DESIGN OF SHARP MDFT FILTER BANKS WITH PERFECT RECONSTRUCTION IN THE SPT SPACE
In the modified discrete Fourier transform (MDFT) filter banks, the presence of an inherent structure which leads to the cancellation of all odd aliasing components is the advantage over discrete Fourier transform (DFT) filter banks. We need sharp transition width filters in many applications in wireless communication. The complexity of sharp filters can be reduced by using frequency response masking (FRM) approach. A design is proposed in this paper to realize multiplier-free FRM based MDFT filter banks with perfect reconstruction (PR). Canonic signed digit (CSD) representation is used to make the filters in the filter bank multiplier-free. The number of non-zero bits is reduced in the conversion process to minimize the number of adders and shifters required for the filter implementation. Hence, the performances of the MDFT filter bank with PR may degrade. The performance of the CSD represented filters in the filter bank, are optimized using harmony search algorithm. This design reduces the implementation complexity which may lead to low power consumption, low chip area and high speed of operation for the realization of the uniform filter banks.
MDFT filter banks with PR,frequency response masking, canonic signed digit,harmony search algorithm,gravitational search algorithm,artificial bee colony algorithm.