Machine Learning: a revolutionary step for radio network optimization
Networks that can think and act for themselves power the next step in digital transformation
Advanced self-organizing capabilities within the network will be critical as part of the necessary shift towards autonomous operations. This evolution will push the earlier generations of SON to their limits.
Operators have achieved promising results with their existing automation tools for network optimization, however, the complexity of 5G has raised the bar much higher. 5G networks co-exist with earlier generations from 2G and 3G to 4G, amounting to thousands of base stations and millions of cells. In addition, 5G leverages multiple frequency bands and dynamic technologies such as massive MIMO and beamforming as well as other advanced software features for efficiency enhancements.
To put it simply, the complexity is too much for humans alone to manage.
Beyond the need for human intervention
While SON is all about automation, we can take it to a whole new level with cognitive capabilities.
The reality of today is that some operators have chosen to run SON functions in a way that requires human intervention by experienced optimization engineers. It can be hard to find such experts who fully understand all the dimensions of radio networks. Think of tasks like setting the network KPIs for performance, analyzing the conditions at cell sites and network issues, implementing corrections and parameter updates, and finally determining if the actions were effective.
This need for human judgment at different stages of network optimization could create bottlenecks that slow down the launch of new services. Human judgment is also prone to error, which can cause serious impact on the network performance and related customer experience in the dynamic and highly complex 5G environment.
We can do better! Moving beyond manual-driven automation and letting autonomous operations take the lead ensures efficient operations, accuracy in network optimization and faster time-to-market.
Nokia Cognitive SON is ready for the future of automation
At Nokia, we have seen that the era of automation is here and we are taking it to the next level. We know that trust in automation is a must. Nokia Cognitive SON represents the next level of automation and a transformational approach to using Machine Learning.
With the advanced cognitive abilities, our SON can take over the functions that previously required human intervention. All the operator needs to do is to set the objectives, and Cognitive SON handles the rest – powered by Machine Learning.
It will analyze and learn to understand the network context for each cell and identify the issues impacting performance. It will apply the right actions autonomously and evaluate their efficiency – much faster than ever possible for a human engineer. This type of fully autonomous, intelligent, closed-loop system will be essential for successful network operations in the 5G era and beyond.
However, we must be humble during the transformation as it takes time to build trust in fully autonomous network optimization. It is clear that a human-driven approach belongs to the past, having been replaced to various degrees by existing SON implementations. Now, we are entering the era of holistic automation. To make the transition easier, Cognitive SON provides extensive visibility and open controls for experts to monitor and influence its operation on the journey to adopting a fully autonomous system.
Let me conclude with some truly impressive results from our recent live trials with Tier 1 operators. When comparing Cognitive SON to manual operations, it was able to:
- optimize RAN 160 times faster
- balance cell load 95% faster
Cognitive SON, powered by Machine Learning, makes the future of automation a reality in today’s radio networks. Get ready for this exciting transformation!
