Research On Perception Information Transmission Of High-speed Railway Internet Of Things Based On Deep Reinforcement Learning In Tunnel Scenario

As the most sustainable mode of transportation,High-Speed Railway(HSR)has become the main method of the national economy and a popular means of transportation,as well as the national key infrastructure and an important basic industry.Similarly,due to the rapid progress of information technology,intelligent technology has made breakthroughs,and human society will surely step into the era of intelligence after experiencing industrialization and information technology.With the rapid development of infrastructure and emerging technologies,intellectualization has become an inevitable trend in the future development of HSR.The HSR needs to adopt advanced technologies such as cloud computing,Internet of Things,artificial intelligence,and big data,and improve the modernization and intellectualization level of HSR through continuous technological innovation.We will promote the transformation and upgrading of high-speed railways into smart railways.In addition to the basic characteristics of high-speed movement,intelligent HSR also needs to have efficient safety guarantee ability,a powerful information management platform,and provide comfortable user services.Intelligent HSR needs to combine lots of advanced technologies and the fifth Generation(5G,5th Generation)train mobile communication technology to achieve the above characteristics.Moreover,the safety of HSR operation is the best important.To ensure the reliability and safety of HSR operation,intelligent HSR uses Io T technology to carry out real-time monitoring of different facilities and equipment in the HSR system,the travel status of HSR,and the surrounding environment,to realize timely discovery and effective treatment of emergencies,to improve the operation safety.The HSR Io T realizes the comprehensive perception of the travel state through perception technology,identification technology,positioning technology,network and computing technology,data processing technology,and intelligent decision technology.In addition,the perception technology builds a Wireless Sensor Network(WSN)by deploying a large number of wireless sensor devices to perceive the temperature,humidity,equipment wear degree,and other information during travel,and transmits the information reliably and effectively to the intelligent control center through a combination of wired and wireless communication to realize the monitoring,scheduling,and control of HSR.However,on the one hand,due to the complex travel scenes of the viaduct,open area,urban area,tunnel,U-shaped slot,and so on,the characteristics of wireless channels are different.Especially in the tunnel scenario,the wireless signal propagation mechanism is special because of its complex and special channel characteristics.On the other hand,the internal environment of HSR travel is also relatively complex,and the problems such as pedestrian movement and excessive occlusions also cause the perception data transmission channel to change with time.Most of the current research on perception network transmission is limited to the network’s throughput performance but ignores the timeliness of information,and does not take the overall transmission time as the main optimization objective.Therefore,designing a reasonable transmission strategy to minimize the overall transmission time of perception information is the focus of improving the safety of HSR travel.To solve the above problems,this paper research minimizing the transmission time of HSR Io T perception data in the tunnel scenario,a carriage is seen as a single network and uses the Access Point(AP)which sets in each carriage to forward the perception data to the remote Base Station(BS).Based on the Deep Reinforcement Learning algorithm,the algorithm is constructed and optimized to minimize the overall transmission time.Then,for the multi-network scenario,which is the multi-carriage scenario,the cooperative neural network model is designed by using the algorithm of Multi-Agent Deep Reinforcement Learning,and the states of other carriages are introduced as variables to improve the performance of the algorithm and further reduce the transmission time.The main contributions of this paper can be summarized as follows:1.Aiming at the timeliness of the perception data of the HSR Io T,the transmission system model of a single network in the tunnel is designed to realize such basic tasks as downlink energy transmission,control instruction transmission,uplink data return,and cross-zone switching by combining wireless energy transmission technology.The remote transmission of the perception data is carried out to the baseband processing unit in the tunnel through the AP.The deep Reinforcement Learning algorithm is used to find perceptual data transmission strategy.2.For the multi-network scenario,the Multi-Agent Deep Reinforcement Learning algorithm is designed by using the Markov decision process,and its state space,action space,reward function,and loss function are constructed.The adjacent carriage area mechanism is introduced,which means the perception information in this area can be transmitted by any AP of the adjacent two carriages.A cooperative neural network model is designed to improve efficiency and algorithm performance.3.For the above transmission strategy,through theoretical derivation,combined with the Deep Q Network algorithm(DQN)and Multi-Agent Deep Reinforcement Learning algorithm,simulation and comparison experiments were carried out based on Python.The stability and effectiveness of the proposed system and algorithm were verified by simulation results,and the optimal transmission strategy was obtained.The research provides new ideas and methods for improving the travel safety of HSR and reducing the transmission time of intelligent HSR Io T perception information.Meanwhile,it explores the performance optimization strategy for the combination of intelligent HSR and the next-generation mobile communication technology in the future.The content of innovation is to expand the existing research foundation and explore the future scenario,which lays a foundation for the subsequent research.