Advanced Spectrum Sensing Techniques for Robust Train-to-Train Wireless Communication Systems: A Comprehensive Evaluation

In the context of full automation of train operation, flexible train management, and improved traffic capacity, the Next Generation Train Control and Monitoring System (NG-TCMS) represents a key technology. The system will be based on wireless technologies at both the backbone and consist levels, thereby creating Wireless Train Backbone (WLTB) and Wireless Consist Network (WLCN) solutions. These solutions are being developed within the framework of several European projects. The chosen technology will have to guarantee robust train-to-train (T2T) communications between consists in the case of virtual coupling of two trains. In order to be able to detect any type of unwanted signal that may disturb wireless transmission, a specific mechanism based on spectrum sensing should be developed. To this end, this paper presents a comprehensive evaluation of advanced spectrum sensing techniques tailored for robust T2T wireless communication systems. Performance evaluation was carried out using real LTE and Wi-Fi 802.11ax signals in a doubly selective channel. By implementing spectrum sensing techniques on Software-Defined Radio (SDR) platforms, this study investigates the efficacy of eigenvalue-based methods under dynamic vehicular propagation conditions, including variable train speeds and distances between trains. The research identifies the most effective spectrum sensing technique among Maximum-Minimum Eigenvalue (MME), Energy to Maximum Eigenvalue (EME), Simplified Predicted Eigenvalue Threshold (SPET), Mean-to-Square Extreme Eigenvalue (MSEE), Maximum Eigenvalue to Geometric Mean (MEGM), and Covariance Absolute Value (CAV). The findings indicate that SPET demonstrates superior performance in various scenarios, making it a promising candidate for real-world railway communication systems.