Research On Adaptive Reconstruction Algorithm Based On Compressed Sensing

As a new technology of signal acquisition and processing,Compressed Sensing(CS)performs data compression while acquiring data,saving software and hardware resources and data processing time.For sparse signals,the signal sampling rates based on CS are much lower than the traditional Nyquist sampling method.All these advantages make it be applied in many fields.Therefore,it is highly concerned by many academics.CS theory exploits the sparsity of signals or transforms them into sparse domains and then accurately reconstructs signals by solving optimization problems.As the key issue of CS theory,reconstruction algorithm should reconstruct the original signal under the premise of known measurement matrix and measurement signal efficiently and accurately.Most reconstruction algorithms require the sparsity of a known signal as a priori condition,however,the sparsity of the signal is hard to obtain in practice.Therefore,it has more actual sense to make research on adaptive compressed reconstruction algorithm.The channels in Orthogonal Frequency Division Multiplexing(OFDM)systems have sparse characteristics in the time domain,so the problem of channel estimation in OFDM systems can be modeled as a sparse reconstruction problem in CS.This thesis focuses on adaptive compressed sensing reconstruction algorithm,also makes research on the application of the algorithm in OFDM system.The main contributions of this thesis are as follows.1)Based on a large number of researches on the traditional compressed sensing reconstruction algorithms,an adaptive compressed sensing reconstruction algorithm is proposed,called Enhanced Adaptive Stagewise Orthogonal Matching Pursuit(EASt OMP).The proposed algorithm employs the backtracking and introduces an index parameter based on the original threshold of the existing stagewise orthogonal matching pursuit algorithm,which can get the final support set and the adaptive estimation of sparsity more efficiently,obtaining better performance of signal reconstruction.The simulation results show that,the proposed algorithm can get higher signal reconstruction quality with less increase of computational complexity,such as that the signal accurate reconstruction probability is improved by 30%~40%,as compared with other relevant algorithms without noise in measurement signals.2)Under the condition that the measurement signals are disturbed by noise,the Enhanced Adaptive Stagewise Orthogonal Matching Pursuit algorithm is applied to the signal reconstruction.The algorithm can improve the signal reconstruction quality by 3~5d B with less increase of computational complexity,as compared with other relevant algorithms.3)The Enhanced Adaptive Stagewise Orthogonal Matching Pursuit algorithm is applied to the channel estimation of OFDM system,realizing the high-performance of channel estimation under the unknown channel sparsity in the actual scenario.The simulation results show that the EASt OMP algorithm proposed in this thesis can still perform well in channel estimation under the premise of unknown channel sparsity,and its performance is also superior to other algorithms by decreasing the Mean Square Error(MSE)of the channel estimation 1~3d B averagely.