IEEE SYSTEMS JOURNAL, VOL. 17, NO.

2, JUNE 2023 2465

Realization of a Robust Fog-Based Green

VANET Infrastructure
Qutaiba I. Ali , Member, IEEE

Abstract—This article proposes an efficient employment of a Many studies have been conducted on VANETs that have been
self-powered fog-based vehicular ad hoc network (VANET) in- combined with cloud computing to increase their functionality
frastructure. Miscellaneous techniques and algorithms are sug- [4]. Generally speaking, the vehicles use time and energy when
gested to help the realization of such framework. In the current
work, we decided to enhance the network architecture of the uploading to and downloading from the cloud [5]. The current
traditional VANET by adopting the concept of self-powered fog cloud computing paradigm can hardly meet the needs of location
computing concepts for better networking, computing, and storage awareness, mobility assistance, and latency as the density of cars
performance. The green fog layer consists of three components: rises in metropolitan areas.
a self-powered edge server, wireless solar routers, and a new de- With the idea of fog computing (FC), a solution based on
vice resulted from the integration between a solar-powered smart
camera and a solar-powered road side unit in order to create edge cloud computing has been put out to resolve problems
a better sensing mechanism of the road traffic. A proper power between the vehicular nodes and the main cloud [6]. FC can
management strategy is suggested to be installed locally in the self- address the communication transmission delays. Additionally,
powered devices to decrease their power utilization and lengthen geo-distributed fog servers with low latency can be used to
the lifetime of their batteries. On the system level, the design steps send the safety alerts to targeted destinations that are located
concentrate on building a sustainable, secured, reliable, and scal-
able communication infrastructure and this was done by adopting throughout various geographical regions. The cloud computing
several approaches, such as VANET fog clustering, malicious nodes concept is extended to the network edge by FC [4]. In close
detection, and a combination of various security methods. The proximity to vehicular nodes, FC is a scaled-down kind of
performance of the different methods and algorithms suggested in cloud computing with capabilities akin to those of the cloud. A
this article is evaluated using different simulation and experimental multitiered design for FC places vehicle nodes at the network’s
tools to discover their impact on enhancing the robustness of the
fog-based Green VANET. edge. The fog platform is situated halfway between the data
centers and the vehicle nodes in the traditional cloud settings.
Index Terms—Fog computing (FC), green infrastructure, power Utilizing FC’s benefits in a dispersed cloud environment is the
management, road side unit (RSU), smart camera (SC), solar
energy harvesting, VANET.
key driver behind its implementation in VANETs. Low latency
is provided between the vehicles and the cloud through FC. FC
can also assist highly mobile vehicles [6].
I. INTRODUCTION Road side units (RSUs), a type of VANET infrastructure
EHICLE networks are important because they provide component, were traditionally placed next to wired electrical
V safety, convenience, accommodation, and higher produc-
tivity by sharing valuable data. As innovation emerged signifi-
sources like traffic signals [4], [5]. The covered region and
services offered by the RSUs are, however, constrained by such
cantly in the past decade, vehicle networks had been developed an assignment. RSUs powered by renewable sources must be
and became well known with the development of cloud com- implemented in order to overcome this restriction [7]. In this
puting. Cloud computing [1], [2] is the dominant component regard, green VANET solutions suggest that RSUs can obtain
for vehicular ad hoc networks (VANETs). The huge amounts the energy necessary for their operations from the environment
of traffic data produced by smart vehicles, such as video and around them, particularly solar energy. By allowing the RSUs to
sensor data, may not be effectively handled by the conventional be placed anywhere without regard to the availability of a power
centralized VANET technology. Additional servers are needed source, such plans enable the VANET architecture to span a
in different locations in order to gather and process a significant larger area. In order to transport and route the data packets of
volume of real-time traffic data. Cloud-based VANETs may be a their clients (the vehicles and other RSUs), the RSUs are also
suitable option in these kinds of circumstances. Various mobile advised by this arrangement to establish an ad hoc network [7].
devices have recently used cloud computing to handle sophis- Many research activities address various issues in order to
ticated computations that are difficult to complete locally [3]. develop an efficient, secure, reliable, and scalable infrastructure
for vehicle networks. The author in [7] suggested the Green
VANET infrastructure making use of RSUs powered by so-
Manuscript received 7 February 2022; revised 20 May 2022 and 31 August
2022; accepted 16 October 2022. Date of publication 8 November 2022; date of
lar energy and armed with different security and reliability
current version 8 June 2023. solutions. For better network management, Ahmed et al. [8]
The author is with the Department of Computer Engineering, Mosul Univer- propose a load-balancing algorithm to map sensor data, vehicles,
sity, Mosul 00964, Iraq (e-mail: [email protected]).
Digital Object Identifier 10.1109/JSYST.2022.3215845
and data centers performing tasks. A dynamic convergence

