Robust Decentralized and Distributed Estimation of a Correlated Parameter Vector in MIMO-OFDM Wireless Sensor Networks

An optimal precoder design is conceived for the decentralized estimation of an unknown spatially as well as temporally correlated parameter vector in a multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) based wireless sensor network (WSN). Furthermore, exploiting the temporal correlation present in the parameter vector, a rate-distortion theory based framework is developed for the optimal quantization of the sensor observations so that the resultant distortion is minimized for a given bit-budget. Subsequently, optimal precoders are also developed that minimize the sum-MSE (SMSE) for the scenario of transmitting quantized observations. In order to reduce the computational complexity of the decentralized framework, distributed precoder design algorithms are also developed which design precoders using the consensus based alternating direction method of multipliers (ADMM), wherein each SN determines its precoders without any central coordination by the fusion center. Finally, new robust MIMO precoder designs are proposed for practical scenarios operating in the face of channel state information (CSI) uncertainty. Our simulation results demonstrate the improved performance of the proposed schemes and corroborate our analytical formulations.

Automated summary

This paper proposes an optimal precoder design for decentralized estimation in a MIMO-OFDM WSN, develops a rate-distortion theory based framework, and introduces distributed precoder design algorithms using the ADMM method. New robust MIMO precoder designs are also proposed for scenarios with CSI uncertainty, demonstrating improved performance through simulation results.