Research On Wireless Channel Estimation Algorithms Based On Compressed Sensing

With the development of science and technology, people’s demand for wireless communication is growing in rapid, and the requirement for communication quality and speed and multimedia services becomes more and more. Orthogonal Frequency-Division Multiplexing（OFDM） technology can reduce the influence of the channel frequency-selective fading in wireless communication and effectively improves the utilization of the spectrum due to the orthogonality of the subcarriers. Multiple Input Multiple Output（MIMO） technology can achieve the spatial reuse improved the transmission efficiency and system capacity effectively. However, in the OFDM or MIMO-OFDM system, in the process of signal transmission, the signal may generate power attenuation and time delay due to constructions and obstacles nearby and could generate inter-symbol interference（ISI） and Inter Carrier Interference（ICI）. So, it is really important to achieve the channel state information（CSI） correctly in order to efficiently recover the transmitted signal. Meanwhile, the research results indicate that the tap coefficient of multipath is sparse in high dimensional space. But because the traditional channel estimation method can’t make full use of this feature, we need to fand a new method which can improve the accuracy of estimation based the feature.Different from Nyquist sampling theorem, the compressed sensing（CS） an efficiently reconstruct the original signal by a few sampling signal which is the dominant power in the sparse signal. Therefore, compressed sensing can improve the sampling efficiency largely. At present, the CS technology has been applied in various fields such as wireless communication, image processing, pattern recognition and atmosphere, etc. In this paper, we apply the CS technology to the sparse channel estimation in order to estimate the channel state with a few pilot symbols.In this thesis we mainly describe the channel estimation methods based compressed sensing in OFDM and MIMO-OFDM systems. The work is divided to the two parts:First, we applied the Fast Bayesian Matching Pursuit（FBMP） for the channel estimation in OFDM system. FBMP algorithm sets up the bayesian prior model according the bayesian principle, solve the parameters model by using the prior knowledge about the channel and maximum posterior probability, and reconstruct the sparse channel accurate.Finally, in MIMO-OFDM system, this thesis proposes a channel estimation method based on Modified Hybrid Optimized Smoothed l₀ norm（MHOSL0）. MHOSL0 algorithm uses hyperbolic tangent function instead of gaussian function to approximate the l₀ norm and combine the steepest descent method with modified newton for hybrid optimization. This method achieves fast convergence rate and improves the reconstruction accuracy.