1937-9234 © 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 2466 IEEE SYSTEMS JOURNAL, VOL. 17, NO. 2, JUNE 2023

method is proposed to help identify vehicle system load fac-

tors and compare their termination criteria. The authors in [9],
[10], [11], [12], and [13] suggest enhancing the performance
and security of the VANET infrastructure by integrating mod-
ern networking techniques such as software-defined networks
(SDNs), cloud computing, and artificial intelligence methods.
The authors in [14] presented a blockchain-based methodol-
ogy to preserve users’ privacy while simultaneously boosting
vehicle security. The proposed methodology keeps the vehicle
data at the servers, and related hash values are stored in the
blockchain. Mohammed et al. [15] present an approach that
achieves data dissemination based on clustering and avoids a
broadcast storm. A developed clustering approach is dependent
on forwarded messages among vehicles. Truong et al. [16]
presented FSDN, which integrates SDN and edge computing
by considering various factors, e.g., physical medium, mobility,
and capability. Finally, The study in [17] provides an integrated
approach, in which relationships between green infrastructure
connectivity, resilience potential, and conservation strategies are
addressed. Fig. 1. Work plan of the FGVANET.
The majority of the cited references offer a variety of sys-
tematic solutions for the VANET architecture and networking
environment, but they do not account for the difficulties in security methods. Simulation models and experimental testbeds
putting their ideas into practice on actual platforms, particularly are proposed to measure and validate the design outcome of the
embedded systems. Even though our previous work in [7] studies suggested FGVANET.
the implementation demands of self-powered VANET infras- The rest of the article is organized as follows. Section II dis-
tructure, the contribution was limited to a traditional VANET cusses the architecture and details of the proposed system while
infrastructure concept and devices, such as RSUs. Our efforts in Section III explains the principles of prototyping the suggested
this work are directed on addressing various issues with the de- system and demonstrates the proposed experimental work to
ployment of an improved fog-based Green VANET (FGVANET) validate the performance of the suggested system. Finally, Sec-
infrastructure, including following. tion IV includes the main conclusions extracted from this article.
1) What are the necessary enhancements on the devices,
algorithms, techniques, and solutions in order to achieve II. SUGGESTED SYSTEM ARCHITECTURE
an efficient, reliable, secured, and self-powered system? This section describes the design, roles, and modes of use of
2) Is it feasible to integrate such a complex system into the proposed FGVANET infrastructure components, algorithms,
platforms with minimal resources? What features of these and techniques. As mentioned earlier, the combination of these
platforms are suggested? methods creates the intended FGVANET architecture.
3) What are the system’s evaluation criteria for network
efficiency, dependability, usability, power consumption,
and resistance to various threats? A. Fog-Based VANET Infrastructure
Our plan to respond to the above challenges is achieved after The major goals of VANETs are to increase traffic efficiency
passing two design levels: device level and system level, see and driver safety. The majority of VANET applications rely
Fig. 1. In the device level, we suggest that the functionality and on the regular and ad hoc transmission of various messages
architecture of the RSU must be enhanced to extend its sensing between nearby vehicles and between vehicles and roadside
capability [by combining it with a smart camera (SC) features] communication units nearby. Huge amounts of data are gen-
and to maintain its robustness and availability against the differ- erated by this communication paradigm, which necessitates
ent failure events. Also, wireless solar router (WSR) is presented storage, compute, and communication resources [10]. Huge
in this work to create the backbone of the ad hoc network which advancements have been made in the communication, com-
connects the different components of the VANET infrastructure. puting, and automotive industries in recent years. Because of
All these devices were assumed to be energized by a solar power this, VANET cloud computing has emerged as a way to meet
source while utilizing an efficient power management strategy. the needs of VANETs in terms of computational, storage, and
The system-level design philosophy focuses on creating a scal- networking resources. Since VANET applications have specific
able, secure, reliable, and scalable communication infrastructure mobility, low latency, and location awareness requirements, FC
to address the various systemic challenges that were brought up is a standard that encompasses cloud computing and related
when constructing the FGVANET architecture. This was done services to the proximity of a network [11]. Applications and
by adopting several approaches, such as VANET fog clustering services for the VANET heavily rely on FC. In this article, the
(VFC), malicious nodes detection and a combination of various suggested FGVANET makes use of cloud and FC architecture to

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 ALI: REALIZATION OF A ROBUST FOG-BASED GREEN VANET INFRASTRUCTURE 2467

