Content uploaded by Maryam Arabshahi
Author content
All content in this area was uploaded by Maryam Arabshahi on Jun 10, 2024
Content may be subject to copyright.
Optimizing Public Transit Networks:
A 5G/6G Approach to Smart Trac Solutions
Maryam Arabshahi 1, Sanket Partani 2
1German Research Center for Articial Intelligence (DFKI)
2RPTU University of Kaiserslautern-Landau
Abstract
Public transportaon (PT) systems play a vital role in urban mobility, yet the lack of synchronizaon
between trip planning resources oen leads to user frustraon and potenal shis towards private
vehicle usage, exacerbang trac congeson and environmental concerns. This paper proposes
leveraging 5G/6G networks to enhance communicaon within PT networks, addressing issues of
metable accuracy, route management, and environmental monitoring. Through connuous data
transmission from vehicles to the mobile communicaon network, aided byAI algorithms in Network
Data Analycs Funcon (NWDAF) precise arrival me predicons, real-me route adjustments, and
environmental data capture are facilitated. This innovave approach aims to improve the reliability
and accessibility of PT, ulmately fostering sustainable urban development and enhancing societal
well-being.
Introduction
Public transportaon services in cies have mulple methods for planning trips - mobile applicaons,
paper metables at bus stops, digital displays indicang arrival mes, or mapping services like Google
Maps. However, these resources oen lack synchronizaon, leading to frustraon among users.
Discrepancies may arise between planned trip details and real-me bus arrivals, parcularly when
disrupons such as road construcons occur, temporarily rendering certain bus stops inaccessible.
Unfortunately, these changes are not promptly reected in mapping applicaons or ulity-developed
apps. Previous studies have indicated that users experience more dissasfacon from unexpected
delays compared to ancipated delays [5]. This can potenally steer them towards private vehicle
usage instead of PT, which in turn exacerbates trac congeson, extends commute mes, and ele-
vates levels of air polluon, contribung to environmental degradaon and public health concerns
[2]. Moreover, trac congeson and polluon entail economic ramicaons such as producvity
losses, increased fuel consumpon, and infrastructural damage [9]. The next generaon of wireless
communicaon technologies, i.e., B5G/6G, are being developed with sustainability at its core. This
implies that these networks should not only enhance communicaon but also catalyze sustainable
development across various sectors. NWDAF, introduced as a pivotal component of the 5G Core
Network (CN) architecture, includes advanced data analycs capabilies to meet the demands of
emerging applicaons and services. This 3GPP standardized method collects data from across the
communicaon system, enabling comprehensive analysis and seamlessly integrates AI/ML models.
We aim to leverage these technologies to opmize PT services in any urban region. By providing
real-me and accurate arrival me predicons to the users, we aim to reduce the inconsistencies
and consequently, provide the user with beer planning, travelling and re-scheduling opons.
Figure 1. NWDAF in 5G/6G Core monitors the PT services
Methodology
Data Collecon:
A boom-up approach is used to predict the arrival mes of PT services. This approach primarily
entails gathering vehicular trac informaon such as posion, speed, etc. from all (acve)
connected vehicles, including PT services in the city.
Predicve Analycs with NWDAF:
With the availability of real-me data at the CN, NWDAF is leveraged to predict changes in PT
schedule for each stop in the city and subsequently, transmit this informaon to the users.
NWDAF can accurately predict bus arrival mes (a(t)) by factoring in the current coordinates
(C(t)) of the bus at me t, constant minimum travel me between consecuve stops (m) and any
delays from the trac condions (T(t)) in the city – signalized juncons, trac density per lane
during peak/o-peak hours, etc. (Formula 1)
a(t) = f(C(t), m, T (t)) (1)
Graph-Based Trac Predicon:
Modelling and predicng trac behaviour has always been a complex task. We propose to
predict the trac condions by realizing the city roads as a graph, with juncons as nodes and
streets as arcs. It can be further divided into manageable sub-graphs by clustering nodes based
on their coordinates (see Figure 2). Based on the latest informaon from all connected vehicles,
the me of the day and the locaon of dierent buildings in the city, trac density on all arcs can
be predicted for each sub-graph and consequently, any delays in the PT schedule can be
computed.
Acknowledgemnet
The authors acknowledge the nancial support by the German Federal Ministry for Educaon and
Research (BMBF) within the project Open6GHub {16KISK003K}.
