This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.DOI

A Comprehensive Study on IoT

Based Accident Detection Systems
for Smart Vehicles
UNAIZA ALVI1, MUAZZAM A. KHAN2, (Senior Member, IEEE), BALAWAL SHABIR1, ASAD
W. MALIK1, (Senior Member, IEEE) and MUHAMMAD SHER RAMZAN3
1
National University of Sciences and Technology (NUST), Islamabad, Pakistan
2
Department of Computer Science, Quaid-i-Azam University, Islamabad, Pakistan
3
Faculty of Computing and Information Technology (FCIT), King Abdulaziz University Jeddah, Saudi Arabia
Corresponding author: Muazzam A. Khan (e-mail: [email protected]).

ABSTRACT With population growth, the demand for vehicles has increased tremendously, which has
created an alarming situation in terms of traffic hazards and road accidents. The road accidents percentage
is growing exponentially and so are the fatalities caused due to accidents. However, the primary cause
of the increased rate of fatalities is due to the delay in emergency services. Many lives could be saved
with efficient rescue services. The delay happens due to traffic congestion or unstable communication to
the medical units. The implementation of automatic road accident detection systems to provide timely aid
is crucial. Many solutions have been proposed in the literature for automatic accident detection .The techniques
include crash prediction using smartphones, vehicular ad-hoc networks, GPS/GSM based systems, and
various machine learning techniques. With such high rates of deaths associated with road accidents, road
safety is the most critical sector that demands significant exploration. In this paper, we present a critical
analysis of various existing methodologies used for predicting and preventing road accidents, highlighting
their strengths, limitations, and challenges that need to be addressed to ensure road safety and save
valuable lives.

INDEX TERMS GSM, GPS, Accident Detection, IoT, Smart cities

Nomenclature
Acronyms MC Micro-controller
ML/AI Machine Learning/Artificial Intelligence
AC Alcohol Check
Acc Accuracy N No
N/A Not Applicable
Accel Accelerometer
ADANN Accident Detection using Artificial Neural Network P Pressure
ADFL Accident Detection using Fuzzy Logic PDCNN Pedestrian Detection with Convolutional Neural Network
AL Alcohol RF Radio Frequency
APSVM Accident Prediction using Support Vector Machine RP Rasberry Pi
AR Automatic Reporting S Speed
D Distance SB Seat Belt
DBA Driver Behaviour Analysis SBC Seat Belt Check
DBT Deceleration Based on Traffic SC Speed Check
signals DC Drowsiness Check SDO Safe distance from
DDDD Drowsy and Drunk Driving Detection Obstacles SH Smart Helmet
DDRL Drowsiness Detection using Representation Learning STF Switch to Terminate False
DIDCNN Distraction Identification with Deep Convolutional Neural Net- alarm Temp Temperature
work UAS Using Accelerometer Sensors
DLC Driver License Check UCSA Using Cell phone, Speed and Accelerometer
FPR False Positive Rate sensors UDS Using Distance Sensors
H High UGG Using GPS and GSM
IR Infrared Sensor UGGARFT Using Gyroscope, GPS, android app and RF transmitter
L Low ULS Using Limit Switches
LS Limit Switches UPS Using Piezoelectric Sensors
UPST Using Pressure Sensor and Tilt
M Medium
angle USSG Using Shock Sensors and
GPS

VOLUME 4, 2016 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

UVS Using Vibration focuses on using the accelerometer sensor, by monitoring

Sensors VVibration the vehicle speed and report an accident as it reaches below
VBA Vehicle Behaviour
Analysis Y Yes the threshold point [13]. Another approach aims on
preventing drunken driving by installing alcohol sensors on
the steering wheel and not allowing the driver to drive if
I. INTRODUCTION
he’s intoxicated by measuring the alcohol content in his
Due to rapid growth of world population, the demand for oxygen [3], [14].
vehicles has increased tremendously, resultantly problems In this approach, data is obtained from various sensors
of traffic congestion and road accidents has also increased. and event logs to extract prominent features for collision
The general population’s life is under high risk, if any detection model. Various intelligent computational models
accident occurs there’s a long reaction time which are used to detect accidents. These models include nearest
increments the number of deaths, therefore an automatic neighbor, neural networks and regression trees [15].
accident detection system must exist to overcome this Vehicle behaviors can be analyzed given its position and
situation. Statistics show that leading cause of death by velocity values, and can be helpful in the detection of
injury is road accidents [1]. There can be multiple causes of accidents. To detect an accident and to distinguish it from
road accidents, some of them are, driver negligence due to normal cases, different machine learning algorithms like
drowsiness [2], driving while intoxicated [3], over speeding Support Vector Machine (SVM), Artificial Neural Network
[4], [5] etc. Some studies show that weather conditions can (ANN) and Ran- dom Forests (RF) are implemented on
also contribute towards the severity of an accident such as traffic data [16]. Ma- chine learning techniques can also be
fog, rain, high winds. High winds can directly influence the utilized to determine the severity of accident. Different
vehicle which may deviate the vehicle from road, or algorithms such as k-means clustering, SVM under
indirectly due to obstruction dangers present on the roads reinforcement learning by fetching real time data like
such as trees, walls etc. [6] velocity which is obtained by means of vibration sensors
Road crashes can be seen as a collision between any on- installed in vehicle and distance which is obtained by means
road vehicles, obstacles or pedestrians. The survival rate of of ultrasonic sensors [17]. Accidents can also be detected by
victim is highly reliant on how long an ambulance takes to pre-trained surveillance cameras installed on the highways
reach the site of the accident and then carry the patient to [18], as illustrated in Figure 1. Another ma- chine learning
the hospital. In most cases of road accidents, the injuries approach uses fuzzy logic. Data like number of vehicles in
are not severe and the life of the victim can be rescued, each zone, speed of cars in particular lanes etc. is
however due to late arrival of the rescue teams, the injuries collected and then decisions are made accordingly. In this
turn deadly. Thus, the main goal is to identify where the technique, a situation is detected as an accident whenever
accident occurred, send the information to the rescue teams there’s some sort of disturbance in surrounding lanes [19].
in considerably less time, so that they can take the
necessary actions, to save the life of victim [7].
Intelligent Transport Systems (ITS) based on Internet of
Things (IoT) are getting popular and can be seen as a
solution to improve the road safety. One effective technique
to reduce traffic hazards and save precious lives could be
to reduce the response time after an accident has occurred.
Significant research has been carried out to address this
issue and to minimize the response time following an
accident [1]. Dif- ferent approaches are used for this
purpose. In this context, VANET (Vehicular Ad-hoc
Network) can be utilized [8], [9], in which every moving
vehicle acts as a node. On occurrence of accident, the alert
messages are communicated via RF (Radio Frequency)
module [10]. One approach uses limit switches to detect an
accident, GSM (Global System for Mobile
Communications) is used to send an alert message and
location of accident is traced by GPS (Global Positioning
System) module [11]. Smartphone based systems that use FIGURE 1: Real time accident detection using surveillance
an android app to detect vehicle crash are also proposed. cameras
These systems measure change of tilt angle by means of an
accelerometer sensor, speed by means of GPS and send an Many systems focused on taking preventive measures,
alert on detection of accident [12]. Some systems focus on along with detection to avoid accidents such as controlling
preventive strategy because at the end, goal is to save lives. the speed, checking driver license etc. Various approaches
This system particularly focuses on the safety of two wheel- have been used for this such as smartphones,VANETs, vi-
ers and checks if the driver is drowsy [2]. One technique bration sensors, pressure sensors, GPS, GSM and various
2 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

machine learning algorithms. Even though there is literature the objects are capable of sharing resources and data with
available on various strategies for accident detection and other objects. The data collected from the objects is further
prevention; however, no comprehensive survey exists. This analyzed to extract the useful information [22]. The smart
paper aims to fill this gap by critically reviewing the literature objects are equipped with some sort of sensors, processing
related to accident detection, prevention, and reporting sys- powers and an ability to communicate to other objects. IoT
tems, to provide a broader perspective of existing has provided many benefits to improve the quality of life,
techniques so that effective systems can be developed that where smart objects are used to simply our daily errands
can utilize the strengths while addressing the challenges in [23]. The major characteristics of IoT are as follows:
the current systems.
The rest of this article is arranged as follows. Motivation
A. INTER CONNECTIVITY AND SERVICES
behind this work is discussed in Section II, Section III
covers the IoT characteristics, IoT applications in various As long as IoT is concerned, IoT devices can be inter-
domains are discussed in Section IV , Section V covers the linked with the worldwide information and communication
typical challenges in an IoT environment, the emerging IoT foundation [21], [24]. Things-related services such as
tech- nologies are discussed in Section VI, Section VII privacy protection, semantic consistency between physical
covers the literature review of various accident detection things and their virtual interlinked items within the constraints
and prevention techniques, Section VIII presents a of objects are offered by IoT. Technologies continue to
comparative study of different techniques, Section IX develop both in the real world and in the information world
presents recommendations based on the analysis and finally to provide things- related services [21], [24].
the paper is concluded in Section X.
B. HETEROGENEITY
II. MOTIVATION BEHIND THIS STUDY The IoT devices are heterogeneous in nature. Various
According to the World Health Organization (WHO) report, devices can communicate with each other due to this
nearly 1.35 million people died in road accidents, making property. The devices that make use of different hardware
road traffic injuries the eight leading cause of death glob- platforms, net- works, technologies can communicate with
ally [20]. The number of fatalities associated with road acci- each other due to this feature [21], [24]. IoT system
dents is extremely high, thus, measures must be taken to includes connectivity of various devices, platforms and
im- prove road safety. Most injuries incurred by accidents operating systems inter- linked by different protocols [25].
are not serious, and the victim’s life can be saved if rescued IoT provides connectiv- ity between different devices which
timely. However, it takes additional delay to manually have different power constraints, vendors and are designed
notify the emergency teams due to poor communication for different purposes. The goal of IoT is to provide
mechanisms, thus, victims are left unattended for a long seamless machine to machine, machine to human and
time, resulting in an increased death rate. human to human connectivity, so it must be able to connect
The consequences of road accidents are not just constrained different things and networks [26].
to the loss of human lives yet, also incorporate the destruction
of property, traffic blockages, and immense economic loss. C. DYNAMIC CHANGES
Thus automatic accident detection systems are the need of The IoT devices are dynamic i.e. rapidly changing in nature.
time, which can speed up the rescue operations and limit The state of devices varies with time. For example, the
the causalities after the mishap and numerous lives can be devices may be in sleep/wake state,
saved. This paper features existing mechanisms to detect connected/disconnected. The number of devices at a time
accidents, its working, and limitations. Furthermore, can also change [21], [24]. Some new devices can be added
accident prevention methodologies, accident contributing in the network, while other might leave the network. IoT
factors are highlighted as well. This study critically reviews devices are able to adapt to the changing needs. For
existing literature on accident detection and prevention instance, security cameras can adapt their modes according
techniques, with the objective that smart systems can be to the time of the day [25]. The device’s connectivity with
developed with improved accuracy and better strategies to other devices may vary during different time intervals such
control accident causing factors while watching out for the that it’s connected to one set of devices at one instance,
existing challenges in the current systems. and to another set at another instance, so to ensure security,
efficient cryptography systems and good security protocols
III. INTERNET OF THINGS CHARACTERISTICS are needed [26].
Internet of Things (IoT) refers to a network of things. These
things could be any object that we see in our daily life. D. ENORMOUS SCALE:
These objects are not limited to electronic devices or some
The number of devices that are inter connected and need to be
high end technology products but could include objects that
tracked and handled would be at least an order of magnitude
we won’t normally think of them as electronic like dustbins,
greater than the internet devices currently connected. It’ll
chairs, clothes etc. [21]. It’s a network of smart objects
become even more complex to manage the massive amount
in which
of data produced by these devices and using this data and
it’s interpretation for different application purposes [21],
[24].
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
VOLUME 4, 2016 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

