DEPARTMENT OF COMPUTER SCIENCE
STRENGTHENING CYBER DEFENCE
GUIDED BY
Mr.R.Dharmaraj
M.Sc.,M.Phil.,B.Ed.,Head in department
of Computer Science,
Paavendhar College of Arts & Science,
M.V.South. PRESENTED BY,
M.VIJAYARAGAVAN
BSC.COMPUTER SCIENCE
Intrusion Detection System (IDS) is meant to be a software
application which monitors the network or system activities and finds
if any malicious operations occur. Tremendous growth and usage of
internet raises concerns about how to protect and communicate the
digital information in a safe manner.
Nowadays, hackers use different types of attacks for getting the
valuable information. As the internet emerging into the society, new
stuffs like viruses and worms are imported. The malignant so, the
users use different techniques like cracking of password, detecting
unencrypted text are used to cause vulnerabilities to the system.
Hence, security is needed for the users to secure their system from
the intruders. Firewall technique is one of the popular protection
techniques and it is used to protect the private network from the
public network.
IDS are used in network related activities, medical applications,
credit card frauds, Insurance agency. Many intrusion detection
techniques, methods and algorithms help to detect these attacks.
EXISTING SYSTEM
The IDS can be distinguished on the basis of where the detection
is taking place and how or by which technique it is being
detected.
The IDS is classified into two niche segment one being Network
Intrusion Detection System (NIDS) and the other being Host
Intrusion Detection System (HIDS).
The first system mentioned helps in the analysis the incoming
networking traffic whereas the HIDS functioning is based on the
activity of the operating system.
DISADVANTAGE
Computational complexity is high
Time complexity can be occurred
Difficult to handle streaming of data
Need hardware to detect the intrusion
PROPOSED SYSTEM
The proposed methodology for the Intrusion Detection
System (IDS) employing a hybrid architecture of
Convolutional Neural Networks (CNN) with Long Short-Term
Memory (LSTM) algorithm involves a systematic approach
to enhance the accuracy of intrusion detection
This ensures that the subsequent model training is
performed on representative and appropriately formatted
data. The spatial feature extraction phase employs a CNN
architecture designed to analyze the spatial
characteristics of the network traffic data.
Through convolutional and pooling layers, the CNN learns
to identify spatial patterns indicative of potential
intrusions. Following this, the temporal relationship
modelling phase incorporates an LSTM network to capture
sequential dependencies in the data.
The success of the Deep learning architectures lies
in using fast learning algorithms and efficient
solutions.
The development of GPU accelerated computing has
led to the increase in their development and lead to
faster convergence of the algorithms.
Reduce the false positive rate and improve the
accuracy
Time complexity can be reduced
HARDWARE REQUIREMENTS
Processor : i3 9 Gen Processors
RAM : 8GB
Hard disk : 512 GB
Compact Disk : 8 GB
Keyboard : Standard keyboard
Monitor : 15-inch color monitor
SOFTWARE REQUIREMENTS
Operating system : Windows OS
Server Side : Python 3.7.4(64-bit)
or (32-bit)
IDE : Pycharam
Libraries : Pytroch Tensorflow,
KERAS
MODULE DESCRIPTION
DATASET ACQUISITION
PREPROCESSING
FEATURES EXTRACTION
CLASSIFICATION
PERFORMANCE EVALUATION
EVENT CORRELATION
DATASETS ACQUISITION
The KDD Cup dataset, utilized for benchmarking
intrusion detection issues, is used in our experiments.
The dataset is a gathering of simulated crude TCP
PREPROCESSING
Data pre-processing is an important step in the
[data mining] process. The phrase "garbage in,
garbage out" is particularly applicabl .
FEATURES EXTRACTION
Feature extraction is a general term for
methods of constructing combinations of the
variables to get around these problems while
still describing.
CLASSIFICATION
Data Collection: Collect a dataset of network
traffic data that includes both normal traffic
and DDoS attack traffic.Data Preprocessing.
PERFORMANCE EVALUAT
In this module, performance can be evaluated
in terms of accuracy rate.
Proposed work provides improved accuracy
rate than the existing systems.
EVENT CORRELATION
Perform correlation and contextual analysis to
identify patterns and relationships between
events
SYSTEM TESTING
System testing evaluates the entire system as a whole,
including all integrated components and modules. It verifies
that the system meets the specified requirements
White box Testing
The term 'white box' is used because of the internal
perspective of the system. The clear box or white box or
transparent box name denote the ability to see through the
software's outershell into its inner workings.
Black box Testing
Black box testing is a technique of software testing which
examines the functionality of software without peering into
its internal structure or coding.
Integration Testing
Test the integration of different components or modules within
the autism detection system. Verify that these components
interact correctly and exchange data as intended
Performance Testing
Evaluate the performance of the autism detection algorithm under
different conditions, such as varying input data sizes, processing speeds,
and computational resources. Measure key performance metrics, such as
processing
Error Handling
Test the error-handling mechanisms within the autism detection
algorithm to ensure that it gracefully handles unexpected
situations or errors.
LEVEL :1
LEVEL :2
LEVEL :3
LEVEL: 4
LEVEL:5
Intrusion detection plays an important role in the
network security as the applications and their
behavior are changing day to day.
Network intrusion detection has extensively
researched in recent years and many techniques
have been proposed including machine learning
and deep learning techniques.
As a result, there increased the need for accurate
classification of the network flows. Here we have
proposed deep learning model using CNN with
LSTM based feature selection for the accurate
classification of intrusion detection. In this paper
Future work in hybrid algorithm-based intrusion
detection presents several promising avenues for research
and development.
Firstly, exploring novel architectures and optimization
techniques tailored specifically for intrusion detection
could enhance the performance and efficiency of hybrid
models.
This could involve investigating techniques such as
attention mechanisms, graph neural networks, or meta-
learning approaches to capture complex relationships
within network traffic data more effectively.
THANK YOU
