System Design Specification

Of
Eco-Friendly Smart Traffic Controlling
System
RUNIX

Name Registration Number Index Number

D.H.A.S. Kumarasinghe ASP/17/18/092 4109

D.T. Karunarathne ASB/17/18/121 4050

S.H.U.D. Siriwardana ASP/17/18/103 4099

A.D.R.P. Wijesinghe ASP/17/18/113 4107

R.M.V. Deshan ASB/17/18/016 4033

2022/10/25
 Abstract

The present era is marked by rapid development and advancement in technology. The
traffic signal, which is the core of the traffic system, is one of the most essential areas that
requires improvement. This demand is intensifying with the development of smart cities
in Sri Lanka. Unfortunately, traffic is currently controlled by very old traffic signals in
many cities. These traditional traffic signals have many problems, including ineffective
time management at intersections; they are not immune to some environmental
conditions, such as rain; and they have no way of prioritizing emergency vehicles. New
technologies such as Vehicular Ad-hoc Networks (VANET) and Internet of Vehicles
(IoV) enable vehicles to communicate wirelessly with their surroundings and with
specialized infrastructure. In this paper, we propose a new traffic management system
based on existing VANET and IoV, suitable for future traffic systems and smart cities in
Sri Lanka. In this paper, we present the architecture of our proposed Intelligent Traffic
Management System (ITMS) and Smart Traffic Signal (STS) controller. Demands of
future smart cities for equity, reduction of commuting time, provision of equitable traffic
flow, reduction of traffic congestion, reduction of air pollution caused by vehicular
carbon dioxide, reduction of foreign exchange of wasted fuel imports for vehicles, and
prioritization of emergency vehicles. We present local traffic management at an
intersection based on the simulation results, which show that the proposed system
outperforms the conventional management system and can be a candidate for the traffic
management system of future Sri Lankan smart cities. Our proposed adaptive algorithm
not only significantly reduces the average waiting time (delay) but also increases the
number of vehicles served. In addition, we present the implemented hardware prototype
for STS.

ii
 Acknowledgement

This is a project being carried out as our third-year final project in the Faculty of Applied
Sciences, Rajarata University of Sri Lanka. The project of an eco-friendly traffic control
unit obviously would not have been developed even this much without the support of
some individuals and organizations. As group members, we would like to express our
gratitude to the supervisors of our group, Mrs. I.J. Hewarathne and M. Mithsara, who are
lecturers in the Faculty of Applied Sciences, for the guidance and the comments they
gave on the project work. Mrs. I.J. Hewarathne deserves special recognition for coming
up with the project idea. We appreciate all the lecturers and the supportive members of
academic and nonacademic staff in the Department of Computing of the Faculty of
Applied Sciences, Rajarata University of Sri Lanka.

Also, an appreciation for all the team members for their understanding, collaboration, and
immense effort put forth in achieving each and every objective of the project.

ii
i
 Table of Contents

Abstract...........................................................................................................ii
Acknowledgement.........................................................................................iii
List of figures..................................................................................................v
List of Tables...................................................................................................v
CHAPTER 1 – INTRODUCTION..............................................................1
1.1 Problem to be addressed............................................................................1
1.2 Objectives of the Project...........................................................................2
1.3 Project Deliverables..................................................................................2
1.4 System design approach (Process models, Patterns ex. MVC, SCRUM).3
1.5 Standards to be followed...........................................................................4
1.5.1. Standards for SDS Document Preparation.........................................4
1.5.2. Standards for Software Coding...........................................................4
1.6 Organization of the SDS...........................................................................4
CHAPTER 2 – ARCHITECTURAL DESIGN..........................................6
2.1 System architecture...................................................................................6
2.2 Objects and communication......................................................................7
2.3 Processes and special algorithms..............................................................9
2.4 Tools, techniques, libraries, 3rd party tools and implementation
environment………………………………………………...……………...13
CHAPTER 3 – HARDWARE DESIGN....................................................14
3.1 Hardware components and detailed design…….
……………………….14
3.1.1 Relevant Functions of Components………………………….….
….15
3.2 State machines……………………………………………...……….….15
3.3 Hardware component communication and interfaces….
……………….16
CHAPTER 4 – RESEARCH DESIGN......................................................17
4.1 Study objectives-based literature review.................................................17
4.1.1 Background.......................................................................................17
4.1.2 Real-time traffic controlling systems................................................17
4.2 Formalizing high-level implementation components..............................19
4.3 Data extractions, sample design, test data sets, training data sets...........19
4.4 Non-functional aspects............................................................................19
4.4.1. External requirements......................................................................20
4.5 Proposed validation methods and measurements....................................20
APPROVAL.................................................................................................21