E. SAFETY: combustion gases can be monitored to generate an early

While we avail numerous benefits from IoT, an important alert. Protecting Wildlife – Wild animals can be traced and
concern ‘safety’ should also be considered. Whether we are their location can be determined through tracking
creators or receivers of IoT, we must ensure safety. It includes collars,which use GPS to track location and GSM for
the safety of our physical well-being as well as our private communication. [21], [22], [24], [27].
data. This also includes ensuring the safety of communication
endpoints, networks, data being communicated across the
paths, thus creating a security standard that will be able to
scale [21], [24].

F. CONNECTIVITY
Connectivity entitles attainability and compatibility.
Network availability refers to be able to receive on the
network and compatibility refers to be able to use and
produce data [21], [24]. New possibilities for IoT can be
generated by bringing together the routine objects and
connecting the smart gadgets and applications [25].

IV. IoT APPLICATIONS

IoT has numerous and diverse applications which cover al-
most all areas in our day to day tasks. It covers many
domains such as transportation, agriculture, healthcare,
waste man- agement, supply chain, environment and energy
as shown in Figure 2. Some of the applications of IoT are as FIGURE 2: IoT Applications
follows:

A. IoT IN HEALTHCARE C. ROLE OF IoT IN INDUSTRIES

IoT has numerous applications in healthcare. It can be used Explosive and Dangerous Gases – IoT can be utilized to
to monitor and indicate various health indicators. detect gas leakages in manufacturing areas, in the vicinity
Patient Monitoring – Patients can be monitored for of chemical plants and indoor mines. Harmful gases and
various conditions like heart rate, blood pressure, glucose oxygen levels in chemical plants can be monitored and
level etc. inside hospitals. controlled to ensure safety for both workers and products
Medical Cold Storage – Various medicines, vaccines [21].
could be stored and their conditions can be monitored, like Water, Oil and Gas levels Monitoring – The levels of
when they’ll expire etc. oxygen, water and oil can be monitored in storage tanks and
Fall Detection – Assistance could be provided for elderly containers.
or handicapped people if they fall, timely assistance could Maintenance and Repair – Early predictions can be made
be provided to them. for component malfunctions and maintenance service can
Dental – Toothbrush connected by means of Bluetooth to a be set automatically before the actual component failure
Smartphone app can analyze brushing habits. [21]. Managing fleet of cars – The fleet of cars can be
Physical Activity Monitoring – Sensors placed on some monitored for any corporation. It can monitor it’s
gadget like watch, that a person could wear, it can monitor performance and can process the data to choose the one
how many steps he walked, heart rate etc. [21], [22], [24], that needs mainte- nance [27].
[27]. Temperature Monitoring – The temperature inside indus-
tries, mines and other work places can be monitored to
B. IoT IN SMART ENVIRONMENT generate an alert, if the temperature exceeds to ensure the
Weather monitoring – Climate changes such as humidity, overall safety of workers and workplace.
pressure, temperature and rain can be monitored. Ozone Detection – The presence of ozone levels could be
Water Quality Monitoring – Quality of water can be detected in food industries during the process of drying
moni- tored, that is, whether it’s safe for humans to meat. Indoor Air Quality Surveillance – The quality of
consume it. air can be monitored inside workplaces to ensure general
Prevention of Natural Disasters – River levels could be safety of workers and goods [28].
monitored for early flood detection, similarly land slides Rescue Operations – Factory workers who may have been
and other natural disasters could be predicted. stuck in underground units due to natural disasters such
Air Quality Monitoring – Quality of air can be monitored as earthquake, land slides, explosion or some other natural
and measures could be taken to control the emission of CO2 calamity may be saved, thorough deployment of IoT re-
and other toxic gases by vehicles, factories and industries. sources that can accurately track their location. [29].
Forest Fire Detection – Different fire conditions such as
4 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

D. IoT IN AGRICULTURE
owner’s bank account associated with that automobile [32].
IoT has vast applications in agriculture, some of them are as Driving Insights – Different sensors like accelerometer,
follows: Gy- roscope, GPS etc. can be used to analyze the driving
Green House Monitoring – The climate conditions can patterns of the driver.
be monitored and controlled for green house to maximize Geo Fencing and Speed Monitoring – The system could
the production of vegetables and fruits, and their quality indicate the driver if he has gone out of the bounds of
can also be monitored by providing favorable prede- fined geographical area or is over speeding by
environmental conditions. comparing it with the predefined threshold limit.
Animal Tracking – Animals grazing in open fields and Driver Identification – Bio metrics can be used to
pastures can be identified and located. authenti- cate the driver. Bio metrics may include,
Air Quality Monitoring – The quality of air can be moni- fingerprints, face recognition or voice. The driver voice
tored to detect any harmful or toxic gases that may emit could also be used to provide voice commands to the
from waste material [21]. navigation systems. Bio metric data can also be used for anti-
Field Monitoring – The condition of fields can be monitored theft protection to ensure maximum security [33].
via different sensors, the data can then be processed, and
the farmer could be informed that a particular piece of land V. IoT CHALLENGES
requires special care. IoT improves the quality of our life, due to its numerous
Pest Control – Different mechanisms can be incorporated applications. However due to it’s security and privacy
to control the pests in crops, to ensure quality of crops [27]. issues, it’s likely that it’ll be less adopted by the users [34].
Water Management – The water can be managed IoT has vast uses and benefits in different sectors and solves
efficiently to minimize water wastage by making use of many problems, but still it has various challenges and
different sen- sors. limitations. One of the main challenge faced by IoT is
Soil Management – The condition of soil can be monitored maintaining privacy and security of users’s data [35], [36].
such as measuring the PH levels, salinity, moisture content IoT is a large scale network, which includes many
etc. so that the farmer sow the seeds according to the soil manufacturers, industries, and it may vary in different
level. applications according to the user’s need. Such large scale
RFID tags and Sensors – Using RFID (Radio-Frequency deployment of service, needs to be in the boundaries of a
Identification) tags and sensors will help identifying and certain standard. IoT will be developed in a step by step
recognizing the diseases that occurred in plants or crops. procedure. Various challenges faced by IoT need to be
The farmer can access the information from a remote addressed [37]. These challenges may include power
location and can take the necessary actions, to save the consumption, architecture challenge, het- erogeneity,
crops [30]. mobility, interoperability etc. [35], [37], [36]. Some of the
challenges of IoT are discussed below:
E. IoT IN AUTOMOTIVE INDUSTRY
Vehicle maintenance – The on board units installed on cars A. MOBILITY
can provide diagnosis of the car condition and can help the
One of the major issue to implement IoT is mobility,
driver to find faults in the engine or some other parts. The
because IoT is expected to offer services to the mobile users
safety measures could also be ensured, like seat belt check,
as well. It continuously needs to connect the users, in order
drowsiness detection, over speeding etc.
to provide better services. It is unable to provide the
Vehicle Tracking – The locations of vehicles can be
services to the mobile users when it needs to transfer from
tracked easily by simply installing IoT based trackers on the
one gate to another [35]. Mobility is one of the prominent
vehicles. Entertainment with Connected cars – The
characteristic of IoT devices, thus the devices need to join
connected cars become a center of Infotainment. It creates a
the nearby networks without any previous configuration.
pleasant experi- ence for the user. The apps on dash board can
Thus, good security mechanisms need to be implemented to
provide updates on traffic conditions, information about the
make the IoT devices compatible with mobility [36].
current trip, latest games and internet add the fun element to
the trip [31].
B. POWER CONSUMPTION
Collision Detection – In case of a vehicular mishap or
Power drainage of devices is one important challenge in
accident, vibration sensors installed in the cars can detect
IoT. IoT is concerned with how it can interlink things in a
the accident. The information can then be communicated to
com- patible fashion, while watching out for the energy
provide aid to the victims.
constraints because communication is one of the most
Fire detection – The fire sensors installed in the vehicles
power consuming task [21]. Computing is involved in every
can be used to detect the presence of flame or fire. On
aspect of human lives, so power consumption becomes an
detection of fire the fire rescue system could be triggered or
unavoidable issue. Some mechanisms should be introduced
concerned authorities can be notified.
to have IoT devices which consume less energy. In order to
Automated toll and fine payments – Every vehicle will have
use the IoT devices, it’s need of time that problem of
an RFID tag and on every toll booth, the sensors will scan
storing power in devices should be solved [38]. The
the car and a predefined amount will be deducted from the
energy capacity of IoT devices

