Signal to noise ratio improvement by likelihood ratio techniques under model and environmental mismatch

The performance of signal processing methods under model and environmental mismatch are discussed. The problem of signal detection in a random multipath channel is addressed. The performance of four processing methods derived under low signal-to-noise ratio (SNR) conditions are analyzed and compared. Then, the multipath localization problem is considered for the case where there is a lack of precise knowledge of the propagation environment. Multipath localization based on a combination of a matched field type processing technique and recent superresolution techniques in adaptive array processing is reported. Simulations for localization are performed based on data obtained from the generic sonar model (GSM).<>