AI-powered method reduces next-gen wireless network errors

Researchers at South Korea’s Incheon National University have developed a way to improve next-generation wireless networks, enabling faster, more reliable connections by simplifying how large amounts of signal data are managed and using artificial intelligence (AI) to predict and correct errors. Their findings, published in the journal IEEE Transactions on Wireless Communications, promise significant benefits for applications including high-speed travel, satellite communication and disaster response.

As 5G and 6G networks expand, they promise a future of incredibly fast and reliable wireless connections. A key technology behind this is millimetre wave (mmWave), which uses very high-frequency radio waves to transmit huge amounts of data. To make the most of mmWave, networks use large groups of antennas working together, called massive multiple-input multiple-output (massive MIMO).

However, managing these complex antenna systems is challenging. They require precise information about the wireless environments between the base station (like a cell tower) and your device. This information is called channel state information (CSI). The problem is that these signal conditions change rapidly, especially when moving — in a car, train or even a drone. This rapid change, the ‘channel aging effect’, can cause errors and disrupt your connection.

“To address the rapidly growing data demand in next-generation wireless networks, it is essential to leverage the abundant frequency resource in the mmWave bands,” said Incheon’s Associate Professor Byungju Lee. “In mmWave systems, fast user movement makes this channel aging a real problem.”

Lee and his team have now developed an AI-powered solution, dubbed ‘transformer-assisted parametric CSI feedback’, which focuses on key aspects of the signal — including angles, delays and signal strength — instead of sending all the detailed information. By concentrating on these key parameters, the system significantly reduces the amount of information that needs to be sent back to the base station.

The team leveraged AI, specifically a transformer model, to analyse and predict signal patterns. Unlike older techniques like CNNs, transformers can track both short- and long-term patterns in signal changes, making real-time adjustments even when users are moving quickly. A key aspect of their approach is prioritising the most important information — angles and delays — when sending feedback to the base station. This is because these parameters have the biggest impact on the quality of the connection.

Tests showed that the team’s method significantly reduced errors (over 3.5 dB lower error than conventional methods) and improved data reliability, as measured by bit error rate (BER). The solution was also tested in diverse scenarios, from pedestrians walking at 3 km/h to vehicles moving at 60 km/h, and even high-speed environments like highways. In all cases, the method outperformed traditional approaches.

This breakthrough can provide uninterrupted internet to passengers on high-speed trains, enable seamless communication in remote areas via satellites, and enhance connectivity during disasters when traditional networks might fail. It is also poised to benefit emerging technologies like vehicle-to-everything (V2X) communications and maritime networks.

“Our method ensures precise beamforming, which allows signals to connect seamlessly with devices, even when users are in motion,” Lee said. It should thus set a new benchmark for wireless communication, ensuring the reliability and speed required for next-generation networks.

Image credit: iStock.com/metamorworks