Estimation of False Alarm Probabilities in Cell Averaging Constant False Alarm Rate Detectors via Monte Carlo Methods

摘要

Monte Carlo Methods are introduced and used to estimate false alarm probabilities. The estimation of the latter is important in the context of performance analysis of Constant False Alarm Rate (CFAR) radar detection processes. A CFAR detector estimates the clutter level, producing a threshold, and a target is declared present if the statistic representing the test observation exceeds this threshold. The latter is adjusted adaptively, so that the rate of false alarms is held constant. Hence, in a radar analysis context, the performance of a CFAR process can be determined from whether it maintains a constant false alarm rate. In order to compare the performance of a number of different CFAR schemes, in a common clutter environment, we need to estimate these false alarm probabilities. This can be done quite easily using a basic %Monte Carlo estimator. However, the latter may require a very large number of iterations in order to produce a reasonable estimate. To reduce this number of iterations, importance sampling techniques can be used. To illustrate these techniques, we consider the simple case of cell averaging CFAR in a Gaussian environment, with square law detection. This enables comparison of estimators with an exact result.