On the convergence of a new joint DFE & decoding procedure for blind decision directed LMS equalization

Adaptive equalization is a classical technique for mitigating ISI in unknown or time varying channels. The decision-feedback equalizer (DFE) is considered to be an efficient approach in many types of channels where linear equalizers fail. Unfortunately, it suffers from error propagation phenomena. In order to reduce such effect, the present paper deals with the joint application of both equalization and decoding process in the receiver. The main contributions are the proposal of a new joint DFE and decoding configuration which works in a blind decision directed (DD) context and the study of the corresponding error surfaces behavior, and the assessment of the convergence rate. The proposed technique is shown to be effective for the so-called bad channels class, which makes the conventional DD-DFE ill-converge when its weights are initialized with zeros