To improve the poor performance of distributed operation and non-scalability of centralized operation in traditional cell-free massive MIMO, we propose a cell-free distributed collaborative (CFDC) massive multiple-input multiple-output (MIMO) system based on a novel two-layer model to take advantages of the distributed cloud-edge-end collaborative architecture in beyond 5G (B5G) internet of things (IoT) environment to provide strong flexibility and scalability. We further ultilize the proposed CFDC massive MIMO system to support the low altitude three-dimensional (3-D) coverage scenario with unmanned aerial vehicles (UAVs), while accounting for 3-D Rician channel estimation, user-centric association and different scalable receiving schemes. Since coexisted UAVs and ground users (GUEs) cause greater interference, we ultilize user-centric association strategy and minimum-mean-square error (MMSE) channel state information (CSI) estimation to obtain the estimated CSI of UAVs and GUEs. Under the CFDC scenarios, scalable receiving schemes as maximum ratio combing (MRC), partial zero-forcing (P-ZF) and partial minimum-mean-square error (P-MMSE) can be performed at edge servers and the closed-form expressions for uplink spectral efficiency (SE) are derived. Based on the derived expressions, we propose an efficient power control algorithm by solving a multi-objective optimization problem (MOOP) between maximizing the average SE of UAVs and GUEs simultaneously with Deep Q-Network (DQN). Numerical results verify the accuracy of the derived closed-form expressions and the effectiveness of the coexisted UAVs and GUEs transmission scheme in CFDC massive MIMO systems. The SE analysis under various system parameters offers numerous flexibilities for system optimization.

翻译：为改善传统无小区大规模MIMO系统中分布式操作性能不佳与集中式操作可扩展性不足的问题，本文提出一种基于新型双层模型的无小区分布式协作大规模多输入多输出（CFDC massive MIMO）系统，利用后5G（B5G）物联网环境中分布式云-边-端协作架构的优势，提供强灵活性与可扩展性。进一步利用所提出的CFDC massive MIMO系统支持无人机（UAV）低空三维（3-D）覆盖场景，同时考虑三维莱斯信道估计、以用户为中心的关联及不同可扩展接收方案。由于共存的无人机与地面用户（GUE）导致更强干扰，我们采用以用户为中心的关联策略及最小均方误差（MMSE）信道状态信息（CSI）估计，获取无人机与地面用户的预估CSI。在CFDC场景下，边缘服务器可执行最大比合并（MRC）、部分迫零（P-ZF）与部分最小均方误差（P-MMSE）等可扩展接收方案，并推导出上行频谱效率（SE）的闭式表达式。基于所推导的表达式，我们提出一种高效功率控制算法，通过深度Q网络（DQN）求解同时最大化无人机与地面用户平均SE的多目标优化问题（MOOP）。数值结果验证了推导闭式表达式的准确性，以及所提共存无人机与地面用户传输方案在CFDC massive MIMO系统中的有效性。不同系统参数下的SE分析为系统优化提供了多种灵活性。