Data-Driven Analysis of Multi-Channel Router Operation in Railway Communication : A Machine Learning Approach
Estifanos, Nardos Negussie (2024)
Estifanos, Nardos Negussie
2024
Tietojenkäsittelyopin maisteriohjelma - Master's Programme in Computer Science
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2024-12-31
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-2024122711688
https://urn.fi/URN:NBN:fi:tuni-2024122711688
Tiivistelmä
This thesis employs advanced data analysis techniques and machine learning to compare different implementations of Multi-Channel router operation modes in railway communications and identify the optimal mode, as well as assess its resilience under restricted network conditions, ultimately providing insight into the viability of utilizing existing public networks as an alternative to dedicated Future Railway Mobile Communication System (FRMCS) infrastructure.
The evolution of railway communication systems is witnessing a significant transition from GSM-R to FRMCS, driven by the growing demand for data capacity in modern railway operations. This shift presents unique challenges, particularly in Finland where GSM-R has already been decommissioned. The Digirail project, as an initiative tending to implementation of Finland’s ERTMS, set out to explore public mobile networks with Multi-Channel routers utilizing three Mobile Network Operators (MNOs) to meet FRMCS specifications while maintaining interoperability across Europe.
Initial measurements across Finland’s railway network in 2022 evaluated two primary operation modes: best quality (one network selected based on network metric) and packet replication (packet is sent through all networks, using the first one to arrive and discarding the rest), compared against both GSM-R (subset-93) and FRMCS draft SRS values. A subsequent investigation covering the Pasila-Imatra route (250 km) examined nine distinct implementations: five packet replication and four best quality modes from five different router vendors which, in no particular order, were Alstom, Funkwerk, Goodmill Systems, Hitachi, and Siemens.
This research utilizes industry standard approaches such as: exploratory data analysis (EDA), Multi-Criteria Decision Analysis (MCDA) techniques, including TOPSIS and CLUS-MCDA, combined with Artificial Intelligence for performance simulation under reduced carrier availability.
The results obtained demonstrate that Packet Replication 1 (Multi-Path TCP implementation of vendor 2) has consistent top performance with 100% compliance to the Digirail’s target value, and similarly ranking top from analytical analyses. Following this, using Packet Replication 1 data, the simulation models predicted performance values under reduced MNO setups, and, particularly with the omission of the 3rd MNO, better delay values were observed, which validates the robust and versatile nature of the selected mode.
The research contributes to the evolution of railway communication by providing data-driven insight on performance of varying implementations of packet replication and best quality modes. In addition, it demonstrates the feasibility of maintaining acceptable performance levels with reduced network availability. These insights help to understand the benefits of utilizing public mobile networks (4G/5G) for rapid and cost-effective transition to the future.
The evolution of railway communication systems is witnessing a significant transition from GSM-R to FRMCS, driven by the growing demand for data capacity in modern railway operations. This shift presents unique challenges, particularly in Finland where GSM-R has already been decommissioned. The Digirail project, as an initiative tending to implementation of Finland’s ERTMS, set out to explore public mobile networks with Multi-Channel routers utilizing three Mobile Network Operators (MNOs) to meet FRMCS specifications while maintaining interoperability across Europe.
Initial measurements across Finland’s railway network in 2022 evaluated two primary operation modes: best quality (one network selected based on network metric) and packet replication (packet is sent through all networks, using the first one to arrive and discarding the rest), compared against both GSM-R (subset-93) and FRMCS draft SRS values. A subsequent investigation covering the Pasila-Imatra route (250 km) examined nine distinct implementations: five packet replication and four best quality modes from five different router vendors which, in no particular order, were Alstom, Funkwerk, Goodmill Systems, Hitachi, and Siemens.
This research utilizes industry standard approaches such as: exploratory data analysis (EDA), Multi-Criteria Decision Analysis (MCDA) techniques, including TOPSIS and CLUS-MCDA, combined with Artificial Intelligence for performance simulation under reduced carrier availability.
The results obtained demonstrate that Packet Replication 1 (Multi-Path TCP implementation of vendor 2) has consistent top performance with 100% compliance to the Digirail’s target value, and similarly ranking top from analytical analyses. Following this, using Packet Replication 1 data, the simulation models predicted performance values under reduced MNO setups, and, particularly with the omission of the 3rd MNO, better delay values were observed, which validates the robust and versatile nature of the selected mode.
The research contributes to the evolution of railway communication by providing data-driven insight on performance of varying implementations of packet replication and best quality modes. In addition, it demonstrates the feasibility of maintaining acceptable performance levels with reduced network availability. These insights help to understand the benefits of utilizing public mobile networks (4G/5G) for rapid and cost-effective transition to the future.