Design of Sparse IIR filter with Positive Realness and LASSO

カテゴリ: 部門大会

論文No: SS2-6

グループ名: 【C】2023年電気学会電子・情報・システム部門大会

発行日: 2023/08/23

タイトル(英語): Design of Sparse IIR filter with Positive Realness and LASSO

著者名: 岩切 遼（広島大学）,中本 昌由（広島大学）

著者名(英語): Ryo Iwakiri (Hiroshima University),Masayoshi Nakamoto (Hiroshima University)

キーワード: IIR filter|Positive Realness|LASSO

要約(日本語): Digital filters are digital signal processing circuits used in many fields such as audio/acoustic processing and image processing. Especially when implemented in small terminals, low-power operation is strongly desired. Digital filters are classified into FIR filters and IIR filters. IIR filters can achieve the same accuracy as FIR filters with a smaller order, but they must be designed in consideration of stability conditions. In this study, we also consider IIR filters in which some of the coefficients are zero coefficients. Such a filter is called a sparse filter, and it has higher performance characteristics than a non-sparse filter in which all the coefficients are real numbers. In this design, in addition to an algorithm that generates zero coefficients, stability is also guaranteed.