Blind deconvolution based on cumulant fitting and simulated annealing optimization
Jacek Ilow, Dimitrios Hatzinakos, A.N. Venetsanopoulos · 2003
Higher order cumulant analysis is applied to the blind identification/equalization of linear moving average (MA) channels with QAM data. To identify the MA parameters of channels, a higher order cumulant fitting approach is adopted. The set of channel parameters is obtained by simulated annealing optimization, which overcomes the problem of a multimodal cost function in channel identification. The coefficients of a linear equalizer are obtained using the channel inversion algorithm. Batch type as well as batch recursive algorithms are investigated. The feasibility and convergence behavior of the proposed algorithms are examined by means of computer simulations.>