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
VOLUME 4, 2016 5

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

is limited and they have to be replaced after some time, tuators attached to various daily life objects [45]. IoT has
however some devices consume a lot of power and they turned the traditional internet where only human to human
can’t be recharged. Low bandwidth connections should be services were offered into a network where real life objects
used, in order to extend the battery life [36]. can communicate and exchange data [46]. Different wire-
less and wired platforms can be used by the smart devices
C. SECURITY AND PRIVACY for communication purpose. Wireless IoT utilizes different
To provide and avail services through out the day in an IoT wireless protocols for communication [47]. These technolo-
environment, the things and people are connected with each gies include LoRaWAN, Near Field Communication
other. However, the communication over internet is prone (NFC), ZigBee, 6LoWPAN, Bluetooth Low Energy (BLE)
to security breaches, since the devices are not equipped and Z- Wave [44], [45], [47]. Some prominent wireless
with good security mechanisms. Different devices are communica- tion protocols used in IoT are discussed below:
connected with each other and data is shared among them
instantly, thus a mechanism is needed to ensure data A. LPWAN:
integrity and confidentiality. [39]. The network of IoT not Low Power Wide Area Network (LPWAN) is a wireless com-
only consists of things, but also crucial data and high value munication technology. The main characteristics of
gadgets which further creates hurdles to ensure security. LPWAN technologies are large communication range, long
The main issues arise due to implementation through battery life and low cost devices [44]. In the LPWAN
remote clouds which are connected with other interlinked technology, there are two categories, namely, Long Range
systems, user’s personal data etc. [40]. In order to make IoT (LoRa) and narrow band (NB-IoT) [48]. Lora is an
more usable in real world it’s security issues must be unlicensed long range low power wireless technology
solved. The IoT network is estimated to deal with a which improves network performance, reduces device cost
exponentially growing number of inter-connected devices. and supports large number of devices [46]. NB-IoT is a
These objects will exchange infor- mation; thus their licensed narrowband technology which provides improved
interactions must be secure to ensure data integrity. The performance in terms of range, reliability, QoS (Quality of
heterogeneous nature of IoT, where different types of Service) and latency [49].
devices, located at different places will interact, further
makes it complex for the deployment of an efficient and B. LoRaWAN:
scalable security algorithm [41]. Long Range Wide-Area Network (LoRaWAN) is a Low-
Power, Wide-Area Network (LPWAN) which has adequate
D. INTEROPERABILITY IN capacity and communication range with low power con-
HETEROGENEOUS ARCHITECTURE sumption and cost [44]. It offers core IoT requiremnts such
Interoperability is prominent challenge as far as IoT as secure communication, mobility etc. It also reduces the
network is concerned. This is mainly due to heterogeneous complexity involved in communication due to
nature of the IoT devices which have varying protocols, heterogeneous nature of IoT by providing seamless
data formats, platforms etc. [42]. Internet connectivity interoperability [45]. It’s a wireless communication
requires that the connected devices are able to communicate protocol developed by Lora Allience [46]. LoRaWAN
in the language understood by all of them, thus IoT systems standard supports two security layers, to ensure application
need to handle the interoperability issues [21]. IoT and network security. Device au- thentication is supported
platforms connect vari- ous devices that may include by network layer security, whereas security of application
different sensors, access points etc. Each platform may data is ensured by application layer security [44]. LoRaWAN
make use of different formats for availing different services classifies end devices into three cat- egories namely A, B and
and resources. Therefore, it’s quite a challenge to provide C. Class A devices are in sleep mode most of the time and are
services via such diverse platforms [43]. IoT energy efficient. LoRa gateway sends time synchronized
interoperability can be classified into device inter- beacons to Class B devices which enables them to open
operability, syntactic interoperability, networking interoper- extra receive windows. Class C devices are able to receive
ability, semantic interoperability and platform interoperabil- data at any time except for the time when data is being
ity. Many approaches have been proposed by researchers transmitted [45]. A typical LoRaWAN network has three
and industry to handle the interoperability issues, but still types of entities, namely, Gateways (GWs), End Devices
these approaches don’t cover all aspects of interoperability. (EDs), and Network Server (NS). LoRaWAN proved
The collaboration between different vendors could also help efficient in typical IoT applications such as smart meter-
in solving the interoperability issues [42]. ing and environmental monitoring [50]. LoRaWAN
provides high flexibility, scalability, security and
VI. PROMINENT IoT TECHNOLOGIES throughput [48].
IoT consists of a large network of interconnected things that
communicate and exchange data. IoT is able to transform C. NB-IoT:
any real life object into a computing device that can sense NB-IoT is a licensed narrowband technology and is time
and communicate [44]. It consists of a large network of slotted synchronous protocol [49]. It supports low complexity
heterogeneous devices consisting different sensors and ac- and low throughput Internet of Things. NB-IoT is built
using existing LTE (Long Term Evolution) characteristics.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
Its net-
6 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

work architecture is based on LTE network with little modi- systems also focused on taking preventive measures to pre-
fications to cater the needs of large network of users. NB- vent accidents, so the strategies used to detect and prevent
IoT requires a minimum bandwidth of 180 kHz to perform accidents are discussed separately. Figure 3 summarizes the
[51]. It has four downlink physical layer channels, namely, accident detection and prevention techniques discussed in
broadcast channel, synchronization channel, control channel this study.
and the data channel [52]. NB-IoT is preferred by
applications that require high QoS and low latency [49]. A. ACCIDENT DETECTION TECHNIQUES
Millions of people die in road accidents every year, most
D. BLUETOOTH LOW ENERGY (BLE): of the time accidents are not serious, but there is still a
Bluetooth Special Interest Group (SIG) developed huge loss of life due to delay of emergency services. So a
Bluetooth Low Energy (BLE) technology in the 4.0 version system capable of analyzing a situation and able to detect
of the Bluetooth. It’s also called Bluetooth Smart. It’s it as an accident would be very helpful in providing timely
energy ef- ficient and has low device costs [44]. It’s main assistance. Various strategies used to detect accidents are
characteristics include low energy consumption, easier discussed in this section.
setup and installa- tion, and supporting star topology [47].
It’s protocol stack consists link layer, physical layer and 1) Conventional Accident Detection
Logical Link Control and Adaptation Protocol (L2CAP). Techniques a: Using VANET (Vehicular Ad-
The stack is classified into Controller and Host. The hoc Network)
Controller is responsible to implement the link and physical Manuja M et al. [10] proposed that problems of traffic con-
layer, whereas the Host is responsible to implement the gestion arise due to vehicle failure or due to accidents in no
upper layers [44]. The low en- ergy consumption feature of network area. They proposed a system to solve this problem
BLE makes it a suitable protocol for many applications based on VANET. In this system every moving vehicle is
such as wearables, health monitoring apps, home appliances considered as a node. The alert messages are transmitted
etc. [45]. using RF module, and alert messages are received by the
moving vehicles that are in the range of RF module. Finally,
E. ZigBee: the vehicle in the network area receives the message, then
ZigBee is a wireless communication technology defined by the received message is transmitted to the base station. The
the ZigBee Alliance. It provides seamless interoperability alert message contains four types of messages. These are
between devices from different vendors. The ZigBee protocol detected by piezo electric sensor, MEMS (Micro-electro-
stack is build up on IEEE 802.15.4 standard. To ensure mechanical systems) sensor, flame sensor, and temperature
secure communication it provides two types of keys sensor.
namely, network key and multiple link key. The network The system identifies vehicular mishap in considerably
key is shared by all the nodes in network, whereas the less time and alerts the rescue teams along with the location
multiple link key is only shared among the communicating of mishap. A switch is also added in the system, which
nodes [44]. ZigBee supports different types of topologies the driver can use to stop sending alert message in case
like tree, star and mesh network topology. It’s suitable for when injuries are minimum or there’s no serious loss. The
those applications and devices that require low data controller gets the input from sensors and sends the accident
consumption, good battery life and maximum security [47]. alert information to the road side unit and then message is
sent to the rescue team. WIFI and GPS are used to find the
F. NFC: location of vehicle. In previous systems, GSM module
Near Field Communication (NFC) is a short-range failed to communicate in no network area, it’ll provide
communi- cation technology, which operates on coverage in network and no network area. A switch is also
electromagnetic fields at a frequency of 13.56 MHz and a available to stop sending of message in case of no serious
distance of about 10cm. The devices in a NFC network damage. However, the major limitations that could be
communicate by generating fields [44]. NFC contains a tag addressed in the future are VANETs have security and
which has data that can be read only or the device can privacy issues, also they have highly dynamic topology due
rewrite it later on. NFC operates in three modes [47]. NFC to which routing issues may arise.
technology makes it simpler to exchange digital content, B. Shabir et al. [54] presented a survey of different
making transactions and connecting electronic gadgets with a conges- tion control techniques for VANETs. To address
simple touch. It’s configuration and setup is easier to the network congestion issue in VANETs three different
implement and it does not require line of sight [53]. approaches are discussed which are proactive, reactive and
hybrid congestion control strategies. To address network
congestion, latency and throughput are two important
VII. LITERATURE SURVEY
parameters. The techniques are categorized in six major
In this section different techniques for collision detection
categories which include rate- based, power based, hybrid
and reporting are discussed. Most of the systems focused
strategies, CSMA/CA (Carrier Sense Multiple Access with
primarily on mechanisms for accident detection, while some
Collision Avoidance) based, prioritizing and clustering-
based strategies. The discussions show that proactive
techniques are best to control congestion
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
VOLUME 4, 2016 7

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

Accident Messaging and Detection Systems

Characterised As

DetectionPreventionHybrid ML/AI Based

VANET 1.DBA ULS ADFL

DC
UVS AC UDS APSVM
SBC
DLC DDDD
UPS ADANN

UAS 2.VBA UCSA DIDCNN

SC
UPST SDO SH DDRL
DBT
USSG PDCNN

UGG

UGGARFT

FIGURE 3: A Taxonomy of Accident Detection and Prevention

Techniques

in VANETs since they proactively handle the congestion in A typical scenario for reporting emergency situations in a
network, moreover priority based approaches also show VANET environment is demonstrated in Figure 4. The
bet- ter performance if they are used along with hybrid main issue in VANETs is, it’s highly dynamic topology
congestion control techniques. due, to which different problems like network congestion,
In [55] its discussed that network congestion is a se- frequent disconnections etc. may arise. To address these
vere issue in VANETs, since VANETs have highly issues, mul- ticast routing techniques can be used in
dynamic topology. Congestion in VANETs can really affect VANET protocols. Different VANET multicast routing
the per- formance of applications using them, especially protocols are reviewed in this study. The performance of
emergency services like rescue teams, ambulances who routing protocols is analyzed based on the routing
need to com- municate timely in case of emergency techniques used by the protocols. The multicast routing
situations like road accidents, mishaps, security alerts etc. techniques are classified as proactive, reac- tive and
The main applications of VANETs include safety and non- flooding techniques. The goal behind development of
safety applications. The safety applications can reduce the multicast routing methods in VANETs is to address issues
chance of accidents by providing early warnings to the like network overhead, avoiding loops and adaptability in
vehicles. Time criticality is one of the main concern in highly dynamic topology. The VANET multicast routing pro-
safety applications, however the congestion in network can tocols are categorized into geocast and cluster-based
degrade it’s performance. routing. The multicast routing protocols for VANETs are
A VANET is a type of MANET (Mobile ad hoc net- compared on various parameters such as real time, location
work) which considers vehicles as mobile nodes.VANETs based, map based and spatiotemporal. The study shows that
can be used in ITS to ensure convenience and road safety. use of multi-
8 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

monitored, so it might drain a lot of battery.

