Adaptive acoustooptic filter.

The design of optimum systems, in the classical sense, requires a priori some knowledge of the signals to be encountered. As a result, such systems perform poorly when the appropriate characteristics of the input signals are not known a priori sufficiently well or are timevarying. An adaptive processor has the ability to selfoptimize by continually monitoring its performance and updating its parameters. Adaptive techniques have been applied to both spatial and temporal filtering domains. Specifically, adaptive techniques have been applied to antenna array processing by Appelbaum,' Widrow,2 and others.3'4 Applications to time domain problems include Lucky's work on data redundancy removal, 5 Sondhi's adaptive echo canceler, 6 Widrow's work on adaptive noise suppression, 7 Morgan and Craig's adaptive linear predictor,8 and more.9-1 ' In the work on time domain problems cited above the adaptive filtering scheme is based on the orthogonality principle.1 2 The basic idea behind the scheme is to control a variable filter so as to minimize the correlation between the input signal and the residual signal, which is the difference between the input signal and the filter output. A particular implementation of this scheme uses the transversal filter architecture, which consists of a tapped delay line, variable filter weights, and a summer that produces a weighted sum of delayed versions of the input signal as its output. The linearity and parallel nature of the transversal filter arrangement, commonly called the correlation cancellation loop (CCL) system, make the optical implementation possible. The advantages of optical processors in terms of bandwidth and the large effective number of taps make such an implementation attractive. For example, acoustooptic devices (AODs), which can serve as optically tapped delay lines, are superior in terms of bandwidth to CCD implemented delay lines that are currently used in analog adaptive filters.' 3 "1 4