PLAGIARISM SCAN REPORT

Words 1980 Date October 23,2018

Characters 12609 Exclude Url

10% 90% 9 78
Plagiarized
Plagiarism Unique Unique Sentences
Sentences

Content Checked For Plagiarism

ABSTRACT To implement the traditional IDS (Intrusion Detection System) in an active network using Machine learning which keep tracks of the possible
threats and notify the user about potential attacks or simply deals with the problem without bothering user based on the trained model. In this era of
computers network plays very important role of connecting all the systems together. Due to increase in network usage, innovative network is required
and to answer to this SDN is popular now a days. There were many network security concerns are there with the addition of SDN it is multiplied by two
folds. We propose a solution to this by configuring a signature based and a flow based IDS. The former works just like the other signature based IDS but it
cannot detect the unknown or zero-day attacks. So the later flow based IDS with the use of machine learning is to be developed to detect wide number of
attacks. CHAPTER 1 INTRODUCTION This chapter details the discussion on problem definition, motivation, objective of the work, and scope of the project.
PROBLEM STATEMENT AND INTRODUCTION Machine Learning based Intrusion detection system is used to mitigate the limitations of the traditional
Signature based IDS by identifying the zero day attacks and by effectively reduces the human effort. AIM AND OBJECTIVE Internet is a global public
network whose growth and potential has led to subsequent changes to the traditional business model of various organizations .With the easy access to
internet service ,the rate of internet user is beyond reproach and with huge crowd comes great responsibility i.e securing the user data from being
hacked or misinterpreted .And so Intrusion Detection System commonly known as IDS is employed to look after such malicious activities and prevent
them from damaging the our data . Machine learning is something which nullify the need of operator which look over the whole operation because the
model is trained beforehand which allows to operate the system solely .Researchers predict that by as soon as 2020, 85% of the customer interaction will
be managed without human, so it is beneficial to get used to it from now. Along with physical world (for example AI-powered robot) but application of
Machine Learning in virtual world i.e. software world are noteworthy. One of the most prominent idea would be to infuse IDS along with ML which means
no operator has to constantly look for any potential attack and if such situation may occurs, the machine single-handedly will deal with the problem
without concerning the user/client operator. PROBLEM SPECIFICATION Machine Learning based approach to the IDS Can be used to detect wide number
of attack Can be also used to detect previously unknown attack SDN provides much simpler design of network unlike traditional complex design of
classical network designs. 1.4 MATERIALS / TOOLS REQUIRED Anaconda Spyder IDE NSL-KDD Dataset Python libraries like SciKitLearn, Pandas, Numpy,
Matplotlib etc. A VMware with four different Linux operating systems installed on it. Open Daylight Controller + Mininet to create network of computers
Configured Snort IDS to mimic signature based IDS. CHAPTER 2 LITERATURE REVIEW Paper Summary Gain Machine Learning Based Intrusion Detection
System for Software Defined Networks by Atiku Abubakarand and Bernardi Pranggono at 2017 Seventh International Conference on Emerging Security
Technologies (EST), ed: 978-1-5386-4018-0/17/$31.00 ©2017 IEEE The paper present machine learning (Neural Network) based intrusion detection for
SDN. The model IDS are built on the existing signature-based IDS architecture as flow-based IDS to detect anomaly-based attacks in the SDN
environment. The Pattern Recognition is used in this paper due toits performance accuracy rate as compared with the other type of neural network
model. The paper is based on Machine Learning based IDS and provide very meaningful insights that how it is to be done which helps us to carry forward
the work. T. A. Tang, L. Mhamdi, D. McLernon, S. A. R. Zaidi, and M. Ghogho, "Deep learning approach for Network Intrusion Detection in Software
Defined Networking," in 2016 International Conference on Wireless Networks and Mobile Communications (WINCOM), 2016, pp. 258-263. In this paper,
They have implemented a deep learning algorithm for detecting network intrusion and evaluated our NIDS model. Although their results are not yet good
enough to be adopted in any commercial product or an alternative solution for signature-based IDS, their approach still has significant potential and
advantages for further development. By comparing the results with those of other classifiers, They have shown the potential of using deep learning for
the flow-based anomaly detection system. In the context of the SDN environment, the deep learning approach also has potential. They have provided a
detailed analysis and accuracy from different machine learning classifier which helps us to improve them. “Software Defined Networking Definition,”
Available: https://www.opennetworking.org/sdn-resources/sdn-definition, [Accessed 04 Aug. 2018]. It provides detailed description of how SDN works and
its overview, learning, history etc. We came to know more about SDN and its application and how it works. “Software Defined Networking Definition,”
Available: https://www.opennetworking.org/sdn-resources/sdn-definition, [Accessed 04 Jul. 2016]. Their work provides a practical solution for construction
of better IDS based on ‘Outlier Detection method’. Thus, with their results and analysis, we can say that, they have analyzed the TCP header information
that Allows to deal with anomaly detection in fast incoming traffic in real time with lower false positive detection rate. They have provided description of
how they perform the analysis based on different machine learning algorithm and how we can modify it and improve the accuracy. M. Tavallaee, E.
Bagheri, W. Lu, and A. A. Ghorbani, "A detailed analysis of the KDD CUP 99 data set," presented at the Proceedings of the Second IEEE international
conference on Computational intelligence for security and defense applications, Ottawa, Ontario, Canada, 2009. This paper describes the dataset we’ve
used in one project. It talks about every features of the dataset and what encoding has been done on the data. It helps us to understand and analyze the
