Adaptive NLMS diagonally-interpolated Volterra filters for network echo cancellation

This paper presents a novel approach to implement adaptive Volterra filters with reduced complexity aiming at echo cancellation applications. The proposed approach is based on combining two different strategies in order to reduce the computational burden of adaptive Volterra structures, namely shortening the number of kernel diagonals and using a sparse-interpolated setup. As a result, an efficient diagonally-interpolated Volterra implementation is achieved, offering a controllable trade-off between complexity and performance. Simulation results considering a real-world network echo cancellation problem attest the effectiveness of the proposed approach.