Semicoherent Detection Against a Spoofing Opponent,

Many problems arise in radar and sonar applications which can be regarded as problems in detection theory or parameter estimation. Such problems have aspects of game theory associated with them when an intelligent opponent is assumed. This study concerns itself with the detection problem of radar and sonar where one is trying to distinguish between white noise and a short pulse signal imbedded in the white noise. The Doppler frequency shift of the signal is assumed unknown. It is required to distinguish between the desired pulsed signal and an unwanted one when an intelligent opponent is present who can control the Doppler. The unwanted signal return may be encountered, for example, from the combined return from birds or insects in the case of radar or fish in the case of sonar. Often an estimate of Doppler can be used to distinguish targets of this type from a target one wishes to track. If the target one wishes to track is controlled by an intelligent opponent who can choose Doppler in some randomized or nonrandomized manner, e.g., by controlling the aspect angle with respect to the observer, then the development of best detection criteria requires the application of game theory. Against an intelligent opponent, only randomized decisions may be optimum. This is not the case in the normal radar or sonar detection theory problem. Author