Improved Variable Step Size LMS Algorithm Based on Artanh Function

The LMS algorithm with a fixed step size cannot simultaneously achieve excellent performance on convergence rate and steady state error in practical applications. This paper proposes a variable step length LMS algorithm (IAVSLMS) based on artanh function, and constructs a new relationship between step length parameter and error signal by improving the parameter model structure, removing small square operation and adopting the error correlation function. The new algorithm is more computational and conducive for hardware implementation. The simulation results show that the new algorithm performs well in convergence speed, steady state error, and tracking performance.