A robust fusion rule using Piece-Wise Linear function in wireless sensor networks

The problem of decision fusion in wireless sensor networks is investigated in this paper. Based on the parallel fusion model under fading and the noise channels of the generalized Gaussian and the Cauchy models, we develop a robust fusion rule. By utilizing high and low signal-to-noise ratio (SNR) approximations, we obtain both high and low SNR alternatives respectively to the optimum likelihood ratio based fusion statistic for both noise models. To overcome the near-optimality of both alternatives for only limited ranges of SNR values, we propose the piece-wise linear fusion statistic (PWL-FS) that combines both high and low SNR results by using a piece-wise linear function. Performance evaluation is performed through Monte Carlo simulations.