GBDTC Member wins best paper prize at latest IEEE Symposium

The 24th International Symposium on Communications and Information Technologies (ISCIT 2025) provided a forum for exchanging ideas, discussing research and presenting chip, circuit designs and applications in communications and information technologies. The event was held in Hanio, Vietnam, organised by IEEE CAS Vietnam Chapter and hosted by VNU Information Technology Institute.

The paper presented was titled “Antenna-Independent Deep Learning for Self-Interference Cancellation in in-Band Full-Duplex MIMO Systems”. Full-duplex wireless systems promise a major leap in mobile technology by allowing devices to send and receive signals at the same time. This capability could significantly increase data speeds and improve the efficiency of wireless networks. 

However, there is a major challenge: when signals are transmitted and received simultaneously, the system can interfere with itself, a problem known as self-interference. The issue becomes even more complicated when multiple antennas are involved, which is common in modern high-speed networks.

Traditional approaches to cancelling self-interference struggle to keep up as the number of antennas increases, making them impractical for future networks such as 6G. Solving this problem is crucial to enable 6G mobile networks with both high-speed communication and radar-like sensing capabilities, which could support applications like smart transport, disaster monitoring, and advanced industrial automation.

To address this challenge, we have developed a new method that uses deep learning in the “beam” domain. In this context, a beam corresponds to the signal directed toward each user. Unlike antennas, which can number in the dozens or hundreds; the number of active users (and therefore beams) is much smaller. By focusing on beams rather than antennas, the system can efficiently learn interference patterns and model complex interactions without measuring every channel individually.

The results are impressive. By working in the beam domain, the method keeps computational demands low, scales well with larger antenna arrays, and achieves strong interference cancellation. This innovation brings practical full-duplex wireless networks one step closer, paving the way for the high-speed, sensing-enabled 6G systems of the future.