A Machine Learning Approach for offload Decision Making in Mobile Cloud Computing

Mobile cloud computing encompasses a computational offloading framework that helps deploy the compute-intensive task to the remote server to save energy and increase the mobile device's performance. A decision engine is a significant component in the offloading framework, which helps decide when to offload the task to the remote or cloud server. The decision engine's accuracy should be high for the flawless execution of the application during the offloading process. A technique has been proposed by performing a stack ensemble approach on machine learning techniques like the Gaussian approach, multi-layer perceptron, k-nearest neighbors, and linear regression. It considers the various dynamics of the environment like task size, bandwidth, device battery, and device mobility. The proposed model performs better than other decision-making algorithms in terms of execution time and CPU utilization and achieves higher accuracy in making decisions while offloading the compute-intensive task to the remote server.