Content uploaded by Mahdi Razzaghpour
Author content
All content in this area was uploaded by Mahdi Razzaghpour on May 31, 2024
Content may be subject to copyright.
1
Mass Platooning: Information Networking Structures for Long
Platoons of Connected Vehicles
Mahdi Razzaghpour, Babak Ebrahimi Soorchaei, Rodolfo Valiente, Yaser P. Fallah
Investigating Vehicle-to-everything (V2X) communication, we dive into the concept of vehicle platoons, a key innovation in
transport systems, introducing a new era of cooperative driving. This new approach is designed to enhance fuel efficiency and
improve overall traffic flow. Crucially, the success of this system relies on keeping vehicles at closely monitored distances, particularly
at high speeds, which depends on rapid and reliable data exchange among vehicles through a wireless communication channel that is
intrinsically unstable. The possibility of improving platoon efficiency through wireless data exchange is clear, but addressing network
issues such as data loss and delays is crucial. These problems can compromise platoon functionality and need careful handling
for real-world applications. Present platooning models also struggle with forming ’long’ platoons with multiple vehicles due to the
limited range of Vehicle-to-Vehicle (V2V) communication. Quick and efficient traffic information sharing is crucial to ensure vehicles
have adequate time to respond. Given the safety-critical nature of these communications, both reliability and ultra-low latency are
essential, particularly in platooning contexts. To address these challenges, we suggest a distance-based, network-aware relaying policy
specifically for long platoons of connected vehicles. The results of our simulations indicate that this relaying approach significantly
decreases communication breakdowns and narrows the error gap between vehicles, all achieved with only a slight increase in
computational demand.
Index Terms—Cooperative Adaptive Cruise Control (CACC), Information Flow Topology, Long Platoon, Multi-Agent Systems
(MASs), Multi-hop Broadcast, Piggybacking
I. INTRODUCTION
Annually, the United States witnesses over 30,000 fatalities
due to highway accidents. On average, Americans lose 97
hours yearly to traffic jams, leading to an economic burden
of approximately $87 billion in 2018 alone, which translates
to about 1,348 per driver. From an environmental perspective,
congestion led to the waste of 44.3billion liters of fuel
globally [1], [2]. While road expansion offers a partial remedy,
its feasibility is constrained by high costs and limited available
land. An alternative solution lies in transitioning from indi-
vidual driving to cooperative driving strategies. Specifically,
the concept of platoon-based driving involves a convoy of
vehicles sharing a common route, where each vehicle closely
and consistently follows the preceding vehicles, thus forming
a platoon, as depicted in Figure 1, represents a form of
cooperative driving that can significantly mitigate these issues.
The conventional cruise Control (CC) systems allow drivers
to set a desired speed, which the vehicle then maintains
automatically. Adaptive Cruise Control (ACC), leveraging
local sensors like radar and LiDAR, dynamically adjusts the
vehicle’s acceleration to maintain a predetermined distance
from the vehicle ahead. However, due to the limitations of
these sensors, ACC cannot function at intervals less than
one second, leading to challenges in maintaining consistent
inter-vehicle distances, known as string stability issues [3].
To overcome these limitations, Cooperative Adaptive Cruise
Control (CACC) technologies are being developed. CACC
distinguishes itself by utilizing not just on-board sensor data
Connected & Autonomous Vehicle REsearch Lab (CAVREL), Department
of Electrical and Computer Engineering, University of Central Florida, Or-
lando, FL, USA. mahdi.razzaghpour@ucf.edu
This material is based upon work partially supported by the National Sci-
ence Foundation under Grant No. CNS-1932037. Article processing charges
were provided in part by the UCF College of Graduate Studies Open Access
Publishing Fund.
Fig. 1: The diagram presents the communication framework
among vehicles, where dashed lines represent the transmission
of data between them. The notation dnsignifies the gap
between the nth vehicle and the one directly in front of it.
but also incorporating kinematic data from other vehicles
within the platoon, shared via Vehicle-to-Vehicle (V2V) com-
munication, to enhance stability. Research outcomes suggested
that the implementation of ACC might not substantially affect
lane capacity. Conversely, CACC demonstrated a potential to
substantially enhance lane capacity once its adoption reached
moderate to high levels [4], [5]. The key to this improvement
lies in the vehicles’ ability to exchange information, enabling
them to predict the movements of adjacent vehicles [6].
Consequently, a CACC-equipped vehicle can maintain a closer
following distance to the vehicle ahead without compromising
comfort or safety, thereby preserving the integrity of the
platoon’s string stability [7], [8].
The collective behavior of Connected and Automated Ve-
hicles (CAVs) rely on each vehicle’s mutual awareness of
their surroundings. It is presumed that vehicles acquire accu-
rate, real-time information about their surroundings through
sensors, communications, and precise digital mapping [9].
A transitional period is anticipated where both cooperative
and autonomous driving modes will share roadways. In this
context, the role of driver behavior becomes pivotal for the
development and functionality of connected vehicle technolo-
gies [10], [11]. Furthermore, it is essential to explore how
the integration of these driving modes affects road safety,
traffic efficiency, infrastructure capacity, and fuel consump-
2
tion. Future advancements are expected to include vehicular
networking strategies that incorporate social dynamics among
vehicles [12], [13].
The cooperative management of the longitudinal motions in
a group of CAVs can lead to several potential advantages for
the transport system [14]:
•An increase in road capacity is achievable through min-
imizing the distances between vehicles [15].
•Reductions in energy usage and emissions of pollutants
are possible by limiting unnecessary variations in speed
and lowering the aerodynamic resistance encountered by
following vehicles [16], [17].
•Improvements in driving safety could occur, given that
system malfunctions may be less frequent than human
errors, currently the predominant cause of vehicular ac-
cidents [18]–[20].
•Increased customer satisfaction is anticipated as vehicles
in a platoon navigate autonomously, offering a smoother
and safer ride while preventing the cutting in of other
vehicles due to shorter inter-vehicle distances.
A platoon of CAVs functions as a Cyber-Physical System
(CPS), merging computation, communication, and physical
processes in both cyber and physical domains [21]. Within
such platoons, the physical dimension is tied to vehicular
motion, while the cyber dimension encompasses communica-
tion among vehicles and between vehicles and infrastructure
[22]–[24]. The foundation of these distributed systems is the
exchange of information, which is essential to achieve coop-
erative situational awareness. This exchange process outlines
the method by which platoon vehicles communicate, primarily
through broadcasting or beaconing protocols, which have
been extensively explored within the vehicular networking
community. The prevailing approach involves regular beacon
broadcasts to all vehicles in range, enhancing cooperative
awareness. Model-Based Communication (MBC) emerges as
an innovative approach for enhancing communication scalabil-
ity and alleviating channel overload [25]. Its main objective is
to use a more adaptable broadcasting packet content structure,
presenting the parameters of the combined vehicle dynamics
and driver behavior models, diverging from the rigid Basic
Safety Message (BSM) format specified by the J2735 standard
[26]. For representing vehicle dynamics in the MBC frame-
work, various modeling techniques are available, with non-
parametric Bayesian inference methods, especially Gaussian
Processes (GPs), showing potential for accurately representing
the combined dynamics of vehicle motion and driver behavior.
Due to the constraints imposed by sensing and communica-
tion technologies, each vehicle controller typically has access
to information only from its immediate vicinity. Consequently,
controllers rely on localized data to achieve platoon-wide
outcomes. Effective platoon management relies on seamless
transmission of data between adjacent vehicles and from the
lead vehicle down the chain. Numerous studies have explored
strategies to enhance the communication reach from the pla-
toon leader to its followers, especially beneficial for long pla-
toons [27]. Although small platoons may experience minimal
packet delays and losses even at high data transmission rates,
longer platoons face increased risks of these issues under the
same conditions due to the inherent limitations of wireless
communication over extended distances [28]. The effective-
ness of wireless communication diminishes as the length of the
platoon increases, imposing a cap on the maximum number of
vehicles that can be effectively connected [29]. Consequently,
devising efficient communication protocols or algorithms is
critical to ensure the effective dissemination of information
within the platoon system [30], [31]. Despite advancements in
optimizing network configurations for Multi-Agent Systems
(MASs), the development of a V2V based framework is
crucial for the fast and reliable distribution of critical safety
messages throughout a long platoon. For the preservation
of control system integrity, the wireless V2V network must
ensure that the maximum Inter Packet Gap (IPG) does not
exceed a predefined limit. Understanding packet propagation
dynamics and identifying the principal determinants of delay
within such a network are essential for accurately assessing
communication latencies in platoon settings.
The size of the platoon significantly influences the efficacy
of its communications. Our objective is to maximize platoon
size, facilitating the efficient transit of a greater number of
vehicles. This research employs distributed relaying through
multi-hop communication for enhancing the broadcast link,
enabling platoon members to relay packets from the leader to
the rest of the platoon members. This paper’s core technical
contributions are centered on examining how different commu-
nication information structures and Information Flow Topol-
ogy (IFT) affect the scalability of homogeneous vehicular
platoons in fixed formations, alongside the integration of this
communication strategy with vehicular control mechanisms to
consistently preserve inter-vehicle distances, irrespective of the
platoon’s length. Our key contributions include:
•Assessing and benchmarking the performance of ex-
tended platoons with respect to various communica-
tion strategies, considering both operational efficacy and
network-centric metrics.
•Introducing a method for selecting distributed relays that
incorporates the quality of the communication links and
spatial parameters to minimize packet delivery times.
•The proposed approach seeks to diminish latency while
considering the channel’s capacity constraints.
II. RE LATE D WORK
The management of collective dynamics within a network of
CAVs hinges on the vehicles’ capability to perceive their own
and each other’s states, like the distance between vehicles
and their velocities. These insights are garnered through inter-
vehicle sensing and communication mechanisms. The effec-
tiveness of cooperative applications is directly linked to the
success rate of vehicular information exchange and the extent
of the network that receives this information. This section
focuses on CACC and its dependence on V2V communication.
