Research On The Compressed Estimation Of Channel State Information Based On Orthogonal Frequency Division Multiplexing System

Compressed sensing theory has been developing rapidly, which provides a new signal acquisition method to make a breakthrough on signal sampling with a rate much lower than Nyquist rate for sparse signals. Since channel impulse response has typical sparse structure, we research the compressed channel sensing methods for OFDM system to greatly reduce the pilot payload. In this thesis, we studied the compressed channel estimation methods in OFDM system, which can be used to save pilot cost and improve performance.The performance of compressed channel estimation is verified at the beginning of this article. Under the assumption that channel delay spread is shorter than cyclic prefix, the channel multipath search area could be lessened for channel reconstruction. So that the computation complexity is decrease, and the estimation performance is enhanced. Random pilot is not convenient for reality communication system. Based on the minimum correlation coefficient rule, the matrix with minimum correlation coefficient among the candidate matrix set is chosen to be the sensing matrix. Simulation results reveal that the performance of deterministic pilot approaches that of random pilot. We also design the optimal deterministic pilot patterns for LTE system with different bandwidth.Channel usually varies rapidly. Channel tracking technology is applied to estimate the real-time channel parameters. Only in this way can we ensure the receiver reliability. A new channel tracking estimation method is proposed in this article. The new method combines compressed channel sensing and Kalman filter update the time varying parameters iteratively, as well as the feature of OFDM system. Initial values are optimized to fast the convergence rate and enhance the tracking performance. Simulation results show Doppler frequency shift estimation in certain circumstance during the iteration process.For multi-antenna system, take full advantage of the feature determined by direction angle and elevation angle in antenna matrix, we introduced three spatial angle estimation methods, including combined MUSIC, step MUSIC and sparse dictionary based compressed angle estimation. We studied and proved the difference of performance and complexity among them. Since channel is sparse in both delay domain and spatial angle domain, we briefly introduced a channel state information estimation method in delay-angle of departure-angle of arrival demensions according to existing literatures. The mean square error lowerbound of the new method is lower than that of traditional ML training-based methods, and the pilot cost is greatly reduced.