d: Using Accelerometer Sensors:

In [4] its discussed that,“one fourth of the people involved
in accidents are motorcyclists”. Rash and careless driving
are the primary causes of causalities in bike riders. A
system is proposed which uses helmet as an apparatus
for acci- dent detection and reporting. Sensors, processors
and cloud computing infrastructure are used to build the
system. On accident detection details are sent to the
emergency contacts using Cloud based service. GPS tracks
the location of vehi- cle. A GPS, microcontroller and an tri-
axial accelerometer are present on the helmet. The
probability of accident is cal- culated by continuously
monitoring orientation of head and helmet’s position. If the
pre-fetched speed limit is exceeded, a text message as an
alert is sent to the emergency contacts. The system on the
FIGURE 4: Accident reporting scenario in VANET
helmet connects with the cloud server using RESTful
architecture communicating through HTTP (HyperText
Transfer Protocol). The system is cost effective as
casting in VANET routing protocols reduces fragmentation, compared to other systems, also on occurrence of accident,
power consumption and transmission overhead. The setup for text messages are repeatedly initiated until acknowledged.
multicast routing protocol for VANET may vary according to The major drawback of this approach is that there is a possi-
different environments, hence a single protocol may not be bility of false alarm if the driver drops the helmet by
suitable for every type of VANET environment, because the mistake , additionally, more storage and computational
design may change according to the needs and functionality capabilities are needed because the system needs to
of that particular environment. communicate with the cloud services.

b: Using Vibration Sensors: B.S.Anil et al. [58] presented a solution for collision
D.B. Tushara et al. [56] proposed a solution for accident de- detection. The system detects the accident using flex and
tection that uses a micro controller to control operations like accelerometer sensors and the location of accident is com-
detecting and reporting. The system focuses on minimizing municated via GSM modem to the emergency services.
the action time after an accident has occurred. A number is Alert message contains information about location, vehicle
pre-fetched on the system, to which an alert is sent on number and time of accident occurrence. The current
occur- rence of accident. The accident is detected using a situation of passengers is monitored through real time video
vibration sensor. The system has high probability of transferred by the in mounted camera. Accelerometer
generating wrong output because of its reliance on single monitors speed of vehicle. Sensor’s output is compared
sensor only. The weakness of the system is that it generates with the pre-fetched value and if it exceeds the defined
an alert message on occurrence of accident but accident threshold value, it indicates the occurrence of accident. The
location is not shared. camera on the receiver module is connected with a screen,
on which live situation of vehicle can be seen. The major
c: Using Piezoelectric Sensors: advantage of this approach is that the accident severity
U. Patil et al. [57] put forward a solution in which the could be known and medical aid could be provided
system will sense accidents and will inform the nearest accordingly due to in mounted cameras in the vehicle.
police station and rescue teams. GSM technology is used However, this also has privacy concerns due to live video
to communi- cate alert messages to the emergency services. feed transmission.
The system continuously tracks the vehicle and can
immediately alert the emergency services in case of any R.K Megalingam et al. [59] proposed a solution to
mishap. Renessa’s microcontroller is used in conjunction smartly detect vehicle collision and alert the rescue teams.
with GSM modem and GPS receiver. GSM is used for The sys- tem can be placed inside a vehicle in which an
communication purpose which sends an alert message accelerometer sensor detects an accident, this output of
containing the location which is provided by GPS. It’s main sensor is moni- tored by a microcontroller. The RF (Radio
modules consist of piezoelectric sensor, GSM and GPS. Frequency) trans- mitter module is interfaced with the
The system tracks the location con- tinuously so in case of microcontroller which transmits the information to the
mishap the location is communicated timely. The primary emergency services. The RF receiver module at the
drawback of this approach is that, no switch is available to emergency service will receive this information. The
cancel sending of alert messages in case of no serious emergency services on receiving the information would
damage, secondly location is continuously take appropriate actions to provide timely medical aid. The
major drawback of this approach is that

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
VOLUME 4, 2016 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

RF module has limited range, moreover it relies only on one the RF receiver is present in the traffic lights. When the traffic
sensor, so it has single point of failure. lights sense RF signal, they turn green to allow the ambulance
reach the accident spot as fast as it could. The emergency
message is also sent to the victim’s family.

g: Using shock Sensors, GPS:

E. Nasr et al. [61] presented a solution in which the system
informs the PSO (Public Safety Organization) on
occurrence of accident. Shock Sensor detects occurrence of
accident, the signal is processed and then sends the
geographic location to PSO. The system has different
phases which are as follows:
(a) Registration of vehicle
(b) Registration of passenger
(c) Accident monitoring through PSO headquarters
In Vehicle Registration phase, the person’s vehicle is reg-
istered. An IoT device is installed on the system. After
FIGURE 5: Smartphone based accident detection system installation of device, the person gives the vehicle ID to
the person who’s responsible for registration of vehicles in
the PSO headquarters’ database. Its main modules consist
of NFC reader, shock sensor, GPS, and cellular IoT. These
e: Using Pressure sensor, Tilt angle: modules communicate with each other to identify an
In [12] a solution based on android application for accident accident and report it to the PSO headquarters with the
detection and reporting is presented. Outward force experi- exact location of accident. As the shock sensor detects
enced by the vehicle is monitored by means of an external accident, an alert through HTTP request is sent along with
pressure sensor, speed and tilt angles are measured by GPS the location.
and accelerometer sensors of the phone. Bluetooth on the
phone receives the data from sensors. As accident occurs h: Using GPS, GSM:
the speed of vehicle decreases abruptly. So an accident will In [5] a system is proposed, that considers the speed as one
be detected if vehicle speed changes abruptly and values of of the major cause of accident. It uses a GPS receiver to
pressure and tilt angle exceed the pre-fetched threshold monitor the speed and detects an accident based on monitored
limit. A switch which the driver can use to stop sending of speed. The GPS module continuously monitors the speed
alert message in case when the accident isn’t severe or in and compares with the previously monitored speed every
case of false alarm is also present. A smartphone based second using a micro-controller unit.
accident detection system is demonstrated in Figure 5.
However, the research is subject to some limitations such
as magnitude of force experienced by the smartphone
would not be the same as of the magnitude of force
experienced by the vehicle, moreover smartphones have
battery constraints.

f: Using Gyroscope, GPS, android app and

RF transmitter:
H.Kumar et al. [60] proposed a solution for accident detec-
tion and reporting which requires no manual interaction of
human before or after the occurrence of accident. Its main
module consists of GPS, Raspberry Pi, android application
and Gyroscope. The system also stores the blood group of
the driver. As soon as the vehicle meets an accident, the
gyro which is placed in the vehicle measures the rotation FIGURE 6: An accident detection system using GPS and
angle and if it exceeds the threshold limit, it will send an GSM
alert to the active server. The app is placed in the
ambulance which continuously receives the information Whenever the system identifies that the speed is less than
from server, the app helps the ambulance to navigate to the the pre-fetched threshold limit it will detect the situation
accident spot using the route which has less traffic by using as an accident. The location is detected by using a GPS
Google maps API (Application Programming Interface). To module. An alert message is sent to the emergency services
avoid traffic congestion, the ambulance transmits RF signal using a GSM module. Figure 6 illustrates a GPS/GSM
time to time, based accident detection system. The problem of the system
is that

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

10 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

it relies only on a single sensor for accident detection which like traffic conditions, human negligence etc. contribute to-
has high chance of generating false positives. wards road accidents. The major inputs to the model are
In [62] a solution to find the position of vehicle based on number of vehicles passing in a lane, speed, road width and
ARM micro controller is presented. To get the position of road surface condition. Accident data used in the model was
vehicle, the intended person initiates a request via SMS. obtained from the authorities, for the road that had
The request is received by the GSM module and is increasing accidents in the past years, to improve the
acknowledged. The request is then processed by the Spartan accuracy of the model. The model maps the crisp inputs to
processor. The processor sends the command to GPS corresponding crisp outputs using fuzzy rules based on
module of the system. The GPS module tracks the location conditions. Every input variable is defined by the
of vehicle and replies the request with location coordinates membership functions. In fuzzification process the fuzzy
of the vehicle. Vehicle’s location is sent to the user with variables are constructed by deriving membership function
longitude and latitude’s values. The weakness in the for inputs and outputs and linguistic representation of these
literature lies in the fact that external environmental factors membership functions. Tri- angular and trapezoidal
can have greater impact on the performance of system. functions are used in this model. The results of the model
show good accuracy in predicting acci- dents. The factors
2) ML/AI Based Accident Detection Techniques like road lighting and weather conditions were ignored in
Transportation system plays an important role in human this study.
life; however, where it provides many facilities, it has many
risks associated with it as well [63]. The road traffic b: Accident Prediction using Support Vector
accidents are increasing every year, so an effective Machine (SVM):
mechanism is needed to minimize its frequency [16]. B. Pan and H. Wu [66] proposed an approach to detect
Crash prediction models have been very famous to accidents by means of mobile sensors and SVM. Their ap-
ensure road safety particularly on highways [64]. Different proach particularly focused on detecting accidents occurring
models have been presented for crash prediction in the in urban roads, since they are more prone to accidents, due
domain of Machine Learning (ML) and Artificial to involvement of many flow disruptive entities like bus
Intelligence (AI), namely, Artificial Neural stops, traffic signals etc. as compared to freeways where the
Networks (ANN) [65], Support Vector Machine (SVM) traffic flow is not disturbed. This approach uses VANETs in
[66], [64], fuzzy logic [19], [63], Genetic Programming which every vehicle is able to collect its own traffic data
[67], Ran- dom Forest Classifiers [16] etc. The real time like location, identity, speed, lane state with the help of
data is fetched in the models and is compared with the mounted sensors on vehicles. The traffic data is then
previously collected accident data, which helps in collected by RSU (Road Side Unit) for further processing
differentiating a normal situa- tion from an accident [17]. which is sent by On Board Units. Three traffic variables are
The key approaches for accident prediction in the ML/AI considered to detect accident which are vehicle’s speed,
domain are discussed below: acceleration and lane- changing state. Wider range of
monitoring can be provided by means of on-board mobile
a: Accident Detection Using Fuzzy Logic: sensors.
A. Alkandari et al. [19] proposed a solution to detect acci-
dents in range of traffic lights using fuzzy logic technique c: Accident Detection using Artificial Neural Network:
which detects accidents. The system consists of two sub In [18] a system is proposed that can detect an accident
systems namely, Detection System and Action System. The from the video footage obtained from the CCTV cameras
system is based on Webster Method with a little variation. installed on highways. Each frame of video is given as an
The system collects data about different zones including input to the Convolutional Neural Network (CNN) model,
number of cars in a lane, speed of cars in particular lanes which is able to distinguish frames of video into accident
etc. The disturbance in normal traffic flow is the main and non-accident. Raspberry Pi 3 B+ Model is used as a
indicator that an accident has occurred. The main elements remote computer that can be placed on the CCTV cameras.
of the system are crisp inputs/outputs, membership A Pi Camera is used in the system for the purpose of
functions, fuzzy rules and linguistic variables. The demonstration to obtain the video data. Inception v3 model
linguistic variables which are, cross ratio, zone status, is used to detect acci- dents by prior training it, on two
accident status and section speed help in determining a different set of video data consisting of accident and non-
situation as an accident. The output of the system is accident frames. The model is based on CNN which is
generated by applying the fuzzy rules to the linguistic useful for image classification, object detection etc. The
inputs, depending upon the input and rules a suitable action proposed model uses both CNN and Long Short-Term
would be taken to improve the traffic flow. The results Memory (LSTM) layers. The inception v3 can work on
show that the system is able to identify most of the accident heterogeneous convolutions which enables it to extract
scenarios with good accuracy. more features from the images. The model is implemented
In [63] an accident prediction model using fuzzy logic is on Raspberry Pi by using Tensor Flow, Open CV and
presented. The relation between accidents and the factors Keras. Each video frame is run through the model and then
contributing it are non-linear thus using fuzzy logic is a prediction is given whether the frame is accident frame or
preferable choice for non-linear relationships. Many factors not. A threshold limit is set and if the prediction
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
VOLUME 4, 2016 11