upload their data [23] with RSUs, such as 802.11 access points.
The introduction of these RSUs has been shown to alleviate the
problem of network partition, which significantly improves the
dissemination of VANET information and data aggregation [24].
On the other hand, traffic and route surveillance through SC,
which are currently regularly used to monitor traffic and detect
incidents, is one of the major services offered by modern trans-
port systems [25], [26]. The low-costs and low-power CMOS
sensors acquire images of the needed quality to continue image
processing under different illumination conditions, due to their
processing capabilities, their high dynamic range, and high bit
resolution. SC do not only capture pictures or video sequences,
but also conduct high-level image processing, such as on-board
motion and face recognition and communicate the (compressed)
video data as well as video information collected via a network.
New video quality and better video analysis results are provided
by the SC when compared to existing solutions. Moreover,
they offer flexible video transmission and calculation through
a fully digital interface in scalable networks using thousands of
Fig. 2. Fog green VANET architecture. cameras [27], [28]. To gain the advantages of both techniques,
this article deals with the possibility of integrating the RSU
functionality into the SC working domain and vice versa. This
enhance the performance of the different VANET applications. suggestion extends the sensing and data-gathering capabilities of
The proposed architecture consists of three layers: cloud layer, the traditional RSUs which affects positively on the data analysis
fog layer, and edge device layer, as illustrated in Fig. 2. and decision-making strategies. To the best of our knowledge,
Data aggregation, data mining, analytical optimization, stor- this suggestion is introduced for the first time in the literature.
age, batch processing, and computation of complicated infor- The second member of the FGVANET is the wireless solar
mation [18] are the major advantages of the cloud layer and are router (WSR). The main functions of the WSRs are creating
beyond the computing capabilities of edge nodes. The compu- a multihop ad hoc network to route the different kinds of data
tational workloads transmitted via edge nodes to the cloud are packets of the fog layer to their intended destinations and to
not latency-sensitive. Fog layer ensures that the smart vehicle connect the SC-RSU nodes to the edge server (ES) and the upper
layer is connected to the cloud layer. To this end, wireless layers of the network.
communication protocols, including 802.11p, 3GPP, 3G, 4G, Finally, in this article, we assume that the ES is also powered
LTE, and 5G, are used by vehicles (edge devices). The aim is by solar energy which provides an ease of installation and
to provide low latency, awareness of the location, emergency placement in the most appropriate location in the fog cluster. Our
management, caching, discovery of content and computing, and research work at this point (still under development) includes
to increase service quality, as it is close to vehicles and used for selecting a powerful platform, designing its energy harvesting
reinteraction [11]. Fog layer components were usually confined module, and suggesting an efficient power management strategy.
close to wired power sources, such as traffic lights [19], [20].
This task, however, limits the area covered and the services
provided by that layer. It is important that sustainable controlled B. Suggested Power Management Strategy
devices are initiated to overcome this constraint. In this article, In this article, the harvested energy stored in the battery cells of
we recommend that the fog layer components collect energy the various FGVANET nodes is managed using the duty cycling
from the adjacent climate, in particular solar power, to do their concepts. Our main goal is to reach an ideal energy-neutral point
work. This offer allows these nodes to be established anywhere at each node when the energy supply and energy demand are
without recalling the accessibility of power supplies and thus equal. Our solution takes into account adaptive control strategies
the VANET framework covers a more comprehensive district. for modifying duty cycling with an energy profile that is known
We also propose that the fog layer components form an ad hoc in advance. Also, the proposed method deals with event-driven
network to transmit and transfer the information packets of the adaptive duty cycling, which adjusts the duty cycle based on
various components to their destination. the observed arrival patterns of events and hence governs the
The edge devices in our system are vehicles. They are assumed wake/sleep scheme of the nodes based on the available energy
to be connected to the FGVANET infrastructure via a new device (AE) and remaining mission time.
suggested in this article we called RSU-SC platform. RSU and Duty cycling is a strategy for energy conservation in which the
SC are two of the most common VANET devices. The static embedded nodes alternate between an energy-demanding status
RSUs can be used as a gateway to the Internet or for broadcasting (active) and a low-energy state (sleep) on a regular basis [19].
security messages [21]. Static RSUs can be used on the roadside While nodes in low-energy states are restricted to certain func-
separately and are able to sense and communicate in a broad tions in order to conserve energy, nodes in energy-demanding
range [22]. The vehicles can either access their stored data or states are able to perform all the usual tasks of an embedded
Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 2468 IEEE SYSTEMS JOURNAL, VOL. 17, NO. 2, JUNE 2023

