On Verify and Validate a Next Generation Automotive Communication
Network
a
Sebastian Brunthaler, Thomas Waas and Markus Kucera
OTH Regensburg, Regensburg, Germany
Keywords:
Automotive, Ethernet, In-car, Vehicle Communications, Vehicle Networks, Car-to-Car, Autonomous Driving.
Abstract:
As a result of the enormous growth in data traffic for autonomous driving, the conventional in-vehicle network
is no longer sufficient and requires new types of network concepts in a vehicle. This part of the automobile
is known as the next generation communication network. Since the new car-systems can be extended by
various services at any time, the network must adapt dynamically to new requirements wherever possible.
For example, data flow must be configured dynamically between new services. Also data rates will be much
higher in the future than today. This is one of the main reasons why we need to search for new technologies
for data transfer in vehicles. This is based on an in-vehicle ethernet network. The process of configuring
networks automatically has been discussed several times in recent years. One of the next steps is verifying and
validating the automatic configuration process during the development of the new communication network.
This research paper identifies several ways to ensure the automatically generated network configuration leads
to a secure system. To achieve that, other parts of the company’s enterprise IT architecture and network
technologies, the conventional vehicle network and other options for verification and validation are analysed.
1 INTRODUCTION
With the introduction of autonomous driving in road
traffic, vehicles must be able to process significantly
more data than before. For example a few years ago,
a car used to have one or two sensors, like cam-
eras, which used to transmit data through the system.
In the future, several cameras are expected to pro-
cess high-resolution data by self-learning ADAS (Ad-
vanced Driver-Assistance Systems) and autonomous
driving systems. Also, the resolution of images is
constantly increasing as the quality of images in-
creases. All these data has to be transmitted efficiently
via the new network and forwarded to their destina-
tion in real time, if required. This is necessary be-
cause a vehicle must be able to classify objects and
people within milliseconds. As an example, this is a
way to avoid accidents caused by incorrectly detected
elements. Therefore, this will be even more difficult
in the future since, according to a prediction, more
than 70% of the world’s population is going to live
in urban areas by 2050. Hence, the conurbations cre-
ate complex driving situations for autonomous driv-
ing due to an increasing number of traffic. It presents
a new dimension of influences and their complexity
a
OTH Regensburg, Continental Automotive GmbH
the automobile will have to cope with in the future.
The current vehicle network is not designed for the
enormously increasing requirements. Rigid bus sys-
tems such as CAN
1
, LIN
2
and FlexRay
3
are no longer
able to transmit the increased data volumes reliably
and in the time required. Additionally, it is impor-
tant to switch to a new vehicle network that has the
functionality to respond to dynamic changes in the
communications environment. For this, the Ethernet
technology would be chosen. However, this is not
going to replace the bus systems completely. The
main application area will be handling the commu-
nication between the central part of the network and
the connection of sensors, for instance, with high data
throughput. For illustration, figure 1 shows the evolu-
tion from a currently rigid network to a server archi-
tecture. In traditional architecture, each computing
unit is already configured statically in the beginning
and usually remains consistent. In a domain archi-
tecture on the other hand, these computing units are
usually consistent connected to each other with so-
called servers. Thus, they can be configured freely
and extended in their functionality. The next step is
1
https://ieeexplore.ieee.org/document/4678912
2
https://ieeexplore.ieee.org/document/1286317
3
https://ieeexplore.ieee.org/document/4677484
Brunthaler, S., Waas, T. and Kucera, M.
On Verify and Validate a Next Generation Automotive Communication Networka.
