A new approach to blind channel identification for mobile radio fading channels

A new algorithm for blind channel identification based on the super-exponential-algorithm (Shalvi and Weinstein, 1993) is introduced. Combining the impulse response of the equalizer with the channel impulse response leads to an iterative algorithm with fast convergence qualities and short block length on the assumption of independent, identically distributed (i.i.d.) data. The algorithm is robust with respect to system overfit (supernumerarily assumed channel coefficients converge to zero) and influence of additive white Gaussian noise even in mixed-phase moving average (MA) channels. Despite of the use of fourth order cumulants the complexity of the algorithm is rather simple compared with alternative blind methods. The implementation on a signal processor (TMS320C40) performs nearly in real-time under GSM conditions.