Low footprint delayless subband adaptive filter

Adaptive filters are widely used for modelling an unknown system. Performance and complexity of adaptive filters are major problems when the length of adaptive filters becomes long. Delayless subband adaptive filters are introduced to reduce the complexity of adaptive filters while removing the processing delay inherent in subband processing. In this paper, a new adaptation method for a delayless subband adaptive structure is proposed which substantially reduces the memory foot print and to some extent computational load of the delayless algorithm. This is achieved by removing the adaptive weights in subband and performing adaptation in full band by use of subband data. Additionally, by modifying the adaptation algorithm with minimal computational overhead, performance of the proposed algorithm is improved in respect to speed of convergence. Suggested method is compared against two other delayless algorithms and similar or faster speed of convergence is achieved with the proposed method which has lower complexity.