RADTO: A Resource-Aware and Dynamic Task Offloading Strategy for Mobile Edge Computing

With the rapid evolution of technology, the demand for mobile devices capable of executing computationally intensive and time-sensitive applications has surged. However, these devices are often constrained by limited storage, finite battery life, and insufficient processing power, making it difficult to meet such demands effectively. Task offloading, which involves transferring computational tasks to more capable infrastructures, has emerged as a promising solution to these challenges. In this paper, we propose a novel task offloading strategy based on Deep Reinforcement Learning (DRL) that dynamically accounts for the computational capacities of both user devices and servers. This approach aims to optimize execution time while adhering to energy consumption and bandwidth limitations. Experimental results demonstrate that the proposed strategy outperforms existing methods, achieving significantly lower execution times without violating energy and bandwidth constraints. Furthermore, the strategy incorporates partial offloading capabilities to enhance flexibility and maximize resource utilization.