Energy-efficient scheduling of distributed estimation with convolutional coding and rate-compatible punctured convolutional coding

Abstract

The problem of distributed estimation of an unknown noise-corrupted parameter in wireless sensor networks, with a fusion centre, is considered. Convolutional coding and rate-compatible punctured convolutional coding are used to protect the transmission of sensor observations and to reduce the impact of noise channels. Two novel kinds of power scheduling based on different encoding methods are derived for minimising the total power consumption. The formulas of the proposed power scheduling suggest that local sensors with poor observation qualities should decrease their quantisation levels. The levels should also be related to coding methods, channel qualities, a given mean squared error (MSE) and local signal-to-noise ratios. Finally, simulation results show that not only these two schemes are energy-efficient but also they can achieve comparable performance of a given MSE.