Research On Transmission Schemes For Massive MIMO Based Low-Resolution ADCs

In the fifth generation mobile communication systems, massive multiple input multiple output (MIMO) system is widely regarded as a disruptive technology. By setting up multiple antennas at the transmitter or receiver, we can effectively exploit the space resource and improve the resource utilization in time and frequency dimensions. At the same time, the use of multiplexing and diversity technology will increase the transmission rate of information and reduce the error rate. However, a large number of antennas make the hardware design and implementation of massive MIMO system very complicated. Especially in the system, each receive antenna needs an analog to digital converter (ADC) unit, so a large number of antennas means the need for a large number of analog to digital converters. The exponential growth of the overhead and power consumption lead to high-resolution analog to digital converters becoming the main bottleneck of massive MIMO systems. One way to solve this problem is to use low-resolution ADCs. Low-resolution ADCs have many advantages, including the reduction of the circuit complexity and power loss as well as the improvement of the realizability. By using low-resolution ADCs, a mixed-ADC architecture is especially suitable for massive MIMO system. This thesis studies massive MIMO transmission technology based on low-resolution ADCs, especially the mixed-ADC structure and multi-level low-resolution ADCs.Firstly, we study the transmission technology of massive MIMO systems. The wireless communication environment and the basic principle of MIMO are described detailedly, including the fading characteristics of wireless channels, spatial multiplexing and spatial diversity. The MIMO transmission and MIMO channel are modeled and the channel capacity of SISO, SIMO, MISO and MIMO systems are calculated. We introduce channel estimation method, especially the channel estimation based on training sequence. The signal detection technologies in MIMO systems are introduced, including three kinds of linear detectors and four nonlinear detectors.Secondly, we introduce the mixed-ADC design scheme of massive MIMO system. A mixed-ADC structure, where most of the antennas are equipped with low-resolution ADCs, while a few antennas have full-resolution ADCs, are studied from the view of the practical application. We introduce the system model of the proposed scheme, including the uplink transmission model of multi user scenarios and mixed-ADC structure. Then we deduce a set of generalized Bayesian detectors, which are derived from Bayesian inference and developed into a family of detectors under a unified framework. To describe the design scheme of mixed-ADC, we prove the effect of quantization step on the performance of the system and advantages of mixed-ADC structure. We optimize the assignment scheme for the antennas used high-resolution ADCs based on mixed-ADC structure. The performance of MIMO system based mixed-ADC structure is evaluated by simulation and numerical results, which show that quantization step with optimal value, mixed-ADC structure and optiming the assignment scheme for the antennas used high-resolution ADCs can improve system performance.Finally, we propose an scheme of multi-level low-resolution ADCs in massive MIMO system, that is replacing 7 bit ADC with 1-3 bit low-resolution ADC in the mixed-ADC structure. System model is introduced, including uplink transmission model and multi-level ADC structure in multiple user scenarios. Then we study the SE(state evolution) equations, which are used to characterize the performance of Bayesian detector derived by GAMP algorithm, including state evolution analysis of generalized Bayesian detector and state evolution expression of three Bayesian detectors, namely DQ-optimal, PDQ-optimal and linear. Besides, we discuss application scheme of replacing high-resolution ADC with a number of low-resolution ADCs, especially replacing 7-bit ADC with 1-3 bit ADC. We present a power consumption model which shows that multi-level low-resolution ADCs has smaller power consumption. The simulation results show that the theoretical results of state evolution analysis and computer simulation results are basically identical. Furtherly, replacing high-resolution ADC with multi-level low-resolution ADCs can achieve better system performance, and reduce hardware cost and power consumption effectively.