Data-driven echo cancellation for a multitone modulation system

The problem of echo cancellation in a multitone modulation (MTM) scheme is addressed. A general model for the near-end echo is derived and is used to identify candidate data-driven echo canceller (DDEC) structures. The stability, steady state performance, and associated system complexity of an adaptive DDEC based on the stochastic gradient approach is developed. It is further shown how the symmetry in the derived echo path model can be exploited to enable a reduction in canceller complexity, enhancing convergence speed without sacrificing final SNR. Simulation results are provided that confirm the analytical predictions.<
>