Framework for Big data analytics in 5G

C-RAN functional split in 5G

September 6, 2018

Game theory in wireless networks

September 6, 2018

Published by Armin Okić at September 6, 2018

Tags

The huge increase of generated data in the network it is also bringing a big increase of the metadata used to track what is happening in the network. This kind of data can be fully characterized as a big data as it is fulfilling all 4Vs of Big data, Volume, Variety, Velocity, and Variability.

Once Big data is collected and stored it is necessary to process it, which for results has extraction of knowledge which can be used later to make actions in the network, like network optimization in the different levels of the network, enabling specific services to the users, like video caching techniques closer to the user, or to predict cells behavior in order to pre-initialize resources in the network.

The proposed framework for big data analytics in the mobile networks (Figure 1) consists of four following blocks [1]:

Big Data collection – Data can be collected from UEs (from user applications or control signaling), the RAN (cell-level data from eNodeBs), the core network (user bearers/services) and from Internet service providers.
Storage management – Needs to have a scalable capacity as well as scalable performance. Pre-processing of data is necessary before storage, to remove unusable, incomplete and redundant data.
Data Analytics – Collected data is multi-sourced, heterogeneous, real-time and voluminous, which leads to different challenges in data processing and the extraction of actionable knowledge to be used for the design of adaptive schemes for network optimization.
Network optimization – Optimization results are implemented by the control functions in the RAN. It is possible to perform user-level optimization or the prediction of traffic variations in a local area or over network coverage. This approach can be used to optimize the resource management in heterogeneous networks by predicting the users’ dynamics and requirements or predicting the intercell interference. It also can be used for optimal localization of cache servers or for Quality of Experience modeling.

Figure 1 Big Data-driven network optimization framework [1]

All four blocks can be scopes of the future research because every of the listed components can be observed and optimized separately taking certain assumptions from the other blocks. From the other side, more complex analysis can consider all the blocks jointly focusing on the specific scenario.

For sure that Big data analytics will be one of the key elements in 5G and beyond networks which will enable network operators to better understand the behavior of whole network as well as of the partial components, to predict events and outcomes, or to monitor rapid fluctuations. The possibilities of Big data analytics are uncountable and challenges in this scope are numerous.

References:

[1] Zheng, Kan, et al. “Big data-driven optimization for mobile networks toward 5G.” IEEE network 30.1 (2016): 44-51.

Armin Okić

He received his BSc and MSc degrees from the University of Sarajevo, Faculty of Electrical Engineering Sarajevo, Department of Telecommunications. He was working in BS Telecom Solutions company in which he cooperated with network operators maintaining different telecommunication’s platforms. Currently, he is a Ph.D. student at the Polytechnic University of Milan, Department of Electronics, Information and Bioengineering in the area of Telecommunications and Early Stage Researcher at the European Marie Curie Innovative Training Network in the SPOTLIGHT project. His research area is covering Resource Management in Heterogeneous networks as well as an appliance of Machine learning and Big data analytics in the future 5G networks and beyond.