S.A. Klimov, V.M. Kosyakov, M.A. Sviridov

The algorithms of super Rayleigh resolution based on methods of digital spectral estimation, multichannel analysis, and also various versions of statistical decision method require high computational cost that very rigid demands are made on to program and temporary resources for their realization. The purpose of this article is to develop economic, from the point of view of multiplication-addition operators, quadratic objective evaluation procedure which is often used in digital signal processing according to super Rayleigh resolution methods, and also to estimate its computational efficiency for the number of multiplication-addition operators depending on discreteness of objective function evaluation, number of resolved signals and number of signal parameters. The way to reduce the computational cost in quadratic objective evaluating is considered. The way is based on properties of a quadratic matrix; it allows us to carry out the matrix factorization according to Kholetsky method. Estimates of computing efficiency of the standard and suggested algorithms are presented. From the results of analysis we conclude that computational complexity of considered algorithms depends significantly, first of all, on number of resolved signal parameters. The following deciding factor is the number of resolved signals and after all discreteness of objective function evaluation. The relative gain in efficiency can reach more than 100 %.