Speeding up the Throughput of an NLMS Adaptive Filter
Yu-Zhe Huang, Chia-Yu Yao, Ping-Yi Lee · 2024
Adaptive filters are widely used in communications, signal processing, and system identifications. The computational bottleneck of the normalized least mean square (NLMS) adaptive filter is the division. Previous works have shown that using the reciprocal approximation for the divisor not only works well for the NLMS algorithm but can also improve the adaptive filter’s overall throughput rate. This work designs a lookup table (LUT) reciprocal approximation method to replace the division for the NLMS adaptive filter. The experimental results show that the adaptive NLMS filter using the proposed approximation consumes less power and resources than the ones using the standard SRT division and the previously reported CSD reciprocal approximation on an FPGA platform.