Detection of a random signal in a multi-channel environment: a performance study
Performance of the optimal (likelihood ratio) test and suboptimal tests, including the normalized cross correlator and two energy detectors are compared for problems involving non-gaussian as well as gaussian statistics. Also, optimal one-channel processing is compared to optimal two-channel processing for equivalent total signal-to-noise ratios. Receiver operator characteristics (ROC) curves obtained by a combination of simulation and analytic methods are used to evaluate the performance of the processors. It was found that two-channel processing helps the detection performance the most when the noise levels are uncertain. This was true for all signal and noise densities studied. In cases where the noise levels and channel attenuations are known, or when only the attenuations are uncertain, the performance using optimal one-channel processing was close to that found using optimal two-channel processing. When comparing optimal processors to the three suboptimal processors, it was found that when the noise level in each channel is very uncertain, the performance of the normalized cross correlator is much closer to the optimal than that of either of the energy detectors. If, however, the noise levels are know with a fair degree of certainty, the performance of the energy detectors improves considerably, in some cases approaching the optimal performance.
- Research Organization:
- Duke Univ., Durham, NC (USA)
- OSTI ID:
- 7001295
- Country of Publication:
- United States
- Language:
- English
Similar Records
Multibaseline gravitational wave radiometry
Correlation between model observer and human observer performance in CT imaging when lesion location is uncertain
Related Subjects
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND TECHNOLOGY
ELECTRONIC EQUIPMENT
EQUIPMENT
GAUSSIAN PROCESSES
MEASURING INSTRUMENTS
MULTI-CHANNEL ANALYZERS
OPTIMIZATION
PERFORMANCE
PULSE ANALYZERS
RADIATION DETECTORS
SIGNAL-TO-NOISE RATIO
SIGNALS