Probability of alarm detection in a search and rescue system

Diagnostic systems are often based on the detection of weak periodic signals buried in noise. As an illustrative example, a maritime search and rescue system is described. Here, alarms consist of frequency-modulated signals where the carrier frequency may vary in a certain range due to Doppler shifts. The signals are strongly corrupted by additive noise and by fading. For a given averaging time, the probability of missed alarms is computed and minimised, comparing different receiver structures and detection schemes. The theoretical investigations were verified in practical tests with striking accuracy.