Coefficient Design And Structure Optimization Of Parallel FIR Filters

FIR digital filter is the core component of digital signal processing system,which is widely used in some areas such as digital communication and image processing because of its ideal linear phase response.Coefficient design and structure design are two main directions in FIR filter design;in this thesis,the design of FIR digital low-pass filter is carried out from the two aspects mentioned above.The optimal design of finite impulse response filter can be abstracted into a problem of calculating the extreme value of multidimensional continuous function,so it can be solved by genetic algorithm.Because the traditional genetic algorithm is easy to fall into local solution,we propose a new strategy to improve the algorithm in the probability and operator of crossover and mutation,and it is proved to be feasible by trial function.Under the least squares(LS),minimax(MM)and minimum mean square error(MMSE)criteria,we have realized the optimized design of digital FIR low pass filters based on the improved genetic algorithm.Specifically,under least square criterion,minimum stop band attenuation of the 40 order FIR filter designed by improved algorithm is 39.65dB,filtering performance is improved by 7.6%than the designed result of ant algorithm.Hardware consumption increases linearly with the parallelism in traditional filter parallel architecture.As a first step,we use fast convolution algorithm to reduce the number of sub filter;then,in processing of realizing low parallel filter,the number of multipliers in circuit is reduced by employing more sub filters with symmetric coefficients and high parallel FIR digital filter is obtained by iterative method.The proposed 6-parallel 72-tap filter which is the iterative structure of 2-parallel filter and 3-parallel filter costs 156 multipliers,saving 12 multipliers than modified FFA structures.It shows that the proposed structure can effectively reduce the hardware cost.