Cooperative Intelligent Transport Systems E-Book
126,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Fachliteratur
- Serie: ISTE Consignment
- Sprache: Englisch
The advent of the automated and connected vehicle will require the implementation of high-performance communication systems: Cooperative Intelligent Transport Systems (C-ITS). However, controlling and managing these C-ITS is complex. A number of points need to be jointly considered: 1) a high level of performance to guarantee the Quality of Service requirements of vehicular applications (latency, bandwidth, etc.); 2) a sufficient level of security to guarantee the correct operation of applications; and 3) the implementation of an architecture that guarantees interoperability between different communication systems.
In response to these issues, this book presents new solutions for the management and control of Intelligent and Cooperative Transport Systems. The proposed solutions have different objectives, ranging from increased safety to higher levels of performance and the implementation of new, more energyefficient mechanisms.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 513
Veröffentlichungsjahr: 2024
Ähnliche
Table of Contents
Cover
Table of Contents
Title Page
Copyright Page
Preface
PART 1: Introduction to Cooperative Intelligent Transport Systems
1 Local Interactions for Cooperative ITS: Opportunities and Constraints
1.1. Introduction
1.2. Ephemeral local interactions: concept and examples
1.3. Local interactions serving cooperative ITS
1.4. Role of infrastructure in cooperative ITS services
1.5. Conclusion and prospects
1.6. References
2 Evolution of Use Cases for Intelligent Transport Systems
2.1. Introduction
2.2. Vehicular communication technologies
2.3. Evolution of use cases
2.4. Challenges and future services of V2X
2.5. Conclusion
2.6. References
PART 2: Optimization of Data Transmission for Cooperative Intelligent Transport Systems
3 Towards an Optimization of Data Transmission in Cooperative Intelligent Transport Systems
3.1. Introduction
3.2. Context
3.3. Experimental evaluation of the performance of the C-ITS message broadcasting system
3.4. Discussion of the main causes
3.5. Recommendations and research avenues
3.6. Conclusion
3.7. Acknowledgments
3.8. References
4 Efficient Hybridization of C-ITS Communication Technologies
4.1. Introduction
4.2. Related works
4.3. Definition of a heterogeneous network architecture and design of a protocol stack
4.4. RL for selecting the mode of communication
4.5. Performance evaluation
4.6. Conclusion
4.7. References
5 Using SDN Technology to Control C-ITS: Towards Decentralized Approaches
5.1. Introduction
5.2. Context
5.3. Application of Blockchain to SDVN architectures
5.4. Optimization of Blockchain technology for SDVN architectures
5.5. Future research avenues
5.6. Conclusion
5.7. References
6 Application of Network Slicing in C-ITS Systems
6.1. Introduction
6.2. Vehicle-to-everything (V2X) communications
6.3. Presentation of V2X technologies
6.4. Network slicing for 5G-V2X
6.5. Conclusion
6.6. References
PART 3: New Approaches to Data Processing in Cooperative Intelligent Transport Systems
7 A Novel Cloud Approach for Connected Vehicles
7.1. Introduction
7.2. State of the art
7.3. The GeoVCDN approach
7.4. Analytical model
7.5. Evaluation
7.6. Simulation results for data utility
7.7. Use case study
7.8. Conclusion
7.9. Acknowledgment
7.10. References
8 Optimal Placement of Edge Servers in C-ITS Systems
8.1. Introduction
8.2. Context
8.3. State of the art
8.4. OptPlacement: efficient edge server placement
8.5. Conclusion
8.6. References
9 Risk Estimation: A Necessity for the Connected Autonomous Vehicle
9.1. Context and objectives
9.2. Estimation of risk local to the ego-vehicle: some existing metrics
9.3. Development of communication strategy to extend risk: CBL and CBL-G
9.4. Computation of cooperative risks: extended local risk and global risk
9.5. Impact of global risk and anticipation of risky situations
9.6. Discussion
9.7. Conclusion and prospects
9.8. References
10 Resilience of Collective Perception in C-ITS – Deep Multi-Agent Reinforcement Learning
10.1. Introduction
10.2. State of the art
10.3. Mathematical modeling of the cooperative driving environment
10.4. Multi-agent learning with DRL for selection and exchange of perception data
10.5. Simulations, results and evaluations
10.6. Conclusion
10.7. References
PART 4: Securing Cooperative Intelligent Transport Systems
11 Distance-Bounding Protocols
11.1. Introduction
11.2. Relations between threats for DB protocols
11.3. Overview of existing protocols
11.4. References
12 Context-Aware Security and Privacy as a Service for the Connected and Autonomous Vehicle
12.1. Introduction
12.2. Security, privacy and trust of connected and autonomous vehicle applications
12.3. Security and privacy architecture
12.4. Self-adaptive selection of network access technologies
12.5. Main research works to be conducted
12.6. Conclusion
12.7. References
13 Vehicular Wireless Communications: Risks and Detection of Attacks
13.1. Introduction
13.2. General characteristics of wireless communications for connected vehicles
13.3. Characteristics of wireless communications
13.4. Susceptibility of communications and risks incurred
13.5. Attack detection
13.6. Conclusion
13.7. References
List of Authors
Index
End User License Agreement
List of Tables
Chapter 2
Table 2.1. Comparison of ITS technologies
Table 2.2. The potential requirements for NR V2X (3GPP 2018)
Chapter 3
Table 3.1. The number of messages exchanged during the experiment
Table 3.2. Number of events displayed on smartphones
Table 3.3. The number of ITS-G5 and cellular messages received
Chapter 4
Table 4.1. Road traffic parameters
Table 4.2. Configuration of standards
Chapter 6
Table 6.1. Default values for AIFS and CW at the access layer level
Table 6.2. Connectivity requirements for V2X autonomous driving applications
Table 6.3. Simulation parameters
Table 6.4. Configuration of link parameters
Chapter 7
Table 7.1. Mechanism parameters
Table 7.2. Network parameters
Table 7.3. Communication parameters
Table 7.4. Dissemination parameters
Table 7.5. Probability of RSU
Table 7.6. Probability of segments
Table 7.7. Number of hops
Table 7.8. Computer configuration
Table 7.9. Variables and default values
Table 7.10. Network load – 411 vehicles
Table 7.11. Comparison of network load – vehicle range 150 m
Table 7.12. Comparison of network load – five RSUs
Table 7.13. Comparison of network load – RSU range 150 m
Table 7.14. Network load comparison – data update frequency
Table 7.15. Comparison of network freshness – vehicle range 150 m
Table 7.16. Comparison of network freshness – 267 vehicles
Table 7.17. Comparison of network freshness – RSU range 100 m
Table 7.18. Comparison of network freshness – eight RSUs
Table 7.19. Performance of approaches.
