Precoder Design considering Decentralized Estimation over MIMO-WSN based on Stochastic Models

What is it about?

New precoder matrices are designed to estimate an unknown vector parameter considering stochastic channels for MIMO-WSN with decentralized sources. Two types of wireless sensor networks (WSNs) are considered, with and without total power network limitations. Within this context, two closed-form relations are derived for the precoder matrices and mean square error (MSE) of the unknown vector parameter is obtained. To cancel any operations at a fusion centre to estimate the aforementioned parameter, a distortionless constraint is added. In simulation section, we compare the MSE performances of the proposed method in stochastic multiple-input multiple-output (MIMO) channels with those in deterministic MIMO channels. The simulation result indicates that the MSE performance in the stochastic channels is decreased by almost 1.3 dB at low SNRs. Therefore, reliability of the unknown vector parameter estimation is enhanced using the MIMO model. Ultimately, effects of bandwidth and number of sensor nodes on the MSE criterion are also analysed.

Featured Image

Why is it important?

The main contribution of this paper is to investigate a model closer to reality. For the first time, precoder matrices are designed considering MIMO-WSN based on stochastic models, to minimize mean square error of the unknown vector parameter under total power and no power restrictions, without a requirement of any operation at the fusion center. Designing precoder matrices is a convex optimization problem with power and distortionless constraint conditions