Abstract—Mobile Edge Computing (MEC) is a technology that enables mobile devices to transfer computationally demanding tasks to nearby servers. MEC significantly lowers the local processing load by allowing a variety of complicated mobile devices tasks to be transferred to the network system’s edge so that they can be executed by the servers at the edge. In this research, we used a wirelessly powered MEC system, such that every Wireless Device (WDs) computational task is either completed locally or entirely offloaded to a MEC server. However, allocating resources for communication and computing in an edge-cloud environment efficiently is still difficult, though, particularly when there are several mobile devices and edge servers. So, an improved deep learning-based reinforcement model is suggested for Improved Deep Reinforcement Online Offloading (IDROO) as a scalable method to address this issue. This framework uses a deep neural network to learn the binary offloading decisions from experience and has additional output layers for resource allocation (transmission power and Central Processing Unit (CPU) frequency) along with the task offloading and it also optimizes the joint loss function that accounts for the offloading of tasks and efficiency of allocating resources. The simulation of the suggested model is done by taking number of users as 10, 20, 30, and the numerical result shows that the model can achieve near optimal performance in dynamically changing environmental conditions. Specifically, IDROO improves computation rate by approximately 18–25% and reduces energy consumption by around 15–20% compared to existing methods, while achieving faster convergence.
 
Keywords—task offloading, data computation, edge-cloud environment, bit-error, allocating resources, task offloading, Improved Deep Reinforcement Online Offloading (IDROO)
 
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).