A delayless IFIR adaptive filter structure with adapted filterbank

In many applications such as acoustic echo cancellation and wideband active noise control, the least mean square (LMS) based adaptive filters with hundreds of taps are used, resulting in lower convergence and high computational complexity. In recent years researchers have developed adaptive filters based on subband techniques to improve the convergence rate and reduce the computational complexity. One such approach is IFIR structure based adaptive filters also called filter bank adaptive filters(FBAF). Though the technique improves the convergence rate, it introduces a delay in the signal path. FBAFs usually use fixed paraunitary perfect reconstruction (PR) filterbanks for preprocessing the data. An arbitrary FIR transfer function can he modelled by a delayless IFIR structure with no conditions on the interpolating filler coefficients. The interpolators can be designed using optimization techniques with knowledge of input signal statistics to improve the convergence rate. In this paper, we propose an algorithm to adapt the interpolators themselves and the model filters simultaneously to reduce the mean square error and hence require no offline optimization procedure to design the interpolators.