A. Cooperative Adaptive Cruise Control (CACC)
During the 1990s, the concept of vehicle platooning gained
widespread interest across both academic and industrial sec-
tors, especially after the launch of the California Partners for
3
(a) PF (b) PLF
Fig. 2: Typical IFTs for a platoon: (a) Predecessor Following (PF); (b) Predecessor Leader Following (PLF)
Advanced Transit and Highways (PATH) program [32]. The
primary objective in managing an autonomous vehicle platoon
involves ensuring uniform speed across the platoon while
maintaining a specified distance between each vehicle. For
the nth vehicle, characterized by its position xn(t), velocity
vn(t), and gap with the vehicle in front dn, as shown in Figure
1, the goal is to achieve the desired separation d∗
n.
lim
t→∞ |xn(t)−xn−1(t)|=d∗
n,for n= 1, . . . , Nv−1(1)
where Nvis the number of platoon members. Another critical
objective is the mitigation of disruptions or shock waves within
the platoon, ensuring vehicles move at the same speed:
lim
t→∞ |vn(t)−v0(t)|= 0 ,for n= 1,· · · , Nv−1(2)
with v0(t)representing the lead vehicle’s speed. Detecting
variations in the velocities of leading vehicles is crucial to
diminish traffic shock waves, which result in ”stop-and-go”
patterns [19]. The goal of minimizing variations in acceleration
is in line with the objectives in terms of fuel efficiency and
ride comfort.
lim
t→∞ |an(t)−a0(t)|= 0 ,for n= 1,· · · , Nv−1(3)
with a0(t)representing the lead vehicle’s acceleration.
Direct transmission of precise state data allows vehicles to
maintain closer proximity without endangering safety. Studies
have established that platoons can attain both asymptotic
and string stability, provided that time headway exceeds a
certain minimum threshold. Moreover, it has been found that
increasing the count of leading vehicles a vehicle responds to
can lower this threshold [33]. Simulation results reveal that
employing velocity feedback significantly improves platoon
dynamics by reducing members’ acceleration rates. Incorporat-
ing position feedback further smooths out reactions to changes
in the leader’s acceleration. The findings suggest that feedback
from the vehicles that precede directly, rather than only from
the leading vehicle, may also be effective [34]. In essence, the
efficacy of platoon-based cooperative driving hinges on the
network configuration and the control mechanisms in place,
which integrate communication, computational, and physical
elements.
1) Information Flow Topology (IFT)
In distributed MASs, IFT critically influences the communi-
cation links among agents, significantly affecting data sharing
dynamics. The incorporation of wireless communication in
CAV systems has introduced a variety of IFTs, posing unique
challenges and opportunities for the design and analysis of
such systems [35]. The selection from this broad spectrum of
design options necessitates meticulous evaluation to enhance
system efficacy. It has been identified that implementing
CACC significantly boosts traffic flow and safety, independent
of the chosen IFT.
In real-world applications, IFT often varies over time due to
the dynamic nature of communication links and the changing
presence of vehicles in the network. Figure 2 illustrates
two IFTs commonly used in CAVs, Predecessor Following
(PF) and Predecessor Leader Following (PLF). Additionally,
more complex topologies like r-Predecessor Following (rPF)
and r-Predecessor-Leader Following (rPLF) are applicable, as
shown in Figure 1, where rindicates the number of directly
communicated predecessors. In comparative terms, PLF and
rPLF demonstrate superior performance over PF, with the
choice between PLF and rPLF hinging on the availability of
communication links. PF’s limitation in accessing information
from distant vehicles restricts its communication benefits. By
contrast, PLF enables CACC-equipped vehicles to receive
advance traffic updates from the leader, enhancing anticipatory
responses for traffic flow stabilization. rPLF, expanding on
the communicative reach of CACC, allows following vehicles
to leverage the data of preceding vehicles for improved traf-
fic management. However, given communication constraints,
there’s a limit to platoon scalability, making platoon man-
agement essential. rPLF serves as a preferable choice with
ample communication resources or in lower vehicle density
settings, whereas PLF is deemed more suitable under restricted
conditions [19], [36].
2) Formation Geometry
Within the framework of platoon systems, the Formation Ge-
ometry aspect refers to the preferred spacing between vehicles,
frequently referred to as the gap policy in various studies.
Predominantly, three gap policies are widely adopted: Constant
Distance (CD), Constant Time Headway (CTH), and Non-
Linear Distance (NLD) policies [37]. The selection among
CD, CTH, and NLD policies hinges on the criterion of vehicle
speed affecting the desired inter-vehicle spacing. The CD
method is effective for its capacity to boost traffic throughput
by maintaining a reduced, fixed gap between vehicles, in
contrast to the CTH approach.
B. Vehicle-to-Vehicle (V2V) Communication
In this work, we examine the mode-4 operation that facilitates
direct communication among adjacent users equipped with
Cellular Vehicle-to-Everything (C-V2X) technology, through
periodic transmissions of BSM. Generally, BSMs range from
180 to 300 bytes, with the potential to extend up to 1400 bytes
for specific service messages. The emphasis on communication
stems from the critical requirements for low latency and
high reliability, as communications have shown to surpass
sensors in significantly improving platoon safety. Enhancing
traffic flow stability can be achieved by accessing detailed
4
information from vehicles further ahead [36]. The distributed
nature of these systems means that the following vehicles
experience varying communication delays and are connected
through different numbers of V2V links within V2V-based
platooning frameworks. Such disparities in communication
quality can markedly affect the functionality of CACC sys-
tems. Specifically, excessive communication delays or in-
sufficient transmission rates can compromise string stability
and other critical performance metrics for designated time
gaps [28], [38]. Vehicular network topologies are inherently
dynamic and complex, marked by diverse uncertainties such as
communication delays, packet losses, and transmission errors
[39]. These factors collectively influence the efficacy of CACC
systems [40]–[42]. Interestingly, packet loss can effectively
be seen as introducing randomness into the IFT, altering the
communication dynamics within the platoon.
In the context of vehicular networks, the relay of traffic
data across multiple hops necessitates scalability solutions to
mitigate network congestion and minimize the transmission
of unnecessary data, thus optimizing network capacity. A
promising method for disseminating traffic information in-
volves embedding compressed data within regularly transmit-
ted BSMs, a technique known as piggybacking. However, this
approach does not guarantee packet delivery as the broadcast
mode lacks an acknowledgment (ACK) mechanism, leaving
the broadcasters unable to detect collisions, leading to potential
data loss without notification.
Addressing issues of channel congestion and packet loss
in V2V communication has inspired various strategies aimed
at maintaining network load at manageable levels and mini-
mizing packet losses to acceptable rates. Yet, many of these
solutions do not account for the distinct demands of specific
applications, which may necessitate customized IFT. This
oversight could particularly disadvantage applications like
platooning, where a consistent and dependable exchange of
information is crucial for operational integrity. In light of this,
there have been efforts to devise methods that enhance the
communication reach from the platoon leader to its members,
with the aim of strengthening the cohesive functionality of the
platoon [43]. However, these solutions often require mecha-
nisms such as handshakes, which may not be feasible for all
vehicular network applications.
1) Model-Based Communication (MBC)
To address packet losses in wireless vehicular communication,
one strategy is to minimize dependency on V2V communica-
tions by smoothly transitioning to alternative solutions. For
instance, the development of a velocity estimation algorithm
aims to compensate for communication disruptions by lever-
aging transmitted models. Instead of simply sending beacon
messages with updates on their position, vehicles can send a
comprehensive situational awareness map to their neighbors.
To compensate for packet losses, Ploeg et al. have leveraged
onboard sensors for estimating the acceleration of the vehicle
ahead, a measure typically reliant on V2V exchanges. For
periods of brief communication disruptions, techniques such
as the Kalman filter have been employed to estimate this
acceleration, enhancing CACC management [44]. Implement-
ing these strategies has shown to significantly improve string
stability. However, these CACC design strategies generally
do not incorporate the uncertainties associated with vehicle
states, behaviors, and communications [6], [25], [45]. Despite
acknowledging the uncertainty in communication delays, the
possibility of messages not reaching their intended recipients
has not been explicitly considered. The adoption of GP for pre-
dicting vehicle positions offers a novel approach by modeling
the variability within a driver’s behavior and the overall driving
environment [46]–[49]. Additionally, GPs have been proposed
for planning overtaking maneuvers in autonomous racing
scenarios [50]. GP, which generalizes multivariate Gaussian
distributions to an infinite dimensionality, presents several
advantages for system identification, including:
•The simplification of physics-based models thanks to
GP’s data-driven approach
•The use of Bayesian inference, which leverages marginal
likelihood to mitigate the risk of overfitting
•GP models are particularly suited for situations where
the data is scarce relative to the number of variables,
effectively managing data insufficiency and measurement
noise
In our study, the velocity of each cooperative vehicle over
time, indicated as vn(t), is modeled as a GP. This process is
characterized by a mean function, mn(t), and a covariance
kernel function, κn(t, t′), as follows:
vn(t)∼ GP (mn(t), κn(t,t′)) .(4)
Our focus is on integrating the insights derived from the ob-
served velocity data regarding the underlying function, vn(t),
and its future projections. We assume that for each cooperative
vehicle, the mean of the process is zero, mn(t)=0. We
use a Radial Basis Function (RBF) as the covariance kernel
and consider the measurement noises to be independent and
identically distributed (i.i.d.) following a Gaussian distribu-
tion, N(0, γ2
n,noise). Under these assumptions, the covariance
matrix for the observed velocity of the nth cooperative vehicle
can be expressed as follows:
Kn(t,t′) = κn(t, t′) + γ2
n,noiseI(5)
In this context, Irepresents the identity matrix, whose dimen-
sion matches that of the training (measured) data. Calculation
of κn(t, t′)can be performed on the basis of the definition of
the RBF, as follows:
κn(t, t′) = exp(−||t−t′||2
2γ2
n
).(6)
Under the assumptions mentioned above, we can represent
Vobs
n, and the future values, V∗
n, in the following way:
Vobs
n
Vn
∗∼ N 0,Kn(t,t)Kn(t,t∗)
Kn(t∗,t)Kn(t∗,t∗),(7)
In this formulation, tand t∗represent the time stamps
associated with the sets of observation and future values,
respectively. The function Kn(., .)is derived as per the kernel
matrix described in (5).
