TABLE OF CONTENTS Sequential detection is based on the theories of sequential analysis developed and published by A. Wald [1] in 1947. The basic problem is the decision between two hypotheses, in the radar case H 0 for no target present and H 1 for target present . This has been discussed already in chapter 3 section 3.3. The decision has to be based on a series of received signals, which always contain noise and perhaps a target signal. The basic idea and characteristic feature of sequential analysis is that the number of necessary signals or radar pulses is not determined in advance. Instead, after each new signal a decision is derived to terminate the test with a decision or to continue with a further pulse. The advantage of sequential tests is that on average a substantially smaller number of pulses is needed compared to equally reliable tests with a predetermined fixed number of pulses. Equally reliable means the same pair of error probabilities ?, ? for false alarms and target missing, respectively.
Because of the variable test length of the individual tests only a phased-array radar with its beam agility is able to apply sequential detection. Because of its potential advantages there was early interest, and the application of sequential detection for radar was studied by several authors [2 7]. First, considerable gain in mean test length or signal-to-noise ratio (SNR) of about 8 dB was discovered, but this was valid for only one range element. In this case the gain...
