Reciprocity Calibration For Massive MIMO Systems

The fast increase of requirement for information acquisition and exchange are the major motivation to improve the mobile communication technologies. In the new generation mobile communication system, the demand of performance will be much higher for transmission rate, delay, system reliability and so on. As a very promising technique for the next generation mobile communication system, massive multi-input multi-output (MIMO), also called large-scale MIMO, has attracted more and more attention in both academia and industry. By deploying a large excess of antennas on the transmitter, this emerging technique can significantly increase the space resolution, deliver high system throughput and bring huge improvement in energy efficien-cy. In massive MIMO systems, time division duplexing (TDD) mode is more preferred due to the channel reciprocity. With the knowledge of the downlink channel state information (CSI) from the uplink pilots, joint precoding can be performed at the base station (BS). Thus, the BS equipped with hundreds of antennas can communicate with several user equipments (UEs) at the same time-frequency resource. Then, BS is required to transmit pilots through the precoding. The pilot overhead is reduced to the number of UEs and the required feedback was eliminated, which will save great radio resources. Unfortunately, in practice, the whole com-munication channel consists of not only the wireless propagation, but also the transceiver radio frequency (RF) circuits of the antennas at both sides of the link. The RF mismatches will jeopardize the reciprocity of whole channel and lead to a severe system performance loss. Therefore, it is very important to perform the reciprocity calibration for the TDD system. This dissertation investigates the channel information acquisition theory for massive MIMO systems. We mainly focus on the reciprocity calibration, and propose some well performed and feasible calibration methods to ensure the symmetric of the whole channel in the TDD mode.Firstly, we investigate the system performance for massive MIMO systems with RF mismatches. Ac-cording to the signal model of RF gain, we derive the closed-form expressions of the ergodic sum-rates to evaluate the impact of RF mismatches on the system performance, for zero-forcing (ZF) precoding and max-imal ratio transmit (MRT) precoding respectively. Analysis results show that the RF mismatches at the UEs only lead to a negligible performance loss, and the RF mismatches at the BS are the major reason to result in a severe system performance degradation. Both the amplitude and phase mismatches of RF gain at the BS will cause the great ergodic sum-rates loss. Therefore, there is no need to calibrate the RF mismatches at the UEs, but it is important and imperative to perform the reciprocity calibration at the BS. Furthermore, the analysis illustrates that the RF mismatc hes at the BS can be eliminated through the perfect partial calibration.Next, we address the reciprocity calibration for centralized massive MIMO systems. We propose the mutual coupling calibration methods for ZF precoding and MRT precoding respectively. The reciprocity cal-ibration is performed without the involvement of both extra hardware circuits and the UEs. By exploiting the strong mutual coupling between adjacent antennas, the BS can perform reciprocity calibration and compen-sate for the RF mismatches. Then, we build up the model of calibration error, analyze the complexity and give the closed-form expressions of ergodic sum-rates. Results show that, the calibration significantly reduced the performance loss, which is caused only by the amplitude mismatches at the BS and is irrelative with the phase mismatches. Simulation results show that, if we use the ZF calibration coefficients to compensate for the RF mismatches with MRT precoding, the ergodic sum-rates increase but approach a limit lower than the one with MRT calibration coefficients. Therefore, it is imperative to perform reciprocity calibration suitable for MRT precoding. Besides, comparing with the several methods in the least square (LS) framework, the mutual coupling method has low computational complexity.Additionally, we consider the reciprocity calibration for distributed large-scale MIMO systems. When the access points (APs) and UEs are equipped with single antenna, we extend the total least square (TLS) method to the case of partial calibration, where the calibration procedure is transparent for UEs. Then,we present the proof of the equivalence between the TLS method and the LS method. Moreover, we come up with a novel algorithm named as iterative coordinate descent (ICD) partial calibration method to avoid the eigenvalue decomposition required by LS method. The performance of the ICD method achieves almost the LS method and is much better than the Argos method. When the access points (APs) and UEs all have multiple antennas, the block diagonalization (BD) precoding is used by the APs to jointly transmit the signals. RF mismatches at the UEs still bring a severe inter-stream interference (ISI) and significant performance loss, although the IUI can be eliminated by the perfect partial calibration at the APs. We derive the optimization object function expression of the TLS method for the full calibration with BD precoding. Then, we propose a corresponding ICD full calibration method to avoid the high complexity of the TLS method. The ICD method converges very quickly and reaches to the performance of the TLS method.Finally, we investigate signal transmission for distributed large-scale MIMO systems based on partial calibration. When the UEs have multiple antennas, we use a low complex BD precoding method to transmit the downlink signals. To avoid the heavy overhead of the full calibration, we design a scheme for both the downlink and the uplink transmissions with the partial calibration. The ISI can be canceled out at the UEs through the minimum mean square error (MMSE) receiver in the downlink transmission. While in the uplink transmission, the APs can detect each data stream of every UE via the maximal ratio combining (MRC) receiver. Besides, the interference suppression precoding matrix at the APs can be used for both the downlink and uplink transmission, which need to be calculated only once. Our scheme is well performed and reduces the system overhead effectively. Besides, avoiding the channel fluctuation between APs, we propose a scheme to obtain the diversity gain, which is named diversity combination LS method. By performing the calibration with the combination of the calibration signals in different channel coherent time, our scheme improves the calibration performance and robustness.