Performance improvement of intrusion detection system using neural networks and particle swarm optimization algorithms
- PDF / 638,852 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 32 Downloads / 258 Views
ORIGINAL RESEARCH
Performance improvement of intrusion detection system using neural networks and particle swarm optimization algorithms Ahmad Shokoohsaljooghi1 • Hamid Mirvaziri1
Received: 11 August 2017 / Accepted: 2 May 2019 Bharati Vidyapeeth’s Institute of Computer Applications and Management 2019
Abstract Intrusion detection system is considered as a decision-making tool in the networks. Mainly, an intrusion is an attempt to violate the security mechanisms and its rules. The aim of an intrusion detection system is to monitor network traffic and explore unusual behavior that may be an attack. If a package has a different pattern toward its standard behavior, it can be categorized as abnormal conditions and consequently an attack or intrusion. In this paper, a pre-processing is performed on KDDCUP99, NSL-KDD and CIDD dataset to choose a subset of the features, reduce dimension and then normalize data. Combination of particle swarm optimization and neural network algorithms are used to recognize intrusions which can efficiently classify the attacks and reduce the number of false alarms and improve detection rates. Obtained results show that proposed method provides a higher accuracy and performance comparing with other algorithms to detect different classes of attacks. Keywords Multi-layer perceptron Particle swarm optimization Intrusion detection Data mining
1 Introduction It is with the purpose of violating security policies such as confidentiality, integrity and accessibility that one tries to intrude on a network [1]. Some considerable researches & Ahmad Shokoohsaljooghi [email protected] Hamid Mirvaziri [email protected] 1
Department of Computer Engineering, Shahid Bahonar University, Kerman, Iran
have been conducted on detection and prevention of intrusion in the field of information security. Intrusion detection can be a classification system of the network traffic which determines whether the received packet is normal or not [2]. Denning is the first person who studied intrusion-detection systems [3]. From then, numerous methods have been invented onwards to improve the function of intrusion detection systems including statistical, machine learning and data mining methods. A good intrusion-detection system must have minimum false negative rate and a high detection rate [4]. One thing that is quite significant in computer systems is intrusion detection. An intrusion detection system is a software or hardware which monitors the computer network to see if there is any harmful activity or any violation of the management and security policies and then it reports to the network management sector. Intrusion-detection systems must detect and recognize any kind of unpermitted usage of the system, or any kind of abuse or activity by both internal and external users that would harm the computer. Usually, intrusion-detection systems are used along with firewalls and act as a complementary security-wise for them. Traditional intrusion-detection system cannot adapt to new intrusions; thus, no
Data Loading...
