IMPROVING RESPONSE TIME OF TASK AND ADABOOST CLASSIFIERS IN MOBILE FOG COMPUTING

Abstract

The application of computing resources through mobile devices (MDs) is called Mobile Computing; between cloud datacentres and devices, it is known as (Mobile) Fog Computing (MFC). We ran Cloudsim simulator to offload tasks in suitable Fog Devices (FDs), cloud or mobile. We stored the outputs of the simulator as a dataset with features and a target class. A target class is a device in which tasks are offloaded and features of tasks are authentication, confidentiality, integrity, availability, capacity, speed and cost. Decision Tree (DT), Random Forest (RF), Extra-trees and AdaBoost classifiers were classified based on attribute values and the plot of trees was drawn. According to the plot of these classifiers, we extracted each sequential condition from root to leaves and inserted it into the simulator. What these classifiers do is to improve the conditions that should be inserted in the corresponding section of the simulator. We improved the response time of offloading by Random Forest, Extra-trees and AdaBoost over Decision Tree.

Keywords

• fog computing
• decision tree classifier
• random forest classifier
• extra-trees classifier
• adaboost classifier
• offloading
• machine learning