The rapid development of High-Speed Railway (HSR) puts higher requirements on comprehensive perception and reliable transmission in tunnel scenarios. To realize efficient and reliable perception information transmission of HSR in the tunnel, we propose a multisensor perception communication system, which consists of an Access Point (AP) deployed on each carriage for perception information transmission and self-powered wireless sensors. The AP remote transmits the perception information through the leaky cable deployed in the tunnel. We construct an optimization problem for minimizing the transmission time of the whole system’s perception information in the multi-network system and the adjacent area of the carriage. A Multi-Agent Cooperation-based Deep Reinforcement Learning (MA-CDRL) algorithm is proposed to get the optimal scheduling strategy for reducing the transmission time. We construct the CDRL neural network for the algorithm to introduce the states of other APs, resulting in the system making more efficient transmission strategies. In the simulations, the proposed algorithm gets a better performance than the comparison algorithms and is verified in various dynamic HSR scenarios, such as different travel speeds and sensor distributions.

To align with the vision of future intelligent high-speed railway (HSR) networks, integrating sensor monitoring and remote communication are challenging for ensuring the lightweight of train equipment, high-quality transmissions, and dynamic interaction between monitoring and communication. In this paper, we propose a self-powered multisensor monitoring and communication integrated system in HSR. A low-power backscatter communication working framework of the self-powered monitoring system is designed in the monitoring network model, and a finite Gaussian mixture model (GMM) clustering method is used to analyze the communication cell coverage area in the communication network model. Aiming to minimize the total task completion time, we formulate a data monitoring and remote communication problem with the energy transfer constraint, data collection constraint, and transmission data rate constraint. As for the non-convex minimum time optimization problem, we develop a novel option-based hierarchical deep reinforcement learning (OHDRL) method to deal with the complex continuous variation characteristics of the monitoring and communication integrated HSR system. The system learns to select options at a high level, and the action is executed according to the policy of the selected option at a low level. This approach enables us to handle stochastic HSR environments, closed-loop policies, and goals in a temporal abstraction way. Numerical results reveal that the proposed algorithm for the integrated monitoring and communication HSR achieves a significantly higher reward and more stable learning performance than other algorithms in the literature.

In this paper, we study a self-supplied multi-sensor monitoring and communication integrated system in a high-speed railway (HSR), where each access point (AP) simultaneously transmits radio frequency (RF) energy signal to charge sensors, monitors the train working state via backscatter communication technology, and communicates remotely with the base station (BS). Aiming to minimize the total task completion time, we formulate a data monitoring and remote communication integrated optimization problem under energy transfer, data collection and transmission data rate constraints. To reduce the system task's complexity in a high-speed movement scenario, we propose a finite Gaussian mixture model (GMM) clustering method to analyze the communication handover area. For the complex action space in the highly-dynamic HSR communication handover environment, we develop a novel option-based hierarchical deep reinforcement learning (OHDRL) algorithm to deal with the sparse reward and non-stationary problem. An agent learns to select options at higher levels of decomposed subtasks, and the action is executed according to the internal policy of the selected option at a low level. Numerical results reveal that the proposed algorithm achieves a significantly higher reward and more stable learning performance than the traditional Deep Q-Network (DQN) and Double Deep Q-Network (DDQN) algorithms.