Abstract:
The standard model for stereophonic acoustic
echo cancellation (SAEC) consists of a two-input/twooutput system with real random variables, which can be
modeled using four adaptive filters. Recently, we used
the widely linear (WL) model in the context of SAEC. In
this case, we deal with a single-input/single-output
system with complex random variables, which involves
only one adaptive filter. In this paper, the functionality
of the WL-SAEC scheme is shown using the recursive
least-squares (RLS)-dichotomous coordinate descent
(DCD) algorithm. Moreover, a dual path filtering
methodology is used to make the algorithm more robust.