Chapter 8
Table 8.1. Notation
Table 8.2. Classification of RSUs
Table 8.3. Simulation parameters
Table 8.4. Results of edge server placement
Table 8.5. Placement results of other approaches
Chapter 10
Table 10.1. Communication parameters
Chapter 12
Table 12.1. Advantages of context-aware security
List of Illustrations
Chapter 1
Figure 1.1. Operation of Wi-Fi CSMA (IEEE 2009)
Figure 1.2. Structure of frames and operation of LTE-V2X semi-persistent sched...
Figure 1.3. Different modes of DENM message broadcasting.
Figure 1.4. Functioning of pseudonymous certificates.
Chapter 2
Figure 2.1. Vehicular network architecture.
Figure 2.2. Evolution of ITS technologies
Figure 2.3. Emergency brake warning.
Figure 2.4. Lane change warning.
Figure 2.5. Road works warning.
Figure 2.6. Traffic jam information.
Figure 2.7. Green light optimal speed advisory.
Figure 2.8. Bird’s eye view.
Figure 2.9. See-through.
Figure 2.10. Platoon management.
Chapter 3
Figure 3.1. Simplified view of SDVN architecture
Figure 3.2. DIRA network
Figure 3.3. Example of a real-time event on the Bordeaux Métropole network (se...
Figure 3.4. Location of C-ITS messages
Figure 3.5. DENMs and events sent by users
Figure 3.6. DENMs and events sent by the road operator
Figure 3.7. The cumulative number of signed and unsigned CAMs in the proposed ...
Figure 3.8. The size of CAMs exchanged in cluster architecture and C-ITS archi...
Chapter 4
Figure 4.1. The proposed hybrid vehicular communication architecture.
Figure 4.2. Location of the hybrid communication layer in the protocol stack o...
Figure 4.3. Different representations of the environment
Figure 4.4. Vector representation of reception evaluation.
Figure 4.5. Platoon scenario on a highway.
Figure 4.6. Results for agents impacted by congestion
Figure 4.7. Results for agents not impacted by congestion
Figure 4.8. Comparison of RAT selection approaches
Figure 4.9. Percentage of messages duplicated at reception between the two hyb...
Chapter 5
Figure 5.1. Simplified view of SDVN architecture.
Figure 5.2. Simplified view of Blockchain architecture.
Figure 5.3. Example of advanced Blockchain architecture intended for SDVN netw...
Chapter 6
Figure 6.1. V2X communication types.
Figure 6.2. V2X communications.
Figure 6.3. Choice of spectrum in the United States and in the EU (Canis and G...
Figure 6.4. Frequency allocation for ETSI ITS-G5 in the EU
Figure 6.5. ITS-G5 protocol stack including DCC functionalities.
Figure 6.6. High-level architecture of the ITS network
Figure 6.7. C-V2X operating modes.
Figure 6.8. LTE-V2X communication. Mode 3: in coverage and mode 4: out of cove...
Figure 6.9. 5G system architecture for PC5- and Uu-based V2X communication: Al...
Figure 6.10. An example of NS V2X.
Figure 6.11. Management of the lifecycle of an NSI.
Figure 6.12. Network slicing architecture and NS instantiation for autonomous ...
Figure 6.13. Overview of the global architecture of the ITS-G5 system.
Figure 6.14. Protocol stack of the ITS station + EMS.
Figure 6.15. Average RAN latency (density: 450).
Figure 6.16. Average RAN latency (density: 600).
Figure 6.17. Average RAN latency (density: 600; HPS: 5%).
Figure 6.18. Overview of the deployed E2E architecture.
Figure 6.19. Average E2E latency (density: 600; HPS: 10%).
Chapter 7
Figure 7.1. Architecture of an ITS station
Figure 7.2. GeoVCDN exchanges.
Figure 7.3. ICN data in the CAM.
Figure 7.4. ICN packets
Figure 7.5. Reception of an interest.
Figure 7.6. Vehicle sending a CAM.
Figure 7.7. Reception of data.
Figure 7.8. RSU sending a CAM.
Figure 7.9. Diagram of the environment.
Figure 7.10. Dissemination scenario.
Figure 7.11. Segment.
Figure 7.12. Possible segment configurations.
Figure 7.13. Example of segment configurations.
Figure 7.14. Schematic representations of segment configurations.
Figure 7.15. NDN exchanges.
Figure 7.16. Exchanges in RENE.
Figure 7.17. Simulation extract.
Figure 7.18. Network load – number of vehicles – NDN versus VCDN....
Figure 7.19. Comparison of network load.
Figure 7.20. Network freshness.
Figure 7.21. Illustration of an intersection of the simulator.
Figure 7.22. Best pathfinding with one random RSU.
Figure 7.23. Best pathfinding for different numbers of RSUs (2, 8 and 11)
Chapter 8
Figure 8.1. Reference architecture for vehicular edge computing.
Figure 8.2. Example of edge server placement.
Figure 8.3. RSU locations and candidate locations for edge servers on the map ...
Figure 8.4. Roadside unit requests on the roads of Bordeaux.
Figure 8.5. Overview of the methodology used to evaluate “OptPlacement”....
Figure 8.6. The variation in vehicle density across the duration of the simula...
Figure 8.7. Locations of edge servers for different approaches: OptPlacement, ...
Figure 8.8. Results of the “latency” simulations: OptPlacement, Random-Random ...
Figure 8.9. Results of the “latency” simulations: OptPlacement, Top-K, K-means...
Figure 8.10. Results of abandoned tasks.
Chapter 9
Figure 9.1. Classification of driving situations with their criticality level ...