node [20]. In this article, we propose another technique for see Fig. 3(a). In the first one, the number of frames per second
continuing to consume the stored energy to the smaller limits (FPS) of the SC and the service rate of the RSU (SRRSU ) is
of the exhibition with a lesser impact. This technique, called administered by the AE stored in the battery cells. Then again,
“Duty Cycle Estimation (DCE),” includes two stages: dynamic the subsequent algorithm figures the service rate of the WSR
duty cycle assessment and the administration of the behavior (ASRWSR ) as indicated by its AE levels.
during the active time periods. Additionally, Fig. 3(b) shows that a handshaking methodol-
According to the DCE algorithm, the installed nodes of the fog ogy is accomplished between any SC-RSU and its related WSR
layer (i.e., SC-RSUs and WSRs) ought to perform as indicated in request to determine the best SR values which gives longer
by their accessible energy, explicitly, their service rate (SR) is activity time in the VANET cluster. According to this procedure,
determined as a component of their energy plan. In this context, SC-RSU nodes choose the minimum SR value (their estimated
the sleep periods (SPs) are powerfully picked (in each day) SR value, the SR value received from their associated WSR) in
as indicated by many factors like power consumption, climate order not to exceed the safe limits which afford the maximum
conditions, and the measure of the stored energy. Accordingly, network lifetime.
we need to derive a relation among the duty cycling periods, the
SR and the AE. We first start by defining the following terms,
see Fig. 3(a).
r ASR is the average total traffic (measured in bits per C. VANET Fog Clustering
second) sent and received by and from the node. There aren’t many studies being done on the reliability of
r Duty cycling periods: In this study, the time is broken up VANET infrastructure. The majority of these studies concentrate
into discrete time slots. As a result, the duty cycle is the on multipath routing, which offers fault tolerance and allows for
proportion of the overall slot time to the SPs. the use of a backup route in the event that the primary one fails
r Residual energy (RE) from the previous day and expected [11], [15]. Because they do fewer route discovery activities, they
energy (EE) for the following day are added together to may reduce routing overhead. Additionally, the throughput and
form AE. packet delivery ratio can be improved. Although load balancing
The nodes calculate the previously specified AE at the start and congestion avoidance are a few additional uses for multipath
of each working day. It should go without saying that the node routing, their primary goal is to build a multipath routing archi-
must measure the current going to and from the batteries during tecture that offers improved robustness to node failures [29].
the whole workday in order to calculate the RE. Every second To the best of our knowledge, no previous work had addressed
a sample is taken as part of the measurement procedure, and the problem of creating a fault-tolerant, green, and fog-based
the average current values are then computed every hour. We VANET infrastructure, hence it was taken into consideration for
propose that the cloud and control server (CCS) broadcast (to this article. In reality, numerous ideas have been made to study
all nodes) the weather forecasts for this region and the effective the RSU deployment issue, optimizing the placement of RSUs
charging period for this specific day in order to evaluate the to increase performance and decrease deployment costs [7], our
value of the EE. This forecast covers the anticipated weather goal is to suggest concrete actions that should be taken into
(sunny, cloudy, or rainy) and the approximate number of hours account at different levels in order to strengthen the consistency
that can be used for charging. The node can predict the expected of the VANET architecture. The use of VFC is recommended
current value based on its historically recorded current values in here in order to reduce the size of the ad hoc network and
similar weather conditions as a function of the current measuring isolate issues inside each cluster, providing a system that is
process discussed earlier, and as a result, the EE might be more effective, reliable, and secure. Finally, two fault tolerance
determined (the nodes measure and store the average current strategies that provide a variety of ES failure protections are
value of each day with the named climate condition of that shown.
day (i.e., sunny, cloudy, or rainy) in a local database file, then The link between the ES and its related WSRs and SC-RSUs is
the average current value of each weather condition can be the foundation for the proposed FGVANET solutions previously
calculated). The next step is to calculate the SR of the node discussed. However, in these circumstances, it is possible for
on this particular day according to the value of the AE. The the FGVANET services to go down for an extended length of
relation between SR and AE can be derived by determining the time owing to a malfunction or a transmission medium denial of
power consumed according to the node activities. The process service (DoS) assault. To deal with such situations, we suggest
of mapping the SR value to the various rates of the applied segmenting the entire FGVANET infrastructure (which includes
load is the following stage. Since the SR indicates the SR for a the streets of a particular city) into logical fog clusters, with all
medium-sized fleet of vehicles, mapping is required to provide a cluster components linked to the appropriate ESs, as shown in
changeable SR in response to variations in the fleet size served. Fig. 4. Each fog cluster has the responsibility to grant a certain
This stage necessitates that the node is able to forecast future load area of the city to the VANET administration. To clarify the
based on past behavior [7]. Calculating the amount of sleep each implementation steps of this new kind of VFC, we start by
time slot will receive is the final stage. To determine the various characterizing the duties of the main players in this framework
SP values in accordance with the variation in the applied network as recorded in Table I.
load, the same mapping process discussed earlier is used. The arranging system should follow the next criteria in order
Two flavors of this algorithm can be carried out in this context, to achieve the fog VANET clustering objectives in this phase,
one for the SC-RSU nodes and the other for the WSR nodes, see Fig. 4.
Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 ALI: REALIZATION OF A ROBUST FOG-BASED GREEN VANET INFRASTRUCTURE 2469

Fig. 3. DCE algorithm. (a) Pseudocode for SC-RSU and WSR. (b) Handshaking procedure.

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 2470 IEEE SYSTEMS JOURNAL, VOL. 17, NO. 2, JUNE 2023

TABLE I
COMPONENTS OF A VANET FOG CLUSTER

node settings (such as SC-RSUs), and security status.

Under this approach, the development of a comprehensive
understanding of the current status of the entire VANET
system and any potential hazards is managed. This
separation is also useful for reducing the size of the
overall network (i.e., the number of SC-RSUs served by a
specific ES) and for resolving issues within each cluster,
which creates a framework that is more efficient, robust,
and immune.
The suggested VANET clustering method’s functionality to
recover from ES flaws and resume the normal operation of
the VANET clusters is depicted graphically in Fig. 4, while
the VFC algorithm operations and pseudocodes were described
using the block diagram displayed in Fig. 5. The major points
of the suggested logical clustering approach, which ensures the
VANET services remain available in the event of an ES collapse,
were provided in this subsection. However, in approaching
future work, it will be necessary to thoroughly examine the
many networking and security procedures connected to the
implementation of this recommendation.
Fig. 4. VFC infrastructure and components.

1) The ES of each fog cluster is favorable when the maximum D. Securing FGVANET Functionality
number of neighboring-WSRs (N-WSRs) is around. The The messages and their origins must be protected against
number of N-WSRs must be greater than two for the sake multiple attacks because of the numerous transactions between
of cluster robustness. the ES and its associated nodes. The following techniques, which
2) Limiting the number of WSRs served by each ES not were described in our earlier research in [23] and [24], are
to exceed their network bandwidth or their processing advised.
capability. 1) Performing entity authentication in both directions be-
3) Each cluster’s geographic boundaries must be shaped to tween ES and any node.
allow for the presence of the greatest possible number 2) In their actual working environments, the SC-RSUs and
of neighboring clusters. A minimum of two neighboring WSRs may be vulnerable to physical attacks by malicious
clusters must be guaranteed by the boundary-shaping pro- entities. We advise using the method described in [24] to
cess. locate such nodes.
4) Additionally, the greatest possible number of gateway- 3) All packets linked to transactions among ES nodes are en-
WSRs (G-WSRs) between nearby clusters must be guar- crypted and delivered with their HMAC in order to obtain
anteed. Again, the number of G-WSRs among the nearby message confidentiality, authentication, and integrity.
clusters must be more than two in order to prevent the 4) The nodes in each cluster can cooperate with each other
creation of a single point of failure. and with their ES to create a cooperative intrusion detec-
5) By use of a leased physical connection, ESs in various tion system as described in [24].
fog clusters are connected to one another (as well as 5) For data encryption purposes, we recommend advanced
the cloud server) to exchange recurring reports, updated encryption standard (AES), but various sets of AES keys
Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 ALI: REALIZATION OF A ROBUST FOG-BASED GREEN VANET INFRASTRUCTURE 2471

Fig. 5. VFC algorithm.

