While telecommunications companies (and their customers) are understandably excited by the roll-out of 5G, this next generation of networks won’t deliver on its full potential without the help of artificial intelligence and machine learning. But how can networks which (in the majority) are at best hybrid, at worst silo-bound monoliths, handle the demands of AI and ML? This is where self-organizing networks offer a way forward for those telcos willing to embrace digital transformation.

The 5G networks that will become more commonplace as the year 2020 progress will require an agile control layer and a flexible, self-organizing transport network. This means focus on four key areas.

First, operators need to take a realistic approach to what they can achieve with AI in combination with 5G. There are already efforts to bring more advanced machine learning into play from the assurance perspective – which will be a requirement of self-organizing networks.

AI/ML developments on their own won’t do the whole trick. Operators need to be looking at how the constituent parts of their virtual networks operate and cooperate, and build-in as much flexibility as possible when planning transformation projects. This means allowing for ML to play a significant role in “programmatically configurable” processes across as many network layers as possible. Additionally, the ideal self-organizing 5G network will have multiple relations between transport and data plane technologies.

It should be emphasized that tomorrow’s self-organizing 5G networks will generate and receive data in volumes that will make even today’s IoT-related information seem insignificant. So, operators should be seeking to deploy ML for big data management and analysis in a manner that will let the algorithms learn and become more precise in reaching optimized conclusions about network organization.

accessibility, browsing, 5g

With so much hype about artificial intelligence and machine learning, it would be easy to lose sight of the human element in networks. After all, if the goal is to develop self-organizing 5G networks at this level, it might be assumed that input from technicians would be irrelevant. But nothing could be further from the truth.

It’s long been understood that, in all aspects of IT, the end results are only as good as the instructions and information provided. In this way, telcos need to keep in mind the infrastructure requirements of self-organizing networks, which is something that can only be guided by human intervention. Machines, ultimately, are exceptionally good (and fast) with logic – while humans continue to surpass them in terms of intuition.

Today, we’re certainly not where we need to be with self-organizing 5G networks. But we will be, and operators need to bear that in mind. This will mean embarking on high-level digital transformation projects, while at the same time investigating innovative business cases that will maximize the return on 5G network investment. They can start by making incremental changes to the networks, implementing artificial intelligence and machine learning incrementally, never losing sight of the goal of the fully agile, self-organizing 5G networks of tomorrow.

More about big data analytics for telecom: https://www.comarch.com/telecommunications/oss-bss-data-analytics/