Multiple-Element Threshold Signal Detection of Underwater Acoustic Signals in Nongaussian Interference Environments

摘要

In addition to the general development of a weak-signal M-sensor detection theory, optimum (binary) space-time threshold signal detection algorithms are obtained for specific (Class A) nongaussian underwater acoustic noise environments, as well as for the fully canonical cases of general interference and general signal waveforms. These include algorithms for coherent, incoherent, and composite (coherent + incoherent) reception. It is shown that spatial and temporal processing are interchangeable as long as sampling (of the noise data) is statistically independent, in time and in space (i.e., sparse sampling ). It is also estimated that dense sampling - continuous or analog sampling - can give only O(2-3db) improvement over sparse-sampling in the highly nongaussian (Class A) cases, and O(0 db) in gauss noise, as long as large space-time-bandwidth products (J>>1) are employed. Comparisons in structure and performance with suboptimum detectors (matched-filter detectors and clipper-correlators) are provided, with an extensive set of numerical examples illustrating performance for typical noise and signal conditions. While clipper correlators give noticeable improvement O(20-30 db) over the conventional matched-filter receivers (optimum in gauss noise) in these threshold cases, they also can be significantly less effective O(6-10 db) than the optimum algorithm, even when composite detectors are employed. In all cases, increasing the number of independent spatial samples (array processing gain, M>1) over the single-element, or single-beam configurations (M=1), when possible, can give significant improvement in performance.