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

limit exceeds 60%, the GSM module is triggered by the

Raspberry Pi, which sends an alert message to the nearest
hospitals along with the location coordinates. The results
show good accuracy of the system for the trained set of
data, moreover the system is also cost effective. The
limitation of the study is that, the model was trained for
severe accident conditions data, thus the model may not cover
all the accident scenarios where accident may be moderate
or minor.

B. ACCIDENT PREVENTION TECHNIQUES

Accident prevention can be defined as the strategy or an
approach, or actions taken to avoid or stop an accident
before it occurs. Majority of the accidents occur due to
human negligence. These factors are over speeding, traffic
law vi- olations, drunk driving etc. So controlling these
factors can help to avoid accidents and save the precious
lives. Different approaches used to prevent accidents are
discussed in this section.

1) Conventional Accident Prevention FIGURE 7: Workflow of alcohol sensor based systems to

Techniques a: Driver Drowsiness Check: prevent drunk driving
S. Eduku et al. [68] presented a solution to prevent traffic
hazards by proposing a system which uses eye blink sensor
and automatic braking system to slow down the car and However, its main limitation is that there could be a false
bring it in the state of halt, if drowsiness is detected in alarm in case when the person sitting next to driver is drunk
driver. A RF module routs the information to the nearby but the driver is not drunk, moreover external interference
vehicles to alert them that a halt car is there. IR sensors are of air can also affect the performance of alcohol sensor.
used to monitor the eye blink and detect the state of
drowsiness. Infra-red rays are transmitted to the eyes by c: Maintaining safe distance using ultrasonic sensors:
means of a IR transmitter, the IR receiver on the other hand S. Kasera [69] presented a system to prevent accidents
receives the reflected rays from the eye. If the output of IR while driving in hilly areas. Since the roads on the hilly
receiver tends to be high, it implies that the eyes are areas are very steep and curvy. To implement this
closed. Thus the drowsiness is detected by means of mechanism, the system uses ultrasonic sensors. The system
monitoring this high/low output. Its main modules consist alerts the driver going on one side of the road about the
of an alarm that warns the driver if drowsiness is detected, vehicle coming from the opposite direction of the road. The
automatic breaking systems to slow down the car, and bring ultrasonic sensors are placed along one side of the road
it to halt state and RF module to send alert messages to the before curve and a LED light after the curve, so if the
nearby vehicles in range. The main limitation of the system vehicle comes from one end of the curve the sensor will
is that the range of RF module is low. detect it and LED light lights up on the opposite side of the
road. Two ultrasonic sensors and two LED lights are
installed on both sides of the curve. As soon as the vehicle
b: Using Alcohol sensors, GPS, GSM: reaches one end of the curve, the RED light on the opposite
T. V. Narayana et al. [3] proposed a system to prevent acci- side will automatically light up and will remain on, unless
dents using alcohol sensors to prevent drunken driving, and until the sensor on the other side of the curve detects it.
since 70% of the cases of accidents are due to “drunk and It’s a low power consumption system, moreover its
drive”. An alcohol sensor placed in the system, is used to implementation is very cost effective.
monitor the contents of alcohol in blood. A workflow of
such a system is illustrated in Figure 7. Sensor is placed 2) ML/AI Based Accident Prevention
above the steering so that, as the driver breathes the Techniques a: Driver Distraction Identification
sensor could determine the alcohol level. A threshold is set with Deep Convolutional Neural Network
and if the contents are more than the pre fetched limit, the (CNN):
car doesn’t move. GPS detects the location of the vehicle. In [70] it’s discussed that, distracted driving is leading
Messages are sent by means of GSM to the pre-fetched cause of road collisions. Distracted driving can be any
mobile numbers and inform them that the person is highly activity which makes the driver lose his attention and focus
intoxicated and can’t drive. The system focuses on the while driving. Some of these activities include talking on
preventive measures to avoid any hazardous situation. The phone, texting, adjusting the radio etc. The proposed
system would be helpful to public in general as well, since technique iden- tifies distracted driving using CNN. Driver
it won’t allow drunk person to drive. distraction can be
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
12 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

classified into different categories namely, visual, manual and set consists of positive samples (pedestrians) and negative
cognitive. Multiple types of distraction can exist together in samples (background) obtained from the camera mounted
an activity. Distracted drivers can be identified using on the top of car. In order to consider variety of lighting
various measures, this technique particularly focuses on conditions, the data set was equally distributed in different
using visual data to identify distracted driver. A camera is time zones namely, morning, noon, and evening time. After
installed inside the car cabin to identify the driver’s the merging algorithm terminates, final detection list contains
behavior. Deep learning model is used to classify distracted the final results of the detection system. If the height and
driving from normal driving. In this study, a comparison of position are within the margin of tolerance, the result was
pre-trained deep CNN is presented. The data set is divided considered correct. This approach achieves high accuracy
into 10 distinct classes to identify different distraction through automatic optimization of the features and regular-
behaviors like texting, drink- ing, talking to other passenger ization of neural network. The accuracy of the classifier is
etc. The comparison shows that VGG16 with S2 shows improved by using Haar wavelet features.
good performance in terms of accuracy. The models were
pre-trained on the Image net data set, and then retrained on C. HYBRID TECHNIQUES
State Farm data set to identify distracted driving
Road collisions are one of the leading cause of fatalities.
behaviours.
In most cases, injuries are not serious and if victims are
rescued in time, lives may be saved. Many factors
b: Driver Drowsiness Detection using
contribute to accidents, including driver negligence, drowsy
Representation Learning:
and drunk driving etc. Thus a system that can detect as well
K. Dwivedi et al. [71] proposed a methodology to detect as con- trols the factors contributing to accidents will be
drowsy state of a driver using representation learning to helpful in preventing accidents and saving lives. Hybrid
prevent accidents caused by sleep. The model extracts the techniques are techniques which use both accident detection
visual features from the data, the features are learnt by and prevention mechanisms, as shown in Figure 8. Some of
using CNN. The input image is combined with the learned the systems based on hybrid techniques are discussed
weights to produce feature maps. A soft-max classifier layer below.
uses these set of features to classify the frames into drowsy
and non-drowsy. The facial features are derived using CNN
based representation learning technique by which the
complex re- lationships of raw data can be represented by
combining features of features. The model consists of two
convolutional layers along with a hidden layer of sigmoid
function which is linked with regression layer for
classification purpose. The frames are extracted from the
video and are fed to a face detector. The detected faces
are then cropped to square images and then the images are
normalized, which are then fed to a convolutional neural
network. The classifier shows good performance on diverse
data set. The major advantage of the approach is that,
representation learning can capture more intelligent features
from raw input data as compared to manual methods.
FIGURE 8: An accident detection and prevention system
c: Pedestrian Detection with CNN:
A considerable amount of pedestrians die every year in road
crashes. M.Szarvas et al. [72] proposed the use of CNN 1) Using Limit Switches:
as a classifier in a pedestrian detection system, to identify
P. Berade et al. [11] put forward a solution for vehicular
the pedestrians that could help in reducing the number of
mishap detection by means of GPS to determine the mishap
accidents. The algorithm is tested on pedestrians in an
location and GSM to communicate the information to the
urban environment. CNN classifier is compared with the
pre-fetched number. Additionally, the system also identifies
SVM (Support Vector Machine) classifier and results show
if the driver isn’t wearing seat belt or if he is intoxicated.
that CNN wins over SVM in terms of accuracy using
The sensor detects the accident and then PIC
classic im- age features. The candidates’ images are given
(Programmable Interface Controllers) microcontrollers send
as an input to the classifier. A classification score is given to
signals to GPS, GPS tracks the location and then the alert
each candidate by the CNN, candidates with higher scores
message is sent through GPS module. Limit switches are
than the threshold are saved in the raw detection list. The
used for accident detection. Seat belt test will ensure that
list is then sorted by classification scores.
the driver is wearing seat belt and alcohol sensor placed on
Multiple detection merging operation is used to remove
steering will check if driver is drunk. The accident is
re- dundant results for the same pedestrian. The evaluation
detected via limit switches placed at the back and front of the
data
car. A LCD screen displays

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
VOLUME 4, 2016 13

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

messages, in case of seat belt is not worn, or if driver has

consumed alcohol, LCD screen gives warnings. System not
only focuses on accident detection but also on its prevention
by introducing seat belt and alcohol sensors. The use of
PIC micro controllers in this system is very effective since
they consume very less power as compared to other micro
controllers and are also very fast. The findings of this study
have to be seen in light of some limitations, as no switch is
included to cut off the signal to the rescue teams in case the
driver is safe.

