Distributed Estimation over Unknown Fading Channels

In this paper, we consider the data detection problem involved in distributed estimation over unknown non-orthogonal fading channel. In general, when studying the distributed estimation problem, the impairments induced by communication channels are restricted to additive noise, quantization or packet loss. In addition, communication protocols are often of TDMA or FDMA type. Herein, by modulating the local data with doubly spread waveforms, we show that although each node receives a mixture of data transmitted by its neighbors, these data exhibit a trilinear structure, which can be used for separating the neighbors contributions. We state identifiability conditions and study the embedding of the data detection steps in a distributed estimation problem.