TABLE II
DIFFERENT KEYS ARRANGEMENTS

Fig. 6. Experimental testbed.

(or pairs) are necessary to encrypt the packets in dif-

ferent sessions. The required key encryption, purpose,
and target-source pairs are summarized in Table II. We of actual road traffic can be utilized to create the necessary
assumed keys were preinjected into each node by means network traffic for a traffic generator PC to imitate VANET be-
of a tamper-proof device in order to be installed in the field. havior against fog layer nodes and then to calculate the electrical
current that has been drained from these devices. The simulation
package (OPNET IT GURU for academic purposes) is used to
III. PROTOTYPING THE MODEL create a simulation model. This model’s objective is to simulate
This section displays our efforts to implement the suggested traffic patterns that are as accurate to actual circumstances as
FGVANET infrastructure’s suggested solutions. To assess the feasible and to represent a typical FGVANET cluster of 10 WSR
effectiveness of the suggested fog layer in realistic road traffic covering a (4 Km2 ) area of a typical city [20]. The traffic volume
scenarios, an experimental setup similar to that in Fig. 6 is statistics are considered for a four lanes road (in Mosul city/Iraq)
recommended. In accordance with this paradigm, measurements that bypasses the Mosul University Campus [20].

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 2472 IEEE SYSTEMS JOURNAL, VOL. 17, NO. 2, JUNE 2023

A. Model Assumptions TABLE III

TECHNICAL DETAILS OF FGVANET COMPONENTS
The goal is to develop traffic patterns as close to the real
circumstances as possible. In assembling the traffic generation
program, the following assumptions are taken into account.
1) The SC-RSU nodes are considered indistinguishable and
exposed to equal conditions of road transport with an ideal
communication environment.
2) Due to their connection to the vehicles and other WSRs,
the information traffic generated by SC-RSUs is for-
warded via a reasonable routing mechanism to the ES. The
SC-RSUs generate their 1000 bytes of traffic 10 times per
minute and forward them to the ES via WSRs [21]. The
vehicles have been configured to broadcast their 100 bytes
of status packets every second [19].
3) According to our previous study [24], optimized link state
routing (OLSR) protocol gives the best performance in
fixed nodes topology, among other protocols, so that it
is embraced by the simulation model. In our model of
simulation, the OLSR components are managed by a set
of predefined configurations in the OLSR RFC 3626 [23]. the suggested power management and other algorithms on plat-
4) The simulated network is tested with one or two VANET forms having the following features (which were recommended
services: traffic monitoring and web access, According to by [21], [22], [23], and [24]):
the customer conduct recorded by [22], the web applica- 1) Fast wake up time from sleep to active mode (less than 10
tion is expected to be used by 35% of the absolute vehicles. clock cycles).
5) SC are also assumed to be identical and work in continuous 2) Clock stop mode (defined later).
road traffic monitoring mode (the most bandwidth and 3) Data saving during the clock stop mode.
energy-demanding mode) under different video quality 4) Low power consumption (lower than 15 mA during sleep
conditions. mode).
6) Weighted fair queuing is assumed to be used by SC-RSUs 5) Adequate networking and processing capabilities (more
to govern the traffic flow to the network. In this tech- than 18 Mbps) in order to respond to VANETs’ application
nique, the packets are still assigned to different classes demands [21].
and admitted to different queues. The queues, however, 6) Supporting wide range of applications and networking
are weighted based on the priority of the queues; higher protocols in the TCP/IP stack.
priority means a higher weight. The system processes 7) Open source development and programming environment.
packets in each queue in a round-robin fashion with the 8) Reasonable cost.
number of packets selected from each queue based on the According to the above criteria, the recommended plat-
corresponding weight. Video traffic generated by the SC forms of the integrated SC-RSU consist of two parts: the RSU
is considered as the highest priority traffic type followed functionality which can be realized using Ubicom IP 2022
by the traffic monitoring application and finally the web network processor [24], while the SC implementation can be
access application. achieved via CX93610 Evaluation Kit [25]. These two systems
7) WSRs have traffic classification and bandwidth segmen- can be connected using different types of computer buses and
tation features to control the shared bandwidth between they are powered using a solar energy harvesting module. Also,
the traffic received from their associated SC-RSUs and our design assumes choosing the Microtik 433 Routerboard
the outing traffic from the other WSRs (routed traffic) supplied with two wireless interfaces [23] to implement the
according to predefined rules. These rules assume a fair WSR. The technical specifications of these platforms with their
bandwidth segmentation first (50% for SC-RSUs and 50% energy harvesting requirements are listed in Table III [19], [30].
for other WSRs routed traffic). Then, this ratio is recalcu-
lated every few minutes (determined by the administrator) C. Performance and Power Consumption Measurements
according to the real recorded traffic demands and more
bandwidth is reassigned to the most demanding side. The main aim of the next investigations will be to determine
the SC-RSUs and WSR drained currents in accordance with
their various operating methods: transmission, receipt, IDLE,
full-load CPU, SLEEP, and many more. The settings of this
B. Platforms Selection analysis are summarized and the average qualities acquired for
When making an extensive survey, many platforms were distinctive activity modes are listed in Table IV.
founded to support the networking and energy-saving require- The following study is motivated by the investigation of
ments of the suggested FGVANET. It is possible to implement the average daily energy usage of the RSU (as part of the