iv
REFERENCES............................................................................................22

List of figures
Figure 1.1: Agile method……………………………………………………….…………3

Figure 2.1: Component diagram………………………………………………....…..........6

Figure 2.2: Sequence diagram of normal vehicles …………….……………….…………7

Figure 2.3: sequence diagram of emergency vehicles……………………….……………8
Figure 2.4: Flowchart of traffic controlling system……………………………….………9
Figure 2.5: Algorithm of vehicle controlling unit…………………………………….….10
Figure 2.6: Algorithm of traffic lights……………………………………………………11
Figure 2.7: Class diagram of traffic controlling unit………………………………….….12
Figure 3.1: Draft schematic of hardware design……………………………14

Figure 3.2: State machine diagram of normal vehicle process……………………..

……..15
Figure 3.3: State machine diagram of emergency vehicle process………………….……
16

List of Tables

Table 1: Deliverables of the Project- Documentation.................................................... 2

Table 2: Deliverables of the Project- Implementation ................................................... 2

v
 CHAPTER 1 – INTRODUCTION

1.1 Problem to be addressed

What encouraged us to do this research was the environmental damage, fuel wastage and
time wasted in traffic jams in Sri Lanka. In most cities in Sri Lanka, signal allocation is
primarily based on time. The timer system has a disadvantage. Even when a road has very
little traffic, the green sign is allocated to the road until the cost of its timer drops to zero,
while the traffic on any other road faces red. Signals at that location cause congestion and
a loss of time for commuters. We identified the problem that the traffic controller uses for
image processing does not work properly during the night as well as in the face of bad
weather. The existing automatic traffic controllers in Sri Lanka are time limited and are
not based on real-time data. All road lighting along the route is illuminated at some point
regardless of the actual need. This not only wastes electricity but also disturbs the natural
habitats of the surrounding animals. However, traffic congestion still happens, especially
in the bigger cities during peak hours, due to the problems that are usually faced by the
current traffic light systems:

 Loss of man-hours by preventing the time wasted by waiting in traffic jams

Immediately, after office hours, an increasing number of vehicles on the road causes
heavy traffic. This causes a long waiting time. The timer is not intelligent enough to sense
the presence of vehicles, and thus the sequence of traffic lights remains the same
following the preset time on the timer.

 Foreign exchange is being squandered on crude oil imports

Fuel wastage is also responsible for the economic crisis in which Sri Lanka is at present.
We see a huge problem with the wastage of foreign exchange caused by the unrestricted
import of fuel.

 Air pollution by vehicle emissions

The environmental damage caused by vehicular traffic is incalculable. It affects

everything from small animals to humans. Although air pollution is shown to be a minor
problem, it is a big problem in taking the world towards a green city concept.

 An emergency car stuck in the traffic jam

During peak hours, emergency vehicles will also be stuck in traffic flow due to the road
user's need from the above scenario. An alternative solution is needed to solve or reduce
the above problems. The proposed system should be relatively cheap to maintain and yet
reliable to handle traffic flows.

1
1.2 Objectives of the Project

The aim of this project is to build an eco-friendly traffic control system. The main
objectives of the project are as follows:

1) Gathering traffic data at the project's chosen intersection using a demonstration sensor
2) Identify the emergency vehicle using a portable controller.
3) Create a prototype traffic control unit.
4) Create the AI model

1.3 Project Deliverables

Project deliverables always represent the milestones of the entire project that the
developers work on. They are the tangible byproducts of the primary project. The project
deliverables are not the same as the project objectives, but they are closely related to each
other.

