High throughput robust transversal adaptive filter based on t distribution assumption

In this paper, a new robust transversal adaptive filter with high throughput is proposed. The updating of coefficient of the filter is determined based on assuming that the error signal is a t-distribution with a degree of freedom. Small weighting is assigned to the signal portion with large amplitude, while large weighting factor is used for the signal part with small amplitude. By doing so, the effect of large amplitude signal parts to the performance of the adaptive filter is suppressed. To obtain fast convergence, the RLS-like approach is applied to solve the updating step. To reduce the calculation burden, the calculation to determine the updating step is done in four sampling instants. Decomposing the updating algorithm into four steps does it. Our simulation results show that by doing so, we can obtain fast convergence and high throughput algorithm. Thus, the algorithm can be implemented by using a limited number of low-cost processor elements while preserving the throughput