5
Fig. 3: Desired multi-hop forwarder
2) Relaying
In the realm of vehicular communication, single-hop transmis-
sion is predominantly employed for safety-critical applications
where low latency is essential, such as in collision-avoidance
systems that require instantaneous situational awareness [11].
Conversely, multi-hop communication serves applications ne-
cessitating the propagation of vehicle or road data over dis-
tances exceeding the limitations of single-hop transmissions.
This includes a range of safety and efficiency related applica-
tions, from emergency vehicle and post-crash warnings to traf-
fic information dissemination, vehicle tracking, traffic manage-
ment, and road monitoring. Given the paramount importance
of time-sensitive safety applications, research has predomi-
nantly concentrated on single-hop communication paradigms.
However, multi-hop communication, despite its broad appli-
cability, has often been overlooked in discussions surrounding
the distinct requirements and characteristics of both efficiency
and safety operations. This study specifically addresses the
challenge of integrating vital safety communications within a
framework that also accommodates the multi-hop transmission
of general traffic information and other non-critical data [51].
The efficacy of tracking applications and their tracking
precision depend on the volume of vehicle data successfully
transmitted and the extent of the network coverage that re-
ceives these data [52]. Information pivotal to multi-hop safety
and efficiency applications, such as details on traffic jams,
vehicular density, movement patterns, or obstructions, varies
in relevance according to the distance of the recipient from
the data source. For these applications, which can accom-
modate certain delays, the expedited relay of data across
vast distances does not present a clear advantage. Employing
piggyback techniques on beacon signals for data forwarding,
though not the fastest method, significantly reduces network
congestion by circumventing the need to generate additional
packet overhead. In the context of relay selection algorithms,
the existing methodologies can essentially be grouped into
three categories:
•Centralized: leveraging data from all agents
•Decentralized: depending solely on an agent’s individual
data
•Distributed: incorporating data from the agent as well as
its immediate neighbors.
Given the logistical challenges associated with harnessing state
information from all agents for relay selection, decentralized
and distributed approaches are deemed more feasible for
MASs.
Within the framework of a single relay approach, the
Platoon Member (PM) that accurately identifies the Platoon
Leader’s (PL) signal and is situated nearest to the platoon’s end
is designated as the relay. This method optimizes the channel
quality between the chosen relay and the receiving PMs,
facilitating superior reception capabilities compared to other
potential relays. Initial findings indicate that the communica-
tion of data related to objects at the sensory boundaries notably
enhances tracking efficacy [53]. In [54], the effectiveness of
various distance-based content selection strategies for vehicu-
lar map dissemination was examined. The studies revealed that
adopting a probabilistic method, prioritizing distant objects
within the transmitter’s perception map for message inclusion
significantly improves tracking precision.
Limitations of current state-of-the-art algorithms include
the necessity for handshake mechanisms and the aging of
broadcasted information. Yet, there’s a pressing demand for
the fast broadcasting of aggregated traffic data across multiple
hops. A notable study on side-link relays for platooning
introduced two relay strategies that leverage geographical data,
necessitating handshake protocols in which each recipient
must acknowledge receipt or non-receipt (ACK / NACK)
from the sender [55]. An improvement to the relay selection
process was suggested, identifying the optimal relay as the
node capable of reaching every intended recipient within a
specified destination zone [34]. Another approach considered
relay selection based on reception likelihood [43], though this
method incurred additional delay. Importantly, when a Road
Side Unit (RSU) serves as a broadcast relay, it operates without
expecting or processing any feedback, so it doesn’t engage
in re-transmissions in the event of packet loss. This work
concentrates on evaluating the effects of packet relay via RSUs
on application-level outcomes, specifically focusing on main-
taining inter-vehicle distances within platoons, which requires
both infrastructure support and the exchange of time-sensitive
information [56]. A scheme for platoon-centric cooperative
retransmission has been devised, using a 4-D Markov chain
to enable senders to correct errors in previous transmissions
for their platoon neighbors [57]. Additionally, a fundamental
strategy for achieving scalable platooning involves employing
information topologies with a consistent tree depth [58].
From the given illustration, it becomes apparent that select-
ing the most distant vehicle within the communication range
as the next relay might seem ideal for retransmission tasks.
Nevertheless, the reality that vehicles beyond a certain range
might suppress their broadcasts, anticipating that subsequent
vehicles will propagate the message, is complicated by the
fact that Packet Error Ratio (PER) invariably escalates with
distance. This situation requires that the vehicle chosen for the
ensuing transmission must consider both its proximity to the
initial sender and the likelihood of successful message receipt
by vehicles further afield. To satisfy these criteria, a distance-
dependent probabilistic method is suggested. Consequently,
the selection of the next relay does not automatically favor
the most remote vehicle; instead, it is based on a calculated
probability, taking into account the specific circumstances of
potential forwarders.
III. PRELIMINARIES AND PROBLEM FORMULATI ON
The significance of communication topology in coordinating
MASs forms the basis of our exploration in this paper, where
we delve into the role and attributes of a fixed communication
network topology on distributed consensus efficacy across
six distinct network models. The study examines the impact
6
of various IFTs on the performance of CACC systems. A
comprehensive analysis comparing different IFTs is essential
for guiding their selection, applicable regardless of the pla-
toon size. In standard vehicular communication frameworks,
message loss tends to escalate with distance from the source,
adversely affecting tracking performance. Various strategies
have been proposed to enhance the signal reach from the
PL to other members, especially when managing platoons
at the limit of the C-V2X capacity. Platoon management is
essential when the CACC system approaches the maximum
number of vehicles that can be supported by stable C-V2X
communication range. This paper details efforts to reduce
Information Age (IA) using a network-aware approach that
incorporates probabilistic and distance-dependent strategies.
We define the actors as follows.
•Platoon Leader (PL): This CAV controls the platoon’s
speed.
•Platoon Member (PM): A CAV following the PL’s di-
rectives. It also communicates with adjacent vehicles and
relays messages to the next vehicle.
As illustrated in Figure 1, platoon vehicles can simultaneously
transmit and receive data, acting as relays. We categorize
communications into two-fold:
•PL’s broadcasts: Broadcasting by PL enables transmission
of information to PMs within range.
•Direct V2V interactions among PMs: It’s assumed that
PMs communicate through one-hop V2V connections.
A. Channel Model
Researchers adopt a basic communication framework to exam-
ine how factors such as actuation delay, message frequency,
and communication latency influence the minimal viable
spacing between vehicles. For instance, analyses incorporate
a comprehensive IEEE 802.11p simulation model, which,
for specific application needs, allows for the substitution of
different link quality estimators [53], [59]–[61]. The need
for multiple hops increases as the distance between the lead
and the last vehicle in the platoon increases, a scenario more
common in long platoons [62]. For the purposes of channel
modeling in this study, we rely on PER data derived from prior
research [24], [63], utilizing PER as the channel model [64].
Figure 4 illustrates the PER for different vehicle densities at
various distances. As expected, the PER increases with both
an increase in distance and density.
Fig. 4: PER for different traffic flows at various distances
B. Vehicle Model & Model Predictive Control Design
At the heart of these vehicular systems lies the objective to
compute the vehicle’s control inputs through solving an opti-
mization problem framed within a Model Predictive Control
(MPC) framework, which accounts for the future driving pat-
tern of the vehicle ahead. Within these MPC-driven systems,
the forthcoming state of the vehicle ahead is anticipated, and
by solving an optimization problem, the necessary control
inputs for the following vehicle are determined. Consequently,
this approach facilitates a predictive car-following pattern,
enhancing vehicular efficiency by adaptively adjusting the
distance and velocity.
1) Vehicle Model
In this study, we consider a platoon of Nvvehicles, where
n∈ {0,1, . . . , Nv−1}denotes the nth vehicle in the platoon,
and n= 0 represents the platoon leader. As shown in Figure
1, dndenotes the gap between nth and (n−1)th vehicles and
is defined as
dn=xn−1−xn−lv
n,(8)
where xndenotes the longitudinal position of the rear bumper
of the nth vehicle, and lv
nits length. The strategy adopted here
employs a fixed time headway for spacing between vehicles
to enhance string stability and safety. This desired spacing is
articulated as
d∗
n(t) = τnvn(t) + ds
n.(9)
where vn(t)is the speed of the nth vehicle, τnthe time
headway, and ds
nthe standstill distance. The deviation in
spacing from its desired value is ∆dn(t) = dn(t)−d∗
n(t), and
the difference in speed between the nth vehicle and the one
preceding it is ∆vn(t) = vn−1(t)−vn(t). Consequently, the
change in this spacing becomes ∆˙
dn(t) = ∆vn(t)−τnan(t)
and the velocity difference evolves as ∆ ˙vn=an−1−an, where
anindicates the acceleration of the nth vehicle. Incorporating
the driveline dynamics fn, the acceleration rate for vehicle n
is formulated as ˙an(t) = −fnan(t) + fnun(t), with un(t)
serving as the control input. Defining Sn= [∆dn,∆vn, an]T
as the state vector for the nth vehicle, its dynamic behavior
is captured by the state-space model
˙
Sn(t) = AnSn(t) + Bnun(t) + D an−1(t)
=
0 1 −τn
0 0 −1
0 0 −fn
Sn(t) +
0
0
fn
un(t) +
0
1
0
an−1(t).
(10)
For the leading vehicle (n= 0), the term an−1(t)is considered
to be zero. Transitioning to a discrete-time model using a first-
order forward approximation results in
Sn(k+ 1) =
(I+tsAn)Sn(k) + tsBnun(k) + tsD an−1(k),(11)
with tsdenoting the sampling time.
