Adaptive implementation of the optimum radar signal processor

Abstract

In the present lecture, relevant examples of adaptive radar signal processing techniques are surveyed. An adaptive system performs the processing on the incoming signals by using an architecture having time-varying parameters. The corresponding filtering mask is taylored to the actual interference (i.e. clutter) which is real-time estimated by the same input signal. Additionally, the mask is able to track variations of interference power spectrum which may occur during the time. This approach overcomes the inherent limitations of conventional systems based on filters (e.g. MTI, FFT) having predetermined coefficients. These non adaptive techniques suffer poor interference cancellation when the expected environmental conditions significantly differ from the actual ones or when the experienced interferences vary unpredictably during the time. A brief recall of the theory of optimum signal processing is given. To this purpose, the most relevant concepts of the Wiener, Levinson, Kalman and Brennan Reed filtering are pointed out. All these theories provide useful insights and analytical means for the adaptive techniques considered. Five adaptive processors are examined, all sharing a common theoretical background provided by the optimum filtering theory, namely: Parametric Estimator (PE), Gram-Schmidt (GS) orthonormalization algorithm, Direct Matrix Inversion (DMI) technique, Maximum-Entropy-Method (MEM), Kalman Filter (KF). The working principle and the implementation algorithm of each above-mentioned processor will be de scribed and the performance achieved will be compared with respect to the optimum procedure. Also, the problems concerning with their hardware implementation will be mentioned, even though further researches are needed in this area. The revised adaptive techniques seem to be a valuable upgrading of the conventional fixed parameters processor.