Study On Downlink Transmission Method Based On Channel Correlation

With the rapid development of the Internet, the demand for high speed and high throughput wireless network is increasing. To further improve system performance, the large-scale antenna technology has emerged. However, this technology will cause large overhead of channel estimation and CSI feedback in Frequency Division Duplex system, and computational complexity will increases with channel matrix dimension significantly in data processing. Thus, a channel correlation based user clustering method is proposed to eliminate interference between user groups and compress the effective channel dimensions, and finally simplify the complexity of the whole downlink transmission.In real scenarios, users are not distributed uniformly, but in relatively stable clusters. The channel correlation matrix are restricted if users are in poor scattering environment. We study the channel model and analysis the attribution of the channel correlation matrix, and also extend the work to the MU-MIMO system. These features are the basis of user clustering, interference elimination and dimension compression. Based on these features, this article study on a two-stage precoding schemes and the simulation of system performance. The outer precoder is to eliminate the intergroup interference and compress the effective channel by using the statistical information. The inner precoder is doing conventional operation such as user selection and MU-MIMO precoding in individual groups after the outer one. We verify that the scheme is both considering system performance and operation overheads. Further, the downlink transmission scheme which includes user clustering and user selection is designed and verified. We study the matrix clustering by using K-means methods and also make some progress on related group number and outlier users problems. And a novel clustering method which combined with MUSIC is proposed. To improve the whole transmission, we also do the research on the user selection and precoder design.Beside this, we extend the work to interference management in a HetNet. We study the relation between position and interference of 2-tie user groups. Further, a simple interference management solution based on spatial parameters is discussed.