Majority logic fusion of censored decisions in wireless sensor networks with Rayleigh fading

In this paper, we consider a binary decision fusion problem for a wireless sensor network, with flat Rayleigh fading, where the sensors make local binary decisions on the status of a common binary phenomenon. A training-based channel estimator is used at the fusion center (FC) to estimate the complex Gausssian fading coefficients characterizing the (direct) links between sensors and FC. This (estimated) channel state information is used, by proper thresholding, to censor the sensors. On the other hand, a locally optimal decision threshold is considered for binary quantization at the sensors. The average detection error probability at the FC, under the use of a majority logic fusion rule, is selected as a qualitative measure of system performance and is evaluated by simulation. The impact of the choice of the censoring threshold on the average detection error probability, under several values of channel estimation error, channel SNR, and sensor SNR, is studied.