OFDMA Based Two-hop Cooperative OFDMA Based Two-hop Cooperative Relay Network Resources AllocationRelay Network Resources Allocation

Mohamad Khattar Awad, Xuemin (Sherman) Shen

Student Member, IEEE Senior Member, IEEEDepartment of Electrical & Computer Engineering, University

of Waterloo, Waterloo, Ontario, N2L 3G1, Canada

ICC 2008

Speaker: Chan-Ying Lien (HuHu)

2

OutlineOutline

• Introduction• System Model• Problem Formulation• Proposed Algorithm and Complexity Analysis• Performance Evaluation• Conclusions

3

IntroductionIntroduction

• The channel suffers from– Frequency selective fading

• OFDMA

– Distance dependent fading (i.e., large-scale fading)

• Amplify-and-Forward scheme (AF)

4

IntroductionIntroduction

• To further exploit the wireless channel capacity– A RS can cooperate with a SS in the Time Division

Duplex (TDD) scheme.

• The problem of resource allocation in this network is NP-Complete

5

IntroductionIntroduction

• A low complexity resource allocation protocol

– Each SS communicates with the BS • in non-cooperative mode • in cooperative mode with only one of the available RSs

– Users are allocated a sufficient number of subcarriers to guarantee their minimum rate requirements

– Each subcarrier is exclusively allocated to one SS and RS pair

6

System ModelSystem Model

• Consider a single cell scenario– One BS– Multiple fixed RSs– Multiple SSs

RSRSRSRS RSRSRSRS

RSRSRSRS

RSRSRSRSRSRSRSRS

RSRSRSRS

7

System ModelSystem Model

• Full channel state information (CSI)

• SSs and RSs report their CSI to the BS– SS: minimum rate requirements

• The central resource allocation unit at the BS performs the resource allocation– The subcarriers assignments and RSs assignments to ea

ch SS and RS

8

System ModelSystem Model

• TDD• Half-duplex• The SS transmits while the RS and BS receive in

the first half of the time slot• In the second half of the slot, the RS transmits to

the BS

9

Problem FormulationProblem Formulation

• A SSs– A = {s1, … , sa, … , sA}

• B RSs– B = {r1, … , rb, … , rB}

• A subscriber stations share a total of Nsc subcarriers available to the cell– N = {1, … , n, … ,Nsc}

10

Problem FormulationProblem Formulation

• The maximum achievable rate in (bits/sec/Hz) by s

a on subcarrier n with the cooperation of rb is given by

a: MSb: RSd: BSβb : rb’s amplifying gain

11

Problem FormulationProblem Formulation

• The direct transmission maximum achievable rate in (bits/sec/Hz) over both time slots is given by

12

Problem FormulationProblem Formulation

• Binary Integer Programming (BIP)– Satisfying SSs’ minimum rate requirements ca

– yab {0, 1}∈• yab = 1 means that sa is cooperating with rb

– xnab {0, 1}∈

• xnab = 1 means that the subcarrier n is allocat

ed to the pair sa− rb

13

Problem FormulationProblem Formulation

• Because both SS and RS transmit on the same frequency, but in different time slots, and the resource allocation is time independent, the BS d can be represented by a virtual relay rB+1

• Adding a virtual relay station enlarges the set B to B+

• To use a uniform cost function in the optimization problem

where δ(・ ) is the Dirac delta functionβn

(B+1) = 0d = B+1

14

Problem FormulationProblem Formulation

15

Proposed Algorithm and Complexity AnalysisProposed Algorithm and Complexity Analysis

16

Performance EvaluationPerformance Evaluation

300m150m

RSRSRSRS RSRSRSRS

RSRSRSRS

RSRSRSRSRSRSRSRS

RSRSRSRS

A(MS) = 50B(RS) = 6

17

Performance EvaluationPerformance Evaluation

• The large-scale fading is distance dependant and follows the inverse-power law:

D is the distance between the transmitter and receiver in metersκ is the path loss exponent|αn|2 is the nth subcarrier channel gain at the transmitter

18

Performance EvaluationPerformance Evaluation

19

Performance EvaluationPerformance Evaluation

20

Performance EvaluationPerformance Evaluation

21

Performance EvaluationPerformance Evaluation

• 25 SSs (A = 25), 4 RSs (B = 4)

• 64 subcarriers (Nsc = 64)

22

ConclusionsConclusions

• In this paper, the resource allocation for the two-hop OFDMA cooperative relay networks has been addressed.

• Numerical and complexity analysis demonstrate that the proposed algorithm achieves near optimal allocation in relatively short running time.

• In particular, the algorithm achieves about 70% to 90% of the optimum in only 2% to 12% of the optimization tool running time.

23

Thank You