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 ALI: REALIZATION OF A ROBUST FOG-BASED GREEN VANET INFRASTRUCTURE 2473

TABLE IV TABLE VI
INITIAL RESULTS OF SC-RSU AND WSR DAILY-AVERAGE DRAINED CURRENT AND TX TRAFFIC OF THE SC
(COLORED VIDEO)

TABLE VII
DAILY-AVERAGE DRAINED CURRENT AND TX TRAFFIC OF THE SC (GRAY
SCALE VIDEO)
TABLE V
DAILY-AVERAGE DRAINED CURRENT AND TOTAL TRAFFIC OF THE RSU

TABLE VIII
DCE IMPLEMENTATION FOR SC-RSU AND WSR

SC-RSU node) and its networking demands. We can use an

example traffic pattern to generate network traffic according
to the number of vehicles served and the VANET applications
(the safety application is only for traffic surveillance while
multiple applications do not only include road traffic monitoring
but also web browsing).The average depleted current (95%
decrement) for the safety applications, and (75% decrement)
for various applications, as confirmed by the viability of the
assumed duty-cycling methods, during taking the SLEEP mode,
is shown in Table V (this table will be used later for traffic profile of a packet by the WLAN NIC or when the sleep timer expires),
determination of the SC-RSU). The adopted power-saving tech- the board wakes up within a (3–5) clock cycles (depending on
nique makes use of an important feature: “Clock Stop Mode,” the platform) to perform the necessary actions. Tables VI and
where the platform’s mainboard may be disabled along with the VII, which confirm the previous assessment of the helpfulness
system clock, which disables the CPU core clock. When the of the DCE technique and the clock stop mode, also show the
system clock is turned off, the interrupt logic still works, and a performance of the SC under different video quality settings. The
sleep timer might be enabled to continue operating. Using sleep streaming speed of the video is measured in frame per second
timer interrupts or in response to an external interrupt from the (FPS), the video size and the color mode are changed to cover
WLAN NIC, it is possible to return from the clock stop mode a wide range of SC profiles, while the networking and energy
(the sleep mode) to the normal execution. Since the chip is not demands are measured under these circumstances (also these
reset with this technique, the program execution picks up where tables will be used later for traffic profile determination of the
it left off. In this mode, all circuitry on the mainboard—with SC-RSU).
the exception of the external interrupts circuits, the sleep timer, In different energy level situations, the DCE algorithm is
and the program memory—is turned off while the WLAN NIC executed. Table VIII contains the initial results of these tests.
remains on. Whenever an interrupt occurs (due to the reception This investigation focuses on the ability of the DCE algorithm

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 2474 IEEE SYSTEMS JOURNAL, VOL. 17, NO. 2, JUNE 2023

Fig. 7. Pseudocode of SC-RSU traffic profile selection procedure.

TABLE IX profiles (SC video quality and RSU-supported applications) are

SOME POSSIBLE SC-RSU TRAFFIC PROFILES
available to the administrator to select from in order to keep the
network surviving and functioning according to the energy plan
on that particular day.
The centralized green scheduler (CGS), a central control
model designed to decrease the power consumption of RSUs
with limited processing capacity and lengthen battery life, and
the DCE algorithm are then contrasted [19]. The CGS (for each
RSU), which is kept in a control server, is made up of the
forecast, ON/OFF, and assessment algorithms. The power man-
agement module first launches the forecast algorithm to gather
the recorded street traffic data for each RSU at predetermined
intervals, predicts future traffic patterns [based on artificial
to be adapted to various working conditions, where distinctive neural network (ANN) learning], and remotely shifts the RSU
levels of AE and different weather conditions are expected. This nodes between active and power-saving states. Table X shows
table lists the different upsides of SR (for WSR and SC-RSUs) that both systems have a favorable effect on RSU power needs
that have acquired these situations when the initial battery (and hence the network lifetime) while keeping the estimated
charging rate is the RE. In order to find the best traffic profile cost within reasonable ranges, despite the fact that they approach
for each SC-RSU, the procedure shown in Fig. 7 is suggested. RSU power management tasks in different ways.
The execution of this procedure is performed contentiously and
individually by each SC-RSU to select “dynamically” the most
appropriate settings of both the SC and the RSU in response to
D. Assessment of VFC Algorithm
the different energy and network load conditions. The results
listed in Table IX are extracted after running the traffic selection Essential network conditions are assumed and preliminary
procedure with some AE values which represent high, moderate, tests are also executed to evaluate the performance of the VFC
and low energy levels. It is noticed that the suggested DCE algorithm. When building the model, the following concepts are
algorithm can adjust the duty cycling according to the AE levels examined.
and thus the SC-RSU nodes and the WSRs adapt their working 1) The WSRs are assumed to be consistently suited to a
plan while keeping on working in a pre-managed and arranged central ES. The coverage area for each WSR is included
manner. It is noted from Table IX that different SC-RSUs’ traffic (100 m2 ).

