Exploiting advances in technology to drive more sustainable and higher performing telecoms networks is quickly emerging as a key consideration for the modern CTO. For me, artificial intelligence (AI), combined with the potential of O-RAN, present some of the most compelling opportunities.
Mobile telecoms businesses have high ambitions, with more than 60% of the sector by revenue committing itself to Science Based Targets to cut carbon emissions rapidly over the next decade. It’s clear that this requires investments in new technologies such as AI, with lower Scope 3 emissions that use the full capabilities of underlying technology, ever more efficiently.
At the same time the world is not standing still. Data consumption continues its inexorable growth, driven by a stream of new applications promising higher quality of customer experience and new industries looking to capitalise on the capabilities of wide area network connectivity to reduce their own emissions. So, a key question we must ask ourselves is – how does the telecommunications industry reduce its carbon emissions at the same time as handling ever increasing amounts and diversity of data?
O-RAN and AI – An AI use case in telecoms to drive energy efficiency and automation
Fortunately, there are technology trends that will bring a stream of new innovations to support the drive to net zero. O-RAN will be central to this. Its open interfaces will allow different vendors to innovate in their particular niche of the overall network solution, creating thriving markets at the subsystem level.
At the same time, more telecoms companies are looking to extract more value with O-RAN interfaces using machine learning to analyze the data exposed. Research from the Cambridge Consultants 5G testbed (which we expand upon in our O-RAN white paper), found that the use of advanced algorithms can enable a much higher degree of automation, presenting opportunities to maximise service performance across the network and transform service user experience.
By increasing levels of network automation, O-RAN and AI also have the potential to provide more efficient use of infrastructure, enabling network cost reductions and driving more efficient fault finding and network operations.
This means it will be possible to achieve higher quality of experience using far fewer network resources, breaking the pattern that has established itself in the mobile industry of simply overprovisioning capacity in order to achieve the desired quality of experience – a strategy that has shown itself to be wasteful of all kinds of resources and has driven commoditisation of network infrastructure.
Results from our research – transforming user experience
The benefits of O-RAN and advanced machine learning extend to the point where it will be possible to directly optimise the network for the individual user/application performance.
We’ve shown in our research that by analysing the data exposed by the network, actual user experience of specific applications can be inferred without affecting user privacy. Understanding user experience, especially when combined with other data sources, therefore provides an ability to create distinct service offerings from standardised network infrastructure.
The benefits of this are clear. We will be able to maximise quality of experience at individual as well as aggregate levels without impacting privacy. We will maximise utilisation and the efficiency of the network, minimising costs and environmental impact. And we will create a platform for innovation and revenue growth through immersive media.
Achieving these goals requires expertise across a range of competence areas. It is necessary to understand cellular network architectures and how these are to be designed to achieve most efficient coverage and experience for a given investment.
It also requires deep understanding of the network data that is derived through O-RAN interfaces and what this data is revealing about the state of the service and the network. Finally, it also requires deep understanding of the nature of machine learning algorithms and how to develop the most applicable algorithms to achieve the desired outcome.
Pictured above: Based on a video conferencing application, the Cambridge Consultants algorithm (yellow) accurately predicts user experience that is matched to the actual user experience (purple). Taken from ‘O-RAN and AI: ready to transform service performance in the optimised network’, Cambridge Consultants, 2023
The evolution to virtualised software-based implementations
The technologies and architectures underlying network infrastructure are undergoing radical change. We are evolving rapidly from bespoke hardware-centric implementations to virtualised software-based implementations and looking forward to AI native technologies in which many aspects of the technology stack are based on non-deterministic, data-based technologies. Our recent whitepaper – O-RAN and AI: ready to transform service performance in the optimised network – explores this point in more depth and presents research Cambridge Consultants has completed to demonstrate the potential of this approach by using AI to infer the user experience. We describe a system designed to estimate real time user experience and provide guidance for how to deploy other network functions to maximise that experience.
