An incremental noise constrained LMS algorithm

While a lot of research has focused on diffusion-based algorithms for distributed estimation over adaptive networks, incremental algorithms have not been studied to that extent. Here we present an incremental scheme-based distributed least mean square algorithm that uses the variance of the additive noise as a constraint. The proposed algorithm is derived and then its theoretical performance analysis is presented. Simulation results for different scenarios are then presented and it is shown that the simulation results corroborate very well the theoretical findings. More importantly, our algorithm outperforms, a recently proposed algorithm, the variable step-size incremental-based LMS (VSSILMS) algorithm.