A Heterogeneous Cloud-Edge Collaborative Computing Architecture with Affinity-Based Workflow Scheduling and Resource Allocation for Internet-of-Things Applications

Cloud-edge collaborative computing (CECC) is a critical way to solve the real-time problems in the Medical Internet of Things (MIoT) and Industrial IoT (IIoT) applications. However, how to efficiently schedule the real-time and non real-time tasks to the cloud or edge servers that are configured with heterogeneous computing resources such as GPUs, NPUs and FGPAs remains a critical challenge. To address this challenge, this paper first defines the capability elements of CECC architecture, and then formally describes the workflow containerized tasks and heterogeneous computing resources of CECC system. Next, it proposes a heuristic task scheduling algorithm based on the affinity between tasks and nodes (physical machines or virtual machines) by matching the task resource requests with the node resource configurations. Later, it generates the initial mapping matrix between tasks and nodes with the affinity sorting result. And finally, it optimizes the mapping results under the constraints of limited resources and task dependency, which consequently generates an efficient scheduling scheme for the real-time IoT tasks in the CECC system. Experimental results demonstrate that the proposed algorithm can effectively increases the utilization efficiency of heterogeneous resources and improves the scheduling performance of real-time IoT tasks.