The emergence of new communication technologies and the proliferation of the Internet of Things (IoT) have created a growing need for real-time processing of large and complex data. In this context, edge computing systems have been introduced as one of the main solutions for resource management and facilitating distributed processing. This research focuses on optimizing resource allocation and task offloading in distributed edge computing networks by developing and implementing a model based on deep reinforcement learning using the MADDPG algorithm. The proposed model intelligently manages the resources and assigns the processing tasks to the appropriate nodes within the network by utilizing the capabilities of deep learning and reinforcement learning. It is designed to minimize processing and data transmission delays while optimizing energy consumption. It demonstrates remarkable flexibility and adaptability under different environmental and network conditions. Simulation results show that the model effectively reduces energy consumption and significantly minimizes processing and transmission delays. In addition, performance eva‎luations in different scenarios, including variations in the number of vehicles and RSUs, show that the model seamlessly adapts to environmental changes and different network conditions, resulting in a significant improvement in system efficiency. By utilizing advanced deep reinforcement learning techniques, this research demonstrates the potential of intelligent resource management in distributed edge computing systems to significantly improve overall system performance. The implementation of this approach could play a transformative role in the development of intelligent networks and improve the efficiency of distributed processing systems.

علي فانيان

محمدحسين منشئي , عارف كريمي افشار