This paper investigates noise reduction performance and performs convergence analysis of a Variable Error Data Normalized Step-Size Least Mean Square (VEDNSS LMS) algorithm. Adopting VEDNSS LMS provides fast convergence at early stages of adaptation while ensuring small final misadjustment. An analysis of convergence and steady-state performance for zero-mean Gaussian inputs is provided. Simulation results comparing the proposed algorithm to existing algorithms indicate its superior performance under various noise and frequency environments.

Estimating a location of mobile phones or sound source is of considerable interest in wireless communications and signal processing. In this letter, we propose squared range weighted least squares (SRWLS) using the range estimate attained from the Taylor series-based maximum likelihood. The weight can be determined more accurately when using the proposed method, compared with the existing methods using the variance of noise. The simulation results show that the proposed method is superior to the existing methods in RMSE as the measurement noise amount of sensors increases.

This letter proposes a new adaptive filtering method that uses the last L desired signal samples as an extra input vector, besides the existing input data, to reduce mean square error. We have improved the convergence rate by adopting the squared norm of the past error samples, in addition to the modified cost function. The modified variable error-data normalized step-size least mean square algorithm provides fast convergence, ensuring a small final misadjustment. Simulation results indicate its superior mean square error performance, while its convergence rate equals that of existing methods. In addition, the proposed algorithm shows superior tracking capability when the system is subjected to an abrupt disturbance.