Communication efficient signal detection in correlated clutter for wireless sensor networks

We study a problem of detecting deterministic signals buried in correlated clutter using wireless sensor networks. We are specifically interested in developing a distributed algorithm over the network to detect the presence of a deterministic signal while keeping low communication delay and energy associated with the distributed computation. In this paper, we deploy a distributed version of the Sparse Matrix Transform (SMT) that decorrelates a signal measured by a number of sensors in order to compute a matched filter. The matched filter represents the sum of the Log-Likelihood Ratios over all the sensors of the two hypotheses corresponding to whether a deterministic signal is present or not. We show through numerical simulations that our algorithm is very efficient in terms of communication energy and delay while sustaining a high Signal-to-Clutter Ratio.