A unified algorithm for detection and tracking of unknown variable signals on a sequence of noisy spatial scans is synthesized on the basis of the adaptation theory for information systems. Signal expansion over a finite orthonormal basis whose dimension is estimated using the results of observation is used for parametric description of an unknown signal. Markov polynomial approximations are used for description of the evolution of the signal position and coefficients of the orthogonal expansion. Recurrent relations for the decisions about signal detection and a posteriori estimates of its position and parameters are obtained in the Gaussian approximation for a posteriori probability densities. A nonlinear problem of estimation and specification at each cycle of the compound parameter vector is solved using an iterative method.
展开▼
