ABSTRACT

Web application attacks are incessantly increasing in number and in severity. The big
data available on the internet motivates hackers to launch new kind of attacks. As
announced by the OWASP organization, injection attack has been the first of the top 10
security vulnerabilities, and SQL injection attack is one of the most important types.
This attack represents a serious threat to the web applications. SQL is a
communication medium between Web application and back end database. So the
mostly attackers use SQL for accessing a database. SQL injection can cause great
harm to the network, resulting in data leakage and website paralysis.

Injection attack is one of the best 10 security dangers declared by OWASP. SQL
infusion is one of the main types of attack. In light of their assorted and quick nature,
SQL injection can detrimentally affect the line, prompting broken and public data on the
site. Therefore, this article presents a profound woodland-based technique for
recognizing complex SQL attacks. Research shows that the methodology we use
resolves the issue of expanding and debasing the first condition of the woodland. We
are currently presenting the AdaBoost profound timberland-based calculation, which
utilizes a blunder level to refresh the heaviness of everything in the classification.

v
 TABLE OF CONTENTS

CHAPTER NO TITLE PAGE NO

ABSTRACT vi

LIST OF FIGURES ix

1 INTRODUCTION: 1

1.1 OVERVIEW OF THE PROJECT 1

1.2 OBJECTIVE OF THE PROJECT 2

2 LITERATURE SURVEY: 3

2.1 RELATED WORK 3

3 AIM AND SCOPE OF PRESENT 6

IMPLEMENTATION:

3.1 AIM AND SCOPE 6

3.2 EXISTING SYSTEM 6

3.2.1 DISADVANTAGES OF EXISTING 6

SYSTEM

3.3 PROPOSED SYSTEM 6

3.3.1 ADVANTAGES OF PROPOSED 7

SYSTEM

vi
 3.4 ARCHITECTURE 7

3.5 REQUIREMENTS 8

3.5.1 HARDWARE REQUIREMENTS 8

3.5.2 SOFTWARE REQUIREMENTS 8

3.5.3 LIBRARIES REQUIRED 8

3.6 SAMPLE INPUT 8

4 METHODOLOGY: 9

4.1 SCEMATIC PROCESS 9

4.2 MODEL IMPLEMENTATION 9

4.3 ADAPTIVE DEEP FOREST 10

5 RESULT AND DISCUSSION 12

PERFORMANCE ANALYSIS:

5.1 RESULT ANALYSIS 12

6 SUMMARY AND CONCLUSIONS: 15

6.1 CONCLUSION 15

6.2 FUTURE WORK 15

REFERENCES 16

vii
APPENDIX 17

A. SOURCE CODE 17

B. SCREEN SHOTS 21

viii
 LIST OF FIGURES

FIGURE NO FIGURE NAME PAGE NO

3.4 Architecture 7
3.6 Dataset 8
4.2 Model Implementation 10
B.1 Graphical Representation of Data 21
B.2 Frequency Distribution Graph 22
B.3 Accuracy score of Algorithms 22
B.4 AdaBoost Predictions 23
B.5 Categorical Graph 24
B.6 Hyper Parameters Tuning 24

ix
 CHAPTER 1
INTRODUCTION

1.1 OVERVIEW OF THE PROJECT:

With the rapid development of Web 2.0 technology, network applications
have gradually become an indispensable part of our lives. At the same time, Web
applications are confronted with more challenges. As announced by the OWASP
organization, injection attack has been the first of the top 10 security vulnerabilities
in 2013 and 2017, and SQL injection attack is one of the most important types
SQL (Structured Query Language) is a standard database programming
language for accessing and manipulating data in database WEB programming
languages (such as PHP,JAVA) provide various methods for constructing and
executing SQL statements . Developers usually construct SQL statements by
concatenating string which is submitted by users from web page. Due to the wide
variety of SQL languages, there are too many encoding methods for constructing
SQL statements, so there is a risk of being attacked anywhere via constructing
SQL statements. In the process of SQL injection attack, the attacker inserts
malicious code fragments into the request parameters, which causes the server to
execute illegal queries, resulting in data leakage and database damage. For
example, the attacker can obtain the username, password and other privacy data
of website users by SQL injection, which seriously threatens the data security. As
shown in, the attacker sends malicious SQL statements into the application
through methods like user input injection, cookie injection, server variable injection
and so on. Since SQL attacks only need to construct structured query statements
without developing additional malwares, SQL injection has become one of the
most frequently used attacks by attackers, and the attacking methods change
frequently with the development of website technology. The structure of SQL
language is changeable and supports different coding methods. Therefore, some
traditional detection methods, such as blacklist mechanism and rule-based
detection mechanism, cannot produce better defensive effect.
In recent years, many researchers have done a lot of work on SQL injection
detection, but the detection scope is usually limited to some subsets of SQL
injection.