Figure 9.2. Modeling of an autonomous driving system drawing a parallel betwee...
Figure 9.3. Uncertainty modeling and computation of the distance of Gruyer (Di...
Figure 9.4. Principle of the CBL strategy for road communication node clusteri...
Figure 9.5. CBL-G, an extension of the CBL strategy by integrating RSUs as gat...
Figure 9.6. Dynamic clustering of a fleet of vehicles equipped with means of c...
Figure 9.7. Frontal dynamic local perception taking into account positioning u...
Figure 9.8. Local dynamic perception map from the point of view of branch i, d...
Figure 9.9. Generation of a queue of vehicles in the Pro-SiVIC platform for an...
Figure 9.10. Local risk, collision probability and TTC in the vehicle queue sc...
Figure 9.11. Local risk and global risk in the vehicle queue scenario
Figure 9.12. All the actors and attributes involved in the construction of new...
Figure 9.13. Proposal of an ontology of the risk and the components/elements/a...
Figure 9.14. Different input variables in the computation of risk metrics with...
Chapter 10
Figure 10.1. Basic architecture of collective perception
Figure 10.2. Main components of reinforcement learning.
Figure 10.3. Format of a CPM
Figure 10.4. The need for collective perception.
Figure 10.5. Image of the driving environment at an instant, t.
Figure 10.6. Circular field of view of an Ego CAV
Figure 10.7. Redundancy level of objects in the network: performance of the ba...
Figure 10.8. Variation in average CAV reward as a function of the learning epi...
Figure 10.9. Variation in average loss as a function of the learning episodes.
Figure 10.10. Comparison of the performance of the proposed method with the dy...
Figure 10.11. Level of consciousness generated by the CP in the network as a f...
Chapter 11
Figure 11.1. Relations between different threat models
Figure 11.2. General structure of DB protocols
Figure 11.3. Distance fraud attack against DB protocols using a PRF
Figure 11.4. Protocol-dependency graph. For a color version of this figure,see...
Chapter 12
Figure 12.1. Five types of vehicular communication.
Figure 12.2. Taxonomy of Internet of Vehicles (IoV) client applications.
Figure 12.3. Integration of our proposal into the C-ITS reference architecture...
Figure 12.4. Interactions between modules of the proposed architecture.
Figure 12.5. Cognitive loop
Figure 12.6. Autonomous cloud.
Figure 12.7. General view of the retained architecture.
Chapter 13
Figure 13.1. Mathematics and data
Figure 13.2. Learning methods
Guide
Cover Page
Table of Contents
Title Page
Copyright Page
Preface
Begin Reading
List of Authors
Index
WILEY END USER LICENSE AGREEMENT
Pages
iii
iv
xiii
1
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
273
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
SCIENCES
Networks and Communications, Field Director – Guy Pujolle
Network Management and Control, Subject Head – Francine Krief
Cooperative Intelligent Transport Systems
Control and Management
Coordinated by
Léo Mendiboure
First published 2024 in Great Britain and the United States by ISTE Ltd and John Wiley & Sons, Inc.
Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms and licenses issued by the CLA. Enquiries concerning reproduction outside these terms should be sent to the publishers at the undermentioned address:
ISTE Ltd27-37 St George’s RoadLondon SW19 4EUUK
www.iste.co.uk
John Wiley & Sons, Inc.111 River StreetHoboken, NJ 07030USA
www.wiley.com
© ISTE Ltd 2024The rights of Léo Mendiboure to be identified as the author of this work have been asserted by him in accordance with the Copyright, Designs and Patents Act 1988.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s), contributor(s) or editor(s) and do not necessarily reflect the views of ISTE Group.
Library of Congress Control Number: 2024936174
British Library Cataloguing-in-Publication DataA CIP record for this book is available from the British LibraryISBN 978-1-78945-180-1
ERC code:PE6 Computer Science and Informatics PE6_2 Distributed systems, parallel computing, sensor networks, cyber-physical systems PE6_7 Artificial intelligence, intelligent systems, natural language processingPE7 Systems and Communication Engineering PE7_8 Networks, e.g. communication networks and nodes, Internet of Things, sensor networks, networks of robots
Preface
Léo MENDIBOURE
ERENA, COSYS, Université Gustave Eiffel, Marne-la-Vallée, France
Automated vehicles could eventually reduce greenhouse gas emissions from transport and improve road safety and traffic flow. However, its safe implementation will require a high-performance communication system that enables vehicles to obtain information from their neighbors and from the infrastructure, such as lane changes, the presence of obstacles, diversions and extended perception.
Cooperative Intelligent Transport Systems (C-ITS), designed to enable these exchanges, will therefore play an essential role in the advent of the automated and connected vehicle. However, their deployment in a highly constrained and mobile environment could prove problematic in terms of guaranteeing quality of service (QoS), as well as the reliability and security of exchanges.
In response to these problems, this book presents new solutions for managing and controlling performance and security for C-ITS. After two introductory chapters presenting the concept of local interactions and the current development of use cases for C-ITS, this book will explore various ways of optimizing the control and management of C-ITS: hybridization of access technologies (cellular, ITS-G5), use of new tools (e.g. artificial intelligence), etc.
May 2024
PART 1Introduction to Cooperative Intelligent Transport Systems
1Local Interactions for Cooperative ITS: Opportunities and Constraints
Jean-Marie BONNIN1 and Christophe COUTURIER2
1 IMT Atlantique, Rennes, France
2 YoGoKo, Rennes, France
1.1. Introduction
Since the advent of wireless communication and its integration into consumer devices, the concept of intelligent environment or pervasive application has emerged. The ability to communicate with all objects in our immediate environment makes it possible to take information or trigger actions. Information collection feeds a context that applications take into account to adapt their behavior to the situation.
For this type of application, direct interaction with objects in the environment greatly facilitates matters, since it is not necessary to rely on a precise location and database to associate information (or objects) with this location. If we need to know the room temperature, all that is needed is to discover a temperature sensor and query it directly. Acquiring the same information when a server is in charge of collecting and exposing the building’s temperature data firstly implies discovering the server that has the information at its disposal, then dialoging with it to retrieve the temperature of the room in which the sensor is located, and finding consequently a way to determine that the location is necessary. The machinery to be put in place is much more complex and yet it seems more intuitive, as the majority of the industry has been built on this model.
