Signal source estimation in sensor networks

We consider the problem of estimating a signal source, modeled as random parameter(s), at a fusion center (FC) based on measurements collected from a sensor network with bandwidth constraint. Each sensor compresses its measurement and sends the information at a finite rate to the FC, whose task is to reconstruct the source. Utilizing the principle of coding with side information we propose a practically simple sequential encoding/decoding scheme that takes advantage of the correlation across sensors´ measurement to reduce total transmission rate. The proposed scheme does not require the knowledge of correlation model at the sensors, and it allows incremental improvement in source estimation quality, as the FC collects data from sensors in a pre-determined order. For the proposed scheme we study the relationship between the sequential order, the rate allocation among sensors, and the observation channel quality for each sensor, that lead to enhanced source estimation accuracy.