Digital twin-assisted and mobility-aware service migration in Mobile Edge Computing

Abstract

Mobile Edge Computing (MEC) is emerging as one of the key technologies to process massive amount of data at the edge of the network for upcoming 6G networks. In the paradigm of MEC, users can offload the computation-intensive and delay-sensitive tasks to the edge servers, since the capabilities of User Equipment (UE) cannot be not sufficient to meet different quality of service (QoS) requirements. However, due to the limited coverage of edge servers, users can move between multiple edge servers. In such cases, service migration is complicated which arises the challenge of where and when to migrate the service dynamically to maintain acceptable QoS. In addition, high uncertainty of the user mobility and different QoS requirements of the services are challenging. Instead of assigning each user to the nearest edge server, we propose a Digital Twin (DT)-assisted service migration approach to minimize the task completion time. The contributions in this paper are threefold: First, a DT edge network architecture is presented that uses the real-time and historical data to determine service migration strategy. Second, the DT classifies the computing tasks as delay-sensitive task and delay-insensitive tasks to meet the QoS requirements. Third, the DT predicts the mobility of users in the future based on the Hidden Markov Model (HMM) and studies the impact of reducing the cost of service migration. Thus, we propose an optimization algorithm to find optimal resource allocation at each edge server. The simulation results shows that our DT-assisted service migration model can achieve less delay and service migration rate compared with the traditional methods.