7
2) Model Predictive Control (MPC)
This research incorporates various constraints related to the
system’s state and inputs, which include limitations on accel-
eration, control inputs, the maximum speed allowed, and the
spacing between vehicles (with the understanding that negative
spacing implies a collision scenario, which is unacceptable).
These constraints are formalized as follows:
amin
n≤an(k)≤amax
n,(12a)
umin
n≤un(k)≤umax
n,(12b)
vn(k)≤vmax,(12c)
dn(k)>0.(12d)
Furthermore, to ensure passenger comfort, the variation in
system inputs is constrained by
tsumin
n≤un(k+ 1) −un(k)≤tsumax
n.(13)
The objective of the MPC for each vehicle involves minimiz-
ing the control inputs over a predictive horizon, stated as
min
un
N−1
X
k=0 h(Sn(k)−Rn)TQn(Sn(k)−Rn)i
subject to: System Constrains, (14)
where unencompasses the control inputs from k= 0 to k=
N−1, where Nis prediction horizon (refer to Table I).
Remark 1: The MPC framework employed here adopts
a single look-ahead approach. Altering the cost function in
(14) allows for the incorporation of an rlook-ahead strategy,
detailed as
N−1
X
k=0 "(Sn(k)−Rn)TQn(Sn(k)−Rn)
+
n−1
X
i=n−r
hcd
ixi(k)−xn(k)−
n
X
j=i+1
(d∗
j(k) + lv
j)2
+cv
ivi(k)−vn(k)2i#,(15)
where cd
iand cv
iare positive scalars, and rindicates the count
of predecessors sharing information with the nth vehicle. This
approach enables a vehicle to adjust its velocity and spacing
in relation to its rpreceding vehicles. Notably, if the number
of available predecessors is fewer than r, the vehicle adjusts
its calculations accordingly.
C. Networking and Relaying Structures
When communication stability is compromised, CAVs au-
tonomously disengage from the PL, reverting to ACC while
initiating a new platoon formation. To counteract this, em-
ploying relay techniques for long platoons becomes crucial.
Furthermore, signals from the PL undergo significant path
loss over extended distances, impeding the ability of PMs
to accurately decode the leader’s signals. Relay technologies
serve as a vital solution to mitigate path loss challenges.
The effectiveness of communication systems is evaluated
based on three primary criteria:
•Reliability and Reachability: In scenarios where a
network maintains full connectivity alongside frequent
dissemination of single-hop safety beacons (BSMs), tech-
niques leveraging piggybacking on these beacons can
attain 100% reachability.
•Scalability: The forwarder ratio, indicative of the propor-
tion of vehicles retransmitting a message against the total
number of vehicles within a single-hop range, is critical.
Scalability ensures that the concept of string stability
remains unaffected by variations in platoon size, allowing
for the seamless integration or disengagement of vehicles.
•Information Age (IA): Defined as the time elapsed since
the last update received from vehicle jin vehicle i, IA
reflects the freshness of the information. Upon receiving
new data on the vehicle j, IA temporarily equals the mes-
sage delivery latency, then increases until the next update.
Maximum IA is influenced by both broadcast frequency
and forwarding delays, with our forwarding algorithm
aiming to minimize IA while considering channel and
network conditions to prevent congestion.
Addressing the challenge of minimizing status update delays
is complex, given that beacons are disseminated via broad-
cast without support for acknowledgments or retransmissions.
Adjustments to the transmission rate or range are typically
managed by single-hop safety protocols. However, our ap-
proach allows for control over message size by including BSM
and additional data within each packet. Studies on Dedicated
Short-Range Communications (DSRC) indicate that augment-
ing a DSRC beacon by 100 bytes barely affects its overall
performance. The relationship between the augmentation of
piggybacked data volume and the increment in forwarding
delay follows a logarithmic pattern [65], and employing pig-
gybacking rather than expanding single-hop beacon coverage
effectively reduces channel congestion [66], [67].
Research shows that the Information Dissemination Ratio
(IDR) initially increases with channel load up to a certain
optimal point before drastically declining [68]. This phe-
nomenon highlights the negative implications of overloading
the communication channel, such as elevated packet loss
rates, which increase IA and diminish network reachability.
Ideally, without channel load constraints, the simplistic flood
broadcasting technique would emerge as superior, character-
ized by immediate retransmission of all received messages by
each vehicle, promising the utmost reachability and minimal
IA, assuming channel capacity isn’t exceeded. Consequently,
an optimal communication strategy should not only achieve
extensive reachability and minimal IA but also regulate the
channel load to hover near the peak IDR value. Nonetheless,
due to bandwidth inefficiencies and the risk of generating du-
plicate packets, flooding is impractical for vehicular networks.
Consequently, vehicular broadcast protocols typically impose
limits on the number of data-forwarding participants to ensure
system scalability.
Given the dynamic nature of network conditions that may
vary for each vehicle, the determination of whether a vehicle
should act as the next forwarder is made on an individual basis,
following a probabilistic approach. Upon receiving a message
for the first time, a vehicle undergoes a Bernoulli trial with a
8
predetermined probability of success to decide on the rebroad-
cast of the message. In the case of a uniform probabilistic
rebroadcast strategy, this probability remains constant across
all vehicles. Hence, each vehicle, performs a Bernoulli trial
with a success probability pi=δ/M, where Mdenotes the
count of vehicles within a single-hop communication vicinity,
and δ∈[1, M ]established in accordance with the specific
demands of the multihop safety or traffic applications, serves
as the reliability factor of the approach. The logic dictates that
an increased δenhances the probability of message rebroadcast
across vehicles, as the success probability is uniformly applied.
Consequently, a higher δnot only boosts the forwarder ratio
and the network’s load but also augments the probability of
beacon retransmission, thereby enhancing reliability at the cost
of increased network load. This underscores why δis referred
to as the reliability factor.
All algorithms for multi-hop broadcasting in vehicular
networks share two fundamental procedures: broadcasting a
message to all nearby vehicles and identifying the subsequent
forwarder for further message propagation. These processes
aim to maximize reachability and minimize both the IA and
network congestion. Broadly, vehicular multicast protocols fall
into two categories: topology-based and position-based meth-
ods. The former selects forwarders based on network topology,
whereas the latter depends on geographic information, such
as the positions of the source, destination, and neighboring
nodes, along with the coordinates defining the multicast area
[69]. An exemplary position-based method is the ETSI Geo-
networking geo-broadcast, a straightforward form of flooding,
known for achieving high packet delivery rates but potentially
leading to a broadcast storm issue. To ensure high packet
delivery while reducing the number of transmitting nodes,
certain nodes are chosen based on topological and connectivity
criteria. The node predicted to have the highest reception
probability is designated as the relay. This selection method
has been shown to provide greater reliability than location-
based relaying, albeit at the cost of increased latency.
In our approach, vehicles probabilistically include the mes-
sages received from the PL in their subsequent beacons.
Specifically, the success probability is calibrated such that,
within each beacon interval, at least one vehicle is guaranteed
to rebroadcast the message, ensuring the mean number of re-
broadcasts exceeds one. To underscore the effectiveness of our
method, it’s crucial to assess how the selection of the next relay
within the single-hop range influences IA. For instance, within
a counter-based rebroadcast strategy, imagine ksequentially
aligned vehicles within the same communication range (refer
to Figure 3). If the foremost vehicle rebroadcasts a message
received from a previous hop, the decision of the remaining
(k−1) vehicles to relay this message or defer to others
significantly impacts IA. In particular, IA is affected differently
if the rearmost vehicle, as opposed to one immediately behind
the leader, rebroadcasts the message, as the former scenario
extends the information further, whereas the latter limits dis-
semination to a shorter range, potentially inhibiting subsequent
relays. Hence, the probability that vehicle iincludes a message
from vehicle jin its next beacon is given by
pi(j) = min ISPF(d(i, j))
1−PERi(j),1(16)
where ISPF(d(i, j)) denotes the Ideal Success Probability
Function for vehicle irebroadcasting [53]. Assuming perfect
channel conditions and no packet loss, the success rate would
mirror the ISPF output, calculated solely on the vehicular
distance since the denominator would consistently be one.
However, in practice, as attenuation and the likelihood of
collisions increase with distance, leading to a higher PER,
the rebroadcast probability is adjusted to counteract reduced
transmission chances. The use of the minimum function en-
sures that the success probability does not exceed one. It’s
noteworthy that despite the variation in success probability in
(16), the Bernoulli trials remain independent, aligning with
the Poisson binomial distribution across N trials in probability
theory.
Initially, the ISPF could align with uniform probabilistic
approaches, assigning identical chances of success across all
vehicles. Yet, this model would inherently favor nearer vehi-
cles due to their lower PER, as the numerator in (16) remains
constant while the denominator decreases as PER increases by
distance. Although PER’s behavior is contingent upon network
conditions, our protocol adjusts the ISPF to counteract PER’s
influence significantly. Preferably, the protocol aims to select
the most distant vehicle within a single-hop range for message
forwarding. By appropriately increasing the ISPF’s numerator
with distance, the adverse impact of PER is neutralized,
enhancing the forwarding probability for distant vehicles. Our
proposed ISPF is defined as
ISPF(d(i, j)) =
1−exp λi×d(i,j)
D(Ptx)
1−exp (λi), λi>0,(17)
where d(i, j)represents the Euclidean distance between vehi-
cles iand j, and D(Ptx)symbolizes the single-hop commu-
nication range (in meters), with Ptx indicating transmission
power. The scale parameter λiis adaptively modified based
on the density and average velocity of vehicles within vehicle
i’s single-hop range. Notably, for λi>0,(1 −exp(λi)) <0,
the function shows a monotonically increasing trend with
distance. In scenarios of free-flow traffic, λimirrors traffic
flow (vehicles per second), with inter-arrival times following
an exponential distribution of parameter λiand average speed
(¯v) is independent of the traffic flow and density. The cal-
culation of λithus becomes λi= ¯vρ, integrating average
speed and density metrics [62], [70]. Specifically, ISLF is
designed to evaluate a PM’s potential as a relay, focusing on
its ability to extend communication range and sustain robust
connectivity with the PL. The rebroadcast probability hinges
not solely on ISPF and distance but also on PER and prevailing
channel conditions. Thus, the farthest vehicle may not always
be chosen as the next forwarder if other vehicles present a
higher forwarding probability under their specific conditions.
