认知无线电网络基于F范数的频谱共享

摘要

针对多用户多输入多输出认知无线电网络的频谱共享问题,提出一种在保证授权用户服务质量要求的前提下,以认知网络容量最大化为目标的基于F范数的频谱共享方法.该方法利用信道矩阵的F范数选择认知用户以获得认知网络的多用户分集增益,并采用两次选择的方式降低算法的复杂度,通过将认知用户的发射信号投射到干扰信道的零空间来避免认知用户对授权用户的干扰,对认知用户的信道矩阵采用奇异值分解方法使其转化为并行独立信道,并使用注水功率分配方法增大系统容量.相对于已有的认知无线电网络频谱共享策略,该方法能够使认知网络获得更大的各态历经容量,降低授权用户的中断概率.仿真结果表明,当认知系统总发射功率为100w时,采用本文方法与文献中已有算法相比,认知网络具有约13%的各态历经容馈增益.%Aiming at the problem of spectrum sharing in multiuser multi-input multi-output cognitive radio networks,a spectrun sharing method based on Frobenius norm is proposed, which is designed to maximize the throughput of cognitive radio (CR) users as well as guarantee the quality of service of primary (PR) users. Multiuser diversity of the cognitive radio network is obtained by user selection, which is according to the Frobenius norm of channel matrices, while twice selection is adopted to reduce the complexity of our method,and then the interference from CR users to PR users is canceled by null-space projection. At last,the singular value decomposition method is employed to derive the parallel spatial sub-channels for each user while the waterfilling power allocation method is used to enhance system capacity. Compared with the existing spectrum sharing strategies, the proposed method can acquire larger ergordic capacity of CR system and reduce the outage probability of PR users. Simulation results show that when the total transmitting power of the CR system is 100w,the proposed method has 13％ ergordic capacity gain of the cognitive network over the existing methods.