Beyond 5G: research advances radio architecture and AI for the future of telecommunications
The growth of 5G in the world continues, with a forecast from Ericsson to reach 1.5 billion subscriptions by the end of 2023. Faster connectivity speeds, lower latency and increased network reliability are some of its benefits compared to previous network generations. Now, scientists are already investigating the next steps: a new type of intelligent communications system that goes beyond 5G. The EU-funded ARIADNE project was created to advance the research on this new system by integrating a high frequency advanced radio architecture and an artificial intelligence (AI) network processing and management approach. This ambitious goal focuses on a long-term vision of pervasive mobile virtual services instead of local network improvements for the current 5G standard.
Achieving technical progress
“As the available frequency spectrum used in current communication networks is being consumed by a significantly growing number of new and usually very demanding applications and services, there is a need to use higher frequency ranges in an efficient way, by applying novel technologies and AI-aided network resource management, to be able to accommodate all the communication needs of the future,” explains Halid Hrasnica, ARIADNE project coordinator. Because of this need, one of the pillars of the project was to develop new radio technologies for communications using D-Band, the frequency range from 110 to 170 GHz. A real-time, point-to-point, line-of-sight link at D-Band was successfully created, operating at the range of 150 to 160 GHz. It was able to transmit data over 226 metres at a high data rate of 13 Gbps per carrier (32-quadrature amplitude modulation), also ensuring an error-free frequency-division duplex communication. The second pillar of the research was advanced connectivity based on passive reconfigurable intelligence metasurfaces (RIS), an innovative technology that could enhance performance, costs and deployment in the 5G/6G networks. The metasurfaces manipulate electromagnetic waves in a controlled and flexible way, significantly enlarging the network coverage. Researchers of ARIADNE built a RIS hardware demonstrator, as seen in the picture, showcasing anomalously reflecting metasurfaces that direct millimetre band waves towards receivers in challenging environments, overcoming obstacles like walls.
The role of machine learning and AI
To overcome challenges of high-frequency communication and to manage the metasurfaces, ARIADNE explored a series of machine learning and AI techniques. “In most cases, the application of these techniques made it easier to model the communication channel and perform the overall network management and planning,” says Hrasnica. Machine learning and AI were used to advance the use of RIS and to optimise the physical layer of D-Band wireless communications, including localisation services and beamforming for better signal focus and coverage. The technologies also improved medium access protocols and network resource allocation, and assisted in network planning, especially in positioning network nodes and metasurfaces.
Next steps
The project considered general use cases, including connections between rooftop locations and at street-level, and RIS-based connectivity without a direct line of sight. It also looked at dynamic ad hoc connectivity in a moving network, besides vehicle-to-vehicle and vehicle-to-everything connectivity, crucial for applications such as autonomous driving. According to Hrasnica, the ARIADNE concept and its achievements need to be further developed to be integrated in future wireless communication networks. “To achieve this, the project’s industry partners are already taking the related concrete actions.”
Keywords
ARIADNE, AI, 5G, telecommunications, artificial intelligence, machine learning, 6G
