Resource Allocation in Energy-constrained Cooperative Wireless Networks

Transcription

1 Resource Allocation in Energy-constrained Cooperative Wireless Networks Lin Dai City University of Hong ong Jun. 4,

2 Outline Resource Allocation in Wireless Networks Tradeoff between Fairness and Throughput Fairness and Throughput in Energy-constrained Cooperative Networks Optimal Resource Allocation in Energy-constrained Cooperative Networks Jun. 4,

3 Wireless Network Medium electromagnetic wave fading channel shared spectrum Terminal cellphone, PDA, laptop, portable not so smart Increasing demand for a large variety of services Voice, data, video, Resource Allocation Limited resources Bandwidth, power, processing capability Jun. 4,

4 Optimal Resource Allocation Maximize Subject to k 1 T k () S ( ) q, k 1,..., k k Single User Power allocation Antenna selection Subcarrier allocation -User Throughput maximization vs. Fairness Effort fairness: Fairness in allocating the resources Outcome fairness: Fairness in utilizing the resources Jun. 4,

5 Example: Opportunistic Transmission Allocate different time slots to different users. Objective: Maximize k 1 T k () Total throughput can be maximized by always serving the user with the strongest channel. The more users, the higher throughput. Multiuser Diversity Jun. 4,

6 Opportunistic Transmission with Fairness Constraint Help the poor -- a disadvantaged user is scheduled when its instantaneous channel quality is high relative to its own average channel condition. Maximize k 1 log( T ( )) k Jun. 4,

7 Tradeoff between Throughput and Fairness Opportunistic Transmission without Fairness Constraint Opportunistic Transmission with Fairness Constraint There is always a tradeoff between fairness and throughput. Jun. 4,

8 Outline Resource Allocation in Wireless Networks Tradeoff between Fairness and Throughput Fairness and Throughput in Energy-constrained Cooperative Networks Optimal Resource Allocation in Energy-constrained Cooperative Networks Jun. 4,

9 Cooperative Networks Diversity gain: p e ~ SNR -L How to achieve diversity gain in wireless ad-hoc networks? source node destination node relay set Node cooperation: more relay nodes, higher cooperative diversity gain. Jun. 4,

10 Multiuser Cooperative Protocol 1 Unfair! Some node may have more chances to be relays. Refuse to cooperate Run out of energy soon Jun. 4,

11 Fairness and Throughput in Energy-constrained Cooperative Networks Number of Nodes k 2 Fair Effort fairness: equal lifetime k 1 Time slots Unfair k k 1 2 at any time slot t higher cooperative diversity gain More nodes: Higher throughput! higher multiuser diversity gain Improved fairness may lead to throughput gains in energy-constrained cooperative networks. Jun. 4,

12 Network Model A wireless ad-hoc network with nodes Each node with an energy constraint of E r S R d Relay region D R Opportunistic transmission Jun. 4,

13 Full Cooperative Protocol Unfair Popular nodes have more chances of acting as relays. They will run out of energy much faster than others. Jun. 4,

14 How to Improve Fairness? Resources required by each node should be no more than what it contributes to other nodes. Power Reward adopted by each node to evaluate the power contributed to and by other nodes. Wk Wk Pk W W j k k k k j R k P k j Power reward increases when node k acts as a relay Power reward decreases when node k employs other nodes as relays Jun. 4,

15 Fair Cooperative Protocol For each pair kdk, ( ), compare W and the sum relay power If W, use relays for cooperation k k Else, no cooperation. Compute the possible throughput. Compare the throughput of all the pairs and select the maximal one. Update the power reward. k k j R k P k j 1 4 x 2 3 With a power reward: Nodes cannot continuously employ relays; Nodes will not continuously act as a relay. Jun. 4,

16 Fairness Indicator Tmin Fairness Indicator: Tmin min{ T1, T2,..., T } T max{,,..., } Equal lifetime: 1 max Tmax T1 T2 T Let d denote the fairness indicator of direct transmission f full cooperative protocol a fair cooperative protocol v 1 d f, 0v 1 a E[ M k ], as Jun. 4,

17 Performance Comparison I: Fairness Indicator Direct transmission Full Cooperation Fair Cooperation E[ M ]/ 0.5 k d f a E[ M k ]/ Fair Cooperation a Jun. 4,

