Optimal Realization of Parallel MAC-Based Adaptive FIR Filter Using a Gaussian Mutation on Quantum-Behaved Particle Swarm Optimization Algorithm

With the increasing prevalence of Digital Signal Processing (DSP) circuits in audio devices, the demand for low-power and high-performance processors has grown due to hardware limitations. Low-power design is particularly important for wireless audio devices with limited battery capacity. Nevertheless, as Field Programmable Gate Array (FPGA) technology advances, many challenging algorithms have recently been implemented to achieve high computing performance for both embedded and real-time applications. Conventional Least Mean Square (LMS) algorithms face challenges in real-time noise cancelation due to short secondary path delays. To address this, an optimized Finite Impulse Response (FIR) filter system using Particle Swarm Optimization (PSO) and Gaussian mutation on Quantum-Behaved PSO (GQPSO) is proposed. This design, implemented in a low-end FPGA, employs pipelined parallel multipliers and partial product-based shift-and-add multipliers for resource efficiency. GQPSO-based adaptive FIR filters are developed in Verilog and synthesized for 8-tap and 16-tap configurations, showing significant performance gains. Post-route FPGA results demonstrate an 89.3% reduction in area utilization and a 97.3% speed improvement over existing architectures, highlighting their effectiveness for real-time noise cancelation.