1
1.2 OBLECTIVE OF THE PROJECT:
Attackers can use SQL injection vulnerabilities to bypass application
security measures. They can go around authentication and authorization of a web
page or web application and retrieve the content of the entire SQL database. They
also use SQL injection to add, delete, and modify records in databases. The main
objective of the project is to detect the SQL injection attacks.

2
 CHAPTER 2
LITERATURE SURVEY
2.1 RELATED WORK:
Konstantinos Kemalis et al, he acquainted another methodology with
characterizing the utilization of SQL weaknesses. This new technique, from one
perspective, utilizes specialized measurements that portray the idea of the SQL
issue made and carried out by the Internet, then again, which controls the most
common way of making an issue that is actually untrustworthy. They have fostered
a SQL injection detection system (SQL-IDS) model that carries out the calculation.
The framework screens Java-based applications and distinguishes SQL pressure
attacks continuously. It reports the aftereffects of a progression of SQL attacks,
which show that understanding the normal issues keeps the framework from
playing out a top to bottom, incorporated examination that doesn't offer a positive
or negative response [1]. V. Naga Manikanta et al, he examined how to
appropriately shield the web application from SQL attacks utilizing a firewall
strategy. That technique utilizes a hearty and vigorous investigation. In the inside
and out examination, he depicted the URL, the structure, the infusion site, and the
basic site highlights. Subsequently, we know about the genuine issues that might
emerge in the materials. In a hearty examination, he tracked the issues
experienced during activity utilizing the firewall archive and contrasted them with
the white rundown of issues. The trial of utilizing a genuine website with two open-
source apparatuses: Web Request and Web Control Framework (w3af) and
GreenSQL. He utilized w3af for a rundown of all GreenSQL-agreeable issues as a
vault fire. The outcomes showed that the execution strategy doesn't need
changing any code on a normal site, however it can recognize a wide range of
SQL attacks utilizing extra intermediary facilitating highlights [2]. Nabeel Salih et al
proposed the SQLI Protocol (WASP) instrument for web apparatuses is being
created for ongoing web applications to recognize SQL attacks in the way
portrayed in this article. Planned materials will then, at that point, be explored and
broken down for better execution. The procedure depends on constant diagnostics
dependent on the information on foreign substances, effective and proficient
dispersion, and linguistic structure in line with distinguishing invalid issues prior to
arriving at the base utilizing Microsoft ASP.NET. The high-level apparatus can

