Abstract: Teachings presented herein combine the relative simplicity of equalization with the performance of maximum likelihood (ML) processing. These teachings are applied to the detection of symbols in a stream of symbol blocks. In one or more embodiments, block-based equalization, including feedforward filtering, suppresses inter-block interference and produces detection statistics for the symbols in each symbol block, and joint detection addresses intra-block interference by jointly detecting the most likely combination of symbols within each symbol block, based on the corresponding detection statistics. The joint detection obviates the need to address intra-block interference within the equalization filters, while, at the same time, the block-based equalization produces detection statistics for each symbol block thereby simplifying the joint detection process.