Machine Learning and 5G

Session: 246

19 Mar 2018 - 13:15 to 14:00

#WSIS

Report

[Read more session reports from the WSIS Forum 2018]

Ms Tatiana Kurakova, ITU-T SG13 counsellor and the Telecommunications Standardisation Bureau (TSB) of the International Telecommunication Union Telecommunication Standardization Sector (ITU-T) started the session by introducing the speaker, Prof. Stawomir Stanczak.

Prof. Stanczak, chairman, ITU-T Focus Group on Machine Learning for Future Networks including 5G (ITU-T FG-ML5G), and head, department of wireless communications at the Heinrich Hertz Institute (HHI), delivered his presentation remotely. He started by explaining the areas of work done by the HHI in data processing and data networking.

According to Prof. Stanczak, the goal of providing mobile connectivity is being achieved. Currently, there is need to connect people and things. Tactile Internet - the next evolution of the Internet of Things (IoT), encompassing human-to-machine and machine-to-machine real-time interactive systems - represents a paradigm shift.

The Tactile Internet entails new key performance indicators (KPIs), such as interaction speed and functional safety and security. In the case of connected cars via wireless infrastructure for better performance (for example, platooning) it would be possible to decrease the distance between cars, optimising traffic flow. The tactile Internet and 5G will play a key role in certain situations, such as driving in dangerous conditions, and augmented driving.

Prof. Stanczak highlighted that there are still technical challenges to realise the potential of machine learning and 5G. These include challenges in performance: many applications require high experienced data rates, zero latency, higher density of connected devices, higher traffic volume density, and higher mobility. In addition there are challenges in efficiency, such as achieving higher spectrum efficiency, lower cost per bit, and higher energy efficiency.

Prof. Stanczak made a comparison between 4G and 5G and presented requirements to achieve spectral efficiency (such as massive Multiple-input multiple-output or MIMO, cell density (small cells) and bandwidth (millimetre wave technologies). This needs to be considered in a new 5G architecture.  

Machine learning in also an important topic in the field of communications. In the context of communication networks, in particular wireless networks, machine learning has been applied to routing and many other aspects of network design. Artificial intelligence (AI) and machine learning (ML) will be used to make self-organising networks feasible. ML can be used for construction of capacity maps, which show what the capacity conditions along a road are, to the benefit of connected cars.
Machines can learn about data traffic and make robust predictions. ML can also be used for energy-saving optimisation. If the volume of traffic is low it would be possible to switch off some traffic elements for optimising energy consumption.

Prof. Stanczak also commented on the work of the ITU-T focus group on Machine Learning for Future Networks including 5G, and its three working groups: Use case, service and requirements; data formats and ML technologies; and ML-aware network architecture.

 

By Marilia Maciel 

Organisers

ITU
 

The GIP Digital Watch observatory is provided by

in partnership with

and members of the GIP Steering Committee



 

GIP Digital Watch is operated by

Scroll to Top