Table 1: Deliverables of the Project- Documentation

Deliverable No Deliverable Name
D1 Team Registration
D2 Project Proposal Submission and Presentation
D3 Revised Project Proposal Submission
System Requirement Specification (SRS) Document Submission and
D4
Presentation
D5 Interim Report Submission
System Design Specification (SDS) Document Submission and
D6
Presentation.
D7 Progress Report Booklet 01 Submission
D8 Progress Report Booklet 02 Submission
D9 Final Dissertation Submission

Table 2: Deliverables of the Project- Implementation

Deliverable No Deliverable Name
D1 Machine learning model trained with dataset
D2 Programmed Arduino UNO
D3 Programmed Arduino UNO with emergency unit
D6 Finalized system

2
1.4 System design approach (Process models, Patterns ex. MVC, SCRUM)

Nowadays, it seems like every technological company uses the agile process for software
development. During the project's entire software development life cycle, the agile
technique incorporates continuous testing and development iterations. The proposed
system follows the agile methodology. There are four main guidelines to follow in order
to create better software utilizing the agile development method. The vehicle discovery
framework is communicated with the Adriano mega microcontroller board and situated in
every one of the headings of an intersection. The whole framework is planned and
recreated utilizing the Proteus workbench. The paper closes with certain features of future
work and other helpful comments.
The Agile model is a combination of iterative and incremental process models. Steps
involved in agile SDLC models are:
 Requirement gathering
 Requirement Analysis
 Design
 Coding
 Unit testing
 Acceptance testing

The reason for using the agile method in our project is the flexibility and adaptability of
it. Because of the flexibility and adaptability, we can change the requirements within
development. Further, continuous iterations in the agile method help to carry on the
implementation in our system without continuing the previous errors. We can explore
creative ways of problem solving without strict guidelines. Because of using the agile
method, we can re-evaluate the project priorities at the end of each iteration. Figure 1.1
shows the agile method used in the project as a system design approach.

Figure 1.1: Agile method

3
1.5 Standards to be followed

This section contains the standards followed by the team during the implementation and
documentation processes within the project.
1.5.1. Standards for SDS Document Preparation

Department of computing provides standards for documentation and for the citations. During the
preparation of the documents regarding the project the standards provide by the department is
referred.

1.5.2. Standards for Software Coding

The coding standards depend on the product intended to be developed. The proposed
system includes AI and machine learning modules, which will be developed using the
Python programming language. Python is one of the widely used languages for
developing machine learning and deep learning models, and it is easy to get started with.
Python is also rich with frameworks and libraries that can be used to make machine
learning tasks much easier. The reason for not choosing a language like Java is that Java
is a more disciplined language, so it’s not as flexible and friendly as Python. In particular,
Python's syntax is very concise and easily readable.

1.6 Organization of the SDS

The organization of the SDS document may flow as follows:

1. Introduction
In this introductory chapter, an introduction to the project, the problem to be addressed,
objectives of the project, both the documentational and implementational deliverables,
system design approach, and the standards to be followed are discussed.
2. Architectural Design
This section will cover the system's architecture, objects and communications that will be
used, processes and special algorithms, state machines and tools, methodologies, libraries,
third-party tools, and the implementation environment. Operational, performance,
security, and cultural needs are the main nonfunctional criteria that influence architectural
design. So, the results from the architectural design contain the system architecture,
hardware, and software specifications.
3. Hardware Design
The Hardware Design Chapter gives a detailed description of the hardware design of the
system with diagrams. Furthermore, here are described the hardware components about to
be used, state machines and hardware component communication and interfaces of the
proposed system.

4
4. Research Design
The study of objectively based literature reviews pertaining to the system, the
formalization of high-level implementation components, data extraction, sample design,
test and training data sets, and non-functional features is covered in the research design
chapter. Furthermore, it suggests system validation techniques and measures.
6. Approval for the document
Finally, the approval for the document will be taken from the supervisors of the project.

