More intelligence in the network: Interview with Dr. Michael Barros from TSSG on network management in 5G

Michael Taynnan Barros Post Doc Researcher

Dr. Michael Taynnan Barros
Post Doc Researcher

The emergence of 5G offers new opportunities and challenges for network management. Eurescom message editor-in-chief Milon Gupta asked someone about it who should know: Dr. Michael Barros from the Telecommunications Software & Systems Group (TSSG) at the Waterford Institute of Technology, Ireland, is chairman of the 5G PPP work group on network management, quality of service and network security.

What are the major challenges for network management in 5G?

Dr. Barros: The main network management challenges in 5G are management of virtual network resources, network scalability, quality of service, flexibility, sustainability, context awareness, and security. On top of that, more researchers are willing to implement the concept of Open Management, which means management policies can be accessed by vertical partners who wish to influence network performance. However, this issue should gain more weight when the new network management solutions will use this concept and make it a standard.

What new issues have emerged or become more relevant in 5G?

Dr. Barros: This question can be answered in two ways: one is related to who is designing the network, and the other one relates to who is managing the network. Controlling a large number of both physical and virtual resources requires more automation and intelligence inside the network. However, for those who concentrate on problems in designing the network, the integration of multiple high-frequency hardware and its relationship with network planning is more important.

What impact will network slicing and virtualisation in general have on network management?

Dr. Barros: They will have mainly scalability, quality of service and security impacts. The main objective of changing the network to aggregate network slicing and virtualization is to allow higher degrees of flexibility in order to obtain higher levels of performance. However, that contributes to an increase of network complexity based on the exponential increase of network entities, which makes it nearly impossible to manage the network with current technologies. On the other hand, by increasing the complexity of both resources and management, new network threats can easily arise. The security of the network management solution needs to be very robust.

Which concepts are discussed in the 5G PPP to address network security?

Dr. Barros: Right now, there is a great amount of effort in anomaly detection and how to do it using machine learning. Because the increased complexity of the network is directly related to the increase of its vulnerability, there must be an efficient, robust and safe infrastructure that can detect and manage network attacks. This is closely related to the concept of network resilience, which has been studied extensively in the network engineering community.

How is cognitive management different from autonomic management?

Dr. Barros: Historically, autonomic management has gone as far as developing completely automated solutions in the network. The concept of self-x was introduced, in which network management is expressed through a mixture of approaches, including self-awareness, self-configuration, self-optimization, self-healing, and self-protection. With the advancements of infrastructure technology for accommodating the next generation of networks, the next level of network management has to incorporate the flexible manipulation of network resources and leverage it with the number of users, network traffic, SLAs, and the required system performance. Machine learning has the capability of adapting an entire system based on historic data, which means that in 5G, the network management will monitor key metrics within the network, understand the configurable parameters, and optimally adjust their values for achieving a superior network configuration, indicated through a set of key performance indicators.

What progress has already been achieved in the 5G PPP towards novel network management solutions?

Dr. Barros: Many novel network architectures that incorporate cognitive network management have been proposed inside the 5G PPP phase one projects. In the CogNet project, there exists a main cognitive component that concentrates on adapting the network configuration based on the type of network traffic and also the different scenarios that this network traffic is in. Other projects have slightly different approaches, but also have the capability of network adaptation. For example, the SONATA project concentrates on network resource orchestration, while 5GEx looks at this problem at a higher level, considering multi-domain networks. And there is the SELFNET project, which looks into this problem with a focus on security issues. The common key feature of all these aforementioned projects is the machine-learning-enabled architecture.

By when will mature 5G network management solutions be available and what will be their main innovative features?

Dr. Barros: I suppose that will depend on the progress of innovation in the 5G-PPP projects across all phases. In the next 5G-PPP phases, it is expected to see more demos with novel features that will close the gap between the research that is being done and the potential commercial interest in this research. The benefits of cognitive network management are very clear to researchers. However, it still needs to be demonstrated how beneficial this technology is by adding the interests of telecoms companies in the solution as well. Nevertheless, the vision is that more and more intelligence is added to the network by learning more about the costumer’s behaviours, application, services and adapting the resources accordingly, based on the 5G KPIs.


Published by EURESCOM | Innovation through collaboration