Research On Channel Tracking Technology Based On Hybrid Precoding Structure In Millimeter Wave Systems

Millimeter wave has become a hotspot for future mobile communication research due to its abundant spectrum resources.At the same time,millimeter wave is closely related to hybrid precoding structure.The research on millimeter wave systems with hybrid precoding structure has important practical significance.Frequent channel estimation brings huge challenges to system overhead,delay,and throughput,therefore,the channel tracking technology which using the time correlation of the channel is prominent.There are great research value about the channel tracking technology of millimeter wave systems.The existing research results of channel tracking technology applied to millimeter wave systems are few,and they are mostly used to anaolg precoding and single RF chain scenarios.This thesis studies the channel tracking technology of millimeter wave systems under hybrid structure,and obtains some results and conclusions as follows.This paper proposes a channel tracking algorithm based on extended Kalman filtering.Firstly,the extended Kalman filter is used to track the channel’s angle information,and an iterative update algorithm is derived to update the combing matrix at the receiver to further improve the tracking accuracy.Experiments show that compared with other algorithm,the proposed channel tracking method reduces the tracking error by 20% and improves the effective tracking time by 30%.When the user speed is extremely low,the algorithm has better tracking performance in large-scale antenna arrays;when the user speed is lower than 20m/s,it is suitable for millimeter wave systems with 16 to 128 antennas;if the user speed is greater than this value,The algorithm is no longer applicable.This thesis proposes a compressive sensing assisted channel tracking method based on precoding optimization.Using the codebook,the channel tracking problem is modeled as a compressive sensing problem,meanwhile,the design problem of the sense matrix is transformed into a precoding optimization problem.Combining with the mutual coherence property and the conditions of precoding optimization deduced in the article,the sparse perception matrix is designed,and the channel state information is accurately reconstructed.The experiment proves that compared with the random sparse perception matrix,the method can reduce the measurements by 50%,and it is suitable for large-scale antenna array scenarios where the user speed is higher than 12m/s,including the above 20m/s which cannot be solved by the CT-EKF method,the two methods have different requirements on the number of system antennas.