PresentazionedottorandiXXIV ciclo

Francesco Pantisano

WiLab at the University of Bologna

- Centre for Wireless Communications (CWC)operates as an independent research programme within the Telecommunication Laboratory at the University of Oulu, Finland.

- Funded by Nokia, Nokia Siemens Networks, Texas Instruments, the Finnish Defense Forces, the European Commission, the European Defense Agency, the European Space Agency, Patria, Elektrobit and Tekes.

- Employs a staff of 110.

WiLab at the University of Bologna

CWC‟s research has two dimensions :- Fundamental long-term research (10-15 years)- Applied research : applied research projects are typically utilized in industry in less than 5 years.

CWC‟s fundamental research serves as the basis for developing the necessary technical competences that include :

- future broadband transmission and radio access network techniques- communication signal processing algorithms and architectures for wireless networks- wireless system planning and resource control- wireless sensor networks

WiLab at the University of Bologna

Intelligent power settings for Hybrid Access Underlay

Femtocells

Francesco Pantisano

Ref: [Agilent : IMT-Advanced - 4G Wireless Takes Shape in an Olympic Year]

Context and Motivation

• Indoor cellular accessthrough Femtocells will bethe main enabler for a dramatical capacity growth .

• Nevertheless, some criticalissues are still open:

• which spectral resources willthey use?

• how to manage cross tierand intra-tier interference?

Context and Motivation

Efficiency Spectrum No. of cells

~3~2

~100

Capacity

gro

wth

pote

ntial

Outline

• Femtocell- to- Femtocell interference avoidance through userscheduling.

•Results

• Macrocell- to- Femtocell interference management.•Results

• Coalition Formation for Femtocell Interference Management•Results

•Goals : exploit the frequency reuse in a highly interfered environment.

guarantee fairness among cooperating femtocells.

Constraints: keep the complexity low, keep the latency low.

Intra-tier interference avoidance through user scheduling.

Scenario

t

LTE-A network

Macro : FDD

Femto: TDD over macro UL band

Centralized Scheduling: Optimum scheduling executed by a „genie‟ by means of List Coloring (branch of

Graph Coloring) based on Overall Conflict Graph and Fairness Constrains.

• Distributed Scheduling:• Scheduling is executed locally by means of List Coloring based on Local Conflict

Graph and Fairness Constrains with neighboring cells.

• It guarantees that a cell will never collide with a conflicted user. Although, other collisions depend on the incomplete knowledge of the neighboring cells scheduling decisions.

Proposed solutions:

LTE compatible scheduler:Each femtocell follows a coordinated sequence of UL and DL transmissions.

Random Scheduler:we compare the performance of our proposed schemes with a random allocations.

Comparisons:

Average Spectral efficiency per femto user vs number of FAPs

Results:

•Goals : maximize the rate for co-channel femtocells

Constraints: cope with the Macro users occupying the same resources and “camping” in proximity.

• Cross tier interference management.

• Existing solutions

Closed Access :

Femto Access Point (FAP) is accessed only by subscribers

Open Access :

Femto Access Point (FAP) is accessed only by subscribers

static

static

Problem : MUE may be camp in proximity of a FAP:

Affects the scheduling of the cluster and is potentially

interfered

Main Idea : IF{ MUE is within the transmission

range of FAPa } AND { SINRF < SINRγ for longer than τ}

Then The MUE is ‟absorbed‟ by and

scheduled by the femto access point.

Proposed solution:

dynamic

The FUEs are scheduled through a 3D list-coloring scheme ( 2D conflict graph + 1D PRB)

When a MUE is absorbed then it is scheduled by the FAP and vacates the PRB

When the MUE is in proximity BUT CANNOT BE REACHED , the FAP/CLUSTER performs a INTRA-BAND handover ( it switch to a different PRB)

The new occupied are decided upon the neighbor cell list and the receiver signal strengthreported by each FUE to the FAP.

The same procedure also applies when the absorption of the MUE would result in a unsatisfying QoS.

For a FAP, scheduling new users is costly in two ways :

- QoS for the existing FUEs

- Potentially, it may lead to move onto new PRB

- HENCE: when the MUE is absorbed by Extending the transmission range , that level of power can be adopted also for the inner FUEs.

Proposed solution:

Results:

Average Outage probability per macro user

Results:

Average SINR per macro user

Femtocells are expected to operate on the same constrained spectrum resources.

Therefore, it is desirable to enable impromptu cooperationstowards a common good.

We address the issue of interference management in Underlay FDD femtocells where:

o FAPs may build coalitions to overcome femto-to-femtointerference

o Within each coalition, transmissions are scheduled according to the previously presented scheme.

Coalition Formation for Femtocell InterferenceManagement

The problem is formulated as a COALITIONAL GAME where FAPs are the players.

Each player aims at maximizing its own payoff ( THROUGHPUT) by choosing a cooperative or non-cooperative strategy.

The equilibrium of the game is studied by means of the Recursive Core Algorithm.

Proposed solution:

Average payoff per femtocell vs. network size

Conclusions:

The presented works have been presented in IEEE conferences ( IEEE PIMRC , IEEE Femtocell workshop, ASILOMAR ).

The results have been documented in EU projects ( BeFemto , NEWCOM++ , UNICS , LoCon ) where WiLab or CWC are partners.

Grazie per l‟ attenzione

A glimpse at GC Theory:

• Vertex : it represents a transmission (UL or DL)

• Edge : is a link between two vertices and represents the effect of a Tx on another one.

• Color of an vertex: it represents the resource unit used , in this case the time slot.

• Color of an edge : it is constraint on 2 Txto coexist.

• Conflicted user: is a user (= vertex) within another cell Tx range.

• Local Conflict graph: includes cell´s users , conflicted users and respective edges.

• Overall Conflict graph : is a Conflict Graph of the overall network

• Fairness constraint : is a bound between 2 FAPs to take turns when scheduling conflicted users.

Appendix I: