Title :
The Diffusion-KLMS Algorithm
Author :
Mitra, Rangeet ; Bhatia, Vimal
Author_Institution :
Discipline of Electr. Eng., Indian Inst. of Technol. Indore, Indore, India
Abstract :
The diffusion least mean squares (LMS) [1] algorithm gives faster convergence than the original LMS in a distributed network. Also, it outperforms other distributed LMS algorithms like spatial LMS and incremental LMS [2]. However, both LMS and diffusion-LMS are not applicable in non-linear environments where data may not be linearly separable [3]. A variant of LMS called kernel-LMS (KLMS) has been proposed in [3] for such non-linearities. We intend to propose the kernelised version of diffusion-LMS in this paper.
Keywords :
least mean squares methods; signal processing; diffusion least mean squares algorithm; diffusion-KLMS algorithm; distributed network; incremental LMS; kernel-LMS; kernelised diffusion-LMS version; nonlinear environments; spatial LMS; Convergence; Kernel; Least squares approximations; Noise; Signal processing algorithms; Supervised learning; Vectors; diffusion least mean squares; distributed adaptive filtering; mercer kernel;
Conference_Titel :
Information Technology (ICIT), 2014 International Conference on
Conference_Location :
Bhubaneswar
Print_ISBN :
978-1-4799-8083-3
DOI :
10.1109/ICIT.2014.33
