Vehicular edge computing (VEC) empowers real-time applications in the autonomous vehicle (AV) domain by positioning edge servers closer to AVs. This proximity reduces latency and energy consumption for task processing. However, effectively managing the offloading of these tasks and allocating resources across dynamic vehicular environments poses significant challenges. Centralized strategies face scalability hurdles, while decentralized approaches often lack cooperative and co-ordinated mechanisms. To tackle these limitations, this paper introduces a novel decentralized framework for optimizing task management and network resource utilization in dynamic vehicular settings. This framework is equipped with a multi-agent deep reinforcement learning algorithm (MADRL) that makes intelligent task offloading decisions. The proposed algorithm considers the diverse computing capabilities of network entities and enhances energy efficiency without compromising latency or task completion rates. The simulation-based performance assessment demonstrates the effectiveness of this framework in reducing energy consumption and improving task completion rates in comparison to existing algorithms.
Decentralized Vehicular Edge Computing Framework for Energy-Efficient Task Coordination
24.06.2024
795743 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
DRL-driven zero-RIS assisted energy-efficient task offloading in vehicular edge computing networks
DOAJ | 2023
|