE WCNC 2005 conference, Vol. 1, pp. 233-238, New Orleans, LA, USA, 13-17 March, 2005,
• A. Elnashar, S. Elnoubi, and H. Elmikati “A Robust Quadratically Constrained Adaptive Blind Multiuser Receiver for DS/CDMA Systems,” IEEE International Symposium on Spread Spectrum Techniques and Applications (ISSSTA 2004), Sydney, Australia 30 Aug 2004 - 3 Sept 2004.
• A. Elnashar, S. Elnoubi, and H. Elmikati “A Novel Adaptive Blind Multiuser Receiver for DS/CDMA Based on combined Inverse QRD-RLS Algorithm and constrained Optimization Approach,” in Proc. ISPACS 2003, Awaji Island, Japan, pp. 423-428, December 7-10, 2003.
• A. Elnashar, S. Elnoubi, and H. Elmikati “Computationally Efficient Real-Time Blind Multiuser Detection for cellular DS/CDMA Based on Inverse QRD-RLS Algorithm and Subspace Tracking,” in Proc. MWSCAS 2003, Cairo, Egypt, December 28-31, 2003.
• A. Elnashar, S. Elnoubi, and H. Elmikati “Robust Adaptive Blind Multiuser Receiver for DS/CDMA Based on combined Inverse QRD-RLS Algorithm and MOE” in Proc. SOFTCOM 2003 conference, Croatia, pp. 512-515, Oct. 2003.
Contributions &Publications