The University of Surrey unveiled an artificial intelligence (AI) model that they say could revolutionise the way telecommunications networks operate.

According to them, the AI model could save up to 76% in network resources when compared to the Open Radio Access Network (O-RAN) system.

The findings, detailed in a study called Deep Reinforcement Learning for Robust VNF Reconfigurations in O-RAN, published in IEEE (Institute of Electrical and Electronics Engineers), demonstrated that mathematically modelling networks with AI can optimise the allocation of computing power throughout the network. 

This means that above improving cost efficiency, the new model can also reduce energy consumption. 

“Our model shows that by using AI, telecommunications providers could use their bandwidth far more efficiently, with only a small additional computational cost,” said Esmaeil Amiri, who led the research at the University of Surrey.

“The model could be adapted for other scenarios—like helping drones conserve their batteries or even reducing latency in remote surgery.”

According to the researchers, this enhancement in network bandwidth capacity is available with minimal computational overhead, unlike other O-RAN systems. 

O-RAN systems have transformed the telecommunications industry by allowing providers to reallocate computing power across their network in response to fluctuating demand without having to modify their hardware at base stations.

However, current technology struggles to swiftly adapt to rapid changes in network demand, limiting its effectiveness.

Recommended reading

• New Commercial Model to Make the Case for Scottish Rural Connectivity
• Edinburgh Telecommunications Provider Launches London Base
• UofG Signs Up to Next-gen Telecommunications Project

“This solution can dynamically adapt to changes in demand, yet with significantly reduced necessity of reconfiguring the network,” said Ning Wang, co-author of the study and Professor in Networks at the University of Surrey.

“This could make our communications networks more robust and more efficient—but the underlying model could have even broader application.”

The proposed scheme is now slated for further testing in the HiperRAN Project. In collaboration with industry partners, the University of Surrey team aims to bring this technology closer to being ready for a wide-scale rollout.

“This research could be implemented easily, helping shape the next generation of telecommunications networks,” said Dr Mohammad Shojafar, another co-author of the study.