COMPUTATION TASK OFFLOADING IN MOBILE EDGE COMPUTING USING DEEP LEARNING

Abstract

Today, in the rapidly evolving world of technology and the internet, no matter where you go, you expect computationally intensive tasks to be completed with less power consumption and minimum delay. This can be achieved by expanding cloud computing capabilities as far as possible from the cell tower. Mobile edge computing (MEC) enables mobile computing near mobile devices, allowing procedures to be carried out at the core network, centralized headquarters, and various network aggregation points. In this paper, we use a deep learning algorithm to make offloading decisions when multiple tasks are executing simultaneously in a cellular network or on a single user equipment (UE). This deep learning algorithm uses the Q-learning method and improves resource management in MEC. The proposed algorithm analyzes CPU utilization requirements for a particular task to determine which tasks must be delegated to the edge server, minimizing power consumption and execution delay. These tasks may include IoT applications, image recognition, image processing, interactive gaming, web browsing, video streaming, smartphones, tablets, robots, and drones.