Robust decision feedback equalizer design using soft-threshold-based multi-layer detection scheme

Most cost-effective DFE implementations suffer from the phenomenon of error propagation, which degrades its bit error rate (BER) performance. To solve this problem, we proposes a soft-threshold-based multi-layer DFE (STM-DFE) technique to reduce the BER. It costs very low hardware overhead compared to the conventional DFE. The key idea is to define a reliable/unreliable region in the DFE algorithm. When the output of the equalizer is in the unreliable region, the STM-DFE does not make a decision on the current symbol, but outputs a log-likelihood ratio to the next symbol. This process continues until the decision becomes reliable, or reaches the maximum stage number. The optimal threshold value between the reliable and unreliable region is analyzed in close-form. Simulation results show that the proposed scheme can efficiently reduce the burst error length (BEL) as well as the BER.