dataset CHAPTER 3 MODULES AND PLAN OF WORK Modules Virtual test bed The Mininet network simulator is installed and configured to create host
system, servers, and OVS on the same OS with ODL. The Metasploitable2 server is hosting different services that are left vulnerable intentionally for
penetration testing purpose, while the Parrot security will be generating attack scenario on Metasploitable2. Furthermore, Snort IDS is installed and
configured on a separate Ubuntu Desktop VM to offer network traffic monitoring, invasion detection by means of NIDS. Network Topology The star
topology is utilized for setting up laboratory network because it is easy to configure looking at the nature of the research and the combination of systems
involved. Also, OVS-switch act as a central hub which is expected to provide optimum performance of the network traffic without overhead in providing
centralize network monitoring. Therefore, failure of a single node will not disturb the entire network. Pattern Recognition of Neural Network Pattern
recognition of neural network is implemented in this model which usually classifies inputs into a set of target categories. The network architecture
consists of three layers( input layer, hidden layer, and an output layer). For better performance, Backpropagation algorithm is utilized to train the
network. Backpropagation algorithm is a training method used by propagation in classification and updating the weight of a network. When an input is
received from the input layer, it is passed on to the following layer, then to the output layer. The resultant output is compared with the certain targets or
desired output, each output result of the neuron is calculated using a function and error value at the output layer. If the output is roughly closed or
directly matches the target, then it is presented as final output, otherwise an error is fire from the output layer toward preceding layers until desired
output is obtained. NSL-KDD Dataset for Training Model The NSL-KDD dataset used in this research is to implement and evaluation/ training of the
proposed model. Evaluation Matrix and Procedure For evaluating the performance of our proposed model we’re going to use standard benchmarks
Accuracy (ACC), Precision (P), Sensitivity (SNS), and F-Measure (F1-score) Periodic Progress Report: First PPR Project: Intrusion Detection System Using
Machine Learning 1. What Progress you have made in Project? We met our internal guide for our discussing our project regarding which domain is best
suitable for us. Our internal guide, Prof. Sneh Vyas, suggested us to refer the recent research related to ML (Machine Learning).In the given timeframe,
we browsed the internet for research available on different standards like IEEE, Springer, IIT-Bombay on ML. Finally, we choose ML as our domain and
“Intrusion Detection System Using ML" as our project definition. 2. What challenge you have faced? While searching for an appropriate domain, we came
across many new domains, which are not part of our syllabus like ML, IOT and AI etc. We were confused about which domain is best suitable for use in
the coming year and had a long discussion with our guide who ultimately made the final call regarding our domain which was appropriate. Then we came
to know about ML after research over various domain so we searched more and more about ML and regarding its functions and applications. 3. What
support you need? In order to successfully implement our project, we required guidance from a well-experienced person who has dealt with IDS and also
has knowledge regarding various implementation of ML. The key challenge we need to overcome is regarding integration of traditional IDS with the
newer technology of ML which can be tricky. 4. Which literature you have referred? We reviewed some theory papers but mainly we explore the
mechanism behind working of ML and IDS by watching tutorials and research papers. Periodic Progess Report : Second PPR Project : Intrusion Detection
System Using Machine Learning 1. What Progress you have made in the Project? Internal guide, Prof. Sneh Vyas, suggested us to explore the
Technologies that we will be using in our project. And we also discussed the technologies like VMware, Rstudio and Open Flow with our internal guide.
Furthermore, we also watched some videos on YouTube related to these software. 2. What challenge you have faced? The only challenged we faced was
regarding the dataset that is needed to train the model for which is to be implemented along with the IDS, Other than that no major issues were
observed. 3. What support you need? We needed some guidance regarding the technologies to be used and where to find the corresponding material
which will help to better understand them. 4. Which literature you have referred? We referred some research paper along with watching the tutorial on
how to implement ML. The research paper we referred are as follow 1. Machine Learning Based Intrusion Detection System for Software Defined
Networks (EST) 2. Network Traffic Classification using Machine Learning Techniques over Software Defined Networks (LJACSA) Periodic Progess Report :
Third PPR Project : Intrusion Detection System Using Machine Learning 1. What Progress you have made in the Project? We gathered the data set
required for our project, before searching for it, we brainstorm a lot about the parameters on which the model for the ids will be trained. We started
searching the appropriate data set which would satisfy the required parameters. And we successfully found a Data Set i.e. KDD-99 which is basically a
data regarding a competition for data miners. The dataset consists of 17000 training and testing data entries. 2. What challenge you have faced? We
faced problems while deciding the parameters on which we would train the machine learning model and also deciding which algorithm to be used for
training. We had to think and search for the types of attacks possible in an interconnected network. 3. What support you need? We required a test bed in
which we would implement the newly developed ML-based ids for measuring the accuracy. And also a high bandwidth Internet Connection.
Sources Similarity

Machine learning based intrusion detection system forCompare text

what progress you have made in the project? our project is almost done. solar model and wind mill is done but still we face some technical problem
regarding speed and output of wind mill. we also refer reference books and some technical magazine for new upcoming techniques and information. 2%
https://www.scribd.com/document/362610094/PMMS-2-Degree-Question-Answer-for-HVDC-PPR-2