A Generalized Linear Filter Approach for Sonar Receivers

Abstract

Detection and resolution are the two fundamental tasks to be performed by a sonar system deployed in a multi-target environment. A matched filter is the optimum detector for a known signal in white noise and an inverse filter achieves the optimal resolution when the received signal-to-noise ratio (SNR) is very high. The main-lobe width of the autocorrelation function (ACF) of the transmitted signal and the received SNR limit the resolution and detection of the matched and inverse filters, respectively. Therefore, some targets cannot be resolved by the matched filter or detected by the inverse filter. One solution to this problem is to create a generalized receiver blending the properties of the matched and inverse filters so that a sonar receiver can vary its properties smoothly to trade detection for resolution. The outputs and the receiver operating characteristics of the proposed generalized linear filter are compared with those of matched and inverse filters to demonstrate the usefulness of the proposed filter.