Evaluating Different Architectures of Automatic Train Operation

Ziyulong Wang*, Egidio Quaglietta, Rob M.P. Goverde

*Corresponding author for this work

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review


Automatic Train Operation (ATO) is well-known in urban railways and it gets increasing interest from mainline railways nowadays to improve capacity and punctuality. For the past few years, the European Union Agency for Railways (ERA) has been developing a set of technical specifications for ATO over ETCS, in which it defines a specific architecture between the trackside subsystem and the onboard subsystem. However, there is no clear scientific argumentation supporting the choice of this architecture. Therefore, this paper proposes and assesses different possible ATO architecture configurations to spotlight operational, technological, and business advantages/limitations of ATO over ETCS architecture from the ERA. The set of analyzed ATO architecture configurations are based on state-of-the-art architectures proposed in the literature for the related Connected Driver Advisory System (C-DAS). Results of the assessment highlight that different ATO configurations might have diverse advantages or limitations depending on the type of railway organization and the technological development of the existing railway signaling and communication equipment.
Original languageEnglish
Title of host publication9th International Conference on Railway Operations Modelling and Analysis (RailBeijing 2021)
Number of pages18
Publication statusPublished - 2021
EventRailBeijing 2021 - Beijing Jiaotong University, Beijing, China
Duration: 3 Nov 20217 Nov 2021


ConferenceRailBeijing 2021
Internet address


  • Automatic Train Operation
  • Connected Driver Advisory System
  • Train Path Envelope
  • Train Trajectory Generation
  • ATO over ETCS


Dive into the research topics of 'Evaluating Different Architectures of Automatic Train Operation'. Together they form a unique fingerprint.

Cite this