• PDF / 2,860,379 Bytes
• 21 Pages / 595.276 x 790.866 pts Page_size
• 69 Downloads / 231 Views

DOWNLOAD

REPORT

---

(0123456789().,-volV)(0123456789(). ,- volV)

ORIGINAL ARTICLE

Detection of anomaly intrusion utilizing self-adaptive grasshopper optimization algorithm Alok Kumar Shukla1 Received: 25 May 2020 / Accepted: 30 October 2020 Ó Springer-Verlag London Ltd., part of Springer Nature 2020

Abstract Due to continued growth in both cyberattacks and network data size, organizations need to develop advanced ways to keep their networks and data secure the dynamic nature of evolving malicious attacks. Nowadays, large number of security mechanisms are installed in the network but it opens the possibility for adversaries to conduct malicious activity in the computer network. To detect potential attacks, intrusion detection systems are important security tools that can help to increase the security posture of computer network. In order to identify new malicious or anomalous attacks, this study developed an opposition self-adaptive grasshopper optimization algorithm based on mutation and perceptive concept. Moreover, reinforcement learning is utilized in support vector machine, named gain actor critic with support vector machine to increase the detection capabilities by identifying new cyberattacks. Extensive experiments are conducted on standard intrusion detection datasets such as NSL-KDD, AWID and CIC-IDS 2017 to measure the performance of the proposed method. It can more reliably detect and classify modern attacks with high accuracy and low false-positive rate. The comparative simulation results demonstrates that the proposed algorithm is more capable than basic grasshopper optimization algorithm and other used evolutionary techniques in terms of detection rate, false-positive rate and accuracy for solving IDS problems. The proposed model has provided high detection rate of 99.71%, accuracy of 99.23% and low false-positive rate of 0.009 in NSL-KDD with six optimal features; in AWID data, high detection rate of 99.11%, accuracy of 99.15% and low false-positive rate of 0.091 with eight optimal features, and high detection rate of 99.61%, accuracy of 99.35% and low false-positive rate of 0.052 in CIC-IDS 2017 data with eight optimal features. Keywords Evolutionary algorithm  Grasshopper optimization algorithm  Intrusion detection  CIC-IDS 2017  Support vector machine

1 Introduction Cyber-threat protection is one of the most noteworthy challenging research areas of information technology nowadays [1]. Unfortunately, most of the existing security mechanism has not been providing sufficient defense facilities in the network environment while maintaining operational continuity and stability of the defense objective in the existence of malicious intruders or cyber-threats [2]. In general, cyber-threats are sets of actions that intend to & Alok Kumar Shukla [email protected] 1

School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, India

negotiate the ideologies of integrity, confidentiality and availability in the computer network [3]. Even though posing security solutions such as firewal