Sonar signal design and evaluation with emphasis on diver detection
Sonar-based underwater surveillance, including the problem of diver detection, is a challenging task. In harbors and coastal areas the environment is often reverberation dominated, due to the numerous backscattering objects and boundaries like ship wrecks, harbor walls, seabed, or the water surface. Reflections from the target and the background are often very similar, except for the fact that the target is typically moving and the background is not. The object movement causes a Doppler e_ect that can be used to improve the separation of moving objects from the quasi-stationary background. Therefore, the ideal active sonar transmit signal would simultaneously provide very good range and Doppler resolution. In this work, existing sonar signal designs are thoroughly analyzed and special emphasis is set to understand the sources of their advantages and disadvantages. Among all the investigated waveforms, frequency modulation (FM) signals have the best properties, but they lack Doppler selectivity that is required to detect small moving targets in reverberation limited environments. This motivates the development of a new design - called cutFM signal. The goal is to create a Doppler selective waveform based on a linear frequency modulated signal. The basic concept is to cut out frequency components from the base signal, in order to obtain a comb like spectrum. The effect of cutting is analyzed in detail and it is shown that the cutting period has to be carefully selected in order to achieve the desired result - a Doppler selective signal. The cutFM signal is compared theoretically and via simulations with corresponding known alternatives. It is characterized by a very good Doppler processing gain and excellent performance in reverberation limited channels. In addition, compared to the known continuous wave (CW) based signals that have equivalent Doppler processing gains, the cutFM signal provides improved range resolution.