Zhiping Ouyang
ABSTRACT
In this paper, we construct a system with moving vehicles, road side units(RSUs) and a base station(BS) in Internet of Vehicles(IoVs). According to Nyquist theorem and Shannon formulation, we can get the energy consumption and execution time for task offloading. Considering the constraints of vehicles in the dynamic environment, task offloading can be modeled as a multiple objective optimization. Due to the lack of the cooperation in deep reinforcement learning and the unstable responses of multi-agent deep deterministic policy gradient(MADDPG) with increasing agent vehicles, we propose two imprvement algorithms MADDPG-MCMF and MCMF-Based MADDPG by combining MADDPG with minimum cost maximum flow(MCMF) in different manners so that they can overcome the shortages of MADDPG. MADDPG-MCMF uses MCMF as the task allocation part of MADDPG and MCMF-Based MADDPG applies MCMF in the reward funciton of MADDPG. Extensive numerical simulations show that the MCMF-Based MADDPG gets a good result in energy consumption and outperforms the MADDPG and MADDPG-MCMF algorithms in term of response latency.
- Md. Zahangir Alam, Abbas Jamalipour. 2022. Multi-agent DRL-based Hungarian algorithm(MADRLHA) for task offloading in multi-access edge computing Internet of Vehicles(IoVs). IEEE Trans. Wireless Commun. 21, 9 (September 2022), 7641–7652. https://doi.org/10.1109/TWC.2022.3160099Google Scholar
Digital Library
- Lu H, Gu C, Luo F, Optimization of lightweight task offloading strategy for mobile edge computing based on deep reinforcement learning. Future Gener. Comp. SY. 102 (July 2019), 847–861. https://doi.org/10.1016/j.future.2019.07.019Google ScholarDigital Library
- Wang C, Wang W, Li W, Adaptive compute offloading algorithm for metasystem based on deep reinforcement learning. Int. J. Pattern. Recongn. 36, 13 (November 2022), 1-28. https://doi.org/10.1142/S021800142252019XGoogle ScholarCross Ref
- Jiang B, Li K, Zhou B, Tao M and Chen Z. 2020. Deep Reinforcement Learning for Distributed Computation Offloading in Massive-user Mobile Edge Networks. The 12th International Conference on Wireless Communications and Signal Processing. IEEE, Nanjing, China, Article 1, 6 pages. https://doi.org/10.1109/WCSP49889.2020.9299723Google ScholarCross Ref
- Wan Y, Qin J, Ma Q, Multi-agent DRL-based data-driven approach for PEVs charging/discharging scheduling in smart grid. J. FRANKLIN I. 359 (January 2022), 1747–1767. https://doi.org/10.1016/j.jfranklin.2022.01.016Google ScholarCross Ref
- Yu Z, Gong Y, Gong S and Guo Y. Joint Task Offloading and Resource Allocation in UAV-Enabled Mobile Edge Computing. IEEE Internet Things J. 7, 4 (April 2020), 3147–3159. https://doi.org/10.1109/JIOT.2020.2965898Google ScholarCross Ref
Recommendations
Policy Adaptive Multi-agent Deep Deterministic Policy Gradient
PRIMA 2020: Principles and Practice of Multi-Agent SystemsAbstractWe propose a novel approach to address one aspect of the non-stationarity problem in multi-agent reinforcement learning (RL), where the other agents may alter their policies due to environment changes during execution. This violates the Markov ...
MADDPG: A task offloading algorithm based on multi-agent deep reinforcement learning in vehicle-edge computing
CNML '23: Proceedings of the 2023 International Conference on Communication Network and Machine LearningVehicle-edge computing has emerged as a new paradigm based on vehicular networking to provide computing resources for mobile vehicles. However, traditional binary offloading methods result in wastage of vehicle and server resources, as well as the ...
CoTEV: Trustworthy and Cooperative Task Execution in Internet of Vehicles
Due to the increasing number of service requests from the vehicles, the load at the road side units (RSUs) increases, which affects the delay-sensitive vehicle services. In Internet of Vehicles (IoV), the vehicles can communicate directly with other ...
Comments