Sample rejection for efficient simulation of intersymbol interference channels with MLSD

A sample rejection scheme introduced previously is generalized for the simulation of multidimensional communication systems. We consider the use of sample rejection (a special case of importance sampling) for efficient simulation of uncoded continuous transmission with periodic trellis-termination over static intersymbol interference (ISI) channels and maximum likelihood sequence detection. Previously proposed rejection regions are applicable only for finite lattices with rectangular or circular symmetries and moderate dimensionality. However, these regions are not applicable or are inefficient if the dimensionality is increased for large block-lengths, or if the lattice symmetries are absent because of the ISI. Hence, we investigate sliding-window (near) maximum-likelihood sequence decoding (MLSD) to resolve the dimensionality problem. In particular, we study the truncated Viterbi algorithm and feedback decoding. We propose several modifications to these two algorithms using sample rejection principles to improve the simulation efficacy for the conventional Viterbi algorithm while achieving near MLSD performance. Finally, numerical examples confirm that feedback decoding and its modifications can be more efficient for simulations of ISI channels and near MLSD than the Viterbi algorithm.