2) Drowsy and Drunk Driving Detection:

S. Nanda et al. [2] presented a system which detects
accidents by means of vibration sensors. The solution
kept in mind the needs of bike riders as they are more
prone to serious harm as bikes aren’t equipped with FIGURE 9: Obstacle distance based collision prevention
preventive measures as compared to other vehicles. GPS system
tracks the location and for communication purpose GSM is
used. The system also has a mechanism for checking if the
driver is sleepy. The system also checks if the rider has alerts in case of abnormal conditions of engine temperature
driving license. The system also makes sure that the or accelerometer. It also has a feature to capture image after
driver hasn’t consumed alcohol. The camera placed on the accident occurrence and send the multimedia messages to
front of bike monitors the traffic signals and automatically pre-fetched numbers. Its main components include speed
lessens the speed. It makes use of many measures like sensor, accelerometer sensor, cell phone interface , tem-
traffic signal detection, checking driver license and perature sensor and Real Time Calendar (RTC) chip. It’s
checking the state of driver to prevent accidents. The environment friendly since junk cell phones are recycled. It
system not only detects but also focuses on preventive can be easily implemented on existing vehicles. The major
measures to avoid accidents thus, saving precious lives. The limitation of this approach is that on occurrence of
major limitation of this approach is, an inexperienced accidents, alert messages have to be sent manually by the
person can use another person’s license to drive, since the user.
system does not authenticate if the license belongs to the
person driving. 5) Smart Helmet
In [14] a system is presented, which uses smart helmet as an
3) Using Distance Sensors: apparatus for accident detection and prevention. The system
In [73] a system is proposed that always monitors the dis- won’t allow the rider to drive unless and until he wears a
tance between vehicles and obstacles that are in front, using hel- met. The pressure sensor and accelerometer sensor are
distance sensors. A system maintaining safe distance from used to check if helmet is worn by the rider or not. The
obstacles is demonstrated in Figure 9. The proposed system alcohol sensors placed on the helmet check the alcohol
alerts the driver to control the speed and reduce the speed content in the drivers breathe. If the contents are more than
by itself when a critical distance comes. Whenever an acci- the pre-defined threshold limit, the bike won’t start, thus
dent takes place for uncertain condition, an email alert will preventing drunken driving. On occurrence of crash the
be sent to the accountable individual with car details. The helmet worn by the rider will hit the ground,the sensors will
system automatically controls the speed in case when detect the motion and tilts of helmet,and will immediately
critical point is reached. Vehicle’s distance is measured by send the site of crash to the emergency contact numbers and
ultrasonic sensors. The system makes use of Raspberry Pi, rescue teams. The system would be very helpful to the
Ultrasonic Sensor, Led Buzzer and Servo motor. On driver as well as the general public, since drunk driving not
occurrence of accident the alert message is sent using only risks the life of driver but also affects the safety of
Raspberry Pi. It prevents accidents by controlling speed other people driving on the roads.
when a critical point is reached. Email alerts are sent in case
of accident, which are slow means of communication, and VIII. COMPARATIVE ANALYSIS
providing medical aid could take time. In this section, various systems discussed in the literature
survey are compared according to the parameters defined in
4) Using Cell phone,Speed and Accelerometer sensors Table 1. The systems are compared in terms of accuracy,
V. Ahmed et al. [13] presented a solution for accident de- their implementation cost to incorporate them in existing
tection and prevention. A system based on microcontroller vehicles, what’s their communication mechanism to send
and an accelerometer sensor along with a recycled cell phone. alert messages to the rescue and emergency teams, are the
The system warns the user in case of unsafe driving and sends messages sent automatically on detection of an accident or
they have to be sent manually by the user etc. Since no system
14 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 1: Comparative Analysis.

Techniques FPR Cost CM Acc AR STF
Nanda et al. [2] L H GSM M Y N
Rao et al. [3] H M GSM L Y N
Chandran et al. [4] M H MC M Y N
Amin et al. [5] H M GSM L Y N/A
Manuja M et al. [10] L H GSM H Y Y
Berade et al. [11] L M GSM M Y N
Faiz et al. [12] H M GSM M Y Y
Ahmed et al. [13] H L GSM L N N
Priyanka C et al. [14] H M GSM L Y N
Ghosh et al. [18] L H GSM H Y N/A
Alkandari et al. [19] L M N/A H N/A N/A
Tushara et al. [56] H L GSM L Y N
Patil et al. [57] H M GSM L Y N
Anil et al. [58] M M GSM M Y N
Megalingam et al. [59] H L RF L Y N
Kumar and Deepak [60] L H GSM H Y N
Nasr et al. [61] H M GSM L Y N
Wen and Meng [62] N/A M GSM M N/A N/A
Gaber et al. [63] M M N/A M N/A N/A
Pan and Wu [66] M H RF M N/A N/A
Eduku et al. [68] L M RF H Y N
Kasera et al. [69] L L N/A M N/A N
Basubeit et al. [70] L M N/A H N/A N/A
Dwivedi et al. [71] M M N/A M N/A N/A
Szarvas et al. [72] L H N/A H N/A N/A
Murshed et al. [73] M M RP M Y N

FPR:False Positive Rate CM:Communication Mechanism Acc:Accuracy AR:Automatic Reporting STF:Switch to Terminate False alarm
H: High L: Low M: Medium N: No Y:Yes
N/A: Not Applicable RP:Raspberry Pi MC:Micro-controller RF: Radio Frequency

is perfectly accurate, there’s always a chance of error in any systems, for accident detection and prevention is presented
system, so the systems are also compared against their chance in Table 2. It’s quiet evident from the analysis that, most of
of generating a false positive, i.e. triggering of alarm in the systems use GPS to detect location of accident, GSM
non- accident circumstances. as a communication mechanism to send alert messages and
Moreover, in some cases, the accident severity may not accelerometer and pressure sensors to detect accidents.
be serious and the driver may not need immediate medical
aid, so in those cases a mechanism should be there to stop
IX. RECOMMENDATIONS
the sending of an alert message, to save the time of rescue
teams. So the presented systems are also compared if Various methods for accident detection and prevention were
they’ve the mechanism to stop sending of an alert message discussed in this paper. The methods included warning the
or not. driver for over-speeding, maintaining safe distance from
other vehicles, avoiding intoxicated and distracted driving
It can be seen from Table.1 that many systems lack accu-
etc. Integration of these systems with the vehicles would
racy, and may not generate accurate results in every
be very beneficial to the society. These systems would be
situation. Moreover, many systems did not have a
effective in minimizing the casualties associated with road
mechanism to stop sending of alert messages in case of
accidents. Additionally, patient history such as blood group,
false alarm.
age, allergies etc. can also be included in these systems to
Most of the systems presented in the literature survey,
provide medical aid accordingly. Moreover, data obtained
relied on some sort of hardware based technologies like
from sensors after an accident has occurred, can be used in
sensors for implementing accident detection and prevention
data mining to deduce important results. Performing
mechanisms. A summary of different type of sensors such
analysis
as alcohol, seat belt, vibration, pressure etc. used in these
VOLUME 4, 2016 15

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 2: Summary of sensor types used for Conventional Accident Detection and Prevention Techniques

Techniques Accel S GPS P D V RFID SB LS AL Temp

IR Nanda et al. [2] J × J J J J J
× × J × × Rao et al. [3] × × J
× × × × × × J × ×
Chandran et al. [4] J × J × × × × × × × × ×
Amin et al. [5] × × J × × × × × × × × ×
Manuja M et al. [10] × × J × × J × × × × J
× Berade et al. [11] × × J × × × × J J
J × × Faiz et al. [12] J × J J × × ×
× × × × ×
Ahmed et al. [13] J J × × × × × × × × J
× Priyanka C et al. [14]J × J J × × × × ×
J × × Tushara et al. [56] × × × × × J ×
× × × × × Patil et al. [57] × × J J ×
× × × × × × × Anil et al. [58] J × J × ×
× × × × × × ×
Megalingam et al. [59] J × × × × × × × × × ×
× Kumar and Deepak [60] × × J J × × × ×
× × × × Nasr et al. [61] × × J × × ×
× × × × × ×
Wen and Meng [62] × × J × × × × × × × ×
× Eduku et al. [68] × × × × × × × × ×
× × J Kasera [69] × × × × J × ×
× × × × ×
Murshed et al. [73] × × × × J × × × × × × ×

Accel: Accelerometer S:Speed P:Pressure D:Distance V:Vibration SB:Seat Belt LS:Limit Switches AL: Alcohol Temp:Temperature IR:Infrared

on the data can give us valuable insights on how most of provide many advantages such as mitigating road collisions,
the road accidents occur, which factors contribute the most identifying precise acci- dent locations and facilitating over
in event of mishaps, which roads are dangerous and the all rescue operations. The
time stamp in which most of the accidents occur. The data
collected from these systems could also help police to find
the crimes of hit and run cases. In addition, the information
about occurrence of accidents can be routed to the vehicles
in range, to avoid any further mishaps, that can be helpful in
reducing chain reaction accidents in which multiple
vehicles are involved in the crash, creating a chaotic
situation.

