Research On Channel And Doa Estimation For Massive Mimo Systems

Massive multiple input multiple output (MIMO) can provide extremely high spectrum efficiency, and is one of the candidates for the fifth generation wireless communication technologies in the future. Because the channel co-herence interval is limited, orthogonal pilots are not sufficient enough to be assigned among cells. Hence, the performance of multi-cell massive MEMO systems is severely impacted by the inter-cell interference (ICI). In order to mitigate the ICI, efficient pilot design schemes and an efficient channel estima-tion approach are proposed in this dissertation. Meanwhile, the beamforming approach can effectively mitigate the ICI, which relies on precise estimates of the angular parameters. However, the computational complexities of the state-of-the-art angular parameter estimation approaches are high. Hence, multiple low-complexity approaches are proposed for estimating the angular parameters of incoherently distributed sources with different kinds of antenna arrays. The main contributions of this dissertation are summarized as follows.1) A semi-blind channel estimation approach is proposed for multi-cell massive MIMO systems. According to the asymptotic orthogonality of the channel vectors in massive MIMO systems, and the column space of the chan-nel matrix is used to make the proposed approach introduce less intra-cell in-terference. The performance of the proposed channel estimation approach with finite number of antennas and that with infinite number of antennas are ana-lyzed. Based on the analysis, multiple Zadoff-Chu (ZC) sequences are used to design efficient pilots that minimize the ICI. Moreover, in order to eliminate the intra-cell interference, the circular shifts of the ZC sequences are designed.2) Pilots for linear receivers in multi-cell massive MIMO systems are pro- posed. First, when linear receivers are employed in massive MIMO systems, the asymptotic received signal-to-interference ratio (SIR) of these systems is derived. In addition, a pilot design criterion is proposed, which is the max-imization of the SIR. After that, the orthogonality of circularly shifted ZC sequences are used to design pilots that eliminate the intra-cell interference. Finally, according to the proposed pilot design criterion, the phases of the ZC sequences are designed.3) A distribution fitting based approach for estimation of the direction-of-arrivals (DOAs) of incoherently distributed sources with large uniform linear arrays is proposed. The proposed approach uses the asymptotic orthogonality of the array manifolds to transform the covariance matrix into a vector. Thus, the number data symbols for estimation is reduced. When the number of an-tennas tends to infinity and the probability density functions of the DOAs of d-ifferent users tend to be orthogonal, it is proved that the estimates can converge to the true values. Moreover, the computational complexity of the proposed approach is lower than that of the traditional approaches.4) An estimation of signal parameters via rotational invariance technique based approach for estimation of the DOAs of incoherently distributed sources with large uniform rectangular arrays (URAs) is proposed. First, the pro-posed approach uses the array structure to obtain coupled functions of two-dimensional (2-D) DOAs. Then, the2-D DOAs are decoupled to obtain the closed-form expressions of the DOAs. Moreover, the estimation performance of the proposed approach with limited number of snapshots is analyzed, and the approximate Cramer-Rao bound (CRB) on estimation errors with URAs is derived. Finally, the computational complexity of the proposed approach is analyzed, and the analysis shows that the computational complexity of the proposed approach is lower than that of the traditional approaches.5) A beamspace transform based approach for estimation of the DOAs of incoherently distributed sources with large uniform cylindrical arrays (UCyAs) is proposed. The proposed approach uses the beamforming weight vectors to transform the received signal vectors from the element space to the beamspace. Thus, the length of the received signal vectors is reduced. Additionally, the search criterion of the proposed approach achieves better estimation perfor-mance than the known criterion. The errors introduced by the transform of the received signal vectors are analyzed. The analysis shows that these errors tend to zeros as the number of antennas increases. On the other hand, the approx-imate CRB on estimation errors with UCyAs is also derived, which is more precise than the known bound. Finally, the analysis proves that the proposed approach is of lower computational complexity than the traditional approaches.