Knowledge Defined Networking: State of the Art and Research Challenges

EasyChair Preprint no. 8514

Abstract

Recently, combining a software-defined network (SDN) with a machine learning (ML) technique to improve network performance has become an even more attractive topic for the research community, thanks to the availability of processing and storage capabilities. Knowledge-Defined Networking (KDN) is a new paradigm that combines SDN, Network Telemetry, network analysis (NA) and ML techniques. SDN provides a complete view of the network through a logically centralized controller. This allows the collection of a large amount of data describing the network behaviour. Network telemetry gathers data generated in the network, such as SNMP, sFlow [4], NetFlow [5] and Syslog data. NA analyses and structures the collected data. Finally, ML algorithms process the structured data, develop knowledge, and exploit this to control the network. This paper presents a brief overview of the KDN paradigm; also, it describes the concepts covered by its architecture and reports on several works done in this field.

Keyphrases: Knowledge-Defined Networking, machine learning, network analytics, network telemetry, SDN

@Booklet{EasyChair:8514,
  author = {Adnane Mzibri},
  title = {Knowledge Defined Networking: State of the Art and Research Challenges},
  howpublished = {EasyChair Preprint no. 8514},

  year = {EasyChair, 2022}}