A new sub-ML detection method for variable-rate space-time block codes

We present a new sub-optimal near-maximum-likelihood (ML) detection method for the family of variable-rate space-time block codes (VRSTBC). The proposed detection method is based on the concept of reliabilities and provides a wide range of performance-complexity trade-offs. The reliability measures are defined with the help of a recent generic ML metric expression in (J. Leuschner and S. Yousefi, 2008). The performance and complexity analysis of the method show a close performance to ML attaining a reasonable level of complexity: 0.5 dB away from ML performance at BER= 10^-5 with %3.7 of ML detection cost.