DOI: 10.5220/0008191601210127
In Proceedings of the 9th International Conference on Pervasive and Embedded Computing and Communication Systems (PECCS 2019), pages 121-127
ISBN: 978-989-758-385-8
Copyright
c
2019 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
121
a zone architecture, containing a central network of
servers. Sensors and actuators are connected to the
network via so-called zone ECUs. These in turn are
connected to the central computing units like servers,
which ultimately process the data. The final state of
in vehicle network is nearly the same as the zone ar-
chitecture. To the central powerful server units, hard-
ware could be added and associated services runs on
one of the server units. Unlike the zone architecture,
the zone controllers are no longer required. The last
two expansion stages of the E/E architecture (Electri-
cal/Electronic Architecture) enable a modular expan-
sion of a vehicle with various services and hardware
through Plug&Play. This is a key aspect in the de-
velopment of a next generation communication net-
work. Vehicles are expected to be more personalized
in the future resulting in a broad spectrum of config-
uration options. Therefore, variety of combinations
requires the automatic configuration of a network to
meet modularity requirements.(Carlson et al., 2019)
Due to the modularity of the system, current
projects such as the A3F project which is explained
in the section 2, aim at a service-oriented network
architecture. This means, applications are supposed
to be installed on a vehicle without having to physi-
cally visit a service point or the manufacturer of the
vehicle. In the future, this will take place through a
so-called App Store on the vehicle. Similar to smart-
phones, this includes a pool of applications to extend
the functionality of a car. The ability of vehicle hard-
ware extension through Plug&Play also implicates the
need of a network reconfiguration from a single unit
accessible to all components within the system. In or-
der to fully integrate and support the hardware func-
tionality into the new network from a software per-
spective, data streams, port rules and additional ser-
vices must be configured and provided for the de-
sired functionality. The A3F project follows the con-
cept of software-defined networking, similar to what
is known from software-defined networks
4
(SDN) or
Software Oriented Architecture
5
(SOA). This are con-
cepts about building a service-oriented architecture
managed from a central location. This principle can
be applied to the desired software extension of the
system. However, this approach also involves risks.
The number of different configurations possible in the
system increases due to software extensions and dif-
ferent applications. For example, 10 services were
available to extend a system with 5 server units, there
were already 1001 possible combinations.
(n + m 1)!
((n 1)! m!)
=
(5 + 10 1)!
((5 1)! 10!)
= 1001
4
https://ieeexplore.ieee.org/document/6994333
5
https://ieeexplore.ieee.org/document/4026892
Since far more than these are planned for the future,
system configuration can no longer be done for each
service individually evaluated and checked for possi-
ble errors in the end. Using SDN’s approach, config-
uration can be done from at least one central location.
As of now, there are several approaches to ensure a
verification and validation process, but no practicable
solution for such complexity.
Another problem is represented by technologies
like CAN, LIN and FlexRay mentioned earlier in this
paper. In the future, these technologies will reach
their limits in an attempt to meet the requirements
as described for a new vehicle network. The data
to be sent can raise up to several gigabit per second
and must be sent in real time from a source Elec-
tronic Control Unit (ECU) to the destination ECU in
a system. For communication within the car, there-
fore it is necessary to use a modern medium. Since
2010, several major automotive manufacturers have
been working on integrating Ethernet into modern au-
tomotive architectures. However, Ethernet should not
completely replace all bus systems but rather comple-
ment them in order to transfer large amounts of data
between the devices as efficiently as possible. But not
only the automobile manufacturers have researched
for the integration of Ethernet, also several universi-
ties and key players in the global automotive industry,
such as Continental, Bosch and Vector, are convinced
of the necessity of Ethernet in the automobile (Stein-
bach et al., 2011)(Bello, 2011)(Eisele, 2018).
2 RELATED WORK
Over the last two years, the A3F research group men-
tioned in section I has developed a network man-
agement system which independently of components
configures a vehicle network. The project is a cooper-
ation between Continental Automotive GmbH and the
University OTH Regensburg. The research group de-
cided to define three expansion stages for in-vehicle
networks in terms of autonomous driving. The last
two states are called Next Generation Networks, pos-
sible network concepts are displayed in figure 1 on
timeline starting in the year 2025.
The first stage is to configure a network at the end
of the manufacturing process line. This represents
the today standard in automotive manufacturing.
Secondly, the vehicle shall be reconfigured during
functionalities have been added. In this extension
the automatic process is used only if the vehicle
is in a safe state. Looking forward, there is still
a need for a precise definition of which state rep-
resents a safe state. An example for a save state
PECCS 2019 - 9th International Conference on Pervasive and Embedded Computing and Communication Systems
122
Figure 1: Future E/E Architecture (Carlson et al., 2019).
could be the idle state of a automobile, like park-
ing position. Though, this topic is not going to be
evaluated further in this paper.
The final expansion stage represents the future of
vehicle networks. Configuration can be carried
out at any time. Thus, the functionality of a ve-
hicle can be extended at any time. A conceivable
scenario would be a new configuration already
created and loaded in background in parallel to the
current configuration. So, the new configuration is
adopted into the network and its components but
only approved after successful parallel operation.