Thus, any PM that decodes the leader’s message is considered
a potential relay, adhering to a decode-forward relay strategy
(refer to Figure 3, where the more red cars have a greater
likelihood of relay).
9
For effective platoon management and vehicle coordination,
we introduce six schemes dedicated to updating information
among vehicles. Furthermore, through the utilization of MBC
strategies that incorporate predictive information, it is feasible
to neutralize the delays encountered by receivers. The develop-
ment of these novel protocols is driven by the unique demands
of their application context.
1) Predecessor-Follower (PF)
The initial scheme, termed PF, serves as the foundational
benchmark against which the subsequent proposed schemes
are evaluated. This approach is depicted in Figure 2a.
2) Predecessor-Leader Follower (PLF)
Adopting a unidirectional leader topology, where information
from the leader is disseminated to every follower, enhances
platoon scalability by ensuring a stability margin that remains
positive and unaffected by the size of the platoon. This
arrangement allows for an agile platoon response in emergency
situations by dynamically prioritizing the leader’s data in
influencing follower behavior, as depicted in Figure 2b.
3) Ten look-ahead IFT
Exploring an approach where each vehicle in the platoon
leverages data from a limited number of vehicles directly
ahead, as opposed to relying solely on the PL, presents
promising outcomes, as depicted in Figure 1. Such a methodol-
ogy, focusing on communication with immediate predecessors
rather than the PL, has demonstrated efficacy. It is established
that adopting a strategy that involves the input of multiple
preceding vehicles for local control can effectively reduce the
time headway gap, ensuring string stability [71]. The result
is a platoon that exhibits faster reaction times and greater
reliability, especially in the face of communication delays.
Specifically, the approach is designed to interact with up to
ten directly preceding vehicles; if fewer vehicles are ahead, it
communicates with all available ones.
4) Virtual Leader Predecessor-Follower (VLPF) with ten
vehicle length
Investigations of the rlook-ahead IFT, as depicted in Fig-
ure 1, where ’r’ denotes the count of predecessor vehicles
considered in the control strategy, indicate limitations in
expanding platoon sizes without increasing the gap between
vehicles. This restriction comes from reduced reception rates
for vehicles located far from the PL [72]. The strategy involves
designating certain PMs as ”virtual leaders,” enabling direct
communication with members beyond the reach of the original
platoon leader’s signal, circumventing the need for multi-
hop communications. Essentially, vehicles following a virtual
leader perceive it as their immediate lead, aligning their speed
and acceleration accordingly to maintain optimal spacing. The
essence of VLPF lies in segmenting extensive platoons into
smaller and more manageable units under the guidance of ap-
pointed virtual leaders, ensuring short distances between vehi-
cles and their designated leaders for robust communication and
facilitating longer platoon configurations. Evidence suggests
that an increase in virtual leader allocation correlates with
reduced collision incidences [73], as conceptually depicted in
Figure 5.
Fig. 5: Depiction of a VLPF platoon showcasing the arrange-
ment of virtual leaders
5) Leader data relaying
Opting for the most distant node within the sender’s communi-
cation range as the relay may compromise message reception
due to propagation losses. A viable strategy to ensure that
platoon members beyond the leader’s communication reach
can still access messages from the leader involves embedding
these messages within beacon broadcasts during subsequent
intervals. In the probabilistic distance-dependent model we
propose, the success probability of message re-broadcast,
determined through Bernoulli trials, varies across vehicles.
For any given vehicle i, the decision to rebroadcast the
received leader message from vehicle jis influenced by the
distance separating the two vehicles and the PER associated
with that distance. PER quantifies the fraction of packets lost
relative to the total expected packets within a given interval,
calculated locally by vehicle ifor messages from vehicle
j, as indicated by P ERi(j)in (16). Tracking the sequence
numbers of received packets can help determine the extent of
packet loss during the assessment period. Studies show that
probabilistic methodologies yield superior mapping precision
compared to deterministic strategies [54]. Our formulation
employs a hybrid probabilistic model that prioritizes rapid
propagation at each relay point, integrating network conditions
and the criticality of the data into the decision-making process.
Thus, it allows for decentralized, collaborative decisions on
retransmissions, emphasizing the role of packet forwarding,
as facilitated by beacon piggybacking, in enhancing platoon
performance by minimizing inter-vehicle distances.
6) Leader relaying with ten look-ahead IFT
Incorporating predictive data from both the PL and preced-
ing vehicles can substantially enhance the string stability of
the platoon [19]. Additionally, the platoon demonstrates an
improved string stability margin even with increased com-
munication delays [28]. Establishing connectivity between
one vehicle and a broad network of others is crucial to
minimize the propagation of spacing errors. A key benefit
of the described communication protocols is their facilitation
of cooperative awareness among vehicles without imposing
excessive demands on network capacity. This is achieved
through a network design that integrates numerous direct
V2V connections alongside a piggybacking relayer, which
distributes the leader’s messages across the entire platoon.
IV. EXP ER IM EN TAL RESULTS
In this analysis, we delve into the operational efficacy of
cooperative platoon systems in a mass network environment,
analyzing the impact of various IFTs on the dimensions and
effectiveness of platoons. To benchmark our approach, which
integrates broadcasting by the PL with V2V communications,
a comparison is drawn against other approaches listed above.
This relaying scheme permits direct communication between
10
Fig. 6: Performance of a 100-vehicle platoon. Each line shows the error, speed, and acceleration of an individual vehicle over
time
Fig. 7: IA for each vehicle ID at different traffic flow rates.
adjacent vehicles via one-hop links, while those not directly
connected rely on multi-hop communication, with intermediate
vehicles facilitating message transmission. This method is
prevalent in decentralized vehicular networks, such as Vehic-
ular Ad Hoc Networks (VANETs). Comparing our relaying
communication strategy with the broadcast-centric approach
is a logical step to highlight the merits of the proposed
system. Simulations serve as a valuable method for validation,
considering the substantial costs and extensive efforts required
for real-world deployments [11]. With a particular emphasis
on cooperative safety, our analysis aims to link communication
patterns with the extent of tracking errors. It is understood
that tracking performance improves with faster delivery of
information to tracking estimators. Therefore, our objective
is to measure the volume of data each vehicle successfully
transmits to its intended recipients, a metric crucial to the
precision of the tracking application.
A. Implementation Details
Within a platoon, the PL sets the speed (V0), acting as the
velocity benchmark for the PMs. PL’s adjustments in speed,
TABLE I: The value of the parameters used in the model and
optimization in the simulations.
Parameter Value Parameter Value
N10 ts0.1s
lv
n5m ds
n2m
amax
n3m/s2amin
n−4m/s2
umax
n3m/s2umin
n−4m/s2
fn10 s−1
either through acceleration or deceleration, act as perturbations
within the platoon system. The simulations run at a step
interval of 100 ms, identified as the standard period for
communication exchanges. To concentrate on evaluating the
effectiveness of our method, we standardize the transmission
frequency and maximum range at 10Hz and D(P tx) = 500
m, respectively. Consequently, the transmission of a piggy-
backed message by any PM requires no more than 100 ms.
Therefore, conveying a message from the lead to a member
situated nhops away incurs a maximal delay of 100 ×n
ms. This latency impacts the platoon’s capacity to maintain
precise and stable spacing between vehicles. In the realm of
broadcasting and multicasting routing, transmission efficiency
and data throughput emerge as critical factors, especially
relevant to vehicle tracking within Cooperative Vehicle Safety
Systems (CVSS). Here, the new messages will make the
previous messages in the queue obsolete, so only the most
up-to-date message is transmitted.
Multiple metrics, such as Channel Busy Ratio (CBR),
delay, PER, and Inter-Transmit Time (ITT) are indicative of
communication reliability challenges. Specifically, CBR and
ITT influence communication quality as evidenced by their
impact on PER, with the probability of communication success
being derivable from an empirical channel model documented
in [64]. Furthermore, to evaluate the implications of unreliable
communication, we reference PER values corresponding to
11
Fig. 8: Relay ratio over time for different traffic flow rates.
traffic densities of 10,15,20,and 30 vehicles per second from
prior research [24], [28], [63] as our comparative benchmarks.
Figure 4 illustrates the PER associated with the specified traffic
densities. For solving the optimization problem, we employ the
CVXPY package within Python, with the Gurobi optimization
suite serving as the solver for mixed-integer programming
problems [74], [75]. The system’s resilience is tested in a
scenario where the PL encounters a speed disturbance. Table
I contains the parameters utilized in the simulations. Each
simulation scenario lasts for 180 seconds, during which the
platoon’s goal is to maintain a desired gap time of 0.6seconds
with the preceding vehicle.
B. Analysis and Results
Our research primarily assesses application-specific perfor-
mance metrics, notably inter-vehicular spacing within a pla-
toon, to gauge the effectiveness of communication protocols.
In contrast to a baseline approach that uses direct V2V
connections, our study explores the benefits of implementing
a relay-based communication strategy. Our protocol requires
just the timestamp, location, and speed data from PL. We
characterize performance in terms of latency and IA, crucial
metrics for delay-sensitive applications. Each act of message
forwarding introduces latency, incrementally delaying the mes-
sage’s arrival at its intended recipient. IA is defined as the time
elapsed between the message’s creation and its reception with
respect to a unique time reference. Subsequently, we assess the
resilience of our CACC system across various communication
links, using inter-vehicle distance as the primary measure
of service quality. During packet loss episodes, the MBC
approach serves as a control signal estimator.