18 Performance Comparison II: Lifetime Number of Nodes Direct Transmission Full Cooperative Protocol Fair FAPCooperative Protocol FAP-S FAP-R Time Slots Jun. 4,

19 Aggregate Throughput Theorem 3 [Dai 09]: The aggregate throughput of an energy- constrained cooperative ad-hoc network with opportunistic transmission is given by T T max max C log atdt ( ) log btdt ( ) T log 1 T T max 2 max max a(t): proportion of nodes competing for the channel b(t): proportion of nodes acting as relays in the relay region <<1 Full cooperation: Fair cooperation: C T log 1 T T f f max 2 max max C T log 1 T T a a max 2 max max C a because C a f f Jun. 4,

20 Performance Comparison III: Aggregate Throughput Direct Transmission Full Cooperation Protocol Fair Cooperation Protocol 30% gain Throughput (bit/s/hz) Time Slots Lin Dai, Wei Chen, Leonard J. Cimini, Jr. and haled B. Letaief, "Fairness Improves Throughput in Energy-Constrained Cooperative Ad-hoc Networks," IEEE Trans. Wireless Commun., vol. 8, no. 7, pp , July Jun. 4,

21 Outline Resource Allocation in Wireless Networks Tradeoff between Fairness and Throughput Fairness and Throughput in Energy-constrained Cooperative Networks Optimal Resource Allocation in Energy-constrained Cooperative Networks Jun. 4,

22 Optimal Resource Allocation Resources: energy & time slots Objective Maximizing the total throughput Fairness: equal lifetime & energy fairness How to allocate? Energy allocation Time-slots allocation Jun. 4,

23 Energy Fairness i E j denotes the energy that node j consumed in transmitting/relaying signals of node i. E E E E E E E E E the total energy allocated to node 1: the total energy consumed by node 1: e C j 1 E j 1 1 A 1 j 1 1 e E j Energy fairness: e e E A C i i total i 1,..., j i Ei E Contribute more, j j1 j1 gain more ji ji Jun. 4,

24 Example: Two-node Cooperation 2 E E 2 h 1 h 2 E E E E2 Energy fairness requires: E1 E2 E1 E2 2 1 E1 E E E E E Suppose h 1 > h 2. E E How to guarantee energy fairness? Jun. Apr. 4, 26,

25 To Cooperate or Not to Cooperate? Node 1: Node 2: State 1: Node 2 helps node 1; State 2: Node 1 transmits alone. Node 1 always helps node 2. cooperation no cooperation Node 1 Node 2 t Jun. Apr. 4, 26,

26 Multi-state Cooperation Define a cooperation Matrix A with e e e C 1 C 2 C a ij a a a a a a a a a E e j i A j e e e A 1 A 2 A C ei E j j 1 i j 1 ae ij A j A should be a doubly-stochastic matrix! Divide cooperation into multiple states; At cooperation state n=1,,n, choose a relay set such that A(n) is a doublystochastic matrix. N N C A e ( n) A( n) e ( n) E n 1 n 1 total1 N A e ( n) E n1 total1 Jun. 4,

27 Multi-state Cooperation 1 3 Destination 2 4 {1,2,3} {2,3} {3} Jun. 4,

28 Performance Comparison I: Aggregate Throughput Direct Transmission Full Cooperation Multi state Cooperation Throughput (bit/hz) Nodes 2 and 3 run out of energy Time (s) Jun. 4,

29 Performance Comparison II: Fairness Performance Compared to Direct Transmission Node 1 Node 2 Node 3 Node 4 Increase in lifetime Increase in throughput Full 104% -36% -26% 74% Multi-state 104% 104% 104% 104% Full 144% -38% -22% 90% Multi-state 87% 55% 67% 76% Effort fairness Outcome fairness Wei Chen, Lin Dai, haled B. Letaief and Zhigang Cao, "A Unified Cross-Layer Framework for Resource Allocation in Cooperative Networks, IEEE Trans. Wireless Commun., July (won the 2009 IEEE Marconi Prize Paper Award) Jun. 4,

30 Open Issues Distributed implementation Power reward + distributed access Generalization to multi-hop cooperative networks Optimal framework Routing and access protocol design Jun. 4,

31 Thank you! Any Questions? Jun. 4,