Coherent max-log decision fusion in wireless sensor networks

Abstract

In this letter, we revisit the problem of distributed coherent decision fusion for wireless sensor networks where the sensors make independent decisions on the hypothesis under test and forward their decisions over noisy fading channels to the fusion center. We propose a novel fusion rule which is obtained by applying the max-log approximation to the optimal fusion rule. Furthermore, we provide a unified performance analysis of binary decision fusion which is applicable to the optimal fusion rule and various suboptimal fusion rules such as the max-log, maximal ratio combining, equal gain combining, and Chair-Varshney fusion rules. Our results show that the proposed max-log fusion rule outperforms all previously proposed suboptimal fusion rules and has a close-to-optimal performance.