Channel Estimation Methods Based On Compressed Sensing For OFDM Systems

With the application scenarios of new wireless communication services being proposed continuously,it has become the primary task of researchers in the field of mobile communication to provide a high-precision,high-speed data service solution.In wireless communication,the accuracy of demodulated signals can be improved by obtaining the channel state information in advance.Compressed Sensing(CS)theory is a new theory which combines compression with sampling.It can effectively solve the problem of high band resource overhead of Orthogonal Frequency Division Multiplexing(OFDM)channel estimation algorithm.Therefore,this paper mainly studies the signal reconstruction algorithm in OFDM-based compressed channel sensing technology.In this paper,the principle of OFDM system and the main content of mathematical modeling compressed sensing theory are described in detail.Firstly,the development of channel estimation technology and compressed sensing theory is described.Then,the characteristics of wireless channel,the generation principle of OFDM system and three important aspects of compressed sensing theory are introduced.Then,the atomic screening rules and algorithm flow of several commonly used algorithms are given and simulated.Then,the influence of the number and structure of pilots and the length of training sequence on the performance of channel estimation is analyzed by simulating the channel estimation algorithm.By choosing the atom that minimizes the residual error after orthogonal projection,combining with gOMP algorithm and adding complex screening steps,an improved piecewise generalized orthogonal matching pursuit algorithm(ISgOMP)is proposed.The performance of the improved algorithm in three aspects,namely,reconstruction error,reconstruction time and reconstruction probability,is verified by MATLAB simulation in two cases of one-dimensional signal input and OFDM system.Aiming at the shortcomings of the generalized orthogonal matching pursuit(gOMP)algorithm,such as too many useless atoms and unstable channel estimation performance,this paper improves the algorithm.Through comparative experiments,it is concluded that channel estimation performance is better when pilot interval is 12 subcarriers and block interval is adopted.Channel estimation performance improves with the increase of training sequence length.When the length is 32,channel estimation process can be completed and signal resources can be saved.The simulation results of the improved algorithm show that ISgOMP can reconstruct the original signal accurately in a short time.The number of measurements needed in the reconstructing process is less than that of the gOMP algorithm and the reconstructing accuracy is higher.The improved algorithm achieves the purpose of enhancing the channel estimation performance.