Recursive Least Squares-Algorithm-Based Normalized Adaptive Minimum Symbol Error Rate Equalizer

The adaptive minimum symbol error rate (AMSER) equalizer is known to have better symbol error rate (SER) performance than the adaptive minimum mean square error equalizer. Furthermore, the normalized AMSER (NAMSER) equalizer outperforms the AMSER equalizer, which can be regarded as the improvement of the normalized least mean square (NLMS) equalizer by incorporating the minimum SER (MSER) criterion. Inspired by that, we propose an improved recursive least squares-based NAMSER equalizer (RLS-NAMSER) that takes the advantage of faster convergence of the RLS algorithm over the NLMS algorithm. The RLS algorithm is first reconsidered from the perspective of optimization problem and an approximate RLS (ARLS) algorithm is proposed which converges faster than the NLMS algorithm. The RLS-NAMSER equalizer is then proposed by combining the ARLS equalizer with the MSER criterion. Simulation results show that the RLS-NAMSER equalizer has better convergence performance than the NAMSER equalizer while having nearly the same steady state performance as the NAMSER equalizer.

Automated summary

The adaptive minimum symbol error rate (AMSER) equalizer is superior to the adaptive minimum mean square error equalizer in symbol error rate (SER) performance. Furthermore, the normalized AMSER (NAMSER) equalizer outperforms the AMSER equalizer by incorporating the minimum SER (MSER) criterion. Inspired by this, the researchers propose an improved recursive least squares-based NAMSER equalizer (RLS-NAMSER) that takes advantage of the faster convergence of the RLS algorithm compared to the NLMS algorithm. Simulation results show that the RLS-NAMSER equalizer has better convergence performance while maintaining nearly the same steady state performance as the NAMSER equalizer.