Fast blind and semi-blind channel identification of mobile radio fading channels using a-priori information

Abstract

An iterative algorithm for fast blind and semi-blind channel identification (no training symbols necessary) based on the super-exponential-algorithm is shown. On the assumption of independent, identically distributed (IID) transmitted data, the algorithm has fast convergence properties. It is robust with respect to system overfit (supernumerarily assumed channel coefficients converge to zero) and the influence of modest additive white Gaussian noise even in mixed-phase moving average channels. The complete algorithm makes use of a-priori information, e.g., from an outer decoding stage (channel decoder) to improve the performance. It is also possible to exploit training symbols. Despite the use of fourth order cumulants, the complexity of the algorithm is rather low compared with alternative blind methods. According to the BER rates, after channel decoding the iterative blind scheme is as efficient as a training sequence based system.