DocumentCode :
2051616
Title :
A comparison of nonparametric statistics for multiple-channel detection of signals in ocean noise
Author :
Cochran, Douglas ; Sinno, Dana ; Gish, Herbert
Author_Institution :
Dept. of Electr. Eng., Arizona State Univ., Tempe, AZ, USA
fYear :
1991
fDate :
14-17 Apr 1991
Firstpage :
3685
Abstract :
The performance of nonparametric multiple-channel detectors based, respectively on the multiple coherence estimate and the generalized coherence estimate in detecting broadband signals in simulated deep-ocean noise are evaluated and compared. Probabilities of detection estimated from simulations are plotted against probabilities of false alarm predicted theoretically from the distribution functions of the detection statistics under the assumption that the channels contain independent white noise. The actual false-alarm performance for both detection statistics in the presence of simulated ocean noise is seen to be significantly different from what is predicted theoretically for white noise. The common practice of establishing detection thresholds for coherence detectors by using theoretically predicted false alarm probabilities is thus shown to be of limited value when colored noise is involved
Keywords :
noise; signal detection; statistical analysis; broadband signal detection; colored noise; detection statistics; false alarm probabilities; false-alarm performance; independent white noise; multiple coherence estimate; multiple-channel detection; multiple-channel detectors; nonparametric statistics; simulated deep-ocean noise; white noise; Colored noise; Detectors; Distribution functions; Oceans; Predictive models; Probability; Signal detection; Statistical distributions; Statistics; White noise;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1991. ICASSP-91., 1991 International Conference on
Conference_Location :
Toronto, Ont.
ISSN :
1520-6149
Print_ISBN :
0-7803-0003-3
Type :
conf
DOI :
10.1109/ICASSP.1991.151076
Filename :
151076
Link To Document :
بازگشت