Harnessing data to optimise network services
The role of mobile network operators is rapidly changing. With social interactions becoming the driving force of the internet, networks increasingly act as a medium for data exchange. “Harnessing this data could help us to critically understand how we live our lives, and therefore how we can optimise infrastructure, such as transport and telecommunications,” explains the DAWN4IoE project’s original coordinator Weisi Guo from Cranfield University in the United Kingdom. “This is what smart cities are all about.” A key challenge here is that traditional networks are not social data aware. What this means is that if you are a network operator, then you have to plan your service based on the best available knowledge – how many people in a given area will likely use your service, for example. What you cannot do is adjust in real time your service delivery’s social context. “Communication networks are not very sensitive to short-term fluctuations, such as if there is a train strike and everyone is trying to text,” adds Guo. “This can lead to bad network reception.”
Understanding consumer behaviour to contextualise network use
The DAWN4IoE project, which was coordinated by the University of Warwick and supported by the Marie Skłodowska-Curie Actions programme, sought to address this challenge. “The project was about proactively looking at how we can use data to better understand consumer feelings and behaviour, and to contextualise what is happening,” explains Guo. This data can then be applied to deliver near-real-time information to optimise network services, by highlighting events that are happening, and explaining why they are happening. “For example, is a particular event an anomaly, or is it likely to be repeated?” says Guo. “And how can a network adjust itself in the future to proactively anticipate an event like this?”
Optimising communication networks
The DAWN4IoE consortium brought together universities and start-up businesses, which provided the software and case study data upon which the researchers could work. One of the things that the project team delivered was a data catalogue of issues that might be useful for a network provider to know about. The team also developed AI-based language models, designed to decipher exactly what it is that people are saying. “English can be a fickle language,” notes Guo. “For example, saying ‘the reception is really bad’ could refer to your telephone service, or a wedding reception you are attending.” AI models were used not only to identify keywords, but to understand phrases from a grammatical and usage standpoint. Design algorithms and mobile edge computation were then developed to apply these findings in ways that could optimise network coverage. “For example, a network operator might want to add more spectrum to a specific train station if there has been a cancellation, and the network is aware that people are texting about it,” adds Guo.
Putting real-time data into engineering practice
Guo believes that bringing real-time consumer data into engineering practice, in an automated and proactive way, is critical to delivering better real-time service. “This is about changing the way we think about network operations,” he says. “The philosophy should be about how we can provide capacity and service that is dynamic and personal.” A critical next step will be to integrate this approach into 6G communication standards and actual practices. “Standardisation is critical to make this happen,” he concludes.
Keywords
DAWN4IoE, wireless, internet, smart cities, AI, telecommunications
