Simultaneous signal detection and classification under a false alarm constraint

An optimal technique for performing simultaneous signal detection and signal classification when the probability of false alarm (P_FA) is constrained to be below a prespecified threshold is presented. A constrained max-min strategy that has the property of ensuring the best minimum level of performance among detection/classification procedures which satisfy the constraint is derived. The procedure compares a set of weighted likelihood ratios to a threshold, and if the threshold is exceeded it chooses the signal class with highest weighted likelihood. The technique is applied to the problem of detection and classification of a change in distribution of an independent sequence