EU research project to optimise 5G slicing
March 3, 2020
Ranplan Wireless has been selected to drive new research into how AI-based data analytics can help to optimise sliced 5G mobile networks.
Project CORRELATION aims to profile mobile users and their consumption of 5G services so that networks can adapt dynamically to meet instant requests for bandwidth and services. By accurately profiling individual users and understanding their data consumption habits in real-time to predict traffic patterns, the research will help operators optimise network slicing performance and ultimately the mobile user experience.
The two-year project will:
- Investigate the correlation of traffic among different services by analysing the usage and profiles of user groups as well as discover the underlying factors behind the results
- Investigate whether service level traffic prediction accuracy can be improved.
“Research on the characterisation and prediction of service-level mobile traffic is still in its infancy, largely because the methodology, especially for 5G, has not been studied adequately,” said Prof. Jie Zhang, Founder and Chairman at Ranplan Group. “This is the same for understanding traffic correlation among different services, while inter-service correlations have not been fully-exploited in service-level mobile traffic predictions. Understanding the correlations between services will allow us to significantly reduce the time used to train deep neural networks and open up the opportunity for dynamic 5G network slicing, which heavily depends on characterising and predicting the spatial-temporal traffic patterns for individual users – both humans and things – in real-time.”
“The CORRELATION project will complement the BANYAN project, a €1.4 million H2020 project led by Ranplan, with other beneficiary/partners of CNR, Italy, Orange Labs, France, and the University of Cambridge, UK. These projects are well-aligned with Ranplan’s strategy of adapting its world leading radio network planning and optimisation technologies to continuous 4G/5G network operation automation,” added Zhang.