5
 CHAPTER 2 – ARCHITECTURAL DESIGN

This chapter discusses the design of an environmentally friendly traffic control system.
Architectural design includes system architecture, objects and communications, processes
and special algorithms, state machines and tools used, techniques, libraries, third-party
tools, and the system's execution environment. Discuss each subsection of this chapter
separately.

2.1 System architecture

System architecture is the conceptual mapping of the formal description of structure,

behavior, and other ideas. Here, the number of vehicles obtained by the range between the
two IR sensors is analyzed and given to the micro control unit. After the microcontroller
sends it to the AI model, the signal light is controlled by it. Here, in the emergency
section (red), a portable device is connected to the receiver and given to the
microcontroller. It provides the necessary access for vehicles in case of an emergency.

Figure 2.1: Component diagram

6
2.2 Objects and communication

For our purposes, there are two main systems in which the system works. One is for
normal vehicles and the other is for emergency vehicles. Figure 2.3 shows that first the
vehicle is captured by the sensor. The vehicle count is then transmitted to the
microcontroller. Then the microcontroller receives the data from all the sensors. After
that, the data is sent to the AI model. The AI model analyzes the traffic flow and takes the
decision. That decision data is sent to the signal light. Based on that data, each signal light
is timed. Then time is given for the traffic light to turn green. Then the traffic light gives
signals to the vehicles.

Figure 2.2: Sequence diagram of normal vehicles

7
In Figure 2.4, which shows the emergency vehicle identification system, the vehicle
driver first obtains the relevant council number. If the number is valid, it will be retrieved
and saved by the portable device. Then the data is sent to the microcontroller, and it
detects the lane number. After sending data to the AI model, it takes the immediate
decision and sends it to the signal lights. Then it changes the color of the corresponding
lane of the traffic light to green. Then the traffic light gives the green light to the driver.

Figure 2.3: sequence diagram of emergency vehicles

8
2.3 Processes and special algorithms

Processes and special algorithms include the algorithms used for the system device. There
are several algorithms and processes involved in our traffic control system. In this section
we have the main processes that are important to our device. Figure 2.5 shows the traffic
control system flowchart can be used to get an idea about the flow of the system.

Figure 2.4: Flowchart of traffic controlling system

9
Here, the vehicle count is obtained by the IR sensor and given to the artificial intelligence
model through the microcontroller. It then identifies areas with high traffic density. Then
time is given for each signal, and signals are given to the respective councilors. In the
event of an emergency, by providing the council number, it will be analyzed by the model
and a green color will be given to the relevant council.

Figure 2.5: Algorithm of vehicle controlling unit

10
An algorithm (Figure 2.8) has also been developed for traffic control light boards. It
colors the bulbs in the traffic lanes with the highest traffic value green and the other low
values red.

Figure 2.6: Algorithm of traffic lights

11
A class diagram depicts the structure of a system by showing the system's classes, their
attributes, operations, and relationships among objects. The following figure 2.8 depicts
the class diagram of the system.

Figure 2.7: Class diagram of traffic controlling unit

12
2.4 Tools, techniques, libraries, 3rd party tools and implementation environment

 Case Tool: Draw.io

Draw.io is a website for drawing software models and flow charts. It supports various
types of UML diagrams like activity diagrams, class diagrams, use case diagrams, and
flowcharts. Draw.io is easy to use because its functionality is simple, and also, we can
work offline.

 Python IDLE
Python has a built-in development environment called IDLE. IDLE is pre-installed in the
Python installation for Windows. Since IDLE is a straightforward IDE, beginners will
find it incredibly user-friendly. With the aid of the Tkinter GUI toolkit and only being
built in Python, IDLE is cross-platform and offers developers freedom. Also, this IDE has
some good features like a multi-window text editor which has features like call tips, smart
indentation, undo, and python colorizing and auto code completion. IDLE has a powerful
debugger with call stack visibility. Developers can search within any window, search
through multiple files, and replace within the Windows editor. Because of these benefits,
we are using Python-Idle to develop our system.

