In this study, 16-channel cosine modulated filter bank is designed. In the proposed design, the coefficients of the prototype low-pass Finite Impulse Response filter with an N=64 filter order is optimized using Artificial Bee Colony Algorithm. In the presented study, different design examples are given with different frequency band values. Amplitude response error and transition band error values are obtained for each design examples. Via these examples, the effects of the change in transition and stopband frequencies on the performance of the designed filter bank is investigated. Simulation results show that determining frequency band values of prototype filter properly improves the performance of the filter bank and decreases the amplitude response error and transition band error values پرونده مقاله

Compressive sensing is a technique that can reconstruct sparse signals under Nyquist rate. This study is about comparison of widely used sparse signal reconstruction algorithms under noisy measurements. Three algorithms, Orthogonal Matching Pursuit, Compressive Sensing Matching Pursuit and Primal Dual Interior Point method are used to reconstruct sparse signal from noisy measurement and performance results are compared. Firstly, a sparse signal is sampled under Nyquist rate and observation vector is obtained. After that, white Gaussian noise is added to this observation vector. Then, sparse reconstruction algorithms are employed to reconstruct the original signal from noisy measurement. These algorithms are tested for various measurement number and sparsity levels. Test conditions are same for all algorithms. Finally some performance metrics results related to reconstructed signal are obtained. These performance metrics are mean squared error, correlation of the reconstructed signal and original signal, reconstruction time of the algorithms and iteration numbers. According to these metrics, when sparsity level is very smaller than measurement number, Orthogonal Matching Pursuit has better results than others. However, when sparsity level is increased and close to measurement number, Primal Dual Interior Point method has better performance than others in terms of reconstruction a sparse signal from noisy measurement. پرونده مقاله