Study on Filter Optimization Algorithm Based on Fractional Fourier Transform

Zeming Zhang,Sheng Zhang,Yuan Zhang,Yuanqin Xia
DOI: https://doi.org/10.1145/3672758.3672805
2024-01-01
Abstract:Adaptive filtering algorithms are a prominent research area in adaptive signal processing. Considering the limitations of the classic Least Mean Square (LMS) filtering algorithm in achieving effective signal mixing separation in strong reverberation backgrounds, this study introduces a Discrete Fractional Cosine Transform (DFRCT-LMS) adaptive filtering algorithm leveraging the Discrete Fractional Cosine Transform's (DFRCT) excellent orthogonality characteristics. To address the issue of fractional order uncertainty in fractional Fourier transform, a method is proposed to establish a relationship curve between fractional order and misalignment, aiding in finding the optimal fractional order. The derivation process and implementation steps of the DFRCT-LMS adaptive filtering algorithm are thoroughly expounded. Through a system identification model, the performance of the proposed algorithm is assessed and compared against other methods using two indicators: feature ratio and learning curve, demonstrating its superior performance.
What problem does this paper attempt to address?