Increasing convergence speed of FxLMS algorithm in white noise environment

A modification of the Filtered-x Least Mean Square (FxLMS) algorithm, aimed to reduce the influence of the secondary path and to increase the convergence speed in the white noise environment, is proposed. Secondary path disrupts the diagonality of the white noise autocorrelation matrix and increases its eigenvalue spread, causing a slower convergence speed of the FxLMS algorithm. We have modified the algorithm by adding the elements of the secondary path impulse response coefficients. In this way the resulting algorithm becomes a self-orthogonalizing FxLMS (SOFxLMS). As a consequence, the eigenvalue spread is reduced and the convergence speed is increased. The convergence analysis of the proposed algorithm and the simulation results confirm the performance improvement.