Figure 6 shows the dynamics of a 100-vehicle platoon
over time, illustrating error, speed, and acceleration for each
vehicle. Initially, there is significant error in inter-vehicle
distances, but this decreases as vehicles adjust their positions,
indicating stabilization. The speed and acceleration graphs also
show large initial variations that converge over time, reflecting
synchronized movements. This convergence ensures uniform
speed and acceleration, maintaining overall stability and safety
in the platoon. A notable finding was that the establishment
of appropriate spacing by platoon members is sequentially
dependent on their preceding vehicles achieving this metric, as
Fig. 9: Relay ratio versus vehicle ID for different traffic flow
rates.
the driving control mechanism relies on the leading vehicle’s
velocity and acceleration.
Figure 7 shows the IA for each vehicle ID at different
traffic flow rates. IA, the time since the last update, increases
with both the distance from the source and the traffic flow
rate. As messages pass through more hops, delays accumulate,
resulting in older information for vehicles further from the
source. Higher traffic flow rates further increase IA due to
network congestion and more frequent packet losses. Thus,
greater distances and higher traffic flows lead to older and
potentially less reliable information for vehicles further down
the platoon.
Figure 8 shows the relay ratio over time for different
traffic flow rates. The relay ratio, indicating the percentage of
vehicles choosing to relay messages, decreases when vehicles
are close together and stationary (around 60-100 seconds in
Figure 6) because the need for relays diminishes. As vehicles
move apart, the relay ratio increases, reflecting more frequent
message relays. While the increase in traffic flows slightly
increases the relay ratio, its impact is less significant than
the effect of vehicle spacing. Thus, the relay ratio is mainly
influenced by vehicle spacing and secondarily by traffic flow
rate, with notable changes during transitions between stopped
and moving states.
Figure 9 shows the relay ratio versus vehicle ID for different
traffic flow rates. The relay ratio, indicating the percentage
of vehicles successfully relaying messages, decreases progres-
sively towards the end of the platoon. Notably, beyond vehicle
ID 80, the relay ratio drops to nearly zero, indicating that
relaying messages further becomes useless. While an increase
in traffic flow slightly increases the relay ratio due to more
frequent retransmissions needed to ensure message delivery,
its impact is less significant compared to the distance from
the platoon leader.
Table II compares errors in distance, speed, and acceleration
for various IFT approaches under different traffic flow rates.
The PF approach shows consistent errors, with slightly higher
acceleration errors at higher flows. The PLF method reduces
errors slightly compared to PF, especially in higher traffic
flows, indicating better stability. The ten look-ahead strategy,
where each vehicle considers data from up to ten vehicles
ahead, shows further error reduction, improving platoon stabil-
ity. The VLPF approach, which uses virtual leaders to extend
communication range, improves error metrics, especially in
acceleration. Implementing a relaying strategy reduces errors
in all metrics, particularly acceleration, indicating more stable
12
TABLE II: Statistics for various IFT approaches with varying traffic flow
Traffic Flow 10 veh/sec 15 veh/sec 20 veh/sec 30 veh/sec
Error (m,m/s,m/s2) Error (m,m/s,m/s2) Error (m,m/s,m/s2) Error (m,m/s,m/s2)
PF 0.1025, 17.1040, 1.5802 0.1022, 17.1043, 1.5914 0.1025, 17.1043, 1.5816 0.1024, 17.1043, 1.5909
PLF 0.0774, 17.0912, 1.5765 0.0773, 17.0918, 1.5883 0.0775, 17.0916, 1.5798 0.0773, 17.0917, 1.5858
Ten look-ahead 0.0714, 17.0527, 1.5409 0.0710, 17.0550, 1.5533 0.0714, 17.0535, 1.5468 0.0713, 17.0539, 1.5521
VLPF 0.0725, 17.0537, 1.5749 0.0722, 17.0542, 1.5954 0.0725, 17.0542, 1.5762 0.0725, 17.0542, 1.5840
Relaying 0.0543, 17.0498, 1.4852 0.0540, 17.0503, 1.4957 0.0542, 17.0503, 1.4886 0.0542, 17.0503, 1.4914
Relaying + Ten look-ahead 0.0460, 17.0431, 1.4892 0.0457, 17.0438, 1.4968 0.0460, 17.0436, 1.4916 0.0458, 17.0436, 1.4953
communication and control. The combination of relaying and
ten look-ahead yields the lowest errors across all metrics and
traffic flows, demonstrating the highest stability and efficiency.
V. CONCLUSION
This study develops a management strategy for CACC-
equipped long platoons, addressing the constraints posed by
the limited range of C-V2X communications. Within such
MPC-driven systems, control inputs are derived by optimiz-
ing based on predicted states of preceding vehicles, thereby
enabling anticipatory following behavior that enhances ve-
hicular efficiency through adaptive speed and spacing ad-
justments. Unlike ACC systems, CACC platoon controllers
that use different IFTs achieve narrower desired time gaps
without compromising stability, thus significantly enhancing
traffic capacity and safety through effective communication.
Research indicates that while ACC systems offer negligible
improvements in traffic flow and stability, CACC systems
present substantial benefits. It is thus essential to restrict the
degradation of CACCs to ACCs to maintain these advantages,
with an emphasis on expanding platoon sizes and communi-
cation scopes. This paper introduces a CACC model based on
specific IFTs, underscoring the need to minimize intervehicle
distances through reliable and low-latency relaying of safety
messages, thus addressing the dissemination of such messages
within the platoon.
The stability and safety of the platoon, pivotal in platooning
applications, hinge on the vehicles’ capability to share status
updates via beacon exchanges. High CBR can precipitate
significant PER and IA, signaling unreliable communications.
This underscores the necessity for meticulously designed com-
munication strategies in long platoons. The paper outlines an
algorithm for rebroadcasting aggregated traffic data across ve-
hicular networks, leveraging piggybacking on existing network
beacons for enhanced latency and scalability. This approach
employs a network-aware, dual-layer, distance-dependent pro-
tocol combined with a mixed probabilistic rebroadcast method
that ensures higher rebroadcast probabilities for vehicles fur-
ther from the sender and in optimal communication conditions,
aiming for rapid hop-to-hop forwarding. Additionally, it has
been discovered that an expanded communication range can
mitigate disturbances, and introducing more communication
dimensions can prevent string instability, commonly associated
with shock waves that detrimentally affect driver and passen-
ger comfort.
Limitations and Future Work: The redundancy in beacon
reception due to other vehicles’ retransmissions can diminish
efficiency, particularly in larger platoons. The study reveals
that reducing update delays grows more beneficial as pla-
toon size increases, necessitating further investigation into
this effect. The influence of different topologies on control
inputs suggests that distributed control is inherently topology-
dependent. Additionally, the impact of topology on disturbance
propagation requires thorough examination.
REFERENCES
[1] Z. Wang, Y. Bian, S. E. Shladover, G. Wu, S. E. Li, and M. J.
Barth, “A survey on cooperative longitudinal motion control of multiple
connected and automated vehicles,” IEEE Intelligent Transportation
Systems Magazine, vol. 12, no. 1, pp. 4–24, 2020.
[2] D. Jia, K. Lu, J. Wang, X. Zhang, and X. Shen, “A survey on platoon-
based vehicular cyber-physical systems,” IEEE Communications Surveys
& Tutorials, vol. 18, no. 1, pp. 263–284, 2016.
[3] J. V. Werf, S. E. Shladover, M. A. Miller, and N. Kourjanskaia, “Effects
of adaptive cruise control systems on highway traffic flow capacity,”
Transportation Research Record, vol. 1800, no. 1, pp. 78–84, 2002.
[Online]. Available: https://doi.org/10.3141/1800-10
[4] S. E. Shladover, D. Su, and X.-Y. Lu, “Impacts of cooperative adaptive
cruise control on freeway traffic flow,” Transportation Research
Record, vol. 2324, no. 1, pp. 63–70, 2012. [Online]. Available:
https://doi.org/10.3141/2324-08
[5] B. van Arem, C. J. G. van Driel, and R. Visser, “The impact of
cooperative adaptive cruise control on traffic-flow characteristics,” IEEE
Transactions on Intelligent Transportation Systems, vol. 7, no. 4, pp.
429–436, 2006.
[6] S. Mosharafian, M. Razzaghpour, Y. P. Fallah, and J. M. Velni, “Gaus-
sian process based stochastic model predictive control for cooperative
adaptive cruise control,” in 2021 IEEE Vehicular Networking Conference
(VNC), 2021, pp. 17–23.
[7] S. Feng, Y. Zhang, S. E. Li, Z. Cao, H. X. Liu, and L. Li, “String stability
for vehicular platoon control: Definitions and analysis methods,” Annual
Reviews in Control, vol. 47, pp. 81–97, 2019. [Online]. Available:
https://www.sciencedirect.com/science/article/pii/S1367578819300240
[8] W. J. Schakel, B. van Arem, and B. D. Netten, “Effects of cooperative
adaptive cruise control on traffic flow stability,” in 13th International
IEEE Conference on Intelligent Transportation Systems, 2010, pp. 759–
764.
[9] B. Ebrahimi Soorchaei, M. Razzaghpour, R. Valiente, A. Raftari,
and Y. P. Fallah, “High-definition map representation techniques for
automated vehicles,” Electronics, vol. 11, no. 20, 2022. [Online].
Available: https://www.mdpi.com/2079-9292/11/20/3374
[10] B. Toghi, D. Grover, M. Razzaghpour, R. Jain, R. Valiente, M. Zaman,
G. Shah, and Y. P. Fallah, “A maneuver-based urban driving dataset
and model for cooperative vehicle applications,” in 2020 IEEE 3rd
Connected and Automated Vehicles Symposium (CAVS), 2020, pp. 1–
6.
[11] A. Jami, M. Razzaghpour, H. Alnuweiri, and Y. P. Fallah,
“Augmented driver behavior models for high-fidelity simulation
study of crash detection algorithms,” 2022. [Online]. Available:
https://arxiv.org/abs/2208.05540
[12] R. Valiente, B. Toghi, M. Razzaghpour, R. Pedarsani, and Y. P. Fallah,
“Learning-based social coordination to improve safety and robustness
of cooperative autonomous vehicles in mixed traffic,” 2022. [Online].