Results
Our experimentaon focused on simulang trac paerns within Kaiserslautern, a representave
city in Germany, during one hour of the morning rush hour. To conduct our analysis, we employed
SUMO (Simulaon of Urban Mobility) [6], a widely-ulized tool for urban mobility simulaon. Ad-
dionally, we ulized data from OpenStreetMapato construct a realisc use-case scenario. Our
simulaon incorporated ve disnct bus lines traversing a network of 50 bus stops within the city.
Through rigorous examinaon of the simulaon output data, we observed signicant disparies in
me delays experienced across the various bus stops. Notably, these delays ranged from a minimum
of -3301.0 seconds to a maximum of 2699.0 seconds. To provide a comprehensive understanding
of the observed delays, we computed the average delay me for each bus stop. The results, illus-
trated in Figure 3, highlight the distribuon of delay mes across the network. Notably, the mean
delay across all bus stops was calculated to be 42.1 seconds. These ndings underscore the inherent
variability in modelling transit performance within urban environments and provide valuable insights
into potenal areas for opmizaon.
Figure 2. Clustered nodes in Kaiserslautern, Germany using K-Means
Figure 3. Bus arrival me delay at 50 bus stops.
Conclusion and Future Work
This paper presents an innovave approach ulizing 5G/B5G/6G technology to enhance PT com-
municaon networks. This is sll a work in progress that aims to address crical challenges such
as RTI accuracy, route planning and management, and environmental monitoring. As part of future
work, we will focus on several key areas. Firstly, we will connue simulang the trac model in the
example city and implement various mobility scenarios. By applying AI algorithms, we aim to idenfy
changes in PT behavior and dynamically adjust routes and schedules to opmize eciency. Addi-
onally, we plan to ulize AI algorithms to idenfy mul-modal hotspots in the city in real-me. This
informaon, combined with data collected from PT by NWDAF, will enable city management to not
only opmize routes and stop locaons, but also provide more modes for travelling, thereby, improv-
ing overall urban mobility. Furthermore, by leveraging insights into PT behavior and city hotspots,
Mobile Network Operators (MNOs) can opmize network slicing and resource allocaon, ensuring
service connuity and improving user experience.
References
[1] American Public Health Associaon et al. The hidden health costs of transportaon. American Public Health Associaon, 2010.
[2] Linda Bailey, Patricia L Mokhtarian, and Andrew Lile. The broader connecon between public transportaon, energy conservaon and
greenhouse gas reducon. Technical report, ICF Internaonal Fairfax, VA, 2008.
[3] Candace Brakewood and Kari Watkins. A literature review of the passenger benets of real-me transit informaon. Transport Reviews,
39(3):327–356, 2019.
[4] Teije Gorris. CIVITAS INSIGHT: Real-me informaon for public transport, 2020.
[5] Hiroyuki Iseki and Brian D. Taylor. Not all transfers are created equal: Towards a framework relang transfer connecvity to travel
behaviour. Transport Reviews, 29(6):777–800, 2009.
[6] Pablo Alvarez Lopez, Michael Behrisch, Laura Bieker-Walz, Jakob Erdmann, Yun-Pang Flöeröd, Robert Hilbrich, Leonhard Lücken,
Johannes Rummel, Peter Wagner, and Evamarie Wießner. Microscopic trac simulaon using sumo. In The 21st IEEE Internaonal
Conference on Intelligent Transportaon Systems. IEEE, 2018.
[7] Sarah Jo Peterson. The Transportaon Research Board, 1920⬔ 2020: Everyone Interested Is Invited. Naonal Academies Press, 2019.
[8] Ramjee Prasad and Rute C Soa. 6G Visions for a Sustainable and People-centric Future: From Communicaons to Services, the CONASENSE
Perspecve. CRC Press, 2023.
[9] David Schrank, Tim Lomax, and S Turner. Urban mobility report texas transportaon instute. Texas: Texas Transportaon Instute, 2009.
[10] Salih Sevgican, Meriç Turan, Kerim Gökarslan, H. Birkan Yilmaz, and Tuna Tugcu. Intelligent network data analycs funcon in 5g cellular
networks using machine learning. Journal of Communicaons and Networks, 22(3):269–280, 2020.
ahps://www.openstreetmap.org/copyright
maryam.arabshahi@dfki.de partani@eit.uni-kl.de https://www.open6ghub.de https://www.dfki.de