边缘智能计算协同机理与智能资源调度算法研究

针对现有边缘计算体系不能较好适应B5G/6G网络大数据、智能化的需求，课题研究边缘智能计算体系——基于分布式边缘计算服务架构为基础，建立云-边-端协同智能计算体系。探索支撑云-边-端协同智能计算体系的两大基本机制——协同机制和资源感知机制，研究资源状态感知方法、全局信息感知方法和分布式协同计算方法，研究如何刻画和实现边缘计算协同及其协同性能增益，研究融合全局资源共享和感知的在线、动态资源调度问题，上述研究可为边缘计算系统的网络结构设计提供理论依据。进一步，研究两类资源调度方法——云-边-端协同智能计算资源调度和多维资源联合调度，解决由于边缘计算节点分布不均衡和计算能力差异、资源的多维度特性和有限性，业务QoS需求的多样性、业务时空分布不均的动态变化特性、计算节点协同关系的复杂性带来的资源调度问题。本课题对推动边缘智能计算理论研究和实际应用具有重大意义和应用价值。

Considering the existing edge computing system cannot support the big data and intelligence demanded in the future B5G or 6G networks, this project emphasizes on the research of the intelligent edge computing system. Specifically, we first build a cloud-edge-device collaborative intelligent computing system based on the distributed edge computing service architecture. Then, to support the cloud-edge-device collaborative intelligent computing system, we explore the collaboration and resource awareness mechanism and research on the resource state-aware, global information-aware, and distributed collaborative computing methods, which can help find out how much collaborative performance gain can be obtained by edge computing collaboration and how to realize the collaboration gain. We study how to improve the performance of the online and dynamic resource scheduling based on the global resource sharing and awareness to provide a theoretical basis for the network architecture design of edge computing systems. Besides, we design two resource scheduling methods including cloud-edge-device collaborative intelligent computing resource scheduling and multi-dimensional resource joint scheduling to overcome the challenges brought by the different computing capacity and the imbalanced distribution of edge computing nodes, the multi-dimensional and finite characteristics of resources, the diversified QoS requirements and the dynamic uneven distribution in time and space of users and the complexity of the collaborative computing nodes. This research will provide theoretical and technical support to promote the theoretical research and practical application of Intelligent Edge Computing.