3
distinguish and shut down all SQLI attacks progressively, bringing about no
blunders, barely any benefits, and no base prerequisites for arrangement [3].
Hasanein D. Rjeib et al explained that web applications for large numbers of our
online movement programs have been added to the web-based help. In any case,
disregarding the protection and security prerequisites of organizations makes them
more powerful as far as security. The security framework proposed in this article
incorporates three methods for forestalling it: how to store put away things, explicit
mistake messages, and how to store them on a Structured Query Language
Injection Attacks (SQLIA). It likewise breaks down the arranged inoculation
process. The normal assurance system is great since it can adequately forestall
SQLIA information base attacks [4]. Abhishek Kumar Baranwal et al, in the
research explained very much planned needle permits admittance to oppressive or
unapproved clients and are principally accomplished through SQL infusion and
Cross site prearranging. In that article, he did a point-by-point investigation of the
various kinds of SQL and XSS attacks, how to recognize and forestall them, the
future, and what we have seen and expounded on the most common way of
battling these attacks. They also, we will make a similar examination of the various
strategies used to battle these attacks. We will then, at that point, show what we
have seen and expound on the future and arrange advancement against these
attacks [5]. Geogiana Buja et al, explained in the research that SQL is extremely
simple to ensure against attacks, and there are as yet countless frameworks on
the Internet that can be powerless against this sort of attack, as there will be a
couple of basic things that are not known. Hence, in this paper, we need to take an
example to comprehend and recognize the shortcomings of the site; SQL input
depends on a characterized, characterized definition. Likewise, search inquiries
can report weaknesses in web applications. Subsequently, the arrangement model
ought to have the option to limit the chance of SQL Injection attacks on the
Internet [6]. Huafeng Zhang et al, in the research showed progress in SQL-based
learning. Not all substances should be investigated and erased. You simply must
be more segregating regarding how you help others. Joining these components
into a pre-arranged format can make the continuous SQL idea clearer. That
utilizes inside and out examination to distinguish SQL attacks in power. We select
the objective ascribes dependent on the SQL attack and afterward take the

4
necessities in the URL or bundle as train data; Train chosen protests and
amassed information utilizing deep belief networks (DBNs) and afterward figure
out how to infuse SQL Injection. Then, track down a genuine instance of
understanding SQL and quit wasting time progressively [7]. Vinaya Kumar et al, in
the research explained that Intrusion Detection System (IDS) is developed widely
by using the machine learning techniques for the detection and classification of
attacks like cyber-attacks in the network and host levels in an instinctive definite
manner. One of the type deep learning models, DEEP NEURAL NETWORK
(DNN) is used for developing productive IDS for detection and classification of
uncertain and incalculable cyber-attacks [8]. Dimitris Mitropoulos et al proposed
research on different security mechanisms at odds with SQL injection attacks. He
proposed a model which highlights the main key debility of the attacks. They
categorized and then analyzed the previous proposed systems and methods
based on the performance, availability and some other characteristics [9]. Yaohui
Wang et al, in the research proposed a SQL injection detection method picking out
PHP, which is the basis of injection analysis automation. This process is the
combination of analysis technology, like behavior model and lexical objects
comparison based SQL for the design and the establishment of a system for the
detection of the SQL injections [10].

5
 CHAPTER 3
AIM AND SCOPE OF THE PRESENT INVESTIGATION
3.1 AIM AND SCOPE:
The aim of the project is to detect the SQL injection attacks. This method
uses the SQL lexicaland grammatical analysis technology to parse the SQL
statement into the SQL grammar tree via stain analysis, and judges whether there
is an SQL injection attack according to the stain state. While, dynamic analysis
detection can only detect pre-defined vulnerabilities by application developers, and
it has no detection capability.
When encountering attacks against unknown vulnerabilities. Parameter filtering is
based on regular expressions and blacklists to filter invalid characters . This
method needs predefined regular expressions, but due to the diversity of SQL
grammar and user input, it cannot meet the detection requirements. Attackers can
adjust the input to bypass detection. Appiah et al. proposed an improved pattern
matching method for signature-based SQL injection attack detection framework ,
which integrates fingerprint identification and pattern matching to distinguish
normal and malicious SQL queries.
3.2 EXISTING SYSTEM:
In the existing application, the malicious attacks are controlled by certain decisions
and programming coding. The decision is based on a certain set of rules. These
sets of rules are designed to detect attack patterns. Other Approaches are White-
box testing, Black-box testing, Static analysis, Model based testing.
3.2.1 DISADVANTAGES OF EXISTING SYSTEM:
The drawbacks of the existing system are, for huge applications, the set of rules
becomes more complex. Test cases are challenging to design without having clear
functional specifications. It is more difficult to maintain the set of rules manually
and test.
3.3 PROPOSED SYSTEM:
An appreciation model can highlight shortcomings in web applications.
Accordingly, this model can lessen the possibilities that SQLIA will deal with your
site. AI and the SVM calculation are utilized to ensure SQLIA control. The answer
for this is to distinguish and keep SQLIA from closing down when you move the
landing page of every application to the test page. Straight and straight areas were