Research Projects

Design of Cellular Architectures for Integrated Terrestrial Non-Terrestrial Networks

We are currently investigating the re-utilization of existing cellular architectures to serve both ground and aerial users reliably. In this context, we are looking at architectures that include the collocation of uptilted and downtilted antennas on the same radio towers. We are utilizing tools from stochastic geometry, reinforcement learning, optimization, and statistical analysis to analyze the coverage probability and achievable rates at UAV-UEs under the collocation strategy, and compare the performance with that under benchmark schemes. Further, the project investigates how different cooperative communication architectures (e.g. cell-free MIMO, C-RAN) can be leveraged to serve both ground and aerial users reliably.

AI-Driven Optimization of Next-Generation MIMO Systems 

The evolving 5G landscape and the ambitious vision for 6G demands breakthroughs in core wireless technologies. At the forefront of 5G and beyond evolution is massive Multiple-Input Multiple-Output (MIMO), a cornerstone technology whose full potential can be unlocked through advanced intelligence. Our group's research direction is keenly focused on applying cutting-edge artificial intelligence, specifically Generative Models and Deep Reinforcement Learning (DRL), to revolutionize key aspects of MIMO system design. Our work primarily delves into enhancing fundamental MIMO functionalities such as channel estimation, beamforming, and codebook design. Through Generative Models, we are exploring novel methods for more accurate and dynamic channel estimation, enabling systems to better understand and adapt to rapidly changing wireless environments by generating realistic channel realizations. Concurrently, Generative Models also aid in optimizing codebook designs by creating context-aware and efficient codebooks that go beyond static pre-definitions. Furthermore, DRL plays a crucial role in enabling MIMO systems to learn and autonomously adapt for optimal beamforming, providing intelligent control to precisely direct signals and mitigate interference. This integrated approach, harnessing the power of these advanced AI paradigms, aims to push the boundaries of massive MIMO system capabilities. Ultimately, our research strives to pave the way for more robust, efficient, and intelligent wireless communication systems for the future.

Integrated Access and Backhaul Design with MIMO Links

Millimeter wave (mmWave) communication enables extremely high data rates for next-generation mobile networks, but its practical deployment remains limited by short-range transmission and high sensitivity to path losses and blockages, necessitating dense networks and expensive infrastructure expansion. Integrated access and backhaul (IAB) provides a flexible, cost-effective alternative, allowing operators to wirelessly deploy additional small base stations (IAB-nodes) and utilize shared spectrum for both access and backhaul. However, shared wirless resources, interference between overlapping access and backhaul links, and harsh propagation conditions especially in dense urban settings can significantly limit the performance. UAV-mounted IAB-nodes offer dynamic positioning and improved line-of-sight potential, making them suited for coverage optimization and adaptability to shifting traffic patterns. On the other hand, ground-based IAB-nodes can be equipped with larger arrays and are not constrained by flight time and energy. Both IAB architectures require advanced beamforming, user association, and resource allocation schemes to address persistent interference and backhaul bottlenecks. This project looks at classical and learning based methods to develop these schemes, and aims to identify the best IAB architecture in different network scenarios.

Theoretical Performance Analysis of Next Generation MIMO

In this line of research, we take a theoretical approach that seeks to establish fundamental performance limits of next generation massive MIMO systems. Our current focus is primarily on continuous array apertures (also referred to as “holographic MIMO"). In contrast to conventional discrete antenna arrays, a continuous aperture array treats the antenna surface as a spatial continuum, enabling extremely high spatial sampling density, hence providing upper bounds in terms of degrees of freedom (DoF) and beamforming gains that can be realized by a surface of given dimensions. In our group, we analyze the DoF and beamforming gains provided by continuous aperture MIMO to guide the design and implementation of large-scale discrete antenna arrays with densely packed antenna elements. We are also interested in developing accurate signal and channel models as well as beamforming solutions for these systems in both narrow and wide band systems.

Page updated

Google Sites

Report abuse