Researchers at Ulster University in Northern Ireland have leveraged artificial intelligence (AI) to create a novel form of Massive Multiple-Input Multiple-Output (ma-MIMO) technology, which could play a crucial role in shaping the next generation of mobile broadband networks.
The team claim their innovation, named MIMONet, has the potential to make future 6G (IMT-2030) networks faster, smarter, and significantly more energy-efficient.
Currently in the early stages of research and development, 6G standards are not expected to see commercial deployment until around 2030, though some countries may conduct early trials by 2028. As part of global efforts to define the final standards and develop prototype solutions, improving MIMO technology remains a top priority. MIMO, which employs multiple antennas to enhance data transmission and reception, is vital for supporting the massive connectivity demands of future wireless networks, including IoT devices, smart cities, and autonomous systems.
The latest breakthrough, MIMONet, was conceived by PhD researcher Yunis Daha from Ulster’s School of Engineering, under the supervision of Dr Usman Hadi. Published in the journal ‘Telecom’, the research addresses a key challenge in wireless communication: how to accurately detect and process signals amid the noise and interference generated by millions of connected devices simultaneously.
Traditional MIMO detection methods often struggle to balance accuracy and computational feasibility, requiring immense processing power that hampers real-time application. MIMONet aims to overcome these limitations by employing a lightweight deep learning architecture capable of learning to separate and identify signals even in complex, noisy environments. This approach results in a network that is not only faster and more reliable but also consumes less energy, making it suitable for energy-constrained devices and large-scale deployments.
“6G will support critical technologies like autonomous vehicles, remote healthcare, and immersive digital environments. For these to function effectively, networks must process vast amounts of data quickly and reliably. Our research demonstrates how artificial intelligence can provide a practical, scalable, and energy-efficient solution, paving the way for highly capable future networks,” said Hadi.
Initial testing indicates that MIMONet surpasses traditional algorithms and even the most advanced AI detectors developed at Ulster, such as AIDETECT from 2023. It consistently delivers superior accuracy across various network configurations — small, medium, and large — while maintaining low computational demands.
The research primarily targets ultra-reliable low-latency communications (URLLC) applications like driverless cars, real-time medical robotics, and smart city infrastructure. Looking ahead, the team plans to expand MIMONet’s applicability to more complex scenarios, including ultra-massive MIMO systems (such as 64×64 and 128×128 arrays), multi-user MIMO, and frequency-selective channels, which are characteristic of real-world 6G environments.
Further development will explore advanced training techniques like transfer learning and data augmentation to enhance the model’s generalisation capabilities. The researchers also intend to evaluate the system’s robustness against non-Gaussian noise and to investigate hardware-level performance, training complexity, and the potential of federated learning for distributed MIMO systems. These efforts aim to bring AI-driven MIMO technology closer to practical deployment in future 6G networks, supporting the next wave of connected innovation.
