machine learning-based ids for software define 5g network · new paradigm called softair toward...
TRANSCRIPT
![Page 1: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/1.jpg)
Machine learning-based IDS for software define 5G
network
2019.09.24SeoulTech
Jose costa Sapalo Sicato
![Page 2: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/2.jpg)
Table of contents
1. Introduction
2. Related work
3. Architecture
4. Intelligent intrusion detection process
5. Experiment result
6. Conclusion
![Page 3: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/3.jpg)
Abstract • Software-defined architecture has many advantages in providing
centralized control and flexible resource management.
• As the focus of network security, intrusion detection systems (IDSs) areusually deployed separately without collaboration
• They are also unable to detect novel attacks with limited intelligentabilities, which are hard to meet the needs of software-defined 5G
• Evaluation results prove that the intelligent IDS achieves betterperformance with lower overhead.
![Page 4: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/4.jpg)
Introduction
• Software-defined fifth generation (5G) architecture will be a crucialtendency in the development of future 5G networks.
• As a result, new network security architecture and systems aredesperately needed to enhance the security of software-defined 5Gnetworks
• As an essential technology in network security, intrusion detectionsystems (IDSs) have received more and more concerns in efficientlydetecting malicious attacks
• To overcome the limitation of traditional IDS, artificial intelligence (AI)has been employed for intelligent detection.
![Page 5: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/5.jpg)
Introduction
At present, there have been a few researches combining IDS and AI.
In this paper, was propose an intelligent IDS based on software defined5G architecture. Benefit from the software-defined technology, itintegrates relevant security function modules into a unified platformwhich are dynamically invoked under centralized management andcontrol.
![Page 6: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/6.jpg)
2. Related Work
• As SDN dynamically manages network configurations and controlspacket processing in a centralized manner, it has well satisfied theevolution demand of cellular networks in the 5G era, which aims toprovide flexible service provisioning mechanisms. Therefore, thecombination of 5G and SDN has attracted a lot of research interest. Anew paradigm called SoftAir toward next generation (5G) wirelessnetworks is introduced.
• However, with the fast development of software-defined 5Gnetworks, the emergence of unknown attacks also poses severesecurity challenges.
![Page 7: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/7.jpg)
2. Related Work
• It also provides network-layer security services such as packetrouting, identity authentication and automated security managementin a global view which facilitates the detection and prevention ofattacks
• A comprehensive survey of existing SDN-based distributed denial ofservice (DDoS) attack detection solutions and present an SDN-basedproactive DDoS defence framework (ProDefense).
![Page 8: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/8.jpg)
3. Architecture
![Page 9: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/9.jpg)
4. Intelligent intrusion detection process
4.1 Random forest is a collection of uncorrelatedstructured decision trees deemed asforest.• If the number of input training data
is N, we take N samples randomlywith replacement from the originaldata.
• For each tree, we choose m (m < M,usually m = M) features out of M-attribute entire set randomly asinput variables withoutreplacement.
![Page 10: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/10.jpg)
4. Intelligent intrusion detection process
4.2 Hybrid clustering-based AdaBoost
• For the first stage, we make apreliminary judgement by adoptingk-means++ to divide the traffic intotwo clusters which most probablyrepresent the normal and abnormalinstances.
• Later, we further partition theanomaly clusters into four mainclasses of attacks using theensemble algorithm AdaBoost.
![Page 11: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/11.jpg)
5. Experiment result
• 5.1 Dataset KDD Cup 1999 dataset- It contains ∼5,000,000 network connections in the training set and nearly 2,000,000 instances in the testing set.
• 5.2 Evaluation metrics
![Page 12: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/12.jpg)
5.3 Performance analysis
![Page 13: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/13.jpg)
5.3.2 Evaluations of the proposed solutions:
![Page 14: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/14.jpg)
• NTs and the error rate ofclassification using the training andtesting dataset, respectively, areplotted. It is shown that the errorrate decreases as NTs become larger.
![Page 15: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/15.jpg)
![Page 16: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/16.jpg)
6. Conclusion
• An intelligent IDS based on software-defined 5G architecture usingmachine learning algorithms.
• It integrates and coordinates security function modules undercentralized management and applies machine learning algorithms todetect intrusions intelligently.
![Page 17: Machine learning-based IDS for software define 5G network · new paradigm called SoftAir toward next generation (5G) wireless networks is introduced. •However, with the fast development](https://reader036.vdocument.in/reader036/viewer/2022070908/5f86a4813b4bd30e553fc09b/html5/thumbnails/17.jpg)
Opinion
• 5G network introduces a slew of cybersecurity concerns andproblems.
• Anomaly-based intrusion detection techniques, thatutilize algorithms of machine learning, have the capability torecognize unpredicted malicious.
• Machine Learning is the field of study that gives the capability tolearn and improve from experience without being programmedexplicitly automatically.