Abstract | ||
---|---|---|
In this paper, the tracking behaviors of the least mean M-estimate (LMM) and normalized LMM algorithms are analyzed in a unified manner in a non-stationary system described by the random-walk model. In the analysis, we consider the presence of impulsive noise and do not impose a specific distribution on the input signal. For predicting the steady-state performance, we provide analytical expressions for the algorithms. Unlike for the stationary case, the steady-state performance for the non-stationary case does not always improve as the step size decreases. As such, the optimal step size is also derived to reach the best steady-state performance. Simulation results support the theoretical findings. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/SSP49050.2021.9513747 | 2021 IEEE Statistical Signal Processing Workshop (SSP) |
Keywords | DocType | ISSN |
Adaptive filters,tracking performance,LMM and NLMS algorithms,impulsive noise | Conference | 2373-0803 |
ISBN | Citations | PageRank |
978-1-7281-5768-9 | 0 | 0.34 |
References | Authors | |
0 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yi Yu | 1 | 0 | 0.68 |
Rodrigo C. de Lamare | 2 | 0 | 0.34 |
Tao Yang | 3 | 160 | 76.32 |
Qiangming Cai | 4 | 0 | 0.34 |