Optimal Minimum Variance Distortionless Precoding (MVDP) for Decentralized Estimation in MIMO Wireless Sensor Networks

In this letter, we present a framework for optimal minimum variance distortionless precoder (MVDP) design towards decentralized estimation of a vector parameter in a coherent multiple access channel based multiple-input multiple-output (MIMO) wireless sensor network. The proposed MVDP scheme yields the optimal minimum variance distortionless parameter estimate at the fusion center without the necessity of any receive processing. A closed form expression is derived for the mean square estimation error of the MVDP and it is demonstrated that it asymptotically achieves the centralized minimum mean square error bound. Further, we derive the optimal decentralized estimation schemes with a total network power constraint (MVDP-T) and per-sensor power constraint (MVDP-P). Simulation results demonstrate the performance of the proposed optimal precoding schemes and also support the analytical results derived.