 Arduino IDE
An easy-to-use, cross-platform tool for programming Arduino is called the Arduino IDE.
Users may select whether they want to store their work on a local computer or on a cloud
platform because it offers both an online editor and an on-premise application. Users
using this IDE get access to community-contributed libraries and get the most recent
support for Arduino boards. Many features in the IDE will aid developers in producing
apps quickly and easily, including board administration, straightforward sketching,
project documentation, straightforward sketch sharing, a sizable library, and support for
third-party hardware. Within our system, frequency generation is required. Using an
Arduino IDE and an Arduino kit, we can archive it.

13
 CHAPTER 3 – Hardware Design

This chapter consists of the hardware design of the project. It includes hardware
components and detailed design, hardware component communication and interfaces.

3.1 Hardware components and detailed design

The following figure depicts the detailed design of the system.

Figure 3.1: Draft schematic of hardware design

The hardware components required for the proposed system are as follows:
 IR sensors
 ZigBee Receiver
 Arduino mega 2560 micro controller
 Arduino UNO micro controller
 Laptop or Desktop
 Breadboards
 LED lights
 Connecting wires
 Keypad
 Display

14
3.1.1 Relevant Functions of Components
 IR sensor
This is used to identify the vehicle count.

 Laptop or desktop
Desktop pc is used to run the application. And it is connected to Arduino uno using USB
cable.
 Arduino UNO micro controller
The Arduino Uno is a microcontroller. It has 14 digital input/output pins, 6 of which can
be used as PWM outputs), 6 analog inputs, a 16 MHz ceramic resonator
(CSTCE16M0V53-R0), a USB connection, a power jack, an ICSP header, and a reset
button. This is used to control the data taken by the IR sensor and the lighting of the LED
bulbs. It can be connected to the computer using a USB cable.
 ZigBee receiver
This receiver is used to establish a connection between the portable control and the
microcontroller.

3.2 State machines

A state diagram, also known as a state machine diagram or state chart diagram, is an
illustration of the states an object can attain as well as the transitions between those states
in the Unified Modeling Language (UML). Our system's state machine diagrams of
normal vehicles and emergency vehicles are shown in Figure 3.1 and Figure 3.2.

Figure 3.2: State machine diagram of normal vehicle process

15
 Figure 3.3: State machine diagram of emergency vehicle process

3.3 Hardware component communication and interfaces

Hardware component communication is an architecture used to interconnect two devices

together. It includes the design of the plug and socket, the type, number, and purpose of
the wires, and the electrical signals that are passed across them. Two types of interfaces
are present in the system in order to communicate between devices in the system. The
USB interface and wireless communication interface are those interfaces.
 USB Interface
USB is a common interface used to communicate between devices often. It can give a 5V
output to the device. In the proposed system, USB is used to connect a desktop PC with
an Arduino Uno, and it is also used as the power supply for the Arduino.
 Wireless Communication
Wireless communication is a popular way to communicate between devices today. It is
also easy to communicate between devices wirelessly. In the system, we install a wireless
receiver in the emergency traffic detection module for communication between the
microcontroller and the portable controller.

CHAPTER 4 – RESEARCH DESIGN

This chapter consists of the review of the literature surveys conducted related to the
project idea. A literature survey is basically a summary of research papers that were

16
found, including the technologies, hardware components, and methodologies used by
other researchers in projects similar to the proposed system.

4.1 Study objectives-based literature review

4.1.1 Background

The background to this project was the concept of eco-friendly smart cities. We are of the
view that such a project could provide a more sustainable solution to the current
problems. We also based the motivation for this project on the fact that fuel scarcity is
currently prevailing in Sri Lanka.

