Relay Selection Study Based On SFD Successive Relaying

Cooperative relay techniques can increase throughput and extend coverage of cellular network.Cellular users located on the margin can make connection with base station via relay,and two D2 D users far away with each other can also make communication via relay.There are two main relaying protocol,full-duplex(FD)and half-duplex(HD).Traditional HD is used broadly for its simple applycation,but it has a inherent drawback which cannot take full advantage of slot and frequency spectrum.The protocol of space full-duplex(SFD)can make up this drawback.SFD has been studyed by many scholars,while they are always based on instantaneous benefit.For wireless communication environment is dynamic,buffer state and channel state change with execution of action,and have time correlation,so we need to think about take foreseeable action.This paper propose a relay selection method based on SFD,while considering buffer's finiteness,environment's dynamics and states' correlation.The main contributions are shown as follows:(1)For single user multi-relay system,the paper make a problem model about maximize system long-term average capcity,considering action and states' time correlation,and solve it under dynamic system environment.Firstly,we analyze system's state and action,formulate the capcity maximization problem using Markov-Decision-Process(MDP),then use history data to estimate state transition probability,and make action policy through learning.(2)Extend SFD to multi-user multi-relay system,the paper make a problem model about maximize system total long-term average capcity subject to user fairness,considering action and states' time correlation,and solve it under dynamic system environment.Firstly,we analyze system's state and action,formulate the capcity maximization problem using Centralized-Markov-Decision-Process(CMDP)considering coupling about buffer employment for users.Then use TMDA and auction theory to solve it.Through simulation,we can find that both single user and multi-user system can obtain better capacity performance based on foreseeable action than short-sighted action.While in multi-user system fairness and profit has a Ramsay rule.