X. CONCLUSION AND FUTURE WORK

The number of casualties associated with road collisions is
growing rapidly. If victims are rescued in due time, several
lives may be saved. We discussed various strategies which
focused not only on accident detection but also on its pre-
vention. These strategies utilized various sensors such as
accelerometer sensors, shock sensors, pressure sensors etc.
and various machine learning techniques such as neural net-
works, support vector machines, representation learning etc.
for accident detection. Various strategies for accident
preven- tion were also addressed, which include detection
of drunk and drowsy driver, regulating vehicle speed,
maintaining safe distance from obstacles etc. Once the
accident is detected, the information is communicated to
emergency services to provide timely aid. Such systems

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access
integration of these systems with vehicles would be dent prevention and detection,” in 2018 Fourth International Conference
on Computing Communication Control and Automation (ICCUBEA),
somehow expensive yet will give various advantages. 2018, pp. 1–6.
However, the sys- tems we discussed were all reliant on [3] T. V. N. Rao and K. R. Yellu, “Preventing drunken driving accidents
some kind of hardware or software based technology and using iot,” International Journal of Advanced Research in Computer Sci-
ence(IJARCS), vol. 8, no. 3, pp. 397–400, 2017.
there is a possibility that those sensors or devices can
[4] S. Chandran, S. Chandrasekar, and N. E. Elizabeth, “Konnect: An internet
themselves be destroyed in the accident and can generate of things (iot) based smart helmet for accident detection and notification,”
erroneous readings and results. So such frameworks are in 2016 IEEE Annual India Conference (INDICON), 2016, pp. 1–4.
required which are less reliant on some kind of hardware [5] M. S. Amin, J. Jalil, and M. B. I. Reaz, “Accident detection and reporting
system using gps, gprs and gsm technology,” in 2012 International Con-
or software. ference on Informatics, Electronics & Vision (ICIEV), 2012, pp. 640–643.
[6] J. B. Edwards, “The relationship between road accident severity and
recorded weather,” Journal of Safety Research, vol. 29, no. 4, pp. 249–
REFERENCES 262, 1998.
[1] F. Bhatti, M. A. Shah, C. Maple, and S. U. Islam, “A novel [7] Nayan Kumar R.N., Navya V.R., Ganashree N., Pranav K.U., and Ajay.N,
internet of things-enabled accident detection and reporting system for “Intelligent vehicle accident detection and ambulance rescue system,”
smart city environments,” Sensors, vol. 19, no. 9, p. 2071, 2019. International Journal of Advance Research, Ideas and Innovations in
[2] S. Nanda, H. Joshi, and S. Khairnar, “An iot based smart system for acci- Technology(IJARIIT), vol. 5, no. 3, pp. 685–687, 2019.

16 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

[8] M. U. Ghazi, M. A. K. Khattak, B. Shabir, A. W. Malik, and M. S.

[28] V. Sharma and R. Tiwari, “A review paper on “iot” & it’s smart appli-
Ramzan, “Emergency message dissemination in vehicular networks: A
cations,” International Journal of Science, Engineering and Technology
review,” in IEEE Access, vol. 8, pp. 38 606–38 621, 2020.
Research (IJSETR), vol. 5, no. 2, pp. 472–476, 2016.
[9] W. Farooq, M. A. Khan, and S. Rehman, “A novel real time framework
[29] M. A. Khan, M. A. Khan, A. U. Rahman, A. W. Malik, and S. A.
for cluster based multicast communication in vehicular ad hoc networks,”
Khan, “Exploiting cooperative sensing for accurate target tracking in
International Journal of Distributed Sensor Networks(IJDSN), vol. 12,
industrial internet of things,” International Journal of Distributed Sensor
no. 4, 2016, Art .no.8064908.
Networks(IJDSN), vol. 15, no. 12, 2019.
[10] Manuja M, Kowshika S, Narmatha S, and Gracy Theresa W, “Iot based
[30] V. N. Malavade and P. K. Akulwar, “Role of iot in agriculture,” IOSR
automatic accident detection and rescue management in vanet,” SSRG
Journal of Computer Engineering (IOSR-JCE), vol. 1, no. 13, pp. 56–57,
International Journal of Computer Science and Engineering( SSRG –
2016.
IJCSE ) ,Special Issue ICFTESH, pp. 36–41, 2019.
[31] V. Subrahmanyam and K. Aruna, “Future automobile an introduction
[11] P. Berade, K. Patil, P. Tawate, and G. M.U., “Intelligent accident identi- of iot,” International Journal of Trend in Research and Development
fication and prevention system using gps and gsm modem,” International (IJTRD),Special Issue | NCETC-17, pp. 88–90, 2017.
Research Journal of Engineering and Technology (IRJET), vol. 5, no. 2, [32] S. Kawthankar and C. Raut, “A survey on smart automobiles using
pp. 1915–1919, 2018.
internet of things for digital india,” International Journal of Recent
[12] A. B. Faiz, A. Imteaj, and M. Chowdhury, “Smart vehicle accident de- Trends in Engineering Research (IJRTER), vol. 3, no. 05, pp. 27–33,
tection and alarming system using a smartphone,” in 2015 International 2017.
Conference on Computer and Information Engineering (ICCIE), Rajshahi, [33] R. K. Bajaj, M. Rao, and H. Agrawal, “Internet of things (iot) in the
2015, pp. 66–69. smart automotive sector: a review,” in IOSR Journal of Computer Engi-
[13] V. Ahmed and N. P. Jawarkar, “Design of low cost versatile microcon- neering (IOSR-JCE), conference on recent trends in computer engineering
troller based system using cell phone for accident detection and pre- (CRTCE-2018), 2018, pp. 36–44.
vention,” in 2013 6th International Conference on Emerging Trends in [34] M. U. Farooq, M. Waseem, S. Mazhar, A. Khairi, and T. Kamal, “A
Engineering and Technology, Nagpur, 2013, pp. 73–77. review on internet of things (iot),” International Journal of Computer
[14] P. C, P.Darshini.C, V.G.Shavi, and S. Begum, “Two-wheeler safety Applications(IJCA), vol. 113, no. 1, pp. 1–7, 2015.
system for accident prevention, detection and reporting,” International [35] O. Helmi, Ehsankanani, M. A. Sokeh, and G. Sepidnam, “The challenges
Journal of Engineering and Computer Science(IJECS), vol. 7, no. 3, facing with the internet of things,” International Journal of Scientific
pp. 23 680– 23 682, 2018. Study (IJSS), vol. 5, no. 4, pp. 527–533, 2017.
[15] M. Ozbayoglu, G. Kucukayan, and E. Dogdu, “A real-time autonomous [36] S. I. Al-Sharekh and K. H. A. Al-Shqeerat, “Security challenges and lim-
highway accident detection model based on big data processing and itations in iot environments,” International Journal of Computer Science
computational intelligence,” in 2016 IEEE International Conference on and Network Security(IJCSNS), vol. 19, no. 2, pp. 193–199, 2019.
Big Data (Big Data), 2016, pp. 1807–1813. [37] S. Chen, H. Xu, D. Liu, B. Hu, and H. Wang, “A vision of iot:
[16] N. Dogru and A. Subasi, “Traffic accident detection using random for- Applications, challenges, and opportunities with china perspective,” IEEE
est classifier,” in 2018 15th learning and technology conference (L&T). Internet of Things journal, vol. 1, no. 4, pp. 349–359, 2014.
IEEE, 2018, pp. 40–45. [38] P. Matta and B. Pant, “Internet-of-things: Genesis, challenges and applica-
[17] P. Amrith, E. Umamaheswari, R. U. Anitha, D. Mani, and D. M. Ajay, tions,” Journal of Engineering Science and Technology (JESTEC), vol.
“Smart detection of vehicle accidents using object identification sensors 14, no. 3, pp. 1717–1750, 2019.
with artificial intelligent systems,” International Journal of Recent Tech- [39] M. A. Razzaq, S. H. Gill, M. A. Qureshi, and S. Ullah, “Security issues in
nology and Engineering (IJRTE), vol. 7, no. 5, pp. 375–379, 2019. the internet of things (iot): a comprehensive study,” International Journal
[18] S. Ghosh, S. J. Sunny, and R. Roney, “Accident detection using convolu- of Advanced Computer Science and Applications (IJACSA), vol. 8, no. 6,
tional neural networks,” in 2019 International Conference on Data Science pp. 383–388, 2017.
and Communication (IconDSC),Bangalore, India, 2019, pp. 1–6. [40] Mandal K and G S Pradeep Ghantasala, “A complete survey on techno-
[19] A. Alkandari, I. F. Al-Shaikhli, A. Najaa, and M. Aljandal, “Accident de- logical challenges of iot in security and privacy,” International Journal of
tection and action system using fuzzy logic theory,” in 2013 International Recent Technology and Engineering (IJRTE), vol. 7, no. 6S4, pp. 332–
Conference on Fuzzy Theory and Its Applications (iFUZZY), 2013, pp. 334, 2019.
385–390. [41] R. Roman, J. Zhou, and J. Lopez, “On the features and challenges of
[20] “ Global status report on road safety 2018. Geneva: World Health Organi- security and privacy in distributed internet of things,” Computer
zation; 2018. Licence: CC BYNC-SA 3.0 IGO.” Available:https://www. Networks, vol. 57, no. 10, pp. 2266–2279, 2013.
who.int/publications-detail/global-status-report-on-road-safety-2018, [42] S. Jabbar, F. Ullah, S. Khalid, M. Khan, and K. Han, “Semantic inter-
Accessed on: May 15,2020 [Online]. operability in heterogeneous iot infrastructure for healthcare,” Wireless
[21] K. K. Patel and S. M. Patel, “Internet of things-iot: definition, characteris- Communications and Mobile Computing, vol. 2017, 2017.
tics, architecture, enabling technologies, application & future challenges,” [43] J. Koo, S.-R. Oh, and Y.-G. Kim, “Device identification interoperability
International Journal of Engineering Science and Computing(IJESC), in heterogeneous iot platforms,” Sensors, vol. 19, no. 6, p. 1433, 2019.
vol. 6, no. 5, pp. 6122–6131, 2016. [44] D. Dragomir, L. Gheorghe, S. Costea, and A. Radovici, “A survey on
[22] T. Mufti, N. Sami, and S. S. Sohail, “A review paper on internet of things secure communication protocols for iot systems,” in 2016 International
(iot),” Indian Journal Of Applied Research(IJAR), vol. 9, no. 8, pp. 24– Workshop on Secure Internet of Things (SIoT), 2016, pp. 47–62.
25, 2019. [45] M. Elkhodr, S. Shahrestani, and H. Cheung, “Emerging wireless technolo-
[23] P. Asghari, A. M. Rahmani, and H. H. S. Javadi, “Internet of things gies in the internet of things: a comparative study,” International Journal
applications: A systematic review,” Computer Networks, vol. 148, pp. of Wireless Mobile Networks (IJWMN), vol. 8, no. 5, pp. 67–82, 2016.
241–261, 2019. [46] O. Khutsoane, B. Isong, and A. M. Abu-Mahfouz, “Iot devices and appli-
[24] S. Lakshmipriya, “Exploring the nuances of internet of things in health cations based on lora/lorawan,” in IECON 2017-43rd Annual Conference
care assisting system,” International Research Journal of Engineering and of the IEEE Industrial Electronics Society, 2017, pp. 6107–6112.
Technology (IRJET), vol. 6, no. 02, pp. 1776–1780, 2019. [47] S. Al-Sarawi, M. Anbar, K. Alieyan, and M. Alzubaidi, “Internet of
things (iot) communication protocols:review,” in 2017 8th International
[25] H. F. Atlam, R. J. Walters, and G. B. Wills, “Internet of things: state-of-
conference on information technology (ICIT), 2017, pp. 685–690.
the-art, challenges, applications, and open issues,” International Journal of
Intelligent Computing Research (IJICR), vol. 9, no. 3, pp. 928–938, 2018. [48] D. M. Ibrahim, “Internet of things technology based on lorawan revolu-
tion,” in 2019 10th International Conference on Information and Commu-
[26] R. Vignesh and A. Samydurai, “Security on internet of things (iot) with
nication Systems (ICICS), 2019, pp. 234–237.
challenges and countermeasures,” International Journal of Engineering
[49] R. S. Sinha, Y. Wei, and S.-H. Hwang, “A survey on lpwa technology:
Development and Research(IJEDR), vol. 5, no. 1, pp. 417 –423, 2017.
Lora and nb-iot,” ICT Express, vol. 3, no. 1, pp. 14–21, 2017.
[27] R. Khan, S. U. Khan, R. Zaheer, and S. Khan, “Future internet: the
[50] M. Luvisotto, F. Tramarin, L. Vangelista, and S. Vitturi, “On the use of
internet of things architecture, possible applications and key challenges,”
lorawan for indoor industrial iot applications,” Wireless Communications
in 2012 10th international conference on frontiers of information
and Mobile Computing, vol. 2018, 2018, Art.no.3982646.
technology, 2012, pp. 257–260.
[51] W. Ayoub, A. E. Samhat, F. Nouvel, M. Mroue, and J.-C. Prévotet,
“Internet of mobile things: Overview of lorawan, dash7, and nb-iot in

