大规模机器类通信信道估计与用户检测技术研究

针对mMTC大连接大规模MIMO系统中的信道估计和用户检测问题展开研究，在基站端配置低精度ADC，基于Turbo迭代思想，提出并联广义Turbo-MMV算法，以降低系统硬件开销和算法计算复杂度。为实现大连接大规模MIMO系统性能与硬件开销的折中，在基站端引入混合ADC架构，充分利用信道矩阵的联合稀疏特性，提出向量形式的广义Turbo-MMV算法，分析信道估计和用户检测的性能界。探索混合精度ADC大连接超大规模MIMO系统模型，提出基于子阵列的广义Turbo-MMV算法，在此基础上，利用深度学习的思想，设计基于模型驱动的信道估计方法，以改善信道估计的MSE性能和用户检测的漏检误检概率性能。本项目为低复杂度mMTC大连接大规模MIMO以及超大规模MIMO系统的研究提供相关技术支撑。

This project mainly conducts research on the channel estimation and user detection in massive machin-type communications (mMTC) massive MIMO systems. Low-resolution analog-to-digital converters (ADCs) are considered at the base station. Based on the Turbo iteration principle, a parallel generalized Turbo-multiple measurement vector (MMV) algorithm is proposed to reduce the hardware overhead and the computational complexity. In order to achieve the tradeoff between the performance of the massive connectivity with massive MIMO and the hardware cost, the mixed-ADC architecture is established at the base station, and the joint sparse characteristic of the channel matrix is fully studied. A vector-wise generalized Turbo-MMV algorithm is proposed to perform the channel estimation and user detection, and the corresponding performance bounds are given for theoretical analysis. Additionally, the massive connectivity with mixed-ADC ultra-large-scale MIMO system model is explored, and a generalized Turbo-MMV algorithm based on subarray is proposed. On this basis, a model-driven channel estimation framework is designed using the knowledge of deep learning to improve the MSE performance of channel estimation and the flase or missing probability performance of user detection. The research of this project will provide relevant technical supports for mMTC massive connectivity with large-scale MIMO and ultra-large-scale MIMO with low hardware cost and computational complexity.