Artificial Intelligence and Machine Learning with SD-WAN
Kytömäki, Joni (2021)
2021
Tietotekniikan DI-ohjelma - Master's Programme in Information Technology
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2021-02-26
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202102011842
https://urn.fi/URN:NBN:fi:tuni-202102011842
Tiivistelmä
The technology sector has been facing a massive artificial intelligence (AI) and machine learning (ML) boom for the last couple of years, and the telecommunications industry is not an exception. Advanced analytics and automation offered by AI/ML fit into the concept of modern networking, and software-defined networking (SDN) and, more specifically, software-defined wide-area networks (SD-WAN), represent as the path towards future networking. The purpose of the thesis is to study how SD-WAN can benefit from AI/ML and demonstrate such use cases in a real SD-WAN network. The scope of AI/ML is limited to modern ML techniques and algorithms.
This thesis is split into two parts. The first part consists of a literature study. The literature study includes a study on how SD-WAN enables ML, a review on how ML can be utilized in an SD-WAN network, and finally, a market review on the current SD-WAN products and their claimed AI/ML capabilities. The second part consists of a case study that includes two proof-of-concept examples. The proof-of-concept examples demonstrate how Elastic Stack ML capabilities can be used together with a Nokia Nuage Networks SD-WAN network to perform analytics. The first example demonstrates a network intrusion detection system, and the second example demonstrates a network performance monitoring solution.
The literature study shows that SD-WAN enables ML-based analytics and automation exceptionally well and that ML has many applications in different sections of the SD-WAN architecture. The market review shows that most of the vendors are trying to implement AI/ML into their products in some way. The case study demonstrates and concludes that Elastic Stack’s ML capabilities can be used on a real Nokia Nuage Network SD-WAN network to get useful analytics about its security and performance. ML-based analytics enable more advanced anomaly detection than traditional threshold-based solutions. However, the study also shows that whether ML-based methods should be used, should be assessed on a case-by-case basis, and the data being analyzed should be known thoroughly.
This thesis is split into two parts. The first part consists of a literature study. The literature study includes a study on how SD-WAN enables ML, a review on how ML can be utilized in an SD-WAN network, and finally, a market review on the current SD-WAN products and their claimed AI/ML capabilities. The second part consists of a case study that includes two proof-of-concept examples. The proof-of-concept examples demonstrate how Elastic Stack ML capabilities can be used together with a Nokia Nuage Networks SD-WAN network to perform analytics. The first example demonstrates a network intrusion detection system, and the second example demonstrates a network performance monitoring solution.
The literature study shows that SD-WAN enables ML-based analytics and automation exceptionally well and that ML has many applications in different sections of the SD-WAN architecture. The market review shows that most of the vendors are trying to implement AI/ML into their products in some way. The case study demonstrates and concludes that Elastic Stack’s ML capabilities can be used on a real Nokia Nuage Network SD-WAN network to get useful analytics about its security and performance. ML-based analytics enable more advanced anomaly detection than traditional threshold-based solutions. However, the study also shows that whether ML-based methods should be used, should be assessed on a case-by-case basis, and the data being analyzed should be known thoroughly.