Understanding and implementation of machine learning using support vector machine for efficient DDoS attack detection
-
Anshu Bhasin
Abstract
Excessive communication over the Internet in the present era has made our privacy vulnerable. With zoomed technology and engineering, it has in turn given wider opportunity to the attackers to penetrate the network just like that of normal users. When attacker’s purpose is to make any specific server or network fail to normal services, it is called network denial-of-service (DoS) attack. Further, distributed DoS (DDoS) attacks are launched through Zombies, which are compromised machines. Recently, for attack detection strategies, most of the researchers and organizations are opting for machine learning (ML) techniques, as these are cost-efficient than humans, when it is about analyzing a huge amount of data. ML in cybersecurity holds the potential to handle areas of prediction, detection, and continuous monitoring. This chapter explores detailed contemporary research and presents meliorated detection mechanism for DDoS attack, based on one-class support vector machine (OC-SVM), an efficient ML technique. More specifically, it focuses on identification of high relevance feature extraction that can exploit the classification capability of OC-SVM for attack detection. The proposed technique includes supervised learning, using NSL-KDD dataset and works adroitly for DDoS attack detection. The empirical results on accuracy and detection rate are compared with other existing methods. False alarm rate and training speed are recorded to project the efficacy of the proposed system.
Abstract
Excessive communication over the Internet in the present era has made our privacy vulnerable. With zoomed technology and engineering, it has in turn given wider opportunity to the attackers to penetrate the network just like that of normal users. When attacker’s purpose is to make any specific server or network fail to normal services, it is called network denial-of-service (DoS) attack. Further, distributed DoS (DDoS) attacks are launched through Zombies, which are compromised machines. Recently, for attack detection strategies, most of the researchers and organizations are opting for machine learning (ML) techniques, as these are cost-efficient than humans, when it is about analyzing a huge amount of data. ML in cybersecurity holds the potential to handle areas of prediction, detection, and continuous monitoring. This chapter explores detailed contemporary research and presents meliorated detection mechanism for DDoS attack, based on one-class support vector machine (OC-SVM), an efficient ML technique. More specifically, it focuses on identification of high relevance feature extraction that can exploit the classification capability of OC-SVM for attack detection. The proposed technique includes supervised learning, using NSL-KDD dataset and works adroitly for DDoS attack detection. The empirical results on accuracy and detection rate are compared with other existing methods. False alarm rate and training speed are recorded to project the efficacy of the proposed system.
Kapitel in diesem Buch
- Frontmatter I
- Preface V
- Acknowledgments VII
- About the Editors IX
- Contents XI
- List of contributors XIII
- Impact evaluation of DDoS and Malware attack using IoT devices 1
- Understanding and implementation of machine learning using support vector machine for efficient DDoS attack detection 29
- Cryptographic method based on Catalan objects and enumerative chess problem 51
- Distributed denial-of-service attacks and mitigation in wireless sensor networks 67
- New techniques for DDoS attacks mitigation in resource-constrained networks 83
- Detection and behavioral analysis of botnets using honeynets and classification techniques 131
- Selected practical and effective techniques to combat distributed denial-of-service (DDoS) attacks 159
- Probability, queuing, and statistical perspective in the distributed denial-of-service attacks domain 173
- Frequently used machine learning algorithm for detecting the distributed denial-of-service (DDoS) attacks 189
- Utilization of puzzles for protection against DDoS attacks 203
- Index 217
Kapitel in diesem Buch
- Frontmatter I
- Preface V
- Acknowledgments VII
- About the Editors IX
- Contents XI
- List of contributors XIII
- Impact evaluation of DDoS and Malware attack using IoT devices 1
- Understanding and implementation of machine learning using support vector machine for efficient DDoS attack detection 29
- Cryptographic method based on Catalan objects and enumerative chess problem 51
- Distributed denial-of-service attacks and mitigation in wireless sensor networks 67
- New techniques for DDoS attacks mitigation in resource-constrained networks 83
- Detection and behavioral analysis of botnets using honeynets and classification techniques 131
- Selected practical and effective techniques to combat distributed denial-of-service (DDoS) attacks 159
- Probability, queuing, and statistical perspective in the distributed denial-of-service attacks domain 173
- Frequently used machine learning algorithm for detecting the distributed denial-of-service (DDoS) attacks 189
- Utilization of puzzles for protection against DDoS attacks 203
- Index 217
