Tham khảo tài liệu 'adaptive filtering applications part 5', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | Perceptual Echo Control and Delay Estimation 111 Fig. 13. Sparseness measure and its impaction on length of adaptive filter The quantity in Eq. 56 represents an energy measure in dBm within an estimated impulse response. Fig. 14 demonstrates the curve for this parameter for two frequency domain algorithms. It can be observed that after seconds estimation of IR energy measure for IPMDF stops fluctuating. Consequently this fact can be used for switching between different adaptation schemes. Fig. 14. Estimated misalignment and energy measure for MDF and IPMDF algorithms 112 Adaptive Filtering Applications Before representing the proposed MDF scheme declare the following statements I. Utilization of sparseness measure 0 c m l for real IR a. Um i. IR is considered dispersive ii. Mt L if fully updated scheme is chosen mostly during initial period iii. L K Mt L if partial updated scheme is chosen b. c m i. IR is considered sparse ii. Mt L. 1- m II. Utilized updating schemas a. non-partial b. X - based selection coefficient-based 2L-vector c. ịt - based selection block-based K-vector III. Utilization of IR energy measure n a. n m - n m-1 An switching to the MDF algorithm b. n m - n m-1 An switching to the IPMDF SC-IPMDF algorithm Finally our proposed algorithm can be described as 1st stage n m - n m-1 An f m Mf 2 L or M K w k m 1 w k m L ut G t k m 91 k m G t k m diag gkN m gkN 1 m . gkN N-1 m g 1 -a Ị 1 a - W -. m 2 L 2-Ệ0Iwj m Ảk const if m N L a const else m N L 1 _ 2. L . f w m l 1 L-4l -4L Wj m 2 J else ệ m Mf 2 L 1 - ệ mỹ or M round K 1 -ỹ m if mod m T 0 1 if l corresponds to M f maxima of xi m l 0 . 2L-1 pi m f z 10. otherwise X f k m Pk m X f k m 91 k m first half of fFT xf k m - v m d -1 Ef m w k m 1 w k m L Gt k m 91 k m .
