New batch type methods and their recursive extensions are introduced for blind equalization of digital communication channels. Based on the underlying cost function of the existing Bussgang and Cumulant Fitting algorithms, Simulated Annealing (SA) optimization is successfully employed to identify and equalize the communication channels.;The existence of local equilibria in the chosen cost functions for blind equalization is demonstrated with particular channel examples. Extensive Monte-Carlo simulations with one and two-dimensional signal constellations, different channel examples, and additive white Gaussian noise demonstrate the ability of the proposed methods to avoid the local equilibria and to identify the correct channel characteristics.;Many of the existing blind equalization algorithms utilize a stochastic gradient approach to minimize nonlinear cost functions. Due to the multimodal nature of these functions, the gradient algorithms might converge to the wrong solution. On the contrary, the proposed methods are based on a global optimization algorithm. The SA algorithm tends to avoid the local minima encountered. The ill-convergence problem is overcome, however, at the expense of higher computational complexity.
展开▼