Generalized MLSDE via the EM algorithm

Generalized maximum likelihood sequence detection and estimation (GMLSDE) is developed in this paper based on the expectation and maximization (EM) algorithm. The GMLSDE couples the estimation of channel parameters and data detection in the framework of the maximum likelihood (ML) criterion and unifies many MLSD/MLSDE structure receivers for different channel models. The GMLSDE clarifies the relation among channel model, receiver structure and degree of optimality. The per-survivor processing (PSP) and per-branch processing (PBP) methods emerge naturally from the EM aspect of the GMLSDE as well.