The difficulty when it comes to building services on direct (we will also use the term “local”) interactions is that this implies standardizing the method of communication, the frequency (or frequencies) used and the message format. For road or city applications, it is therefore necessary to bring many actors to agreement, and to impose choices on the entire ecosystem.
Direct interactions are widely used today for service discovery; for example, Wi-Fi devices continuously scan all frequencies used in the 2.4 GHz and 5 GHz bands to determine if there is an access point available in the environment. The presence of such an access point in no way indicates that the terminal will know how to connect to it, and even in the case where it is able to connect, whether it will be able to obtain a service (an Internet connection). The other technology widely used on consumer terminals is Bluetooth. Again, part of the terminals expose their presence by regularly sending messages at a determined frequency. All Bluetooth devices in proximity are able to see these messages and determine whether or not they know the correspondent. They can then either establish a connection to perform the service (e.g. hands-free kit) by taking advantage of the keying material previously established during pairing, or ask to perform a pairing, which requires the user’s intervention.
It should be noted that even when the two correspondents know each other, whether via Wi-Fi or Bluetooth, the discovery and connection establishment time frame is far too long for services with significant time constraints. We will return to this when we examine how the specificities of ITS-G5 make it possible to significantly reduce the time required to exchange information for road safety-related services.
In the second part of this chapter, we will present the concept of ephemeral local interactions, giving examples of services based entirely (or partially) on this type of interaction. We will describe how the first services that will be deployed in the context of cooperative ITS (awareness) are based on this type of interaction and the advantages/constraints of this approach. Lastly, before concluding, we will explore the place infrastructure holds in the implementation of services, based on ephemeral local interactions.
1.2. Ephemeral local interactions: concept and examples
1.2.1. Examples of services using ephemeral local interactions
Once it has been established that the different devices in interaction use the same communication technology on a subset of frequencies well known to all, it is necessary to specify the type of interaction targeted. Indeed, we will focus more specifically on interactions where no connection is established. When two devices are in proximity, they can “see” each other because of their technology community; they have at their disposal information that is spontaneously sent by their peers without having to go through the time-consuming establishment of a connection. When the communication technology has a fairly short range, simply being in communication and seeing a device gives an indication of co-spatiality that can form an integral part of the service. Therefore, when a telephone receives an advertisement on one of the three Bluetooth Low-Energy (BLE) channels, it knows that it is in close proximity to the tag whose identity is transmitted in the message, in addition to the information contained in the message itself. In a supermarket, the reception of its advertisements makes it possible to locate the mobile as long as the service provider has the precise placement of the tags in the store at its disposal. However, tags can also directly send information that may be used by the smartphone itself, such as a price, promotions or a link to the page describing a product.
Within the framework of the TousAntiCovid backtracking application and other mechanisms for identifying at-risk contacts, developed in the context of the Covid-19 pandemic, this co-spatiality property was used to identify transmission risks (Roca 2022). The complexity of the application comes mainly from the need to protect the privacy of users, while ensuring contact identification that is as accurate as possible. It was therefore necessary to avoid storing the list of contacts but to transmit within the advertisement messages the information required to make it possible to determine a posteriori whether their smartphone had been in contact with that of a contaminated person.
The case of contactless payment applications is rather different, since it instead involves establishing as secure a connection as possible to carry out a monetary transaction. It is therefore absolutely necessary to ensure that we are faced with the right device, and to prove that a valid transaction has taken place. However, the co-spatiality property is used to ensure that the payment card with which the transaction is carried out is in immediate proximity to the payment terminal. NFC (Near Field Contact) technology has been specifically adapted to reduce range and impose “near-contact”. The operation of radio transmissions makes this work quite complex because of the propagation of waves in the frequency bands used. It poses security problems, since it makes it possible, for example, to use relays. It then becomes necessary to go beyond controlling the transmission power to limit the range and to very finely control the transmission time, which also depends on the distance; this makes it possible, when it is excessive, to detect an attempt to relay the signal.
Prior to the emergence of the Bluetooth Low-Energy (BLE) version, applications used RFID technology, which has the great advantage of being able to install devices in the environment at very low cost, capable of “responding” to a request and sending previously configured data. This is generally a simple unique identifier, relatively similar to that of eBeacons in BLE. These RFID tags also have the property of being passive most of the time and of using the energy of the reader, which lights them up to wake up and respond. They therefore do not require a battery but are, however, inactive as long as they are not lit up. Readers also need to consume a fairly significant amount of energy to power the tags remotely.
In the different examples that we have seen, local interactions are essentially used to transmit an identifier, which makes it possible to establish our position by referring to prior knowledge of the position of the various devices. Richer applications make it possible to transmit information that the correspondent can directly use and that is most often linked to the position of the sender (a URL describing a product). This somewhat removes the need to maintain a geographic information system (GIS). In the case of BLE, this information is transmitted regularly, whether or not there is a correspondent to listen to it and to do something with it. The information broadcast in this way forms part of the environment and enriches it. The outlines of what we will call “ephemeral local interactions” are given below.
1.2.2. Characteristics of ephemeral local interactions
The first characteristic of local interactions is that they are established in the event of a close contact supported by a short- or medium-range wireless communication (BLE, NFC, RFID, ITS-G5, etc.).
The examples presented above highlight the opportunistic nature of these contacts. The objects considered evolve within a very large scope. They interact, sometimes ephemerally, with many other objects that they do not know beforehand. As stated above, from the outset, this excludes communication technologies requiring a form of pairing (e.g. Bluetooth) or a connection to a network (e.g. Wi-Fi or cellular networks). Indeed, beyond the fact that the necessary establishment time would often be prohibitive with regards to the applications envisaged, it is simply impossible for objects to memorize specific association parameters for each of these relationships; it would also be even more complicated to memorize the association parameters of all potential interactions beforehand.
As a result, the most suitable technologies to support local interactions are those that allow messages to be exchanged directly without prior configuration. It is of course still necessary to have knowledge of low-level parameters such as the type of technology, the operating frequency, the modulation parameters or security elements, as appropriate. On the contrary, the fact that the sender has no prior knowledge about the recipients (and their addresses) generally requires the use of communications in broadcast (sometimes multicast) mode, rather than specifically targeting a correspondent (unicast).
