Coherence and Noncoherence in Radar Signature Analysis

Using Bayes and maximum likelihood statistical estimation, optimum estimates are derived for the distribution of signal power among components of a radar signature, in the presence of additive noise and phase errors, with various constraints on the coherent processing time. Results are expected to provide the first theoretical and practical basis for the Bayes optimum detection of targets in clutter and signature analysis of composite targets. The target signals are represented as the sum of a set of possible components, each component having amplitude and phase modulations of known form but unknown absolute amplitude and initial r-f phase, and the quantities are estimated as functionals of the distribution of signal power. The processing may involve varying degrees of coherence and noncoherence, due either to phase deviations or to failure of the radar to preserve phase information. (See also RM-5134-PR.)