无线传感器网络中极大极小设置的分布式检测研究

在分布式检测系统中,无线传感信道普遍存在未知的噪声,而基于最大似然函数的融合在有未知的噪声时性能表现较差.为了提高信号检测的准确性,提出一种基于极大极小方法设置的鲁棒融合规则.该规则采用最大似然函数的融合算法的结构形式和Huber极小极大的方法,得到虚警概率和检测概率表达式,适用于信道噪声为非衰减噪声分布、有界方差的噪声分布和混合高斯噪声分布.对多传感器并行分布式检测系统的仿真与分析,表明了该融合算法可提高信号检测的准确性,同时也具有一定的鲁棒性.

Analysis of distributed detection based on maximin setting in Wireless Sensor Networks

In distributed detection system,it is quite common that unknown noise exist in wireless channel layer,and the fusion rule based on maximum likelihood function with unknown distribution noise perform badly.In order to improve the accuracy of signal detection,this paper presents a robust fusion rule based on minimax setting method.The method adopt structural form of maximum likelihood function and Huber′s minimax approach.Finally,the explicit formulas for the detection and false alarm probabilities are derived.The rule is applied to the noise classes of nondegenerate,with a bounded variance and contaminated Gaussian distributions.Numerical simulation results show that the signal detection probability is improved. Meanwhile,the fusion rule has robustness.