Optimization of Multiuser MIMO Cooperative Spectrum Sensing in Cognitive Radio Networks
Optimization of Multiuser MIMO Cooperative Spectrum Sensing in Cognitive Radio Networks
复制标题
DOI：
10.1007/s12559-014-9297-5
复制
发表时间：
2015-06
影响因子：
5.4
通讯作者：
Y. Hei;Wentao Li;Min Li;Zhuo Qiu;Weihong Fu
中科院分区：
文献类型：
--
作者：
Y. Hei;Wentao Li;Min Li;Zhuo Qiu;Weihong Fu
研究了多用户多输入多输出（MIMO）线性协同频谱感知优化问题，其中主用户（PU）和认知无线电（CR）配置了多个天线。协同频谱感知优化通过对接收信号赋予的不同权值进行优化，在给定目标虚警概率的情况下，实现检测概率最大化。确定了MIMO协同频谱传感系统中单天线PU和多天线CR、多天线PU和单天线CR、多天线PU和多天线CR的参数统计特性。针对优化问题的非凸特性，提出了一种基于遗传算法（GA）代替凸方法来寻找最优权向量的方法，该方法不受解域约束和凸性约束。此外，还提供了几种经典的遗传算法交叉算子，研究了它们对传感性能的影响。仿真结果表明，多天线协同频谱感知系统可以显著提高频谱感知的可靠性。此外，遗传算法是解决协同频谱感知问题的一种很有前途的方法。
This paper investigates the multiuser multiple-input–multiple-output (MIMO) linear cooperative spectrum sensing optimization problem, in which the primary user (PU) and the cognitive radio (CR) are equipped with multiple antennas. By optimizing the different weights assigned on the received signals of CRs, the cooperative spectrum sensing optimization aims at maximizing the probability of detection given a targeted probability of false alarm. Statistical characteristics of parameters in MIMO cooperative spectrum sensing systems have been determined for the PU with a single antenna and the CR with multiple antennas, the PU with multiple antennas and the CR with a single antenna, and both the PU and the CR with multiple antennas. Because of the non-convex characteristic of the optimization problem, an alternative approach based on a genetic algorithm (GA) instead of convex approaches is proposed to find the optimal weight vectors without solution domain restrictions and convexity constraints. Furthermore, several classical GA crossover operators have been provided to investigate their effect on sensing performance. The simulation results show that, the reliability of spectrum sensing in cooperative spectrum sensing system can be significantly improved with multiple antennas. Furthermore, the GA method is a promising approach in addressing the cooperative spectrum sensing problem.