The theory for, and demonstration of, information theory applied to radar target identification

The desire for a radar system to be able to identify a target under interrogation has motivated a long standing interest in determining the best possible performance of such radar target identification systems. With the advent of Ultra-Wideband radar, radar target identification became a more feasible task. Experimental performance has usually been presented without relationships to the difficulty of the problem, and design has primarily taken a trial and error approach.;Proofs of the theoretical bases for using information theory for radar target identification are included. The continuous channel, inherent to the radar problem, is used in developing all proofs presented. An intuitive basis for using information theory is also presented from a signal detection theory background. Validity of the theory is demonstrated by using experimental radar target signatures of different bandwidths in Monte Carlo simulations. In addition, a method for computing mutual information is derived and performed on the experimental radar signatures to support the theory.;In this research tools are developed that can aid in the design of identification systems by applying information theory to the radar target identification problem. Specifically, by using the Bayes rate-distortion function, the minimum probability of error can be determined for the radar system in use and for the level of noise the system operates under. Performance changes, with such parameters as bandwidth, can be determined by using the Bayes rate-distortion function and calculations of mutual information for the various parameters.