认知无线网络频谱分配策略的社会最优问题研究

Research on the social optimization of spectrum allocation strategies for cognitive radio networks

结合离散时间排队理论和博弈论,进行了认知无线网络中频谱分配策略的社会最优问题的研究。考虑到主用户(PUs)对次级用户(SUs)数据包的传输中断,建立了一个带有传输中断及中断重传的排队模型,并使用矩阵几何解的方法进行排队模型的稳态分析,给出了次级用户数据包的平均延迟。从不可观察的排队规则出发,令次级用户数据包根据传输所获得的回报及逗留在系统中所花费的代价决定是否加入缓存,揭示次级用户数据包在个人最优策略下的加入率大于社会最优策略下的加入率。针对不同的数据包到达率及服务率,通过数值实验,给出了次级用户接入频谱的合理费用,实现频谱分配的社会最优。

The research was performed by combining the discrete time queue theory and the game theory to socially optimize the spectrum allocation strategies for cognitive radio networks.Considering primary users （PUs） may interrupt the transmission of secondary users （SUs） ＇ packets,a queueing model with a possible interrupted transmission and a retransmission policy was established.By using the method of a matrix geometry solution,the queueing model was analyzed in the steady state,and then the average delay of SUs packets was given accordingly.Based on an unobservable scenario,SUs were supposed to decide whether to join the cache according to the reward from transmission completion and the cost due to staying in the system.Moreover,it was investigated that the join probability of an SUs packet with an individually optimal strategy was greater than that with a socially optimal strategy.For different arrival rates and service rates of packets,a reasonable spectrum admission fee of SUs packets was proposed with numerical experiments,and then the spectrum allocation strategy was optimized socially.