Adaptive channel equalization using generalized order statistic filters

The Ll-filter is applied to the problem of channel equalization. The analysis includes simulations with various additive noise densities ranging from the purely Gaussian case to an impulsive case. The Kronecker product form of the Ll-filter requires only linear computational complexity at each time step (the same as the FIR) and is therefore a very attractive alternative to the FIR. This is because the exact least square (LS) formulation demonstrates that especially in the impulsive environment the Kronecker can do better than the FIR. The adaptive Kronecker Ll-filter is compared to the adaptive FIR, for typical cases and demonstrates good potential for field application of these filters in adaptive equalization. Finally, adaptation stepsize parameter selection implementations are discussed