毫米波小卫星星群作为“单颗大卫星”的替代方案，将在未来空间信息网络的建设中发挥重要作用。然而，组成星群的小卫星存储、计算等资源都受到不同程度的限制，如何对毫米波星群多维且受限的资源进行管理，提高星群整体的通信、计算和数据回传能力，是课题组重点关注的问题。针对毫米波星群中存在的星间链路干扰问题，课题组提出了基于联合波束偏置的波束域干扰抑制方案，提高了星群的星间通信能力；为了降低毫米波星群的数据回传时延和丢包率，课题组提出了基于深度增强学习的数据回传方案，提高了星群的数据回传能力；星上处理是未来毫米波星群承担的重要任务，课题组提出了基于计算迁移的任务调度算法，对星群计算资源进行管理，提高星群计算能力。最后，课题组设计了基于NS3的半实物仿真方案，以更贴近真实的仿真环境对毫米波星群进行研究。初步研究成果表明课题组提出的方案具有较高的可行性，能够为我国未来毫米波星群的发展提供理论依据和数据参考。

As an alternative of satellites with complex functions and large size, millimeter-wave small satellite swarm will play an important role on the construction of space information network. However, small satellites have limited storage space, computing resource and so on. So our concern now focuses on the management of limited resources to improve the ability of communication, computing and data backhaul of millimeter-wave small satellite swarm. Facing to the existing of mutual interference among different inter-satellite links, a novel beam domain-based interference management scheme is proposed to maximize the system sum rate. In order to effectively reduce the average delay and mitigate packet dropped ratio in the process of data backhaul, we proposed a deep reinforcement learning based uplink scheduling scheme for satellite swarm. As the key mission of future satellite swarm, we attach importance to on-board processing of satellite swarm and propose a computational migration based computing tasks scheduling scheme. Finally, a NS3 platform based semi-physical simulation scheme is proposed to verify the feasibility of the proposed schemes. Preliminary results demonstrate that our schemes have high feasibility and can provide theoretical reference and data support for further improvement of millimeter-wave small satellite swarm.