Application of rational orthogonal wavelets to active sonar detection of high velocity targets

This paper is concerned with the problem of detecting moving targets using active sonar. We study a promising class of waveform, the Rational Orthogonal Wavelets (ROWs), designed to be tolerant of multipath Doppler spreading. Recent work in communication has shown that ROWs are effective even on severely spread channels. We apply ROWs for the first time to target detection. ROW pulses are compared to conventional CW and LFM pulses via simulations involving a high-speed target in shallow water. The target is point-like, and ray tracing is used to synthesize Doppler-shifted multipath arrivals in a range-independent environment. The key feature of ROWs that we exploit is the orthogonality they maintain over a wide range of delays and Doppler scales. Thus the maximum-likelihood detector is a simple average of matched filters, and the statistics of the detector under the null hypothesis are known, permitting CFAR operation. In addition, a fast algorithm resulting in rational scaled filter banks is available.