- 期刊
- OpenAccess
Reinforcement Learning in Path Lifetime Routing Algorithm for VANETs
摘要
Continuous dynamic in vehicle movement creates a complex topology environment and connectivity relationship in Vehicular Ad-hoc Networks (VANETs). Consequently, overcoming longer transmission connections and reduced transmission reliability has been a significant problem. During the development of routing methods, fundamental research focuses on identifying and selecting paths with short distances or pulverization of information in environments with high traffic density to transmit packets successfully. As a result, transmission losses are minimized, and transport conditions are avoided; however, these techniques struggle to achieve more stable path routes in real-time environments because of sudden increases in traffic levels. Consequently, a model that effectively monitors the distance between vehicles and assists VANETs in selecting paths automatically and more accurately is proposed. To approximate and extend the lifetime of the communication path, we use an OpenAI Gym environment where a Reinforcement Learning (RL) agent can learn the route using the distance between cars as a model and define policies with higher lifetime rewards. The algorithm used a trained agent with the Proximal Policy Optimization (PPO) and Advanced Actor-Critic (A2C) algorithms in our environment. Applying machine learning on VANETs results in improved network efficiency, which offers real-time routing information in mobile environments. The simulation results show that RL-based routing extends the useful life of the route among the investigated methods between the origin and destination hosts.
延伸閱讀
- 李承駿（2014）。Imitation Learning Based on Inverse Reinforcement Learning〔碩士論文，國立中正大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0033-2110201614004364
- 戴宗明（2018）。Deep Reinforcement Learning Routing for Vehicular Ad-hoc Network〔碩士論文，國立臺灣師範大學〕。華藝線上圖書館。https://doi.org/10.6345/THE.NTNU.DCSIE.017.2018.B02
- 江義淵, 劉有順, & Jia, J. (2016). Parameter Selection in Reinforcement Learning. In 國立東華大學 (Ed.), TANET2016 臺灣網際網路研討會 (pp. 534-538-096). 國立東華大學. https://doi.org/10.6679/TANET.2016.96
- 嚴大益（2020）。Using Reinforcement Learning in assisting feature selection〔碩士論文，國立臺中科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0061-2808202014351100
- Mousavi, A., Araabi, B. N., & Ahmadabadi, M. N. (2014). Context Transfer in Reinforcement Learning Using Action-Value Functions. Computational Intelligence and Neuroscience, (2014), 265-274. https://doi.org/10.1155/2014/428567