Ocean Surveillance Detection Studies. Part 1. Detection in Gaussian Mixture Noise. Part 2. An Investigation of Canonical Correlation as an Automatic Detection and Beamforming Technique
Final rept. 1 Jan-30 Sep 1985
LEE (J S) ASSOCIATES INC ARLINGTON VA
Pagination or Media Count:
In Part I, non-coherent detection of narrowband signals in Gaussian- Gaussian mixture noise is considered. The forms of the optimum detectors are found and evaluated for single and multiple observations. These detectors are sensitive to information on the signal and noise parameters. Also evaluated are several suboptimum detectors which perform well under certain conditions without requiring knowledge of signal and noise parameters. In Part II, a new technique for detecting signals using arrays of sensors is investigated, based on the principles of canonical correlation. The theoretical results indicate that for a sufficient number of sensors, multiple, spatially separate signals, at the same frequency can be individually detected without knowledge of sensor position or using conventional beamforming. If sensor positions are known the algorithm supplies the direction of the signal arrivals. The limited numerical results show that a few sensors can automatically steer on and detect a single signal, but are incapable of resolving multiple signals successfully. Keywords Non- Gaussian noise Detection Probability Correlation Beamforming.
- Acoustic Detection and Detectors