Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A comparative study for Time Series Forecasting within software 5G networks

: Chakrabourty, Pousaly; Corici, Marius; Magedanz, Thomas

Postprint urn:nbn:de:0011-n-6305225 (426 KByte PDF)
MD5 Fingerprint: 0a71edee4d8adf7d1ccdc32b1bbe5174
© IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Erstellt am: 2.3.2021

Wysocki, Tadeusz A. (Ed.) ; Institute of Electrical and Electronics Engineers -IEEE-; IEEE Communications Society:
14th International Conference on Signal Processing and Communication Systems, ICSPCS 2020. Proceedings : December 14-16, 2020, virtual conference
Piscataway, NJ: IEEE, 2020
ISBN: 978-1-7281-9973-3
ISBN: 978-1-7281-9971-9
ISBN: 978-1-7281-9972-6
7 S.
International Conference on Signal Processing and Communication Systems (ICSPCS) <14, 2020, Online>
Konferenzbeitrag, Elektronische Publikation
Fraunhofer FOKUS ()
machine learning; time series forecasting; 3GPP 5G Core; Open5GCore toolkit; failure prediction

5G has a very flexible network architecture due to virtualization and will come with various customisations based on different use cases. 5G also promises to provide intelligent networks with high bandwidth and low latency. One of the tradeoffs for this is the complexity of network monitoring and resource management of 5G; making availability, reliability and performance a challenge. The adoption of Software Defined Networking (SDN) and Network Function Virtualization (NFV) concepts ensure availability of network data and flexibility in architectural decisions for 5G. Because of the availability of data and advanced computing capabilities usage of ML (Machine Learning)/Artificial Intelligence (AI) can be envisaged in the control and management of 5G networks by predicting the load on the network. This article proposes a solution to integrate time-series based predictive analytics with 5G Core and shows a comparative study between two Time Series Forecasting Models-AutoRegressive Integrated Moving Average (ARIMA) and Face-book Prophet. Fraunhofer FOKUS Open5GCore is used as the reference 5G testbed toolkit for validating the proposal.