动态大规模无线传感器网络决策融合

Decision Fusion Under a Dynamic Large-scale Wireless Sensor Network

为了解决动态大规模无线传感器网络的决策融合问题,该文提出了一种表决决策融合规则。假设无线传感器网络中信道的衰减服从瑞利分布,传感器的数量服从泊松分布。在不同的系统参数下,对融合规则的检测性能进行了分析和仿真。结果表明,只要传感器的平均数量充分多,此融合规则即使在较低的信噪比下,也能提供良好的系统检测性能。并且与已有的非参数融合规则相比,此融合规则可以提供更好的检测性能。同时,通过使用决策融合系统的偏移系数,给出了传感器输出的最优接收阈值的获取方法。

In order to solve the decision fusion issue under a dynamic large-scale Wireless Sensor Network （WSN）, a voting decision fusion rule is proposed. It is assumed that the fading of channels submits a Rayleigh distribution and the number of sensors follows a Poisson distribution in the WSN. The analysis and simulation on detection performance of the fusion rule are made for different system parameters. It is shown that this fusion rule can achieve a good system-level detection performance even at low Signal-to-Noise Ratio （SNR）, as long as the average number of sensors is sufficiently large, and this fusion rule can also provide better system-level detection performance than the existing nonparametric fusion rules. Moreover, a method to obtain the optimal receiving threshold of sensor output is given, by using the deflection coefficient of decision fusion system.