Independent tasks scheduling of collaborative computation offloading for SDN-powered MEC on 6G networks

The implementation of mobile edge computing (MEC) and software-defined networking (SDN) over sixth-generation networks is a driving force in the future of cloud computing. It holds significant promise in addressing smart device (SD) resources and battery life limitations. To deal with the variety and resource constraints of SDs while making better use of network infrastructure, collaborative offloading has emerged as a viable technique for improving the ability to schedule independent tasks while alleviating the burden of restricted computation resources and network congestion. Network congestion is a common issue when a network node or link carries more data than it can manage. Existing works frequently overlook the impact of insufficient edge server capacity and network congestion. This paper primarily focuses on an SDN-powered MEC network that utilizes a full offloading policy, which completely offloads all tasks from the SD to the edge server and adopts a collaborative offloading of the MEC network when it is overloaded. The problem also takes into consideration when and to whom to offload the task. To bridge this gap, we first implement a collaborative offloading scheme among MEC servers based on the edge server's resources and neighbors' status to alleviate network congestion. It takes advantage of the computing capacities of edge servers deployed at the network's edge and the SDN controller's global view of the entire network. Then we devise a Deep Q-Network methodology to achieve near-optimal performance, and minimize the total execution time concerning deadline constraints. The experiments reveal that our proposed task scheduling of collaborative computation offloading algorithm can significantly minimize the total execution time more than the existing schemes.