However, this requires more memory space and
computing performance in the components.
Currently the project focuses on stage 2 where
the vehicle has to reconfigure itself after changes in
the system feature functionalities by adding a service
or hardware. Figure 2 shows the concept of a man-
agement system which configured the network from
a central unit in the system. The central icon on
the illustration represents the network manager which
is able to use external input such as network topol-
ogy, message and service parameter from a manifest
into a new configuration for the corresponding vehi-
cle network. The automatically generated configura-
tion should provide usage independent from the vari-
ous communication network topologies, so that man-
ual intervention is no longer necessary. The develop-
ment of the central network management for the sys-
tem was one of the A3F project outcomes. A consid-
erable part of the network manager implementation
is documented in two theses at the University Of Ap-
plied Sciences in Regensburg and can be consulted for
more detailed information (Brunthaler, 2018)(Urban,
2018).
Figure 2: Central Networkmanager.
As mentioned earlier, work has been carried out
in recent years on the automated configuration of a
vehicle network using new vehicle components like
servers. By means of a demonstrator structure a simu-
lation can be realized. This structure has already been
implemented in the last few months. In the labora-
tory setup, a huge part of the network can be extended
and reconfigured with a variety of services without
the need of external intervention. To underline this,
some examples are the configuration of data streams
and port rules at the respective switches or terminals,
such as an ECU or sensor, as well as the provision of
new services in the system. The next step to achieve
the target of this project is to develop a process to ver-
ify and validate the automatically generated network
configuration. To evaluate the process of automati-
cally creating the configuration of a vehicle network,
different approaches can be used for. These are ex-
amined for their feasibility and effort. The following
table 1 shows the investigated topics.
Table 1: Concepts for verification and validation.
Technology Section Source
Independent Verification and Validation 2.1 (Open Alliance, 2017)(techopedia, 2019)
Code Validation concepts 2.2 IEEE(IEEE, 2012)(IEEEa, 2017)
System life cycle Processes 2.2 ISO(ISO, 2015)
Classification in test category 2.3 University
of Prague(Sobotka, 2017)
Automatic and Simulation In-the-loop Test 2.4 (Bock et al., 2008)
Mathematical proof of algorithm 2.5 (Ziegenbalg et al., 2016)
In the following sections, the work on the topics
listed in table 1 are explained in further detail. Only
the points for which broad searches were necessary
are listed.
2.1 Independent Verification and
Validation
Independent Verification And Validation (IV&V)
fully incorporates the concept of Verification And
Validation (V&V). The difference to classic V&V is
that IV&V is performed by a third party company not
involved in development. In this way, software and
hardware can be tested by third parties. The main ad-
vantage of independent testing is the accurate verifi-
cation of the opportunity to accurately verify the spec-
On Verify and Validate a Next Generation Automotive Communication Networka
123
ifications and functionality of the product to be tested
the accurate verification of the opportunity to accu-
rately verify the.(techopedia, 2019)(IEEEb, 2017)
In the A3F project Independent Software Verifica-
tion And Validation (ISV&V) which is part of IV&V
plays a major role. This is usually done according
to previously given methods. In this case, the IV&V
concept could look like this:
ISV&V Concept Planning
System Criticality Analysis
Identification of Critical Components
Selection of suitable methods and audit tools
Specification of Requirements
Verification for: Completeness,
Correctness, Testability
Verification of the Design
Verification of design and conformity between
software requirements and interfaces
Feasibility check, as well as internal and
external consistency
Verification of the Code
Analysis of code metrics
Compliance with code guidelines and
conformity testing
Completeness, Correctness, Consistency
Validation
Identification of faulty or error-prone
components/functions
Compliance with software and system require-
ments
Black- and white box testing
Within the A3F project and the automotive indus-
try, one of the best known institutions for indepen-
dent testing is the Open Alliance with their definition
of the TC1 to TC14 Technical Comittees. The Open
Alliance is a group of several companies interested
in the standardization and testing Ethernet technolo-
gie in modern vehicle networks. Therefore, new test
specifications are constantly being developed to en-
sure uniform protection of the new technology. These
tests are also defined by test models such as IV&V.
