Accession Number : AD0712126

Title :   OPTIMUM TRACKING DETECTORS.

Descriptive Note : Technical rept.,

Corporate Author : DUKE UNIV DURHAM N C ADAPTIVE SIGNAL DETECTION LAB

Personal Author(s) : Hatsell,Charles P.

Report Date : MAR 1970

Pagination or Media Count : 196

Abstract : The purpose of the report is to further the theory of signal detectability for signals with time-varying parameters. Time-varying parameters are modelled by Markov statistics, and the resulting optimum detectors are termed 'tracking detectors.' Tracking-detector design is approached with a two stage philosophy. A first stage contains the bulk of the processing and can be designed without specifying a priori knowledge regarding the unknown parameters; the a priori knowledge is accounted for in the relatively simple second stage. Tracking-detector theory is applied to the parametric signal characteristics of amplitude, phase, and waveshape. By modelling a burst-like signal as a two state Markov process, it is found that the detectability of such signals is a function of the decision threshold setting. Detectability of periodic and transient signals in noise with time-varying power is investigated and it is found that, for at least one noise model, the basic non-linearity of the optimum detector is a limiter.

Descriptors :   (*TRACKING, OPTIMIZATION), (*UNDERWATER SOUND, DETECTION), NOISE, DECISION THEORY, STATISTICAL PROCESSES, PROBABILITY DENSITY FUNCTIONS, INTEGRAL TRANSFORMS, SIGNAL-TO-NOISE RATIO, TRANSIENTS, SOUND TRANSMISSION, THESES

Subject Categories : Acoustic Detection and Detectors

Distribution Statement : APPROVED FOR PUBLIC RELEASE