Detection performance of two efficient source tracking algorithms for matched-field processing

Detection performance is evaluated for two efficient algorithms that track sources moving linearly at constant speed and depth. One of the algorithms uniformly weights the contributions from each ambiguity surface along the tracks examined, and the other weights the contributions in proportion to the expected signal level. It is shown that these essentially “detect after track” algorithms are applicable to the situation where the signal-to-noise ratio is low and the apparent position of a moving acoustic source on a single matched-field ambiguity surface is highly ambiguous. They may be employed to reduce the ambiguity of the source position and to provide an improved detection probability at a given probability of false alarm. Algorithms that examine all possible tracks in multidimensional tracking are computationally prohibitive. However, the linear tracking algorithms described in this paper are based on only those tracks through the strongest peaks in the ambiguity surfaces. If the source track is likely to be detected after tracking, it is shown that the loss in detection performance after tracking, due to the possibility of not examining a track, is small since the probability of examining the source track is high. At high signal-to-noise ratios the performance loss vanishes.