IEEE Computational Intelligence Magazine - May 2022 - 86

device to transmit data to vehicles. Wireless
sensors and RSUs all have limited
coverage radii, which raises the optimization
problem of how to deploy a minimal
number of wireless sensors and
RSUs while guaranteeing full coverage
of the targeted area and network
communication. A PSO-based algorithm
is proposed to optimize the
deployment scheme.
Second, the network routing problem
in vehicular networks has been
widely researched in the literature. The
routing of vehicular networks has certain
special features [16]. One feature is
that relay nodes of the network (i.e., the
vehicles) are mobile, making the network
dynamic. The routing path is thus
unstable and has a limited lifetime. To
address the situation in which the route
may break down at any time, Eiza et al.
[17] proposed an ACO-based algorithm
integrated with situational awareness
that can prepare countermeasures for
quick reconnection. Li et al. [18] used
ACO to construct a routing path for
maximizing quality of service (QoS)
while subjecting the path to the maximum
delay constraint. Sun et al. [19]
used ACO to generate routing paths
under a dynamic environment, in which
buses were treated as primary relay
nodes since they had regular timetables
and travel trajectories, while other vehicles
were treated as secondary relay
nodes. The optimization objective is to
find a routing path with a longer lifetime
and lower delay. In Sun et al.'s
other research [20], they used ACO to
construct routing between vehicles in
adjacent intersections. Another feature
in the routing of vehicular networks is
that certain vehicles may be malicious
(i.e., they transmit incorrect data,
regardless of intention, or even launch
cyber-attacks) in the V2V communication.
Such malicious vehicles will lead
to severe safety threats, as decisionmaking
for traffic planning and management
highly depends on stable and
authentic data collection. To address
this problem, Eiza et al. [21] proposed
an ACO-based routing algorithm.
Plausibility checks are incorporated
into the proposed algorithm to ensure
the consistency of routing control messages.
In addition, Safavat and Rawat
[22] first proposed an effective malicious
vehicle detection method and
then used ACO to obtain a routing
path with low delay, short distance, and
no malicious vehicle. Based on these
studies, ACOs are shown to be popular
because vehicular network routing is a
discrete optimization problem, in
which ACOs are efficient.
Third, the incorporation of edge
computing and 5G in vehicular networks
creates the complex scheduling
problem of computational and communication
resources. Tan et al. [23] focused
on the edge computing scene. A PSO
algorithm is used to generate a solution
representing the caching strategy of
RSUs and the computation allocation
scheme of mobile edge computing servers
for optimizing the overall operating
cost of the vehicular network, which
includes communication, storage, and
computation costs. Chen et al. [24] proposed
a graph-based GA and a heuristic
algorithm to schedule computational
and communication resources for optimizing
resource utilization, QoS, and
support for concurrent requests. Edge
computing also enables vehicles to be
computational nodes. Some studies [25],
[26] considered the influence of the
mobility of vehicles and used EC algorithms
to assign computational tasks to
vehicles to optimize task completion
time. Khan et al. [27] focused on the 5G
communication scene, where the baseband
unit was the basic transmission
device. Each zone controller should be
linked to a baseband unit, and through
the transmission of the baseband unit,
the zone controller can connect to the
software-defined networking controller.
A hybrid-fuzzy logic-guided GA is proposed
to determine links between baseband
units and zone controllers. The
optimization objective includes the
number of baseband units used, delay of
data transmission, and load balance of
the vehicular network.
V2V
Communication
FIGURE 2 Example of a vehicular network with V2I and V2V communications.
B. Intelligent Traffic Light
With the development of the economy
and the manufacturing technology of
cars, the number of cars in cities increases
rapidly, which produces a strong pressure
on the traffic system and leads to congestion.
Besides improving infrastructure,
such as constructing new roads and
broadening the current roads, optimizing
the signal pattern of traffic lights is a
more economical approach to reduce
traffic congestion. Based on the vehicular
86 IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | MAY 2022
V2I
Communication
IEEE Computational Intelligence Magazine - May 2022

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - May 2022