It should be noted that once the initial service discovery phase has been completed, it is possible to establish “traditional” connections in order to deepen exchanges with certain special objects. BLE enables, for example, discovery in opportunistic mode and then switching to dedicated channels in BLE or Bluetooth for more substantial exchanges. The same type of example can be envisaged in V2X, where the detection of a vehicle (or group of vehicles) on the control channel (CCH) can lead to the establishment of special relationships on a service channel (SCH), for example, to process exchanges within a platoon or to carry out a financial transaction (toll).
The characteristics of ephemeral local interactions are therefore as follows:
The information senders and receivers (which can be the same, or two devices with different functionalities) do not know each other beforehand and – often due to mobility – change over time.
They require prior knowledge of a technology and the channels (frequencies) used to broadcast the information. This therefore presupposes regulation or standardization.
There is a spontaneous broadcasting of information – without prior contact – in a predetermined, and therefore most often standardized, format. This information is visible to all devices within communication range and listening.
It follows from these basic characteristics that:
It is difficult to operate on multiple channels, since this presupposes listening to several channels successively, and significantly complicates the encounter between the sent data and a device that is potentially interested.
This leads to a risk of bandwidth overload, since the load cannot be distributed over several channels as is done in current technologies. Therefore, applications using ephemeral local interactions must be limited to fairly simple data (we will not transmit a 4k video stream).
Moreover, since an extension of the range is sought, more robust data encoding is often used, since it is understandable with a lower signal-to-noise ratio and therefore at a longer distance. This reduces the available throughput, so in ITS-G5 an encoding of 6 Mb/s is used, while the technology would allow 27 Mb/s to be reached.
As generalized diffusion (broadcast) or selective diffusion (multicast) is used in most cases, we cannot have an acknowledgment system. Indeed, if the different receivers had to acknowledge each broadcast message, the responses would need to be spread out over a long period of time to avoid collisions. Moreover, this would not be very useful, given that since the sender does not know the list of recipients beforehand, it could not determine that there have been losses;
The securing of exchanges is quite complicated insofar as the assumption is made that there is no prior exchange of information between the protagonists and that they do not know each other beforehand. Setting up cryptographic material to authenticate the sender or to sign (or encrypt) the content is therefore difficult as there is no trusted third party that can be reached at all times. This is all the more difficult because all or part of the devices are mobile and associated with users; therefore, the use of permanent identifiers is prohibited as this would make it possible to track the user’s journey. We will see how different technologies protect themselves from this risk and how the world of cooperative ITS has addressed the need to secure the exchange of sensitive data.
1.2.3. Advantages of ephemeral local interactions
The use of ephemeral local interactions does not make it possible to maintain the usual mode of operation of applications and services; however, it offers advantages that are of great interest.
Therefore, ephemeral local interactions do not require prior knowledge of the protagonists. This mode of interaction is often used in an initial discovery phase before returning to a more conventional mode and establishing a connection. Therefore, a Wi-Fi access point will very regularly send announcements (beacons) that enable the stations to discover its existence. The presence of an access point does not make it possible to determine whether or not the latter can offer a service to the mobile terminal, which, in order to do so, will have to attempt to connect to it and establish a secure session. However, simply receiving beacons in itself enables information to be obtained that is useful to the mobile terminal (Chandra et al. 2007), since this is how smartphones obtain their positioning, through fingerprinting techniques and Geographic Information Systems maintained by service providers. As soon as part of this database is stored locally on the mobile terminal, the latter no longer needs the infrastructure in order to calculate its position.
Bluetooth technologies also use beacons that are transmitted on specific channels listened to by mobile devices (Bluetooth Special Interest Group (SIG) 2016). This enables a terminal to discover the presence of a Bluetooth device, and to connect to it if it has already performed the pairing phase. Here, we are therefore not in the operating principle of ephemeral local interactions, but the diverted use of its beacons to detect the presence of mobile terminals or vehicles (through their hands-free kit) corresponds closely to this logic. Therefore, it is possible, from the roadside, to detect vehicles and the unique identifier of their Bluetooth device, which makes it possible to establish input/output matrices by listening to highway access ramps (Barceló 2013; Boudabous 2021).
One of the main benefits of services based on ephemeral local interactions is that they do not require the prior deployment of a network infrastructure. It is therefore possible to rapidly deploy services without relying on an infrastructure, and to withstand permanent or accidental absences on the network infrastructure. The direct or device-to-device (D2D) mode of 3GPP networks on which cellular versions of V2X technologies are based have also found their first application in the field of emergency tactical networks, which must be able to be set up rapidly, when the telecommunication infrastructure has been harmed by a natural disaster. Depending on the application, we will see that it may occasionally be necessary to rely on an infrastructure to manage the securing of exchanges but without requiring it to be permanently available or involved in the normal operation of the service.
Depending on the needs and the communication technologies employed, using ephemeral local interactions enables very fast interactions that depend solely on the message sending frequency and the traffic capacity of the communication technology used. Therefore, the transmission delay will be higher when using BLE beacons (Yang et al. 2020), which must be able to operate for several months/years on a button battery, and which, as a result, will only send a “beacon” every minute. The maximum detection time of a beacon will therefore be of the order of a minute if we take into account the potential losses and the periods when the receiver is not listening to the correct channel. In the case of ITS G5, CAMs (Cooperative Awareness Messages (ETSI 2014)) are sent by default every 100 ms, which, even considering the potential losses, enables very short latencies.
1.2.4. Suitability of communication technologies for this type of interaction
The communication technologies used must enable an interaction model that is consistent with ephemeral local interactions. Therefore, it is very much possible to build above IP a service that functions through the regular broadcasting of messages and the use of this information to provide a service without any further dialog with the sender. However, while this presupposes prior connection to a network (e.g. Wi-Fi or Cellular), a large part of the advantages described previously disappear. The underlying communication technologies therefore need to be aligned with the interaction model and the targeted properties.
