Measurement of false alarm rate in a synthetic radar environment

Abstract

The false alarm rate is one of the fundamental parameters associated with radar performance specification. The radar designer's task is to set the radar detection threshold as low as possible in order to maximise the probability of detecting targets, subject to the constraint that the rate of generation of false alarms is kept below a specified maximum value. Accurate measurement of false alarm rate, however, may be difficult since, almost by definition, false alarms are rare events and therefore a large number of tests are required to get a reliable estimate. The paper investigates the use of importance sampling to significantly reduce the number of tests needed to achieve the desired estimation accuracy. The theory underlying importance sampling is presented and the processing gains that can be achieved are given, including those for the K distribution which is commonly used to model sea clutter. The use of importance sampling in radar performance assessment is also discussed.