How will AI and Machine Learning enable operators to better manage their networks?
AI and machine learning (ML) will help operators do more than just compete; it will give them the agility they need to win back enterprise customers, many of whom have turned to digital-native providers for cloud-based services.
AI and machine learning will be key to bringing the automation required to manage and optimize networks of the future—the one billion (and growing) number of IoT devices already connected to mobile networks are simply too much for management systems to handle. It is impossible for humans, or manual systems, to monitor and assure these devices, analyse the data they generate, whilst also managing the infrastructure and processes required to optimise network performance and monetisation. This is where AI and ML come in. AI, deep learning and machine learning, are all becoming increasingly important for CSPs wanting to collect, analyse and leverage the data generated by networks and devices.
We’ve already seen the use of AI and ML in managing and improving CSP customer experience, and now we’re seeing these technologies trickle throughout CSP operations. In fact, we’re now seeing the emergence of AIOps, whereby AI is used to automate CSP operations; for example, Etisalat announced that it has teamed up with Microsoft to build a digital platform infused with automation and AI providing a simplified network architecture and operations empowering its subscribers and improving customer experience. We have also recently seen MTN Benin sign an agreement with Ericsson which will bring about great efficiencies and automation and intelligence to their network and will jointly create a world of predictive operations with a focus on customer experience, network quality, performance and automation.
Ultimately, AI, ML and automation will all go towards building the networks of the future—closed-loop, self-healing autonomous networks that can be managed, optimised and configured with no human intervention.
Is it possible to "over automate" a network?
Having good governance around your AI and automation projects is critical to ensure that they operate safely and efficiently. Over automation implies you remove people completely from the networks, but that is not what AI and automation is designed to do. Instead, it is designed to help people do their jobs more efficiently and increase the capability of the networks and the services that operators can provide. Certainly in our life time, networks will not run without human supervision and that is why at TM Forum we have a collaboration group focused on AI governance, ensuring that as we deploy AI and increase automation, we can do so safely. We are developing operational checklists across the lifecycle as well as data sheets ensuring we understand and record everything about the AI models that businesses are using, as well as developing APIs and reference implementations so that we can explain, understand and control AI and autonomous networks in real time. As long as the correct governance procedures are in place, you cannot over automate a network.
You can read our full iterview with Aaron Boasman-Patel, vice president, AI and customer experience at TM Forum, exclusively in this month's issue of CommsMEA.