Signal detection by adaptive filters

Communication engineers are now giving increased attention to detection systems which are able to adjust their own structure so as to be optimum for the particular detection problem of the moment. This paper describes a system which is capable of adapting and optimizing its response to the class of pulse signals whose individual pulses are less than seconds in duration. The analysis and synthesis of the adaptive system is facilitated by the use of an orthogonal function decomposition of the received signal. The use of the orthogonal decomposition permits synthesis of optimum linear filters by various circuit techniques, several of which have been reported elsewhere. The structure of the system utilizing such a decomposition is described in detail. Since the operation of the adaptive filter is based upon signal detection and estimation in noise backgrounds, considerable attention is devoted to the relationship between optimum signal detection and estimation. The methods of statistical decision theory are used. A program to test the validity of the approximations and assess the over-all system performance was carried out by simulation of the system on both analog and digital computers. The results of these experimental runs are described.