Abstract: This invention discloses an optimum form maximum likelihood MIMO detector (SFS ML Detector) that computes the LLR of the most likely received bit sequence from the received signal y, which consist of multiple signal stream from a MIMO channel H and an estimate of H. Through 5 simple steps of QR factorization, computation of partial Euclidean distance, sorting of the partial Euclidean distance, selection of surviving hypothesis by a set of algorithmic rule and computation of the full Euclidean distance of the survivors, and computation of the maximum likelihood from the surviving full Euclidean distance, the LLR of the received bit sequence can be obtained with a significantly lower number of computation and comparison than that is implied in the theoretical form of the maximum likelihood (ML) detector. There is no loss in performance of the SFS ML detector from the theoretical ML detector.