Page 60 - ITU Journal Future and evolving technologies Volume 3 (2022), Issue 2 – Towards vehicular networks in the 6G era

ITU Journal on Future and Evolving Technologies, Volume 3 (2022), Issue 2

Update
Source Destination
4 3 2 1
V2V
V2R v traffic hole V2R

Sensor L 1 L 2 RSU

Fig. 1 – Scenario of update delivery under traf ic light in vehicular networks

The second vehicle is stopped by the red light and a allowing for transmission of multimedia data such as ve‑
traf ic hole appears between the irst and the second hicle images for use in machine learning‑related appli‑
vehicle, blocking the V2V data transmissions and cations [13, 12]. 6G networks allow for the introduc‑
increasing the delivery delay of the second update. tion of the Internet of Things or other real‑time applica‑
While the second vehicle carrying the second update is tion services, such as icial intelligence and big data
waiting at the red light, the third vehicle catches up and computing applications [14]. With speeds up to 1Tbps
forwards the third update to reduce its delay. and packet delays below 100 / , 6G networks can meet
quality of service guarantee requirements [15]. In or‑
In this paper, we investigate the in luence of traf ic lights der to meet high requirements of 6G networks such as
on data delivery in vehicular networks using the met‑
high reliability and high security in a dynamic and hete‑
ric of Age of Information (AoI). Since traf ic lights block
rogeneous vehicular network, Zhang et al. [16] propose
V2V data delivery, we discuss the optimal generation rate
a novel Weight‑Based Ensemble machine Learning Algo‑
at the source by considering the trade‑off between the rithm (WBELA) to identify abnormal messages of vehicu‑
AoI and transmission cost. A red light stops the vehicles lar Controller Area Network (CAN) bus network. Emplo‑
carrying the updates and increases the updates’ delays. ying machine learning into 6G vehicular networks to
Meanwhile, the vehicles moving behind could catch up support vehicular application services is being widely
with the waiting vehicles in forwarding the new updates, studied and a hot topic for the latest research work in
and the old updates at the traf ic light will lose their con‑ the literature. In vehicular networks, network entities
tributions on AoI. Increasing the generation rate can re‑ need to make decisions to maximize network per-
duce the AoI, but also increase the number of such old formance under uncertainty. Mekrache et al. [14]
updates at the red light. We also consider the transmis‑ provide a comprehensive review of research work that
sion cost (energy consumed) [7], as increasing the gene‑ integrated reinforcement and deep reinforcement
ration rate can also increase the cost. Therefore, gene‑ learning algorithms for vehicular networks management
ration rate control at the sensor source should consider with an emphasis on vehicular telecommunications
the luence of ic lights on the data transmissions. issues.
We also propose a Total Average Cost aware generation
rate Algorithm (TACA) for deciding the generation inter‑
val time at the sensor source. Our intensive simulations There are many vehicular applications based on sen‑
verify the proposed algorithm and evaluate the in luence sing and communications between vehicles. Ahn et al.
of the traf ic lights on AoI. [17] present the Road Information Sharing Architecture
(RISA), a distributed approach to road condition detec‑
The remainder of this paper is organized as follows: In tion and dissemination for vehicular networks. Signal‑
Section 2, we review the most related work. We discuss Guru [18] relies solely on a collection of mobile phones
the system model and the traf ic hole problem in Section to detect and predict a traf ic signal’s schedule. For such
3. We analyze the AoI and transmission cost in Section 4. an infrastructure‑less approach, multiple phones in the
We evaluate the ef icacy of the analysis model and the data vicinity use opportunistic ad‑hoc communications to col‑
delivery with traf ic lights in Section 5. The inal section laboratively learn traf ic signals’ timing patterns and pre‑
concludes the paper. dict their schedules. Various advanced sensors (e.g., li‑
dar, radar, camera, etc.) equipped on a representative au‑
2. RELATED WORK tonomous vehicle. Due to the intrinsic limitations of these
sensors, autonomous vehicles are prone to making erro‑
6G and vehicular networks: The 6G vehicular network
neous decisions and causing serious disasters. Networ‑
aims to develop a highly dynamic and intelligent system,
king and communication technologies can greatly
which enables the networks to change the environment
make up for sensor iciencies, and are more reliable,
to satisfy various application requirements and service
feasible and ef icient to promote the information
types, such as enhanced Mobile Broadband (eMBB), ultra‑
interaction, thereby improving autonomous vehicles’
Reliable and Low‑Latency Communications (uRLLCs),
perception and planning capabilities, realizing better
and massive Machine‑Type Communications (mMTCs) vehicle control, and ultimately greatly improving the
[12]. 6G network transmission can reach speeds of 1Tbps, security of autonomous vehicles [19].

48 © International Telecommunication Union, 2022