Minimum BER DFE equalizer in alpha stable noise

This paper addresses the problem of developing a least mean squares (LMS) style algorithm for minimizing bit error rate (BER) for updating linear/decision-feedback equalizers and multi-user detectors operating in impulsive noise environments characterized by an alpha stable distribution. The concepts build on earlier work in a Gaussian noise environment [1] and exploit some of the ideas developed in [2] for dealing with alpha stable noise as a mixture of a Gaussian and a Cauchy distribution. The development exploits the stable nature of the alpha distribution i.e. stability under linear transformation or filtering. An improvement in convergence and BER-performance is achieved by using minimum bit error rate (MBER) algorithm criterion over conventional LMS based design, as apparent from simulation results, thus making MBER criterion ideal for alpha stable noise environment.