Adapted technologies generally use a shared multiple-access channel, which means that to access the communication medium, it is not necessary to obtain prior authorization and that everyone can transmit on the channel. There is therefore competition for access to the radio resource, which is for exclusive use (only one sender at a time). A device seeking to send a frame will listen and wait until the channel is free. There is a risk – and mechanisms to reduce it – that two devices will start sending at the same time. When the messages of two senders overlap in time, there is a collision. Some of the potential receivers do not receive either of the two messages, while the other potential receivers decode one message or the other according to the capturing-effect principle (Roberts 1975; Gezer et al. 2010). However, the senders cannot hear the other transmission and do not notice the collision. Regardless of the organization of the radio channel, as soon as there is no device in charge of organizing the resource (such as a base station in the 4/5G networks), and no prior exchange to decide who is allowed to use the resource (RTS/CTS mode in Wi-Fi), the nodes use the resource and then manage the effects of the collisions. When the transmission is directed towards a specific receiver, it is possible to use an acknowledgment and retransmission system (this is the case for Wi-Fi); however, when it is broadcast, losses cannot be detected by the sender, and this must be taken into account in the very construction of the service.
In the case of ITS-G5, a device listens to the channel and transmits only if it is free for a minimum period of time, thus limiting the occurrence of collisions. This type of self-organized system (CSMA: Carrier Sense Multiple Access) works very well up to a certain level of channel occupancy (approximately 60% of the bandwidth) (Bianchi 2000). It is therefore appropriate to introduce mechanisms to limit the network load. Mechanisms are thus used to vary the message frequency according to the usage scenarios and the load of the radio access network. Of course, reducing the sending frequency increases the average time between two receptions and the time needed to discover a new node; this is why it is important to take into account the usage scenario, such as vehicle speed in the context of ITS. Indeed, the lower the velocity of the vehicle, the less its position changes between two announcements.
Figure 1.1.Operation of Wi-Fi CSMA (IEEE 2009)
In the case of BLE, the very limited bandwidth (typically 1 Mb/s under conventional conditions) will quickly impose constraints on the announcement sending frequency, even if, by design, the messages are limited in size. Since the range is also quite small, this technology could be used for low-velocity applications (pedestrians and VRUs (Vulnerable Road Users) more generally).
In the case of LTE-V2X (Garcia-Roger et al. 2020), the technology developed relatively recently has slightly different properties. Indeed, the radio resource is organized in a frame, with concurrent access to each slot, so that there are several transmissions in parallel on the channel. To limit collisions, due to the very short duration of a slot, it is not possible to listen to decide whether to transmit or not. When the channel is organized by a base station – this is one of the LTE-V2X modes – a station asks to be allowed to transmit before obtaining the resource from the base station. While there can be a collision on the request, the transmission of useful packets is no longer subject to collision. When the channel is self-organized, all the stations share the temporal structure of the frame and must therefore be strictly synchronized. They use semi-persistent scheduling (SPS). This involves listening to the channel and taking advantage of the repetitive nature of regular announcement messages (CAMs) by noting the slots that are occupied in the recent period. For each slot, the station determines the probability that it will be occupied by observing the power level received in that slot for a given duration (1 second by default). It will then select one of the “free” slots for its own transmissions. This selection will then be reviewed regularly. This mechanism works quite well for regular and permanent messages. However, when a collision occurs, it is not detected and lasts until one of the two senders makes the selection again, which becomes critical when the real-time constraints are high.
Figure 1.2.Structure of frames and operation of LTE-V2X semi-persistent scheduling (Haider and Hwang 2019)
Due to the direct communications between the devices concerned, direct interactions allow very short connection times, given that the announcement messages are frequent and that the receivers constantly listen to the correct channel. Mechanisms to time-multiplex sending and reception extend the time frames and the available bandwidth automatically. When everyone listens to the same channel, the connection time, i.e. the average time required to receive the first message that the receiver is able to decode, is directly related to the message sending period and the distribution of losses on the channel. Indeed, if the losses are not distributed uniformly, the calculation of this average time depends mainly on the maximum duration of the consecutive message loss sequences and on the probability of loss. Losses can be due to collisions, other forms of signal interference or masking. The different forms of interference may be considered to be random. On the contrary, as already mentioned above, collisions are positively correlated with the load submitted to the network, hence the importance of controlling the amount of traffic submitted to the network.
1.3. Local interactions serving cooperative ITS
In this section, we will focus more specifically on the use of ephemeral local interactions in the context of cooperative ITS.
1.3.1. Cooperative ITS services
The scope of cooperative ITS (C-ITS) applications is particularly broad. It is traditionally broken down according to the maturity of the technology required. So-called “Day 1” applications are deployable with the technologies currently available. These include the transmission of alerts (slowing down, approaching a priority vehicle, accidents, road works, etc.), signage (on-board display or trafficlight phase) or of presence (position, speed, direction, etc.). In contrast, “Day 2” applications require performance and standardization levels that have not yet been achieved. This covers, for example, driving in convoy (“platooning”), remote driving or vision sharing (“see through”). “Day 1.5” applications are at an intermediate stage: they are feasible for particular cases but their level of standardization does not allow them to be immediately generalized on a large scale. This is the case, for example, with the protection of Vulnerable Road Users (VRU), parking space management or dynamic routing.
As diverse as they are, these applications can be based on the standardization proposed by the “Facilities” layer of the ITS-Station stack. This messaging layer is a kind of middleware between communication layers and applications. This layer is regularly enriched with new functionalities. At this stage, we can cite the following messages as examples:
CAMs (Cooperative Awareness Messages) (ETSI
2014
) are sent regularly (typically every 100 ms) by vehicles to signal their position, speed, direction and physical characteristics. Other vehicles use this information to add the sender to their perception of the environment.
DENMs (Decentralized Event Notification Messages) (ETSI
2014
) are used to signal one-off events such as accidents, construction sites, slowdowns, etc.
SPAT/SPATEMs (Signal Phase and Time/Extended Messages – ETSI TS 103 301) transmit the traffic-light phase status. They are typically associated with MAP/MAPEMs (Map Data/Extended Messages) that describe the geometry of roads and intersections.
IVI (In-Vehicle Information – ETSI TS 103 301) dynamically relays the signage information for on-board display. They replace or complement conventional road signage and are more easily exploitable by the vehicle’s automations.
