New platform facilitates the use of machine learning in NFV optimization2019-02-12
Michel Gokan Khan, PhD student in Computer Science at Karlstad University, has developed the “NFV-Inspector”, a free and fully open source platform for optimizing the infrastructure of the next generation of 5G networks. This platform helps researchers and industry working on Network Functions Virtualization (NFV) to extract information from underlying infrastructure to optimize latency, costs and energy consumption.
“This tool saves a lot of time and makes NFV researchers’ work much easier,” says Michel Gokan Khan. “If you want to do it yourself, you need to manually run different tests many times and it will take months to prepare and process all the results.”
Michel, who has worked in the industry for many years, saw an opportunity to apply his programming skills in the project. During the first months of his doctoral studies, he developed the NFV-Inspector.
“I wanted to develop a system that can be useful for others as well. The architecture of the NFV-Inspector platform is microservice based and modular, which means that it is easy to add more features, optimizations and machine learning plugins, and even new data sources to keep the datasets."
“The platform is released under the MIT license, which means that anyone can reuse and edit the code. My goal is to let everyone in this field to know about the platform and encourage them to use it. Hopefully, they will also contribute to the code and develop more plugins over time.”
Developing the NFV-Inspector has resulted in many awards. Michel, in collaboration with Ericsson, recently won the Best Demo Award at the IEEE NFV/SDN 2018 conference in Italy, and that was against competition including demos from Intel, Nokia Bell Labs, etc. The paper “Analysis and Profiling of Virtual Network Functions: the NFV-Inspector” is co-authored by Javid Taheri, Andreas Kassler and Marian Darula (Ericsson).
Michel also won the Student Travel Grant Award at the IEEE CloudNet 2018 conference in Tokyo for the original NFV-Inspector paper, titled “NFV-Inspector: A Systematic Approach to Profile and Analyze Virtual Network Functions.”
The tool is mainly written in Node.js, Python, and some bash scripts for installation. The databases used are MySQL as the system backend and InfluxDB for time series data storage, but one can use Elasticsearch as well, once the plugin is available. It also supports both Kubernetes and OpenStack cloud computing environments. In a couple of weeks, the first technical documentation will be published.
The NFV-Inspector project forms part of the bigger NFV optimizer project, which is a collaboration between Karlstad University, Ericsson AB and VoerEir AB.
About Network Functions Virtualization
Network Functions Virtualization (NFV) is a network architecture concept that uses IT virtualization technologies to virtualize entire classes of network node functions into building blocks that may be connected, or chained together, to create communication services.