An Energy Efficient CMA Equalizer based on Approximate Computing

Constant Modulus Algorithm (CMA) has found extensive application within digital communication systems. Unlike traditional adaptive equalizers, the CMA equalizer leverages its own statistical properties to recover signals from channel interference, obviating the need for training sequences. However, the significant number of multiplier units required by the CMA algorithm can lead to excessive circuit overhead. Approximate computing offers a promising approach to achieving high performance while mitigating circuit complexity. In this study, we propose an approximation scheme tailored specifically for the CMA equalizer. We introduce two approximate multipliers, strategically integrated into different computational modules to ensure mutual compensation for accuracy. Implementation results demonstrate that the proposed approximate CMA system achieves reductions of 40.10% in area and 18.13% in power consumption compared to the precise system.