CPMs (Cooperative Perception Messages – ETSI TR 103 562) enable actors (vehicles and infrastructure) to dynamically exchange information on their perception of the environment (obstacles, vehicles, pedestrians, etc.).
These standardized messages offer an impressive toolkit to accelerate the development and penetration of applications into the market. However, before they start, application designers must analyze the consequences of the choice of message type on the architecture of their solution. Will the application be dependent on an infrastructure or not? Can this infrastructure be decentralized (RSU) or will it require a central server? The answer to these questions will have a strong impact on the cost and ease of deployment of the solution, as well as on its performance (for example, reaction time).
1.3.2. Benefit of ephemeral local interactions for cooperative ITS
Here, local interactions offer numerous advantages. They are simple to deploy (as their configuration is very lightweight), require little or no infrastructure (e.g. a light connected to an RSU) and offer very good responsiveness (of the order of several ms to several 100 ms). In addition, limiting wave propagation makes it possible to natively reduce diffusion around areas of interest of the different messages.
The benefit is obviously for collaborative perception applications (based on CAM or CPM messages) or for signage (IVI, SPAT, etc.). In the first case, the speed of transmission is decisive. Also, in all cases, the geographical limitation of broadcasting and the ability to exchange information without the need for prior association constitutes a decisive advantage.
The transmission of warnings (typically by DENM message) represents an interesting case. A network access priority and a sending frequency is associated with the message, based on the emergency associated with the event at the origin of the warning. Therefore, DENM messages relating to an extreme emergency (e.g. in the event of a collision) are sent with maximum priority and at a high frequency to ensure that all approaching vehicles dispose of the information early enough to react (in a few ms).
DENM messages are also used to signal road works or other hazards present on the roadway, in which case, the time constraint is generally more relaxed. For example, it is generally not indispensable to signal the presence of road works in under a second. Moreover, this type of information is sometimes transmitted in V2N (Vehicle to Network) via cellular networks. However, use is often made of V2X – and therefore of local interactions – for these messages too. The reason is twofold: the natural geographical limitation of broadcasting, as well as the fact that the vehicles already have a V2X receiver for other types of messages. In this context, imposing a second receiver only for DENM messages would be counterproductive. Moreover, since it may be necessary to cover a geographical area larger than the range of a V2X transmission, ETSI has provided for the possibility of addressing a geographical area directly in the message header, which is named “geocasting” (ETSI 2009). The messages are thus relayed hop by hop to reach the area of interest without requiring any equipment in the road-side infrastructure.
Figure 1.3.Different modes of DENM message broadcasting.
The example of DENMs is also interesting in terms of architecture. It is appropriate to equip a mobile construction site with a mobile RSU to send DENMs, signaling the construction site to approaching vehicles. However, its range will be insufficient to announce the construction site 5 km upstream. For this, the RSU can send the messages in geocasting, but these will not reach their target if there is no vehicle to relay the messages. It is then possible to make use of RSUs that will relay messages as vehicles would do, and without needing to be connected to a network infrastructure; however, this requires RSUs to be specifically deployed if the road is not yet equipped. Lastly, it is possible to make use of a centralized ITS infrastructure and a non-dedicated network infrastructure (e.g. a server that can be reached by a cellular network) so that its message is transferred directly to the vehicles or to the RSUs, which will have to broadcast it downstream. In this case, when no technology meets all the needs, it is useful to combine traditional communications with opportunistic communications.
1.3.3. V2X communication technologies
During the work on WAVE architecture (Eichler 2007; Gräfling et al. 2010) in the United States, several types of services were considered, some of which relied on a connected infrastructure. However, covering all roads with a communication infrastructure before the deployment of services is not a reasonable alternative – especially since even if it had been possible, the communication networks of the time did not allow the time constraints to be respected. The choice was made to use direct interactions, requiring the least configuration to minimize the time needed to acquire critical information.
To do this, it was necessary to reserve a communication frequency that everyone can listen to without a connection phase and without complex configuration. The 5 GHz band was chosen and reserved in the United States, and equivalent bands were reserved in Europe and in Japan (with some differences). Once the frequency was selected, it was necessary to agree on the communication technology used and on the organization of radio resources. The initial choice was to start with a well-mastered technology by adapting it to the mobility and needs of ephemeral local interactions. For this purpose, Wi-Fi (IEEE 802.11) was amended (IEEE 802.11p). In Europe, this technology is integrated into the body of ITS standards developed by ETSI under the name ITS-G5.
It should be noted that Wi-Fi, due to its fully distributed channel access control mode, is well suited to ephemeral direct interactions; indeed, by definition, all those who listen to a channel hear all of the messages and decide based on their perception of the channel, whether or not they can send a Wi-Fi frame. Natively, broadcast is used, which makes it possible to mobilize all stations capable of decoding the message. However, in Wi-Fi, even if all of the stations perceive the frames, the communications are relayed by the access point, requiring the prior deployment of an infrastructure, even if the latter does not need to be connected to the Internet to relay the frames between local stations. Among the adaptations made to Wi-Fi, the choice was made to introduce a new operating mode without infrastructure in Wi-Fi: OCB (Outside the Context of a BSS) mode. The other adaptations concern improvements to radio in order to improve the reliability of transmissions in contexts with high-speed differentials (Doppler effect), and with a lot of interference due to fading (urban environment).
For more than 15 years now, numerous projects (SCORE-F, SCOOP@F (Aniss 2016), InDiD, C-Roads and Intercor) have been deploying, experimenting and testing the limits of ITS-G5 in different contexts and for a wide variety of services, so its operating scope is well known.
It should be noted that many other technologies have been (and are still being) considered for direct communications between two vehicles, in particular, to accommodate much higher throughputs (See Through) or even shorter time frames (Platooning and Cooperative Maneuvers). For example, we can mention highly directive technologies such as millimeter waves or visible light communication. For basic services related to security and the perception of other users, as well as the discovery of available services, vehicular Wi-Fi (ITS-G5 in Europe) has long been the only alternative really considered. Today, the situation is a little more complex; two families of incompatible technologies are in the running.
