An Application Awareness Framework Based on SDN and Machine Learning: Defining the Roadmap and Challenges

Software Defined Networking (SDN) has presented a unique networking paradigm to develop network innovations and address the issues discovered by distributed network architectures. This paper aims to address challenges involved in introducing application aware network framework using an arbitrary per...

Full description

Saved in:
Bibliographic Details
Main Authors: Jahromi, Hamed Z., Delaney, Declan T.
Format: Conference Proceeding
Language:English
Subjects:
Application awareness
SDN
application metric
Computer architecture
Knowledge engineering
Machine learning
Measurement
Monitoring
qoE
KPI
machine learning
Quality of experience
Quality of service
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Software Defined Networking (SDN) has presented a unique networking paradigm to develop network innovations and address the issues discovered by distributed network architectures. This paper aims to address challenges involved in introducing application aware network framework using an arbitrary performance metric and preparing network information for machine learning (ML) analysis. The main goal of this is to automate application specific resource allocation and orchestration.A key facet of the framework is utilizing an application feedback interface to the SDN's Northbound Interface which can receive, during runtime, an arbitrary performance metric from an application and characterizes this in accordance with the network path features, thus making it unique among the literature. The metric describes how well the application is performing in its performance goals. The framework analyses and translates this metric into network features allowing a network manager to calculate the effect of network decisions on application goals.To achieve this the framework utilizes centralized SDN architecture, collects and prepares network information that is better predisposed for ML analysis.
ISSN:2472-8489
DOI:10.1109/ICCSN.2018.8488328