A New Hyperstable Adaptive Recursive Filters Algorithm

Abstract

Recently a simplified hyperstable sequential regression SHSER algorithm for adapting the parameters of the IIR filter has been proposed. The main drawbacks of the SHSER algorithm are; the performance may suffer from a highly nonstationary environments and when the error signal is filtered by a smoothing coefficients vector this vector should be selected in such a way that it fulfills the SPR condition which are not always possible to achieve in practice. In this paper, a new algorithm is suggested to eliminate the drawbacks of the SHSER algorithm. The proposed algorithm works in two phases, in the first phase the error signal is not filtered until it reaches a certain threshold and then the algorithm automatically changes to start the error filtering phase using the estimated system poles vector. This algorithm is called Automatic Hyperstable Adaptive Recursive Filters AHARF algorithm. This algorithm provides very fast convergence rate compared with many existing algorithms, well suited in highly nonstationary environments, and automatically selects the smoothing vector to fulfill the SPR condition. A new variable step-size related to the error signal and the auto correlation matrix is also proposed and adopted in the AHARF algorithm, which allows high convergence rate for all highly nonstationary signals.

Indexing terms:

• Adaptive filtering

Additional information

Notes on contributors

Fawzi M Al-Naima

Fawzi M Al-Naima was born on 22 December 1948 in Mosul, Iraq. He received his BSc (First Class Honours) in Electrical and Electronic Engineering in June 1971 from the University of Newscastle upon Tyne, England, UK and his PhD in Electrical and Electronic Engineering in March 1976 from the same university. He specialized in computer aided network analysis and design. He is having teaching and research experience of about 24 years in England and Iraq.

In July 1989 he joined the Department of Electronic and Communications Engineering, College of Engineering, Saddam University, Baghdad, Iraq, and became a Professor at that department in December 1990. He served as a Head of Department of Electronic and Communications Engineering, College of Engineering, Saddam University from 1992 to 2000. Since April 2000 he has been the Head of Department of Computer Engineering at the same College. His areas of research include optimum design of analog and digital filtering networks, symbolic analysis of large analog and distributed networks and the implementation of digital signal processing techniques in communications.