Available: https://arxiv.org/abs/2211.11963
[13] R. Valiente, M. Razzaghpour, B. Toghi, G. Shah, and Y. P.
Fallah, “Prediction-aware and reinforcement learning based altruistic
cooperative driving,” 2022. [Online]. Available: https://arxiv.org/abs/
2211.10585
13
[14] V. Milan ´
es, S. E. Shladover, J. Spring, C. Nowakowski, H. Kawazoe,
and M. Nakamura, “Cooperative adaptive cruise control in real traffic
situations,” IEEE Transactions on Intelligent Transportation Systems,
vol. 15, no. 1, pp. 296–305, 2014.
[15] J. Lioris, R. Pedarsani, F. Y. Tascikaraoglu, and P. Varaiya, “Platoons
of connected vehicles can double throughput in urban roads,”
Transportation Research Part C: Emerging Technologies, vol. 77,
pp. 292–305, 2017. [Online]. Available: https://www.sciencedirect.com/
science/article/pii/S0968090X17300384
[16] N. Lyamin, Q. Deng, and A. Vinel, “Study of the platooning fuel
efficiency under etsi its-g5 communications,” in 2016 IEEE 19th Inter-
national Conference on Intelligent Transportation Systems (ITSC), 2016,
pp. 551–556.
[17] F. H. Robertson, F. Bourriez, M. He, D. Soper, C. Baker, H. Hemida,
and M. Sterling, “An experimental investigation of the aerodynamic
flows created by lorries travelling in a long platoon,” Journal of Wind
Engineering and Industrial Aerodynamics, vol. 193, p. 103966, 2019.
[Online]. Available: https://www.sciencedirect.com/science/article/pii/
S0167610519305641
[18] J. Axelsson, “Safety in vehicle platooning: A systematic literature re-
view,” IEEE Transactions on Intelligent Transportation Systems, vol. 18,
no. 5, pp. 1033–1045, 2017.
[19] M. Razzaghpour, S. Mosharafian, A. Raftari, J. M. Velni, and Y. P.
Fallah, “Impact of information flow topology on safety of tightly-
coupled connected and automated vehicle platoons utilizing stochastic
control,” in 2022 European Control Conference (ECC), 2022, pp. 27–33.
[20] R. Zheng, K. Nakano, S. Yamabe, M. Aki, H. Nakamura, and Y. Suda,
“Study on emergency-avoidance braking for the automatic platooning
of trucks,” IEEE Transactions on Intelligent Transportation Systems,
vol. 15, no. 4, pp. 1748–1757, 2014.
[21] D. Jia, K. Lu, J. Wang, X. Zhang, and X. Shen, “A survey on platoon-
based vehicular cyber-physical systems,” IEEE Communications Surveys
Tutorials, vol. 18, no. 1, pp. 263–284, 2016.
[22] Y. P. Fallah, C. Huang, R. Sengupta, and H. Krishnan, “Design
of cooperative vehicle safety systems based on tight coupling
of communication, computing and physical vehicle dynamics,” in
Proceedings of the 1st ACM/IEEE International Conference on Cyber-
Physical Systems, ser. ICCPS ’10. New York, NY, USA: Association
for Computing Machinery, 2010, p. 159–167. [Online]. Available:
https://doi.org/10.1145/1795194.1795217
[23] Y. P. Fallah and R. Sengupta, “A cyber-physical systems approach to
the design of vehicle safety networks,” in 2012 32nd International
Conference on Distributed Computing Systems Workshops, 2012, pp.
324–329.
[24] M. Zaman, M. Saifuddin, M. Razzaghpour, and Y. P. Fallah, “Perfor-
mance analysis of v2i zone activation and scalability for c-v2x trans-
actional services,” in 2022 IEEE 96th Vehicular Technology Conference
(VTC2022-Fall), 2022, pp. 1–5.
[25] Y. P. Fallah, “A model-based communication approach for distributed
and connected vehicle safety systems,” in 2016 Annual IEEE Systems
Conference (SysCon), 2016, pp. 1–6.
[26] S. International, “V2x communications message set dictionary,” Society
of Automotive Engineers, Standard Doc J2735, 07 2020.
[27] M. Razzaghpour, S. Shahram, R. Valiente, and Y. P. Fallah, “Impact of
communication loss on mpc based cooperative adaptive cruise control
and platooning,” in 2021 IEEE 94th Vehicular Technology Conference
(VTC2021-Fall), 2021, pp. 01–07.
[28] M. Razzaghpour, A. Datar, D. Schneider, M. Zaman, H. Werner, H. Frey,
J. M. Velni, and Y. P. Fallah, “Finite state markov modeling of c-
v2x erasure links for performance and stability analysis of platooning
applications,” in 2022 IEEE International Systems Conference (SysCon),
2022, pp. 1–8.
[29] V. Vukadinovic, K. Bakowski, P. Marsch, I. D. Garcia, H. Xu, M. Sybis,
P. Sroka, K. WespBowski, D. Lister, and I. Thibault, “3gpp c-v2x
and ieee 802.11p for vehicle-to-vehicle communications in highway
platooning scenarios,” Ad Hoc Networks, vol. 74, pp. 17–29, 2018.
[30] M. Razzaghpour, S. Shahram, R. Valiente, M. Zaman, and Y. P. Fal-
lah, “Optimized control-centric communication in cooperative adaptive
cruise control systems,” 2024.
[31] M. Razzaghpour, R. Valiente, M. Zaman, and Y. P. Fallah, “Pre-
dictive model-based and control-aware communication strategies for
cooperative adaptive cruise control,” IEEE Open Journal of Intelligent
Transportation Systems, vol. 4, pp. 232–243, 2023.
[32] S. E. Shladover, “Path at 20—history and major milestones,” IEEE
Transactions on Intelligent Transportation Systems, vol. 8, no. 4, pp.
584–592, 2007.
[33] Y. Bian, Y. Zheng, W. Ren, S. E. Li, J. Wang, and K. Li,
“Reducing time headway for platooning of connected vehicles via
v2v communication,” Transportation Research Part C: Emerging
Technologies, vol. 102, pp. 87–105, 2019. [Online]. Available:
https://www.sciencedirect.com/science/article/pii/S0968090X18311197
[34] L. Davis, “Dynamics of a long platoon of cooperative adaptive
cruise control vehicles,” Physica A: Statistical Mechanics and its
Applications, vol. 503, pp. 818–834, 2018. [Online]. Available:
https://www.sciencedirect.com/science/article/pii/S0378437118303236
[35] Y. Zheng, S. Eben Li, J. Wang, D. Cao, and K. Li, “Stability and
scalability of homogeneous vehicular platoon: Study on the influence of
information flow topologies,” IEEE Transactions on Intelligent Trans-
portation Systems, vol. 17, no. 1, pp. 14–26, 2016.
[36] T. Ruan, H. Wang, L. Zhou, Y. Zhang, C. Dong, and Z. Zuo, “Impacts of
information flow topology on traffic dynamics of cav-mv heterogeneous
flow,” IEEE Transactions on Intelligent Transportation Systems, vol. 23,
no. 11, pp. 20 820–20 835, 2022.
[37] S. E. Shladover, C. Nowakowski, X.-Y. Lu, and R. Ferlis,
“Cooperative adaptive cruise control: Definitions and operating
concepts,” Transportation Research Record, vol. 2489, no. 1, pp.
145–152, 2015. [Online]. Available: https://doi.org/10.3141/2489-17
[38] Y. Zheng, S. E. Li, K. Li, F. Borrelli, and J. K. Hedrick, “Dis-
tributed model predictive control for heterogeneous vehicle platoons
under unidirectional topologies,” IEEE Transactions on Control Systems
Technology, vol. 25, no. 3, pp. 899–910, 2017.
[39] P. Wang, B. Di, H. Zhang, K. Bian, and L. Song, “Platoon cooperation
in cellular v2x networks for 5g and beyond,” IEEE Transactions on
Wireless Communications, vol. 18, no. 8, pp. 3919–3932, 2019.
[40] T. Nguyen, L. Y. Wang, G. Yin, H. Zhang, S. E. Li, and K. Li, “Impact
of communication erasure channels on control performance of connected
and automated vehicles,” IEEE Transactions on Vehicular Technology,
vol. 67, no. 1, pp. 29–43, 2018.
[41] C. Zhao, X. Duan, L. Cai, and P. Cheng, “Vehicle platooning with
non-ideal communication networks,” IEEE Transactions on Vehicular
Technology, vol. 70, no. 1, pp. 18–32, 2021.
[42] F. Acciani, P. Frasca, G. Heijenk, and A. A. Stoorvogel, “Stochastic
string stability of vehicle platoons via cooperative adaptive cruise
control with lossy communication,” IEEE Transactions on Intelligent
Transportation Systems, pp. 1–11, 2021.
[43] F. Marzouk, J. Rodriguez, and A. Radwan, “Analysis and enhancement
of platoon management in smart city,” in 2018 14th International
Wireless Communications Mobile Computing Conference (IWCMC),
2018, pp. 1301–1306.
[44] C. Wu, Y. Lin, and A. Eskandarian, “Cooperative adaptive cruise control
with adaptive kalman filter subject to temporary communication loss,”
IEEE Access, vol. 7, pp. 93 558–93 568, 2019.
[45] C. E. Rasmussen and C. K. I. Williams, Gaussian Processes for Machine
Learning. The MIT Press, 2006.
[46] H. Kazemi, H. N. Mahjoub, A. Tahmasbi-Sarvestani, and Y. P. Fallah,
“A learning-based stochastic mpc design for cooperative adaptive cruise
control to handle interfering vehicles,” IEEE Transactions on Intelligent
Vehicles, vol. 3, no. 3, pp. 266–275, 2018.