VOLUME 4, 2016 17

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

lpwans standards and supported mobility,” IEEE Communications [72] M. Szarvas, A. Yoshizawa, M. Yamamoto, and J. Ogata, “Pedestrian
Surveys & Tutorials, vol. 21, no. 2, pp. 1561–1581, 2019. detection with convolutional neural networks,” in IEEE Proc. Intelligent
[52] A. Adhikary, X. Lin, and Y.-P. E. Wang, “Performance evaluation of nb- Vehicles Symposium, 2005,Las Vegas, NV, USA, 2005, pp. 224–229.
iot coverage,” in 2016 IEEE 84th Vehicular Technology Conference [73] M. Murshed and M. S. Chowdhury, “An iot based car accident pre-
(VTC- Fall), 2016, pp. 1–5. vention and detection system with smart brake control,” in Interna-
[53] S. Madakam, R. Ramaswamy, and S. Tripathi, “Internet of things (iot): A tional Conference on Applications and Techniques in Information Science
literature review,” Journal of Computer and Communications, vol. 3, pp. (iCATIS2019), 2019.
164–173, 2015.
[54] B. Shabir, M. A. Khan, A. U. Rahman, A. W. Malik, and A. Wahid,
“Congestion avoidance in vehicular networks: A contemporary survey,”
in IEEE Access, vol. 7, pp. 173 196–173 215, 2019.
[55] W. Farooq, M. A. Khan, S. Rehman, and N. A. Saqib, “A survey of
multicast routing protocols for vehicular ad hoc networks,” International
Journal of Distributed Sensor Networks (IJDSN), vol. 11, no. 8, 2015.
[56] D. B. Tushara and P. A. H. Vardhini, “Wireless vehicle alert and collision
UNAIZA ALVI received her bachelor’s degree in
prevention system design using atmel microcontroller,” in 2016 Interna-
Information Technology from University of the
tional Conference on Electrical, Electronics, and Optimization Techniques
(ICEEOT), Chennai, 2016, pp. 2784–2787. Punjab, Lahore in 2019. She is currently pursuing
the master’s degree in Information Technology
[57] U. Patil, P. More, R. Pandey, and U. Patkar, “Tracking and recovery of
the vehicle using gps and gsm,” International Research Journal of from NUST, Islamabad. Her research interests in-
Engineering and Technology (IRJET), vol. 4, no. 3, pp. 2074–2077, 2017. clude machine learning, Internet of Things,
[58] B. S. Anil, K. A. Vilas, and S. R. Jagtap, “Intelligent system for vehicular vehic- ular ad-hoc networks and artificial
accident detection and notification,” in 2014 International Conference on intelligence.
Communication and Signal Processing, Melmaruvathur, 2014, pp. 1238–
1240.
[59] R. K. Megalingam, R. N. Nair, and S. M. Prakhya, “Wireless vehicular
accident detection and reporting system,” in 2010 International
Conference on Mechanical and Electrical Technology(ICMET),
Singapore, 2010, pp. 636–640.
[60] H. Kumar N and G. Deepak, “Accident detection and intelligent
navigation system for emergency vehicles in urban areas using iot,”
International Journal of Engineering and Techniques(IJET), vol. 3, no. 6,
pp. 330–334, 2017.
DR MUAZZAM A KHAN KHATTAK (Senior
[61] E. Nasr, E. Kfoury, and D. Khoury, “An iot approach to vehicle accident
detection, reporting, and navigation,” in 2016 IEEE International Member IEEE) is working as Tenured Asso-
Multidis- ciplinary Conference on Engineering Technology (IMCET), ciate Professor, Department of Computer
Beirut, 2016, pp. 231–236. Science, Quaid-i-Azam University, Islamabad
[62] Z. Wen and J. Meng, “Design of vehicle positioning system based on Pakistan. He received his Ph.D. degree in
arm,” in 2011 International Conference on Business Management and Computer sciences as a sandwich program from
Electronic Information, Guangzhou, 2011, pp. 395–397. IIUI and UMKC, USA in 2011. He received his
[63] M. Gaber, A. M. Wahaballa, A. M. Othman, and A. Diab, “Traffic acci- Postdoc from University of Missouri, KC, USA
dents prediction model using fuzzy logic: Aswan desert road case study,” in 2016. He joined School of Electrical
J. Eng. Sci. Assiut Univ, vol. 45, no. 1, pp. 28–44, 2017. Engineering and Computer Science (SEECS),
[64] X. Li, D. Lord, Y. Zhang, and Y. Xie, “Predicting motor vehicle crashes NUST, Islamabad as an Assistant Pro-
using support vector machine models,” Accident Analysis & Prevention, fessor in 2013 and later on promoted to Associate Dean and Tenured
vol. 40, no. 4, pp. 1611–1618, 2008. Associate Professor in 2017. He has been at the School of Computer
[65] I. Ohe, H. Kawashima, M. Kojima, and Y. Kaneko, “A method for auto- Science, University of Ulm, Germany, and School of Computer and the
matic detection of traffic incidents using neural networks,” in Pacific Rim Electrical Engineering, University of Missouri, USA as a research fellow.
TransTech Conference. 1995 Vehicle Navigation and Information He worked at the Networking and Multimedia Lab, University of Missouri
Systems Conference Proceedings. 6th International VNIS. A Ride into the Kansas City, the USA as a Research fellow. His research interests include
Future, 1995, pp. 231–235. Internet of Things, Next Generation Intelligent Networks, Block Chain
[66] B. Pan and H. Wu, “Urban traffic incident detection with mobile sensors Based Information and Network Security, Vehicular Ad-hoc/Sensor Actor
based on svm,” in 2017 XXXIInd General Assembly and Scientific Sym- Networks, Acoustic/Multimedia Networks.
posium of the International Union of Radio Science (URSI GASS), 2017,
pp. 1–4.
[67] Z. Halim, R. Kalsoom, S. Bashir, and G. Abbas, “Artificial intelligence
techniques for driving safety and vehicle crash prediction,” Artificial
Intelligence Review, vol. 46, no. 3, pp. 351–387, 2016.
[68] S. Eduku, M. O. Alhassan, and J. Sekyi– Ansah, “Design of vehicle acci-
dent prevention system using wireless technology,” International Journal
of Scientific and Reasearch Publications (IJSRP), vol. 7, no. 10, pp. 397–
403, 2016. BALAWAL SHABIR received a bachelor’s
[69] S. Kasera, “Iot based accident prevention system,” International Journal degree in Electrical Engineering from the Air
for Scientific Research Development (IJSRD), vol. 7, no. 5, pp. 832–835, University Islamabad. Later on, he completed his
2019. master’s in Information security from Air
[70] O. G. Basubeit, D. N. T. How, Y. C. Hou, and K. S. M. Sahari, University Islam- abad as well. Currently, he is
“Distracted driver detection with deep convolutional neural network,” pursuing his Ph.D. at NUST, Pakistan. His
International Journal of Recent Technology and Engineering (IJRTE), vol. research interests include cryptography and
8, no. 4, pp. 6159–6163, 2019. network security, privacy, ma- chine learning,
[71] K. Dwivedi, K. Biswaranjan, and A. Sethi, “Drowsy driver detection vehicular ad-hoc networks, internet of things, big
using representation learning,” in 2014 IEEE international advance data analytics, Blockchain, bioinfor- matics,
computing conference (IACC), 2014, pp. 995–999. cloud and fog computing.

18 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3006887, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

ASAD W. MALIK (Senior Member IEEE) is an

Assistant Professor at NUST-SEECS, Pakistan.
Besides, he is also working as Senior Lecturer at
the Department of Information Systems, Faculty
of Computer Science & Information Technology,
University of Malaya, Malaysia. He finished his
Ph.D. with majors in parallel and distributed sim-
ulation/systems from NUST, Pakistan in 2012.
His primary area of interest includes distributed
simulation, cloud/fog computing, and internet of
things.

MUHAMMAD SHER RAMZAN is the professor

serving in Faculty of Computing and Information
Technology (FCIT), King Abdulaziz University
Jeddah, Saudi Arabia. In 1998, he joined Interna-
tional Islamic University, Islamabad (IIUI) as an
Assistant Professor. Dr. Sher received his Ph.D.
degree from TU Berlin, Germany in 2007. From
2008 to 2012 he served as a Professor, Chairman
and later on as a Dean FBAS from 2017-18. He
has authored more than 100 publications. His areas
of research are Information Security, IP Multimedia Subsystem, WSN,
Healthcare, IoT and NGN.

VOLUME 4, 2016 19

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see