Power allocation for Gaussian Mixture model prior knowledge in wirless sensor networks

Abstract

This paper presents power allocation in nonlinear sensor networks for Gaussian Mixture (GM) information source. The observations of sensors are transmitted through independent Rayleigh flat fading channels to a fusion centre (FC). Transmit Power is optimally allocated to sensor nodes so as to minimize the mean square error (MSE) of estimate at FC. Bayesian linear and optimal nonlinear estimators are deployed at FC to compare the proposed optimal and uniform power allocation among sensors. Extensive simulations validate that the proposed Bayesian linear estimator with optimized power gains effectively works for GM prior distribution.