Research On Channel State Information Acquisition Based On Compressive Sensing For5G Systems

For the later IMT-Advanced system (5G), with the development and popular of intelligent end and mobile internet business, the flow of data business has predicted to grow in explosive trend. It has been predicted that the data service will grow to1000manifold until the year of2020. In addition, the International Telecommunications Union(ITU) has proposed to the International Mobile Telecommunications (IMT-Advanced) that5G system is supposed to afford1Gbps and100Mbps user rate in low speed scenario and high speed scenarios, respectively.In the future5G communication system, the architecture of network will transform from traditional cellular network to distributed, heterogeneous network, with denser deployment and smaller cell coverage, which leads to more serious inter-cell interference. The channel state information (CSI) is important for multiple input and multiple output (MIMO) orthogonal frequency division multiplexing (OFDM) systems, because it can eliminate inter-cell and inter-user interference, enable self-adaption and pre-coding, promote the system performance. As many key technologies of5G, like Massive MIMO and coordinated multipoint communication (CoMP), have imposed higher requirements on CSI, especially its accuracy and overhead, how to reduce CSI acquirement overhead and improve CSI accuracy have become important challenges.Generally, the acquirement of channel state information includes channel estimation and channel feedback, there exist many researches in this field. This thesis summarized and analyzed classic methods in the fields of channel estimation and channel feedback. Moreover, in the new scenarios of5G system, the resource consumed in the process of CSI acquirement grows proportionally with CSI quality, transmit antennas and the number of cooperative cells. Therefore, based on parametric characteristics of wireless channel and compressive sensing theory, this thesis researched into creative CSI acquirement methods with regarding to the new technologies and new scenarios in5G systems. Firstly, this thesis researches the sparse property of wireless communication channel, recalled compressive sensing theory and its application, and analyzes its feasibility in compressing CSI. In addition, classic methods in channel estimation and channel feedback is summarized.Secondly, with respect to massive antenna deployment, this thesis utilize channel correlation in spatial domain, combine with compressive sensing theory, propose that complete CSI can be obtained using even less pilots than sampling theorem. After analyzing its feasibility, we formulate an optimization problem based on maximizing channel estimation quality, from which pilot pattern is designed. The simulation results demonstrate that, spatial channel estimation based on compressive sensing can significantly reduce pilot overhead, and using the proposed pilot pattern can greatly increase system performance compared with equal pilots and random pilots.What’s more, with regard to multi-user multipoint coordinated communication scenarios, this thesis proposed a quantized time domain compressed feedback scheme which CSI is compressed before feedback utilizing the sparsity property. Additionally, we compare this scheme with the existing ones on aspects of theoretic analysis and simulation, and evaluate them in terms of average user rate and feedback overhead. Moreover, this thesis proposes a bit allocation algorithm based on branching and bounding to solve the non-negative integer constrain on bits allocated. The simulation results shows that the proposed quantized time domain compressed feedback scheme can efficiently reduce feedback overhead, and able to achieve performance close to perfect CSI case.Finally, a brief summary of this thesis is made and possible future research contents on CSI acquirement based on compressive sensing are pointed out.