A 'MAXIMUM ENTROPY'-BASED NOVEL NUMERICAL METHODOLOGY FOR PROBLEMS IN STATISTICAL ELECTROMAGNETICS

This paper presents the development of a novel 'maximum entropy'-based numerical methodology for the solution of electromagnetic problems, where the inputs and system parameters vary statistically. The application of this methodology to the problem of a plane wave impinging on an array of cylindrical conducting rods with stochastic variations in its parameters is then presented. To address this problem, a statistically signiflcant number of replicas of this array of conductors are constructed. The current proflles in these coupled conductors are estimated by using the Method of Moments (MoM). Upon estimation of the current proflles on the conductors, the monostatic radar cross- section is estimated for each replica of the array. The probability density function is then constructed through the estimation of a flnite number of moments from the available output data subject to the constraint of maximum entropy. The methodology is very general in its scope and its application to scatterers with other geometries such as spheres, spheroids and ellipsoids as well as to other application areas would form the basis of our future work.