In recent decades, technological advances have led to the development of many
new ideas and areas that have made an important contribution to improving people's
quality of life. Autonomous driving is one area that has resulted from these advances.
While attempts have been made in the past to try and facilitate the realization of
the self-driving car, the available technologies have never met the needed
requirements. However, the recently introduced 5G is believed to be a promising
enabler for the fully autonomous vehicle.
The goal of this thesis was to investigate whether the 5G network from 3GPP
Release 15 is capable of providing the needed resources to fully support autonomous
driving and other C-V2X use-case families while simultaneously operating in
extremely demanding (network-wise) road traffic scenarios.
In the theoretical part of the thesis (in the first four chapters), we firstly
introduced some of the most relevant concepts in the autonomous vehicle paradigm,
which would later help us better understand the practical part. We begin by defining
the autonomous vehicle from a communication standpoint, and its communication
needs and services. We continue by listing some of the most important benefits of
autonomous vehicles, as well as some of the most troubling challenges that experts
face, varying from technological and environmental challenges to legislative and
philosophical ones. We also briefly present Vehicular Ad Hoc Networks, Intelligent
Transport Systems, Vehicle-to-Everything and how the further development of such
concepts would have a positive impact on humanity.
The fourth chapter discusses the C-V2X technology, beginning with a brief
overview of its evolution throughout history. It is a technology that uses the existing
cellular network to provide the vehicle-to-vehicle, vehicle-to-infrastructure, vehicleto-network, and vehicle-to-pedestrian connections.
Every new step towards a fully autonomous vehicle brings about new sets of
even more complicated challenges. In the very beginning, the two most relevant
technologies used for V2X were DSRC and the LTE-based V2X. Although the LTE-
22 Abstract
based V2X has multiple advantages compared to DSRC, they both got outclassed, in
terms of performance, once 5G NR-V2X was introduced along with the 5G cellular
network.
V2X applications, such as cooperative sensing and maneuvering, high-density
platooning and teleoperated driving, show hard-to-meet computing and
communication demands, well beyond what LTE and DSRC can provide. Most V2X
safety applications demand ultra-low latency (below 10 ms), ultra-high reliability (near
100%), and a high data rate (in the Gbps range). The 5G cellular network is the only
cellular technology advanced enough to be capable of reaching such performance
thresholds.
In addition, we presented the 5GAA roadmap which shows the results and
predictions in the key priority areas, such as the acceleration of the evolution of cellular
technologies from the LTE-based V2X towards the 5G NR-V2X.
The theoretical part ends by showing how 5G has enabled more C-V2X use cases
that were previously impossible to sustain with LTE as the underlying technology, as
well as how it has introduced newer, more complex use cases with higher
requirements. It quickly became clear that some of the C-V2X use cases are extremely
demanding on the network, to the point where a lot of doubts arose whether they would
be able to operate in real-life situations.
The theoretical research was complemented with a series of simulations in order
to monitor the packet delay and reliability of the simulated 5G network, which would
later be compared to the QoS requirements of the C-V2X use-case families. The
simulations were executed with the help of the OMNeT++ framework 5G-Sim-V2I/N
which enables to simulate 5G V2I/V2N use cases with applications comprising the
whole 5G user plane.
We have simulated two different road-traffic scenarios. In the first scenario, we
simulated a motorway where the cars drive at higher speeds, which can have a huge
effect on the network's performance. In the second scenario, the initiative was to
choose something completely different from the first one in order to observe different
results between two extremes. Hence, the choice for the second scenario was an urban
environment with heterogeneous road characteristics and buildings that act as
obstacles which interfere with the signals.
In both scenarios, two QoS parameters were measured for four different UDP
applications running in parallel, on multiple cars. The running parallel applications
included a V2X application, a VoIP call, a video stream, and a data download/upload,
while the measured KPIs included packet delay and reliability. The challenge was to
monitor the network performance throughout the simulation, and in the end compare
Abstract 23
the final results with the end-to-end latency and reliability requirements of some of the
most demanding C-V2X use-case families, which were said to have been enabled by
the 5G network. The objective was to find out for which of these C-V2X use-case
families the simulated network met the requirements.
We have simulated the 5G network from 3GPP Release 15 and have evaluated
our results in terms of end-to-end-latency and reliability; it quickly became clear that
the network did not meet the requirements needed to support such extreme use-case
families in the simulated scenarios.
These findings brought us to the conclusion that the simulated network is in need
of serious performance enhancements, in terms of lowering its latency and increasing
its reliability, both of which have been promised for future 3GPP releases
|