4.1.2 Real-time traffic controlling systems

K. Nellore et al. [2] classify the road traffic into recurring and non-recurring congestion.
Recurring congestion can be addressed easily as long as the traffic pattern maintains
regularity with respect to time and place. However, the non-recurring congestion is
difficult to identify as the traffic pattern is not fixed and is unpredictable. Non-recurring
congestion occurs due to unexpected situations such as climatic variation, festivals, and
road accidents. To control traffic, traffic density must be detected. Authors N K Jain et al.
[3] describe various traffic monitoring systems. One such technique is Situ traffic detector
technology. Based on the installation of a detector at or below the road surface, this
technique can be further divided into intrusive and non-intrusive technologies. Intrusive
techniques cause traffic disruption, whereas non-intrusive techniques cause little to no
disruption. In order to monitor the traffic, the author suggests a sensor technique, in
which various sensors are attached to the moving vehicles. These moving devices
(vehicles) are made to interact with a central monitoring traffic system. Based on the
sensor data, the central monitoring system takes control over traffic movement. In another
technique, a live video stream is fed from cameras at a junction to collect traffic data.
Using video and image processing, traffic density can be calculated.

Author S. Parekh et al. [4] propose a solution along similar lines for detecting the traffic
density using multiple IR sensors by categorizing the vehicle density on the lanes into
three different levels. By dividing the lanes into blocks and placing the sensor grids under
the roads, we pave the way for erroneous results to creep in. An article in BBC News [5]
accounted for badly maintained roads as one of the reasons for the high number of
accidents. With the roads in that state, relying on sensor grids under the roads would not
only expose the grids to possible damage but also lead to erroneous results when the
standards are not followed. Any modifications to the grids would mean a high
maintenance cost given the fact that the work on and due to the top layer would also
account for the cost.

R. Sundar [6] also proposes a technique using Radio-Frequency Identification (RFID)

sensors. It equips every vehicle with an RFID tag. When in the RFID reader’s range, the
signal indicating its presence is sent. This enables the system to track the number of

17
vehicles and hence the congestion volume. Based on the calculated values, it sets the
lights accordingly. The main highlight of the solution is its ability to provide vehicle
clearance during emergencies. This is achieved by a ZigBee transmitter and receiver at
the vehicle and traffic junction, respectively. To add to this, it is also designed to track
down stolen vehicles by matching the missing vehicles’ RFID tags with the RFID tags at
the signal and, if found, alert the respective authorities. The major challenge in this
solution is to issue a unique RFID tag for every vehicle on the road, including the existing
ones, and also ensure a non-tamperable RFID tag.

S. Misbahuddin et al. [1] propose IoT-based traffic management solutions for smart cities.
The proposed technique controls traffic dynamically through smart phones operated by
onsite traffic officers. To interface with an external circuit, a single-board computer with
networking capabilities, such as the Raspberry Pi, is used. The technique can be further
improved if traffic data is automatically passed to the Raspberry Pi unit, controlling lights
at an intersection so that authorities can make a quick decision. Using a mobile agent
under a vehicular ad hoc network (VANET) to automatically control traffic is proposed
by M. Rath et al. [7]. The function of the mobile agent-based improved traffic control
system (MITS) is triggered when any smart vehicle enters the VANET zone. MITS
counts the number of vehicles and if it exceeds the threshold value, sensors trigger an
alarm. This prevents further entry of vehicles into the network. In order to prevent
congestion, vehicles are also rerouted depending on their choice. An increase in vehicle
ownership increases traffic management problems. The traditional methods of traffic
management include monitoring of vehicle speed through cameras, pollution checks, etc.
The increase in the number of vehicles causes inconvenience to adopting the above
methods.

Author et al. [8] suggests IoV based traffic management system in order to control traffic,
accident detection, theft avoidance etc. However, the author also mentioned the
drawbacks of IoV in traffic management with respect to security and network failure.
Sensors are placed in fixed positions in order to collect the data generated. This data is
fed to a central server in order to control and communicate with different devices. A large
number of vehicles in one area causes jams. According to the authors Kumar et al. [9],
this jam can be reduced if vehicles are guided with rerouting. The proposed method is
demonstrated by dividing street maps into small, distinct maps. In order to find the
optimal path, an ant colony algorithm is then applied. A Fuzzy logic-based traffic
intensity calculation function is proposed to model heavy traffic.