Indeed, cellular network actors have proposed to adapt a variation of LTE’s D2D (Device-to-Device) mode to V2X communications (Garcia-Roger et al. 2020; ESTI 2018). Even more recently, 3GPP published standards to offer more advanced services through the adaptation of the new 5G radio interface to V2X (NR-V2X) (ETSI 2020; Storck and Duarte-Figueiredo 2020). The possibility of implementing high value-added and potentially marketable services is one of the challenges of the battle between the two technology families. It should be noted that V2X cellular communications have an infrastructure-free mode that does not rely on the base station to control access and optimize the radio resource. However, they can also use an infrastructure to offer services and, in particular, with the 5G network architecture that promises guaranteed interaction times of the order of milliseconds, which would make it possible, for example, to organize complex maneuvers from servers in proximity to traffic (Mobile Edge Computing).
Technological differences and the debate regarding the choice of technologies is not what interests us in this chapter. It will be necessary, however, to quickly choose a technology, even if it is not perfect, because this is a prerequisite for the deployment of the first services, based on direct interactions. Indeed, everyone must speak at least one common language, if only to discover the presence of others and exchange, without prior complex interactions, messages related to emergency situations.
Other technologies enable direct interactions; for example, they target access control services (wireless car-key) but it requires prior pairing or complex configuration to manage identifications. Note that it is increasingly possible to use a smartphone and standard communication technology (Bluetooth, BLE or Wi-Fi) to implement these services. In any case, these are not the forms of interaction that interest us.
1.3.4. Properties of C-ITS services built on local interactions
1.3.4.1. Latency and reduced responsiveness
The different C-ITS services do not require the same level of service, but for all, a fairly low latency (less than 100 ms) is sought. It is in fact even shorter for advanced services such as convoy maneuvers (platooning) or cooperative maneuvers. The maximum latency depends on several factors; first and foremost on the communication technology used and how it controls access to the radio resource. Latency also depends on the network load and the state of the radio channel. How the load impacts transmission performance depends directly on the technology used. Therefore, in ITS-G5, the average channel access time increases with the load submitted to the network; however, it is above all the losses due to collisions that end up making the time frames too long. Indeed, the messages are generally transmitted at a certain frequency (10 times per second for CAM messages), which makes it possible to support frame losses while not jeopardizing the service. The losses, which also increase with the load submitted to the network, therefore amount to adding time between messages received successfully.
It should be remembered that due to transmissions in broadcast, it is impossible to use an acknowledgment mechanism and therefore to make transmissions more reliable by retransmitting lost messages. To increase the message transmission rate, LTE-V2X prefers prevention rather than cure, systematically transmitting critical messages twice. While this consumes twice the radio resource, it greatly increases the probability of reception.
The two competing technologies, ITS-G5 and LTE-V2X, use very different mechanisms to control access to the radio resource, and this results in different load behaviors. In both cases, the collision rate increases with the network load, but in the case of ITS-G5, each message competes for access to the channel and can therefore be subject to collisions; in the case of LTE-V2X, the SPS algorithm protects the recurring traffic already established, because the vehicles seeking to send know which resources are occupied. Initial access to the channel is faster in ITS-G5, however, with messages very often delivered in a few milliseconds. In the case of LTE-V2X, sending a new message requires a certain time frame for choosing the resource (Resource Block) and a collision, the probability of which depends on the network load; this can take place on this resource. In this case, messages are not delivered throughout the duration of the collision (the 100 ms re-selection window), which is a real problem for the most urgent warning messages.
While it is difficult to reduce collision delays in a fully distributed mechanism, it is easier to prioritize access to the radio resource for the most urgent messages. Therefore, ITS-G5 allocates a different priority to the message according to their service class, using the Wi-Fi EDCA (Enhanced Distributed Channel Access) mechanism (ETSI 2019).
1.3.4.2. Frugality of exchanges and congestion mitigation
Due to the capacity limits of the radio channel and the number of devices with the potential to send, exchanges must be limited to what is strictly necessary. In addition, it is necessary to favor short messages that free the channel up faster. This is especially true in ITS-G5 because owing to the channel access method (CSMA), a message in the process of being sent cannot be interrupted and prevents vehicles having more urgent messages from accessing the channel. In order to reduce the size of messages and avoid protocol overload due to the TCP/IP stack, the messages have been installed directly above the connection layer with ad hoc protocols1. For the same reasons, certificates used to verify the legitimacy of the sender are transmitted only once per second.
Despite these precautions, the vehicle density may be such that it is necessary to prevent channel congestion by limiting the load submitted to the network. In order to evaluate the network load, and because of the lack of centralized coordination, each vehicle can only have, at its disposal, information that it is able to observe by itself, such as: the proportion of time the channel is occupied and the number of vehicles observable in the vicinity. When it measures that the network is loaded, it can play on time by reducing the number of messages and space by limiting the transmission power to reduce the “cost” of a transmission in terms of spectrum occupancy. Some of the mechanisms apply at the radio technology level, while others involve the upper layers (called cross-layer control) to limit the number of messages submitted to the network. This reduction takes into account an assessment – by necessity, simplified – of the usefulness or criticality of the messages. Therefore, the frequency of CAM messages can be reduced to 1 Hz when the vehicle is slow or stationary.
1.3.4.3. Facilitated privacy
If the contents of the messages are not directly private in nature, the information transmitted by a vehicle makes it possible to trace the latter and thus reconstruct its journey. More generally speaking, each time a permanent or semi-permanent identifier can be associated with a positioning, there is a risk of allowing tracing.
It is therefore indispensable to mask unique and permanent identifiers. Indeed, radio technologies use unique identifiers (MAC address) that are allocated when manufacturing the radio interfaces and are used as source addresses in the frames. The upper layers also use identifiers (certificate, IP address, etc.) in the different layers of the protocol stack used. For several years now, it has been commonly accepted that the different identifiers used must no longer be able to be associated with a particular terminal. To avoid this, regularly changing random identifiers are used (BT or Wi-Fi). However, care must be taken to change all the identifiers present in a message’s different headers at the same time, to avoid two successive identifiers from being associated too easily. Indeed, if the MAC address changes but the identifiers used in the upper layers do not change, an observer will have no trouble attributing the two successive MAC identifiers to the same terminal.
