A subband equalizer with the flexible structure of the analysis/synthesis subsystem

In this paper we review different methods of subband equalization based on oversampled filterbanks and present some new results for subband adaptive systems. Nowadays, there are a lot of applications, which use subband adaptive filters (SAFs). However, SAFs are not clearly understood compared with such mature and acknowledged technologies as OFDM or well-known least-mean-square time-domain algorithms. The advantages and drawbacks of SAFs achieved due to involving of multirate signal processing over conventional adaptive algorithms like least-mean-square (LMS) and recursive least-mean-square (RLS) as well as fast algorithms (e.g., fast LMS) are considered below. The novel approach concerned with flexible time-frequency decomposition is also investigated. The main attention is focused on the oversampled SAFs (both real-valued and complex-valued) because these SAFs let avoid aliasing caused by decimation due to the properly designed analysis and synthesis filterbanks. In the last part of this paper it is shown that, under certain circumstances, the SAF with the flexible structure can be considered as the best choice in the trade-off between performance, computational complexity and processing delay.