One of the specifications is the TC8. This document
contains several tests for electronic vehicle controls
which must be passed prior to installing the compo-
nents on the vehicle. The tests in TC8 are located on
the ISO/OSI layers one, two and three, known as the
data link and network layer. The tests are usually done
by independent test institutes according to the speci-
fications of the Open Alliance.(Open Alliance, 2017)
2.2 System Life Cycle Processes and
Code Validation
The concept of life cycle process is known from other
IT areas and software development. In general, this
demonstrates the applications V-cycle principle which
is shown in figure 3. This principle is based on
the constant monitoring of individual development
stages. In contrast to TC8, this is not only a final
check but a permanent one. This type of verification
and validation should be saved for the development
process of the next generation vehicle network, as it
allows an early detection of errors and malfunctions in
the system or software. Also, processes for code val-
idation should be integrated at this point. That way,
uniform standards can be guaranteed during develop-
ment. The recommended way of implementing this is
using of IEEE standardization (IEEE, 2012)(IEEEa,
2017), code guides and software patterns. The main
advantages of this are that errors can be found more
easily, as well as clearer and comprehensive code.
2.3 Define Classification in Test
Categories
To ensure that the process of automated configuration
creation delivers accurate results, not only the initial
generated configuration, but also the algorithms have
to be tested on its own. To achieve this, an inven-
tory of the test metrics for this type of system must
be made. One possibility is the methods contained in
the publication to verify and validate distributed auto-
motive systems(Sobotka, 2017) released test concept
to define a classification for the individual elements.
In Fig. 4 a model of the classification is represented
graphically. Thus, the implementation and execution
of tests can be guaranteed for all components which
were generated during configuration.
Figure 3: V-Cycle of development.
PECCS 2019 - 9th International Conference on Pervasive and Embedded Computing and Communication Systems
124
Figure 4: Classification for testing (Sobotka, 2017).
2.4 Automatic and Simulation
In-the-loop Test
In order to cope with the growing complexity, the
process of verification and validation must be auto-
mated to the extent possible. The concept Simulation
In-the-loop test is limited to the application layer in
the ISO/OSI
6
model, primarily. Firstly, simulations
and tests are performed for the individual services
to ensure they operate smoothly. This takes place
during development by testing the source code be-
fore it gets uploaded in the so-called appstore. In the
next step, automated tests are designed in the runtime
environment of an application. Here, the complex-
ity mentioned above plays a big role because not only
the system itself grows but also the simulations from
the in-the-loop tests quickly become very complex.
Therefore, they cannot be tested directly in the ve-
hicle in the required time due to the high number
of possible combinations of the various services al-
ready discussed. Hence, tests must be carried out
automatically in advance and their results must
be available in order to compare them with
the configuration created. For this purpose,
test tools such as Jenkins
7
, Ranorex
8
, Gating
9
6
https://ieeexplore.ieee.org/document/1094702
7
https://jenkins.io/
8
https://www.ranorex.com
9
https://gatling.io/
and Tsung
10
can be used. These are well suited for
the development process of various applications. The
application level testing approach appears promising
as it is already used in related areas such as Internet of
Things (IoT)(Murad et al., 2018) and Cloud Comput-
ing. In Cloud Computing, validation is performed at
the level of cloud services, so every service is tested
before it is deployed on a system (Sehgal and Bhatth,
2018).
One way to check the configuration is to transfer
an automatically generated configuration of the vehi-
cle network to a powerful backend which is able to
run the tests or compare with existing results.This ap-
proach could be realized using Over-The-Air Tech-
nology (OTA), for instance. In case of a positive eval-
uation, the configuration can be adopted in the net-
work. An example communication between vehicle
and backend is shown in figure 5. This procedure is
time-consuming and requires a powerful network and
backend, which is associated with high costs. In re-
cent times network expansions used to be a big topic.
The 5G technology mentioned here could provide the
bandwidth necessary for the network. This would en-
able the approach of transmission to a powerful back-
end, in general (Andrews et al., 2014).
2.5 Mathematical Proof of Algorithm
Under the assumption that a computing unit such as
an ECU is implemented correctly and provides cor-
rect results, it is still very difficult to verify the ac-
curacy of the algorithm for automated configuration
generation. Ideally, a theoretical proof for the cor-
rectness of this algorithm should be found. For this
purpose, each sub-function of the automatic config-
uration algorithm is checked separately. This proce-
dure is called a partial verification of correctness. The
next step is to check the entire algorithm for absolute
correctness (Ziegenbalg et al., 2016). Since the full
complexity of the network management functionality
is not yet known, a mathematical proof approach will
be difficult. Also disturbances by other network par-
ticipants are not yet excluded by a formal proof and
must be checked beforehand. Therefore, it is clear
that the mathematical proof is one of the most com-
plex operations for such an algorithm. One option
is the implementation of mathematical and software
checks to test the algorithm itself using the result as
a complete induction verification of the functionality.
