MCG: Mobility-Aware Computation Offloading in Edge Using Weighted Majority Game

Edge computing plays a pivotal role in computation offloading at low latency. However, selecting the appropriate node to offload a computation is a challenge, especially when the user node is mobile. The problem can be stated as follows: (i) a mobile device has to offload a computation, and the user is moving, (ii) a set of edge/fog devices is available nearby the mobile device, then to select the suitable device to offload the computation. To address the challenge, this paper proposes an end-to-end mobility-aware computation offloading framework, MCG, which consists of: (i) a novel mobility prediction module that finds the user mobility pattern, (ii) selection of a set of edge/fog devices based on the predicted mobility and user's current location, (iii) selection of the high majority device from the set of edge/fog devices based on the resource availability and present load of the devices, and (iv) offloading the computation to the selected high majority device. The experimental results demonstrate that MCG outperforms existing mobility prediction modules in terms of accuracy, precision, and recall measures. The theoretical analysis and experimental results illustrate that MCG reduces the latency and power consumption of mobile device during offloading compared to existing offloading strategies.