Proposal and Reliability Evaluation of a Cloud-Based Railway Signalling System with k-out-of-n Logic for Ensuring Safety and Availability  
Author

Takashi Toyama

 

Co-Author(s)

Mamoru Ohara;  Masayuki Arai; Satoshi Fukumoto

 

Abstract Railway signalling systems require extremely high levels of safety and reliability to prevent hazardous events such as train collisions and derailments. Conventional signalling systems mainly rely on fail-safe dedicated hardware architectures to satisfy stringent safety requirements equivalent to Safety Integrity Level 4 (SIL4). However, these architectures involve high lifecycle and maintenance costs, leading to growing interest in
the application of cloud computing technologies and Commercial Off-The-Shelf (COTS) hardware to railway signalling systems. At the same time, cloud-based systems must maintain sufficient availability even under communication disturbances such as message loss and network delays. To address these challenges, we have previously proposed a cloud-based railway signalling architecture in which replicated central units are deployed across distributed data centers and field terminals determine control outputs using k-out-of-n logic. In this paper, we reintroduce the proposed architecture and focus on improving its availability evaluation method. We clarify the limitations of existing discrete-time Markov models and propose an enhanced model
considering temporary postponement of command determination caused by communication failures. Evaluation results suggest that the proposed model may provide a more realistic availability assessment than conventional models.

 

Keywords Railway Signalling, Cloud Computing, COTS, Reliability, Safety, k-out-of-n logic
   
    Article #:  RQD2026-167
 

Proceedings of 31st ISSAT International Conference on Reliability & Quality in Design
August 5-7, 2026