On similar lines, deep Q learning was also deployed in a solution given by X. Liang [10].
It aims at optimizing the traffic light duration by taking real-time traffic information and
dynamically adjusting the duration accordingly. To overcome the large number of states
in the traffic control system in vehicular networks, the solution opts for a convolutional
neural network (CNN) to approximate the Q value. The actions are modeled as a Markov
decision process, and the rewards are the cumulative waiting time difference between two
cycles. The authors also recommend a double-dueling deep Q network for handling
complex traffic scenarios. They also claim a 20% reduction in the average waiting time.

18
A major challenge that the model could face would be erratic traffic in real-time for the
model to respond to the conditions within the stipulated time.

From the previous literature reviewed, an infrared (IR) sensor or many ultrasound sensors
have been used for the traffic control system where existing systems have a deficiency or
challenge. The challenge of traffic detection using many ultrasonic sensors reports the
traffic condition to the microcontroller. Another challenge of using ultrasound sensors on
each road to reduce traffic congestion is its cost Those used in the real world should be
more reliable and cover the desired distance. Instead, using the IR sensor and receiving
data through the receiver seems to be more effective. Compared to other research
projects, our research has given high priority to environmental friendliness. Also, the
research gap between our project and the previous projects can be clearly identified by
creating a system that can be implemented in a very efficient and easy way and by
creating an artificial model in a way that minimizes the traffic jams.

4.2 Formalizing high-level implementation components

According to the research phase of the proposed system, the data obtained by the IR
sensor should be extracted using the microcontroller. The data is then sent through a
special artificial intelligence algorithm. For this approach, we need to use a deep learning
approach. Deep learning is a subset of machine learning methods based on artificial
neural networks.
 AI Model
This model is responsible for analyzing the traffic data received through the
microcontroller and making the right decisions, and it is run by a computer. The real-time
vehicle traffic detection model is based on reinforcement learning (RL) called MARLIN
[11] and a system and method to determine and generate optimal traffic signal actions
based on the state and actions of connected traffic signals using RL. The model detects
traffic vehicles using IR sensors. The data set must be clean and clear for accurate
identification. Test cases should be generated to verify the quality and validity of the
model.

4.3 Data extractions, sample design, test data sets, training data sets

Research datasets begin near traffic intersections previously conducted by team members.
Our project data is stored in a folder to perform. We hope to train the recognition models
on traffic data in just two days. We planned to use this data as a training dataset as well as
a validation dataset. We expect to use 80% of the data for the training process and 20%
for the validation process, as that is the method used in most cases.

4.4 Non-functional aspects

Non-functional aspects essentially explain how the system will operate and establish the
standards that may be used to assess the system's performance in order to meet customer
demands. The non-functional needs found in this project are listed below.

19
• Usability requirements
We employ the technology specified in this article. Our system is completely automated
and requires no specialized computer skills to operate. Basic knowledge is all that is
required. The system will produce an alert if there is any abnormality.

• Efficiency requirements

This model should only be accurate for vehicles and should not detect other objects. To
achieve this requirement, we test the system using a portion of the data set in the test
phase of the rapid method. According to our studies, we have studied [12] the use of a
similar system for traffic detection made using the same board computer, the Raspberry
Pi. The problem was that the processing speed and accuracy rate of those systems were
very low in detecting objects. The accuracy rate was 92.61%, but this was for larger
objects such as people in the 1 to 3-meter range. So, this is not good for applications that
need more speed. In our situation, we need to detect vehicles at a speed of at least 1 to 2
seconds. So, we decided to go for a desktop PC, which is more powerful than the
Raspberry Pi.

• Reliability requirements
The system is proposed to be implemented in order to give the responses immediately to
the end user. All modules perform tasks with approximately 90% accuracy and without
any interruption.
4.4.1. External requirements

 Privacy requirements
The system does not meet the users in any way, and as a special safety measure for the
safety of the sensors and the main system, they are installed in the street lights and in a
special booth at the selected junction.
 Ethical requirement
The lights in the system should not harm the eyes of motorists. Also, during the
installation of lights in the traffic junction, they should be installed so as not to disturb the
traffic. Also, the sensor system should be installed so as not to harm animals and people.
The system should not be confusing to humans. To achieve the goal, the system should
emit light frequencies that match the human eye. Also, this project should take place with
the full permission of the Environment Authority, the relevant police and the Traffic
Control Authority.
4.5 Proposed validation methods and measurements

