Abstract
The surge in data traffic is challenging for network infrastructure owners coping with stringent service requirements (e.g., high bandwidth, ultralow latency) as well as shrinking per-gigabyte revenues. Network softwarization and edge computing are powerful candidates to mitigate these issues. In parallel, there is an increasing demand for network virtualization and container-based services. In this study, we investigate the management of software defined networking (SDN)-based transport network and edge cloud service orchestration. To this end, we use a machine learning (ML)-based design to manage both transport and edge cloud resources of a mobile network effectively. To generate and use real-world data inside our ML platform, we use the Graphical Network Simulator-3 (GNS3) emulator environment. Our emulation results indicate that almost all of the trained ML models can accurately select the correct edge clouds (ECs) (i.e., with high test accuracy) under the considered two scenarios when transport and EC network parameters are considered in comparison to models trained via only transport or cloud-based parameters. At the end of the article, we also provide an evolved architecture where the proposed ML platform can be embedded in an end-to-end mobile network architecture and H2020 5Growth project's baseline management platform.
Highlights
A distributed mobile network environment with integrated edge and cloud computing capability should support many different services and support each service with its own set of dependencies
As network usage increases or decreases with end-user demands, it is expected that the capacity of the services that are deployed to the different edges of the mobile network react these changes rapidly
In cloud & transport-aware strategy, Decision Tree Classifier (DTC) and Random Forest Classifier (RFC) algorithms can label test dataset with the correct Edge Clouds (ECs) based on majority ruling over classification predictions (i.e. 56% and 94% of test dataset are labelled with EC1 in scenario-I and 100% of test dataset are labelled with EC2 in scenario-II for DTC and RFC algorithms respectively)
Summary
A distributed mobile network environment with integrated edge and cloud computing capability should support many different services and support each service with its own set of dependencies. As network usage increases or decreases with end-user demands, it is expected that the capacity of the services that are deployed to the different edges of the mobile network react these changes rapidly In this case, the provided capacity of the Edge Clouds (ECs) that are distributed in the different parts of the Mobile Network Operator (MNO) infrastructure should be reevaluated. The main drawback of current LCM entities in containerbased domains is their lack of monitoring the status of the KPIs in other segments of the mobile network that can affect their decisions This limited perspective may not be a problem when managing network services in a centralized data-center environment since the physical servers are installed adjacently and all resources are assigned to the mobile network from a single point. Impossible to apply rigid rule-based policies, Machine Learning (ML) guidance need to be exploited
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
