MEDIAN DETECTOR FOR NONCOHERENT DISTRIBUTIONS.
NAVAL RESEARCH LAB WASHINGTON D C
Pagination or Media Count:
Sea-clutter measurements using high-resolution radar indicate that clutter cross-section returns do not follow a Rayleigh distribution. The log-normal and contaminated-normal distributions have been suggested, and detection curves for these distributions have been published. In this report, the median detector is proposed as a detector against these clutter distributions, and the theory is developed which is necessary for detecting constant signals in log-normal and contaminated-normal clutter when one has N independent samples. Curves are obtained for the threshold values and probabilities of detection for false-alarm probabilities ranging from 10 to the -2 power to 10 to the -9 power. If one compares the mean and median detectors, the conclusion is reached that the median is better for the log-normal distribution and that the mean is slightly better for the contaminated-normal distribution for small sample sizes. Using the concept of asymptotic relative efficiency, the median is shown to be better than the mean for large sample sizes for both distributions. Author
- Statistics and Probability
- Active and Passive Radar Detection and Equipment