A second possibility is a comprehensive brute force
test of the network. This way, possible anomalies can
be found after a system reconfiguration. This test ex-
amines a large number of cases in which errors can be
10
http://tsung.erlang-projects.org/
On Verify and Validate a Next Generation Automotive Communication Networka
125
Figure 5: Backend communication.
detected. Hence, the chance of detecting all errors is
high but there is no guarantee.
3 CHALLENGES OF TESTING
DYNAMIC NETWORKS
The greatest difficulty in testing the next generation
network lies in the variety of possible combinations
of the individual software and hardware components
that can be found in the system. Due to the modular
extension of a network, there are practically no limits
to the complexity of a system. This is a fundamen-
tal challenge for the approach of automated network
configuration at a central point in the network. How-
ever, the real challenge for the future is the validation
and verification of the created configuration.
3.1 Closely Coupling among Software,
Hardware and Network
Due to the different dependencies between the indi-
vidual components, these must be solved first in order
to run specific tests for application and device level in
the network. Further, before testing the functionality
of the overall system, the individual levels have to be
evaluated. Here, too, automation needs to be applied
to testing.
3.2 Communication Paths
As the network grows, so does the number of pos-
sible communication channels in the system through
which applications can exchange data. New account
points such as switches are emerging which need to be
configured to transfer data quickly and correctly. In
addition, mechanisms such as redundancy can create
complex problems for system verification. Checking
all these paths and their devices is a difficult task.
3.3 Automated Verification and
Validation
Automation is one of the most important issues in the
verification and validation of a new network. Only
by automating tests, it is possible to check the various
configurations. For this purpose, tools must be found
that not only test software automatically, but also ful-
fill test requirements such as the IV&V described in
section 2.1. This will be most likely an extension of
an existing test tool or a combination of several.
3.4 Security
Protection against unauthorized and unwanted mod-
ification or manipulation of an automatically created
network configuration also plays an important role in
verification and validation. If a configuration has been
unauthorizedly modified from outside this should also
be detected by software testing in the system. With
this, the system should be protected internally and
against external manipulation.
3.5 Safety
Not only security but also safety is an important as-
pect. In order to be able to determine exactly which
danger emanates from the new system, an assess-
ment must first be made according to the ASIL lev-
els. These define the dangers for the environment of
the vehicle. In the paper Sicherheit von autonomem
Fahren the different driving situations are considered
on the basis of ASIL and grouped into the differ-
ent subclasses (Wirth and Reif, 2018). Based on
the levels, various safeguards must be created which
will also have an influence on the network, such as
redundant and real-time communication. The term
safety essentially contains two components, func-
tional safety and safety of the intended functionality
(SOTIF). Safety and security in combination are seen
as a unit who’s requirements needs to be met by the
sytem at all times and must be fulfilled by the sys-
tem.(Schnieder and Hosse, 2019)
4 PROMISING APPROACHES
All the approaches presented are important for veri-
fication and validation of the next generation vehicle
network. Nevertheless, in the research area of Group
A3F, a prioritisation has emerged which will be pur-
sued further in the coming months. Primarily, this
will be a combination of Independent Verification and
Validation (IV&V) 2.1 and Software defined Network
PECCS 2019 - 9th International Conference on Pervasive and Embedded Computing and Communication Systems
126
and Architecture. With service-oriented architecture
and testing, this forms a promising verification and
validation approach.
5 WHERE TO NEXT
The next steps in this research project are the exten-
sion of the table to include concepts from other ar-
eas of corporate IT, as well as the detailed elaboration
of the concepts described in promising approaches 4
concepts. After the complete elaboration of the most
promising approaches, feasibility studies have to be
deployed. These have yet to be defined and imple-
mented. It would be conceivable to do feasibility
studies on the existing A3F project demonstrator in
order to obtain well-founded test results. Currently,
the biggest problem with the system is the complex-
ity of the new vehicle network which will be difficult
to estimate. One of the described process for testing
and designing the verification and validation concept
is not sufficient. Therefore, a combination of several
procedures must be put together to solve this problem.
