Sequential binary rank detection

The paper discusses a multichannel sequential signal-detector that operates on the basis of binary quantization of observation rank. Attention is given to various characteristics of the detector, including the dependence of the mean number of observations on quantization threshold and on signal-to-noise ratio for both fluctuating and nonfluctuating signals. The use of a binary rank detector simplifies calculations of the likelihood ratio, though it leads to a loss in useful information. Adaptive procedures can be applied to the detector with relative ease under conditions of a priori ambiguity.