GpLMS: Generalized Parallel Least Mean Square Algorithm for Partial Observations
Résumé
We propose a generalized parallel least mean square algorithm (GpLMS) to deal with partial observation scenarios. GpLMS takes advantage of a two stage parallel LMS architecture to enhance the convergence rate and updates weight vector based on observed entries to obtain a low computational complexity. We compare the results from our proposed algorithm with the state-of-the-arts in an adaptive beamforming context to illustrate its effectiveness.