Operators are transforming their networks so that they can achieve new service and business models. They know they cannot do this without automating the processes that will inform, control and manage these networks. That is why we are hearing so much about AI and Machine Learning in the network right now.
But what is the current level of automation in the network? And will existing automation techniques be the foundation for future automation – or are operators looking at a radical step change in technology to automate the software-define, virtualised, personalised, dynamic, network?
At Mobile World Congress. Yves Bellego, Director, Technical & Network Strategy, Orange Group and Paul Gowans, Wireless Solutions Marketing Manager, Viavi, discussed how automation in the network could develop.
Watch now to find what they think about how operators can evolve or deploy increased network automation.
The EE view
To get another angle, after the event we also asked David Salam, EE, Director of Mobility and Analytics at EE, for his views on how automation will evolve in the network. Read to the end for the full view on how EE is thinking about using network analytics to structure its network.
1. Where do you see the requirements for automation in the network in the near term?
Salam: Turning network data in to insights to better serve our customers remains a top priority for EE.
Today’s networks are complex and evolving fast often in a multi technology, multi-vendor environment. Understanding and correlating end to end service performance in a rapidly changing environment is not easy. The short-term step-change requirement in automation for operators is therefore to enable the shift to real-time monitoring.
This challenge will further increase with the introduction of 5G. Network convergence and slicing will mean that it will no longer be possible to monitor and manage customer experience and service quality just based on network metrics, alarms and KPIs. We will need to understand the context in which our customers use our network and services; and optimise the service experience based on customer’s context. Ease of data collection and ability to handle the growing volume and variety of data will be a key enabler for implementation of predictive techniques, and eventually automation.
Open and standardised data interfaces to allow easy, efficient and real-time collection of data at scale is going to be important in the 5G eco-system.
2. Yves talks about his teams requiring the need to retain control. Do you agree that this is a key concern? How much control do you need over the automated process, how will you structure that human control?
Salam: Automation in networks will depend on three things- 1) real time knowledge of current state, 2) analytics to understand the dynamics, causes and effects, and 3) ability to automate and orchestrate. Hence a lot of things will need to fit together before we can consider free-running automated network actions. It is likely to be a Technology evolution rather than a step change.
I do, however, believe that the operating model does require a step change. Along with access to real-time enterprise and network data, companies need to think about Network Data Science much more as a defined skillset. We need the resources with the right mix of network knowledge and expertise, combined with data science skills to leverage learning and value.
Such capabilities will allow result in a movement away from monitoring platforms to a future where we are monitoring the performance of the algorithms, allowing operators to act more intelligently.
3. Evolution or step change? Do you agree that automation in the network will expand by building on the use cases we have to date – and that automated, predictive algorithms today will be the foundation for the end to end, programmable networks of the future. Do you agree also with Yves that what is required now is industrialised solutions that can operate at scale. Or are we looking at a real step change in technology?
Salam: For us it is more about an evolution as we evolve through predictive analytics (what will happen), to prescriptive analytics (what should I do) and then to cognitive analytics (autonomous control). This may sound simple, but in the future we will need to predict where customers are going next, what service they will want next, how best to deliver that service whilst continuously re-assessing for change. That sounds like quite a step change to me.
This will require hyper-scale big data processing with open source and open standard based data interfaces operating in real-time. And this automation will push towards the edge to optimise seamlessly with and maximise the access network.
This video was produced in association with TMN partner Viavi Solutions.
More information on Viavi Solutions.