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 ALI: REALIZATION OF A ROBUST FOG-BASED GREEN VANET INFRASTRUCTURE 2475

TABLE X
COMPARISON BETWEEN CGS AND DCE METHODS

TABLE XI
VANET CLUSTER SIZE EFFECT

2) Each SC-RSU has been assumed to be identical to the

number of vehicles served.
3) The VANET framework, as shown previously, supports
two applications (road traffic monitoring and web access).
4) The over-(5%) percentage of packet loss reflected network Fig. 8. Average WSR migration delay. (a) Low load. (b) High load.
congestion [5].
We will first look at the advantages of selecting the correct
“WSR Response” message from G-WSR to Foreign ES). Here,
cluster size. The number of WSRs in the FGVANET infrastruc-
we made the following assumptions: either light load (just a
ture has a significant impact on the overall system performance,
traffic monitoring application with a range of vehicle serving
as shown in Table XI. The SC-RSUs, WSRs, the ES, and the
the cluster) OR high load (traffic monitoring plus Web access
communication network all have to process more network traffic
applications with different numbers of vehicles served by the
as the cluster size increases, which results in a decrease in system
cluster). Additionally, we presumptively used a uniform network
performance owing to the added load. More packets are lost due
design, with the ES at the center and the 10 WSRs evenly spaced
to network traffic congestion and the survivor packets experience
out around it. Finally, based on the cluster shape, we assume that
an additional delay in reaching their destinations when the
there will be 2, 5, or 7 hops between the foreign ES and the G-
cluster size exceeds predetermined boundaries. We suggested
WSR. The other cluster, where the moved WSR already exists, is
limiting the number of WSRs in each cluster to 10 based on
assumed on the identical premises. The variance in WSRMD val-
these observations. (this figure reflects the current settings and
ues for various network and road conditions is depicted in Fig. 8.
may be changed under different circumstances).
It is evident that the network traffic conditions and the number
The purpose of the subsequent tests is to assess the proposed
of hops between the source and the destination have an impact
WSR migration method in the event of an ES failure event. The
on how well the suggested fault-tolerant algorithm performs.
average WSR migration delay (WSRMD ), which is defined as
However, in response to the load setting, the clustering algo-
the time required by the clustering algorithm to relink a specific
rithm is possible to recover a specific portion of the VANET
area(s) of the network (i.e., WSRs) to the VANET infrastructure
infrastructure with various delay values. These delay values,
following the occurrence of a certain failure event, is the metric
however, may be deemed sufficient in light of the requirements
used to assess the ability of the VFC algorithm to survive after
of the purported VANET services [5], [8].
the occurrence of various failure events, such as ES collapse.
The WSRMD value is determined by the cumulative delays of
IV. CONCLUSION
the various procedures depicted in Fig. 5, including (“Invite new
WSRs” message from CCS to foreign ES, “Invite new WSRs” In this study, we suggest effective methods for deploying
message from ES to G-WSR, “Invite new WSRs” message self-powered VANET infrastructure. Numerous methods and
between the G-WSR pairs, "Invite new WSRs” message from algorithms to support the implementation of such a system are
G-WSR to WSR, “WSR Response” message from WSR to suggested. This work can lead to a number of conclusions.
G-WSR, “WSR Response” message between G-WSR pair, and The intelligent power management algorithm and its capacity
Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.
 2476 IEEE SYSTEMS JOURNAL, VOL. 17, NO. 2, JUNE 2023

TABLE XII [7] Q. Ali, “GVANET project: An efficient deployment of a self-powered,

