Threshold Detection in Non-Gaussian Interference Environments: Exposition and Interpretation of New Results for EMC Applications

The critical problem of developing optimum weak-signal detection algorithms for general classes of man-made and natural interference is concisely examined. This paper includes a summary of new results for both canonical and specific types of desired signals and electromagnetic interference (EMI) environments. A suitable bias term, missing in most earlier analyses, is essential to preserve the asymptotically optimum (AO) character of these algorithms, as sample size (or observation period ~n) necessarily becomes large, to enable the detection of weak signals with acceptably small probabilities of error. Canonical optimum threshold algorithms are obtained for three principal modes of detection: 1) coherent, 2) incoherent, and 3) composite. The latter 3) is AO, when it consists of the sum of (optimum) algorithms for 1) and 2); it can give noticeable improvement over either 1) or 2) alone. Various (usually) suboptimum algorithms, e.g., simple correlators and clipper correlators, are also considered. Performance measures (n > > 1), e.g., error probabilities, etc., and asymptotic relative efficiencies (ARE's), are noted, along with the useful concepts of processing gain and minimum detectable signal. Interpretations of the algorithmic structure and the implications of these results, generally, for EMC, are concisely discussed.