Parallel LMS-based adaptive algorithms of echo cancellation

摘要

This paper deals with a number of parallel LMS family algorithms in the context of their use in the acoustic echo cancellation. The most complex parts of the LMS-based algorithms were determined and parallelized. A number of parallelization methods were designed taking under consideration varied types of architectures of modern concurrent computing environment, such as GPUs, clusters of workstations and cloud computing.