REFERENCES
Ing. Jan Sobotka, Methods for Verification and Validation
of Automotive Distributed Systems, Czech Technical
University in Prague, August 2017.
Till Steinbach; Franz Korf; Thomas C. Schmidt, Real-time
Ethernet for Automotive Applications: A Solution for
Future In-Car Networks, Hamburg University of Ap-
plied Sciences, Germany, IEEE 2011.
Lucia Lo Bello, The case for Ethernet in Automotive Com-
munications, University of Catania, Italy, 2011.
IEEE, Inc. Standard for System and Software Verification
and Validation, IEEE 1012-2012, edition 2012.
IEEE, Inc. Standard for System, Software and Hardware
Verification and Validation, IEEE 1012-2016, edition
2017.
IEEE, Applying Standard Independent Verification and Val-
idation (IV&V) Techniques within an Agile Frame-
work: Is there a Compatibility Issue?, 2017 Annual
IEEE International Systems Conference (SysCon)
Open Alliance, OPEN Alliance Automotive Ethernet ECU
Test Specification, Open Alliance, v2.0 edition 2017.
ISO, Systems and software engineering System life cycle
processes, ISO/IEC/IEEE 15288, edition 2015.
Sebastian Brunthaler, Erweiterung eines Netzwerkmanager
zur automatischen Konfiguration der sich im Fahrzeug
befindenden Netzwerkger
¨
ate anhand verschiedener
Applikationsparameter, OTH Regensburg, 2018.
Daniel Urban, Untersuchung von Methoden zur Ein-
bindung von virtuellen Umgebungen in ein zuk
¨
unftiges
Ethernet-basiertes Fahrzeugnetzwerk, OTH Regens-
burg, 2018.
Jochen Ziegenbalg; Oliver Ziegenbalg; Bernd Ziegenbalg,
Korrektheit von Algorithmen Korrektheit von Com-
puterergebnissen, Algorithmen von Hammurapi bis
G
¨
odel, 2016.
Johannes Eisele, Zwei Drittel der Weltbev
¨
olkerung werden
2050 in St
¨
adten leben, Zeit Online 2018.
Thomas Bock; Markus Maurer; Franciscus Meel; Thomas
M
¨
uller, Vehicle in the Loop. In: ATZ Automobiltech Z
110 (1), S. 10–16. DOI: 10.1007/BF03221943.
G. Murad; A. Badarneh; A. Quscf; F. Almasalha, Software
Testing Techniques in IoT, S. 17-24. 2018 8th Interna-
tional Conference on Computer Science and Informa-
tion Technology (CSIT), Amman 2018
Dr. Lars Schnieder; Ren
´
e S. Hosse, Leitfaden Safety of the
Intended Functionality, Springer Vieweg, 2019
Techopedia, https://www.techopedia.com/definition/24836/
independent-verification-and-validation–iv&v
Thomas Erl, Service-Oriented Architecture. Concepts,
Technology, and Design, Prentice Hall PTR, Upper
Saddle River 2004
Eike Bj
¨
orn Schweißguth, Entwicklung und
Evaluierung eines SDN-gest
¨
utzten echtzeitf
¨
ahigen
Ger
¨
atenetzwerkes, Springer Vieweg, 2016
Naresh Sehgal; Pramod Chandra P. Bhatth, Cloud Comput-
ing, Concepts and Practices, Springer International
Publishing, 2018
Patrick Wirth; Monika Ulrike Reif, Sicherheit von au-
tonomem Fahren, Winterthur : ZHAW Z
¨
urcher
Hochschule f
¨
ur Angewandte Wissenschaften, 2018
Steve Carlson; Christopher Mash; Christoph Wechsler;
Helge Zinner; Olaf Grau; Natalie Wienckowski, Au-
tomotive Ethernet: Beyond 10 Gb/s Electrical PHYs
call for interest, IEEE 802 LMSC March 2019 Ple-
nary meeting in Vancouver, BC, Canada, 2019
Jeffrey G. Andrews; Stefano Buzzi; Wan Choi; Stephen
V. Hanly; Angel Lozano; Anthony C. K. Soong;
Jianzhong C. Zhang, What Will 5G Be?, IEEE Journal
on Selected Areas in Communications (IEEE Journal
on Selected Areas in Communications), 2014
On Verify and Validate a Next Generation Automotive Communication Networka
127