HRF-ExGB: Hybrid random forest-extreme gradient boosting for mobile edge computing

Abstract

The development of increasingly cutting-edge mobile apps like augmented reality, facial recognition, and natural language processing has been facilitated by the sharp rise in smartphone demand. The increased use of mobile devices like wireless sensors and wearable technology has led to a rapid increase in mobile applications. Due to the explosive growth of the Internet and distributed computing resources of edge devices in mobile edge computing (MEC), it is necessary to have a suitable controller to ensure effective utilization of distributed computing resources. However, the existing approaches can lead to more computation time, more consumption of energy, and a lack of security issues. To overcome these issues, this paper proposed a novel approach called Hybrid Random Forest-Extreme Gradient Boosting (HRF-XGBoost) to enhance the computation offloading and joint resource allocation predictions. In a wireless-powered multiuser MEC system, the starling murmuration optimization model is utilized to figure out the ideal task offloading options. XGBoost is combined with a random forest classifier to form an HRF-XGBoost architecture which is used to speed up the process while preserving the user's device's battery. An offloading method is created employing certain processes once the best computation offloading decision for Mobile Users (MUs) has been established. The experiment result shows that the method reduced system overhead and time complexity using the strategy of selecting fewer tasks alone by optimally eliminating other tasks. It optimizes the execution time even when the mobile user increases. The performance of the overall system can be greatly improved by our model compared to other existing techniques.