Iterative algorithms using the difference of adjustmentvector estimates
Abstract
Stability conditions are analyzed and rates of convergence of the adjusted vector to the optimal one are estimated for stochastic iterative algorithms using the weighted difference of preceding estimates of the adjustment vector. Consideration is given to the class of stationary stochastic algorithms for the adaptive adjustment of noisecompensation systems according to the criterion of maximum SNR. It is shown that, for this class, it is not possible to find procedures having a higher rate of convergence than that of procedures using only estimates of the objectivefunction gradient. The validity of these results is confirmed by a mathematical model for a 10element linear antenna array.
 Publication:

Radiotekhnika i Elektronika
 Pub Date:
 November 1983
 Bibcode:
 1983RaEl...28.2138A
 Keywords:

 Adaptive Filters;
 Algorithms;
 Iteration;
 Radar Filters;
 Signal To Noise Ratios;
 Vector Analysis;
 Antenna Arrays;
 Convergence;
 Digital Filters;
 Mathematical Models;
 Numerical Stability;
 Rates (Per Time);
 Stochastic Processes;
 Communications and Radar