Convergence analysis of adaptive total least square based on deterministic discrete time

Deterministic continuous time (DCT) is a conventional method of studying the minor component analysis (MCA). Unfortunately, DCT is not used in practical systems because of it's strict conditions. Therefore, the convergence condition of AMEX MCA learning algorithm is derived based on deterministic discrete time. Theoretical analysis shows that the total least square solution is not obtained until special conditions between learning factor and autocorrelation matrix of input signal are satisfied. Finally, simulation results show the correctness of the convergence condition.