[47] E. Moradi-Pari, H. N. Mahjoub, H. Kazemi, Y. P. Fallah, and
A. Tahmasbi-Sarvestani, “Utilizing model-based communication and
control for cooperative automated vehicle applications,” IEEE Trans-
actions on Intelligent Vehicles, vol. 2, no. 1, pp. 38–51, 2017.
[48] H. N. Mahjoub, A. Raftari, R. Valiente, Y. P. Fallah, and S. K.
Mahmud, “Representing realistic human driver behaviors using a finite
size gaussian process kernel bank,” in 2019 IEEE Vehicular Networking
Conference (VNC), 2019, pp. 1–8.
[49] Y. Wang, Z. Wang, K. Han, P. Tiwari, and D. B. Work, “Gaussian
process-based personalized adaptive cruise control,” IEEE Transactions
on Intelligent Transportation Systems, vol. 23, no. 11, pp. 21 178–21189,
2022.
[50] T. Br¨
udigam, A. Capone, S. Hirche, D. Wollherr, and M. Leibold, “Gaus-
sian process-based stochastic model predictive control for overtaking in
autonomous racing,” 2021.
[51] A. Raftari, M. Razzaghpour, and Y. P. Fallah, “Error-driven information-
passing policy for model-based position tracking in vehicular networks,”
IEEE Transactions on Vehicular Technology, vol. 72, no. 12, pp. 15 375–
15 390, 2023.
[52] Y. P. Fallah, C.-L. Huang, R. Sengupta, and H. Krishnan, “Analysis of
information dissemination in vehicular ad-hoc networks with application
to cooperative vehicle safety systems,” IEEE Transactions on Vehicular
Technology, vol. 60, no. 1, pp. 233–247, 2011.
[53] A. Tahmasbi-Sarvestani, Y. P. Fallah, and V. Kulathumani, “Network-
aware double-layer distance-dependent broadcast protocol for vanets,”
14
IEEE Transactions on Vehicular Technology, vol. 64, no. 12, pp. 5536–
5546, 2015.
[54] S. M. O. Gani, Y. P. Fallah, G. Bansal, and T. Shimizu, “Message content
control for distributed map sharing in vehicle safety communications,”
in 2017 IEEE 36th International Performance Computing and Commu-
nications Conference (IPCCC), 2017, pp. 1–7.
[55] J. Fu, G. Wu, and R. Li, “Performance analysis of sidelink relay in scma-
based multicasting for platooning in v2x,” in 2020 IEEE International
Conference on Communications Workshops (ICC Workshops), 2020, pp.
1–6.
[56] T. R. Gonc¸ alves, V. S. Varma, and S. E. Elayoubi, “Performance of
vehicle platooning under different v2x relaying methods,” in 2021 IEEE
32nd Annual International Symposium on Personal, Indoor and Mobile
Radio Communications (PIMRC), 2021, pp. 1018–1023.
[57] D. Jia, R. Zhang, K. Lu, J. Wang, Z. Bi, and J. Lei, “Improving the
uplink performance of drive-thru internet via platoon-based cooperative
retransmission,” IEEE Transactions on Vehicular Technology, vol. 63,
no. 9, pp. 4536–4545, 2014.
[58] S. E. Li, Y. Zheng, K. Li, and J. Wang, “Scalability limitation of homo-
geneous vehicular platoon under undirected information flow topology
and constant spacing policy,” in 2015 34th Chinese Control Conference
(CCC), 2015, pp. 8039–8045.
[59] M. Sybis, V. Vukadinovic, M. Rodziewicz, P. Sroka, A. Langowski,
K. Lenarska, and K. Wesołowski, “Communication aspects of a modified
cooperative adaptive cruise control algorithm,” IEEE Transactions on
Intelligent Transportation Systems, vol. 20, no. 12, pp. 4513–4523, 2019.
[60] H. Nourkhiz Mahjoub, A. Tahmasbi-Sarvestani, S. M. Osman Gani,
and Y. P. Fallah, “Composite ¡inline-formula¿ ¡tex-math
notation=”latex”¿α−µ¡/tex-math¿¡/inline-formula¿ based dsrc
channel model using large data set of rssi measurements,” IEEE
Transactions on Intelligent Transportation Systems, vol. 20, no. 1, pp.
205–217, 2019.
[61] E. E. Marvasti, S. M. O. Gani, and Y. P. Fallah, “A statistical approach
toward channel modeling with application to large-scale censored data,”
IEEE Transactions on Intelligent Transportation Systems, pp. 1–12,
2020.
[62] G. H. Mohimani, F. Ashtiani, A. Javanmard, and M. Hamdi, “Mobility
modeling, spatial traffic distribution, and probability of connectivity for
sparse and dense vehicular ad hoc networks,” IEEE Transactions on
Vehicular Technology, vol. 58, no. 4, pp. 1998–2007, 2009.
[63] M. Zaman, M. Saifuddin, M. Razzaghpour, Y. Fallah, and J. Rao,
“On batching acknowledgements in c-v2x services,” in 2023 IEEE 97th
Vehicular Technology Conference (VTC2023-Spring), 2023, pp. 1–7.
[64] M. Killat and H. Hartenstein, “An empirical model for probability of
packet reception in vehicular ad hoc networks,” EURASIP Journal on
Wireless Communications and Networking, vol. 2009, pp. 1–12, 2009.
[65] W. K. Wolterink, G. J. Heijenk, and G. Karagiannis, “Information
dissemination in vanets by piggybacking on beacons — an analysis of
the impact of network parameters,” in 2011 IEEE Vehicular Networking
Conference (VNC), 2011, pp. 94–101.
[66] J. Mittag, F. Thomas, J. H¨
arri, and H. Hartenstein, “A comparison
of single- and multi-hop beaconing in vanets,” in Proceedings of the
Sixth ACM International Workshop on VehiculAr InterNETworking,
ser. VANET ’09. New York, NY, USA: Association for Computing
Machinery, 2009, p. 69–78. [Online]. Available: https://doi.org/10.1145/
1614269.1614282
[67] L. Yang, J. Guo, and Y. Wu, “Piggyback cooperative repetition for
reliable broadcasting of safety messages in vanets,” in 2009 6th IEEE
Consumer Communications and Networking Conference, 2009, pp. 1–5.
[68] Y. P. Fallah, C. Huang, R. Sengupta, and H. Krishnan, “Congestion
control based on channel occupancy in vehicular broadcast networks,”
in 2010 IEEE 72nd Vehicular Technology Conference - Fall, 2010, pp.
1–5.
[69] K. C. Dey, L. Yan, X. Wang, Y. Wang, H. Shen, M. Chowdhury,
L. Yu, C. Qiu, and V. Soundararaj, “A review of communication, driver
characteristics, and controls aspects of cooperative adaptive cruise con-
trol (cacc),” IEEE Transactions on Intelligent Transportation Systems,
vol. 17, no. 2, pp. 491–509, 2016.
[70] S. Yousefi, E. Altman, R. El-Azouzi, and M. Fathy, “Analytical model
for connectivity in vehicular ad hoc networks,” IEEE Transactions on
Vehicular Technology, vol. 57, no. 6, pp. 3341–3356, 2008.
[71] S. Darbha, S. Konduri, and P. R. Pagilla, “Effects of v2v communication
on time headway for autonomous vehicles,” in 2017 American Control
Conference (ACC), 2017, pp. 2002–2007.
[72] K. Lenarska, K. Wesolowski, and M. Sybis, “Application of virtual
leaders in long vehicle platoons operating with cooperative adaptive
cruise control using ieee 802.11p transmission,” in 2020 IEEE 21st In-
ternational Symposium on ”A World of Wireless, Mobile and Multimedia
Networks” (WoWMoM), 2020, pp. 176–178.
[73] M. Won, “L-platooning: A protocol for managing a long platoon with
dsrc,” IEEE Transactions on Intelligent Transportation Systems, pp. 1–
14, 2021.
[74] S. Diamond and S. Boyd, “CVXPY: A Python-embedded modeling lan-
guage for convex optimization,” Journal of Machine Learning Research,
vol. 17, no. 83, pp. 1–5, 2016.
[75] L. Gurobi Optimization, “Gurobi optimizer reference manual,” 2021.
[Online]. Available: http://www.gurobi.com
Mahdi Razzaghpour is a Ph.D. candidate in Com-
puter Engineering at the University of Central
Florida and a member of the Connected and Au-
tonomous Vehicles Research Lab (CAVREL). His
research interests include Reinforcement Learning,
Machine Learning, and deep learning with a focus
on the cooperative driving problem. He received
his M.Sc. degree in Computer Engineering from
the University of Central Florida in 2021 and his
B.Sc. degree in Electrical Engineering from Sharif
University of Technology in 2019 and has worked
as a research intern at Honda Research Institute (HRI).
Babak Ebrahimi Soorchaei is a Ph.D. candidate
in Computer Science at the University of Central
Florida and a member of the Connected and Au-
tonomous Vehicles Research Lab (CAVREL). His
research interests include Computer Vision, Machine
Learning, and deep learning with a focus on the
cooperative driving problem. He received his M.Sc.
degree in Computer Science from the Technical
University of Kaiserslautern in 2016 and his B.Sc.
degree in Computer Science from the University of
Tehran in 2011.
Rodolfo Valiente received the Ph.D. degree in com-
puter engineering from the University of Central
Florida and the M.Sc. degree from the University of
Sao Paulo. He is a Research Scientist with HRL Lab-
oratories. His research interests include computer
vision, autonomous vehicles, robotics, reinforcement
learning, and artificial intelligence.
Yaser P. Fallah is a Professor in the ECE De-
partment at the University of Central Florida. He
received his Ph.D. degree from the University of
British Columbia, Vancouver, BC, Canada, in 2007.
From 2008 to 2011, he was a Research Scientist with
the Institute of Transportation Studies, University
of California Berkeley, Berkeley, CA, USA. His
research, sponsored by industry, USDoT, and NSF,
is focused on intelligent transportation systems and
automated and networked vehicle safety systems.
