Dithered steepest descent algorithms

In this paper we consider a modification to the steepest descent adaptation technique for finite impulse response filters. The technique we propose uses a bilinear multiplicative noise update to ensure the convergence of the filter coefficients to the Wiener-Hoff solution. We demonstrate improvements to the convergence rate as well as an increase in the robustness this adaptation scheme brings. We also briefly discuss the sufficient conditions for convergence.