Container-Based Cloud-Edge-Device Collaborative Multi-Service Data Processing in Distribution Network

Distribution network of smart grid imposes higher requirement on data processing of next-generation consumer electronics devices. Container-based cloud-edge-device collaboration improves quality of service (QoS) by flexible data offloading and parallel data processing. However, the matching among consumer electronics devices, servers, and containers, and container competition further complicates the problem. In this paper, we propose three-dimensional learning-matching-based joint selection algorithm of server and container for multi-service data processing. The formulated problem is firstly reduced to a two-dimensional problem by combination-based dimension reduction. Then, it is solved by the proposed dynamic learning-matching price rising mechanism based on service priority, data size, and upper confidence bound. By avoiding low-priority service preempting container resources of high-priority service, simulation results validate that the proposed algorithm has superior delay reduction performance.