Validation is the verification and provision of objective evidence that requirements have
been met for a specific intended use. According to our system, smooth traffic handling is
the ultimate goal. Since it is difficult to test this method in a real traffic intersection, it is
ideal to validate our method by artificially moving vehicles using a methodical approach.
Our AI model detects vehicle traffic with 92% accuracy. Our system can calculate the

20
traffic flow with 99% accuracy, and it is detected by an AI model. We can confirm it by
manual observation. The final data generated by the system can be displayed by LED
bulbs.
APPROVAL

Signature of the team member:

Name Registration Number Index Number Signature

D.H.A.S. Kumarasinghe ASP/17/18/092 4109

D.T. Karunarathne ASB/17/18/121 4050

S.H.U.D. Siriwardana ASP/17/18/103 4099

A.D.R.P. Wijesinghe ASP/17/18/113 4107

R.M.V. Deshan ASB/17/18/016 4033

Date: 25/10/2022

Approval of the supervisor(s):

I agree / disagree with the scope stipulated in this System Design Specification

1. Name: Mrs. I.J. Hewarathne

Department: Department of Computing
Signature:

2. Name: Miss. M. Mithsara

Department: Department of Computing
Signature:

21
 REFERENCES

[1] S. Misbahuddin, J. A. Zubairi, "IoT based dynamic road traffic management for
smart cities," in 2015 12th International Conference on High-Capacity Optical
Networks and Enabling/Emerging Technologies, HONET-ICT 2015, 2016.
[2] K. Nellore and G. P. Han cke, A survey on urban traffic management system
using wireless sensor networks, vol. 16, MDPI AG, 2016.
[3] N. K. Jain, R. K. Saini and P. Mittal, "A Review on Traffic Monitoring System
Techniques," in Soft Computing: Theories and Applications, Springer, 2019, pp.
569-577.
[4] S. Parekh, N. Dhami, S. Patel and J. Undavia, "Traffic Signal Automation
Through IoT by Sensing and Detecting Traffic Intensity Through IR Sensors," in
Information and Communication Technology for Intelligent Systems, Springer,
2019, pp. 53-65.
[5] BBC, "BBC News," BBC, 10 June 2016. [Online]. Available:
https://www.bbc.com/news/world-asia-india-36496375. [Accessed 5 June 2019].
[6] R. Sundar, S. Hebbar and V. Golla, "Implementing intelligent traffic control
system for congestion control, ambulance clearance, and stolen vehicle detection,"
IEEE Sensors Journal, vol. 15, no. 2, pp. 1109-1113, 1 2 2015.
[7] M. Rath, B. Pati and B. K. Pattanayak, "Mobile agent-based improved traffic
control system in VANET," in Studies in Computational Intelligence, vol. 771,
Springer Verlag, 2019, pp. 261-269.
[8] T. T. Dandala, V. Krishnamurthy and R. Alwan, "Internet of Vehicles (IoV) for
traffic management," in 2017 International Conference on Computer,
Communication and Signal Processing (ICCCSP), 2017.
[9] P. M. Kumar, U. Devi G, G. Manogaran, R. Sundarasekar, N. Chilamkurti and
R. Varatharajan, "Ant colony optimization algorithm with Internet of Vehicles for
intelligent traffic control system," Computer Networks, vol. 144, pp. 154-162, 24 10
2018.
[10] X. Liang, X. Du, G. Wang and Z. Han, "A Deep Reinforcement Learning
Network for Traffic Light Cycle Control," IEEE Transactions on Vehicular
Technology, vol. 68, no. 2, pp. 1243-1253, 1 2 2019.
[11] https://research.utoronto.ca/technology-opportunities/db/smart-traffic-control-
software-powered-machine-learning
[12] Smart Traffic Junction Using Raspberry Pi / G. Mahesh Kumar 2020,
https://www.researchgate.net/publication/347375160_Smart_Traffic_Junction_Usin
g_Raspberry_Pi

22