Distributed Estimation in Heterogeneous Sensor Networks Using Principal Component Analysis

This paper focuses on estimating a common unknown random vector in heterogeneous sensor networks (HSNs). We assume that both observation models and sensor operations are linear. The fusion center (FC) uses the received observations to find the minimum mean square error (MMSE) estimate of the signal. Two cases are considered: Raw data-based fusion and principal component analysis (PCA)-based fusion. We derive the mean square error (MSE) for both cases. It is shown by numerical and simulation results that PCA-based fusion can reduce the transmission requirement while satisfying the system performance requirement.