Robust estimation of sinusoidal signal with colored noise using decentralized processing

Abstract

We develop a new technique for the estimation of the number of signals and their central frequencies by using the decentralized processing, when it is known a priori that the observations consist of a finite number of source signals corrupted by additive colored random noise process. Our decentralized processing scheme is that each sensor estimates the frequency and the covariance matrix of the observations, and sends the estimated results to the fusion center. For the parameter estimation at each sensor, we use the earlier work of [1]. At the fusion center, we utilize the Robust estimation technique to combine the estimates obtained from individual sensors. The overall performance does not deteriorate very quickly even if some of the sensors are destroyed.