The non-causal decision-feedback equalizer (NCDFE) is a decision-aided equalizer that uses not only past decisions, like DFEs, but also future decisions, which usually come from another, classical equalizer. When there are no errors on the decisions, the NCDFE attains the matched filter bound (MFB). In practice, it suffers from propagation of errors. We propose an implementation of the NCDFE based on soft decisions where only the most reliable decisions are fed back: this decreases error propagation and allows performance closer to the matched filter bound.
展开▼