FGVANET SYSTEMATIC FEATURES reliable and secured VANET infrastructure,” IET Wireless Sensors Syst.,
vol. 8, no. 6, pp. 313–322, 2018.
[8] U. Ahmed, J. C.-W. Lin, G. Srivastava, U. Yun, and A. K. Singh, “Deep
active learning intrusion detection and load balancing in software-defined
vehicular networks,” IEEE Trans. Intell. Transp. Syst., vol. 711, pp. 1–10,
2022, doi: 10.1109/TITS.2022.3166864.
[9] A. A. Khan, M. Abolhasan, and W. Ni, “5G next generation VANETs
using SDN and fog computing framework,” in Proc. IEEE Conf. Consum.
Commun. Netw., 2018, pp. 1–6.
[10] R. Bibi, Y. Saeed, A. Zeb, T. Ghazal, T. Rahman, and M. Khan, “Edge
AI-based automated detection and classification of road anomalies in
VANET using deep learning,” Comput. Intell. Neurosci., vol. 2021, 2021,
Art. no. 6262194.
[11] M. Kumar and R. S. Raw, “A novel routing protocol for hierarchical
software defined vehicular adhoc network,” in Proc. 9th Int. Conf. Com-
put. Sustain. Glob. Develop., 2022, pp. 771–775, doi: 10.23919/INDIA-
Com54597.2022.9763267.
[12] K. Zhang, Y. Mao, S. Leng, Y. He, and Y. Zhang, “Predictive offloading
in cloud-driven vehicles: Using mobile-edge computing for a promising
network paradigm,” IEEE Veh. Technol. Mag., vol. 12, no. 2, pp. 36–44,
Jun. 2017.
[13] S. Basudan, X. Lin, and K. Sankaranarayanan, “A privacy preserving
vehicular crowd sensing based road surface condition monitoring system
using fog computing,” IEEE Internet Things J., vol. 4, no. 3, pp. 772–782,
Jun. 2017.
[14] A. L. Shrivastava and R. K. Dwivedi, “A secure design of the smart
vehicular IoT system using blockchain technology,” in Proc. 9th Int. Conf.
Comput. Sustain. Glob. Develop., 2022, pp. 616–620, doi: 10.23919/IN-
DIACom54597.2022.9763216.
[15] S. J. Mohammed and S. T. Hasson, “Modeling and simulation of data
dissemination in VANET based on a clustering approach,” in Proc.
Int. Conf. Comput. Sci. Softw. Eng., 2022, pp. 54–59, doi: 10.1109/
CSASE51777.2022.9759671.
[16] N. B. Truong, G. M. Lee, and Y. Ghamri-Doudane, “Software defined
networking-based vehicular ad hoc network with fog computing,” in Proc.
IEEE Conf. Integr. Netw. Manage., 2015, pp. 1202–1207.
[17] Z. Liu, C. Xiu, and C. Ye, “Improving urban resilience through green
to regulate against various operational VANET circumstances infrastructure: An integrated approach for connectivity conservation in
are crucial to the successful implementation of the “energy the central city of Shenyang, China,” Complexity J., vol. 2020, 2020,
Art. no. 1653493.
harvesting & battery-based” network components. Similarly, the [18] N. Peter, “Fog computing and its real time applications,” Int. J. Emerg.
feathering nature of a WSR, SC-RSU, and VANET application Technol. Adv. Eng., vol. 5, no. 6, pp. 266–269, 2015.
requirement must be taken into account when the characteristical [19] Q. Ali, “Design, implementation & optimization of an energy harvesting
system for VANETS’ road side units (RSU),” IET Intell. Transp. Syst.,
network administration and adaptation to noncritical failure vol. 8, no. 3, pp. 298–307, 2014.
methods is adjusted to ensure a balance between a very reli- [20] Q. Ali, “Enhanced power management scheme for embedded road side
able and decent performed facility is made between prescribed units,” IET Comput. Digit. Techn., vol. 10, no. 4, pp. 174–185, 2016.
[21] Q. Ali, “Event driven duty cycling: An efficient power management
arrangements. The merger of such security strategies, network scheme for a solar-energy harvested road side unit,” IET Elect. Syst.
recovery algorithms, and renewables will open the doors for the Transp., vol. 6, no. 3, pp. 222–235, 2016.
creation of a more efficient and robust VANET infrastructure. [22] Q. Ali, “Green communication infrastructure for vehicular ad hoc network
(VANET),” J. Elect. Eng., vol. 16, no. 2, pp. 1–10, 2016.
Table XII below abstracts the systematic features of the proposed [23] Q. Ali, “Security issues of solar energy harvesting road side unit (RSU),”
FGVANET. IJEEE J., vol. 11, no. 1, pp. 18–31, 2015.
[24] Q. Ali, “Securing solar energy-harvesting road-side unit using an embed-
ded cooperative-hybrid intrusion detection system,” IET Inf. Secur., vol. 10,
REFERENCES no. 6, pp. 386–402, 2016.
[25] A. Mavrinac and X. Cheng, “Modeling coverage in camera networks: A
[1] M. Vouk, “Cloud computing issues, research and implementations,” J. survey,” Int. J. Comput. Vis., vol. 101, pp. 205–226, 2013.
Comput. Inf. Technol., vol. 16, pp. 235–246, 2008. [26] D. Devarajan, R. J. Radke, and H. Chung, “Distributed metric calibration
[2] D. Rountree and I. Castrillo, _e Basics of Cloud Computing: Under- of ad hoc camera networks,” ACM Trans. Sensor Netw., vol. 2, no. 3,
standing the Fundamentals of Cloud Computing in Theory and Practice. pp. 380–403, 2006.
Amsterdam, Netherlands: Elsevier, 2013. [27] M. Taj and A. Cavallaro, “Distributed and decentralized multi camera
[3] M. Chiang and T. Zhang, “Fog and IoT: An overview of research oppor- tracking,” IEEE Signal Process. Mag., vol. 28, no. 3, pp. 46–58, May 2011.
tunities,” IEEE Internet Things J., vol. 3, no. 6, pp. 854–864, Dec. 2016. [28] W. Lei, “Camera sensor activation scheme for target tracking in wireless
[4] A. V. Dastjerdi and R. Buyya, “Fog computing: Helping the internet of visual sensor networks,” Int. J. Distrib. Sensor Netw., vol. 2013, 2013,
things realize its potential,” _e Comput. J., vol. 49, no. 8, pp. 112–116, Art. no. 397537.
2016. [29] J. Cheng, G. Yuan, M. Zhou, S. Gao, Z. Huang, and C. Liu, “A connectivity-
[5] J. Feng, Z. Liu, C. Wu, and Y. Ji, “AVE: Autonomous vehicular edge prediction-based dynamic clustering model for Vanet in an urban scene,”
computing framework with ACO-based scheduling,” IEEE Trans. Veh. IEEE Internet Things J., vol. 7, no. 9, pp. 8410–8418, Sep. 2020.
Technol., vol. 66, no. 12, pp. 10660–10675, Dec. 2017. [30] Q. I. Ali and J. Khalid, “Practical design of solar-powered IEEE 802.11
[6] D. Talia, “Clouds for scalable big data analytics,” Comput. J., vol. 46, no. 5, backhaul wireless repeater,” in Proc. 6th Int. Conf. Multimedia, Comput.
pp. 98–101, 2013. Graph. Broadcast., 2014. pp. 9–12.

Authorized licensed use limited to: Sri Krishna College Of Engineering & Technology. Downloaded on April 28,2024 at 18:14:40 UTC from IEEE Xplore. Restrictions apply.