Adaptive algorithms and their convergence characteristics

The convergence characteristics of the gradient transversal filter (LMS), the one coefficient gradient lattice (GL1), and the least squares lattice (LSL) are explored via two examples. The first investigates their ability to track the frequency dynamics of dual sinusoids whose power levels are widely separated and the second investigates the rate at which each algorithm is able to whiten a fourth order, complex, autoregressive process.