Characterization of optimum polarization for multiple target discrimination using genetic algorithms

In radar scattering problems in which target discrimination cannot be accomplished on the basis of a difference in range, the polarization spectrum (signature) may be the appropriate tool to differentiate between different targets or targets and clutter. This idea stems from the fact that the polarization for which the radar receives the maximum or minimum scattered power (optimum polarization) from a target may be different from the optimum polarisations of the other targets or its surrounding clutter. In this paper we confine our interest to the problem of characterizing polarization combination of the receive and transmit antennas of a radar system for target classification. The motivation for considering the problem at hand stems from the fact that in many cases of practical importance, target classification when there are more than two possible targets is of interest. The complex optimization problem in finding the receive and transmit polarizations of a non-polarimetric radar is addressed using a stochastic optimization method. The application of genetic algorithms for characterization of optimum antenna polarizations is considered to achieve multiple target classification with least error probability. The objective function is defined so that the solution directly specifies the threshold levels for target classification. In this procedure it is assumed that the polarimetric responses of the targets are known.