b. Generative AI/LLM for Communications

Large language models (LLMs) enable both high-fidelity generation (e.g., by generative diffusion models (DMs)) and intelligent sequence-to-sequence generation (e.g., by Transformer). The remarkable generative capability of LLMs is bringing new opportunities to enhance the designs of mobile networks and next-generation multiple-antenna systems. (1) As a popular generative AI technique, DMs are gaining attraction in wireless communications for their ability to handle complex data distributions and optimize system performance in challenging environments in beamforming design and channel estimation. DMs excel in scenarios with strong interference and noise, making them superior to traditional AI methods and conventional optimization techniques. We propose a unified framework for training DMs with supervised, unsupervised and reinforcement manners to multiple antenna communications. The DMs can be applied to both decision-making tasks (e.g., beamforming optimization) and generation tasks (e.g., channel modeling and estimation), offering robust solutions for tasks like channel estimation, feedback, and interference management. The DM-based approach addresses challenges such as imperfect channels and task-oriented communications, demonstrating significant performance improvements in simulations. The potential applications of DMs in communications are vast, ranging from enhancing channel estimation and feedback mechanisms to optimizing resource allocation and interference management. The integration of DMs with emerging technologies like reconfigurable intelligent surfaces (RIS) and next-generation multiple access (NGMA) holds promise for future wireless networks, particularly in 6G. (2) A Transformer-based beamforming design is proposed for pinching-antenna systems (PASS), which efficiently captures inter-user/inter-antenna dependence and bypasses conventional optimization.

c. Mobile Edge Generation (MEG)

Future wireless communications are evolving from information delivery toward generative information reconstruction at the network edge. This research develops mobile edge generation frameworks that leverage generative AI models and distributed knowledge across wireless devices and edge infrastructure to reconstruct rich information from compact transmitted representations. Instead of delivering complete source data over bandwidth-limited wireless links, the proposed paradigm transmits task-relevant generative cues, such as semantic features, tokens, or latent representations, while exploiting receiver-side generative models to recover or synthesize the desired content. We investigate the joint design of generative representations, wireless transmission, and receiver-side generation, enabling source, channel, and generative models to be optimized in an integrated manner. By shifting part of the communication burden from transmission to generation, mobile edge generation aims to substantially reduce communication overhead while supporting efficient, adaptive, and intelligence-native wireless services.

a. Integrated Sensing and Communication (ISAC)

Integrated Sensing and Communication (ISAC) is a promising technique for the next generation wireless network. Unlike the past five generations of wireless network where there is only one envisaged function, i.e., wireless communication, the next generation wireless network expects a paradigm shift to support multiple functions including the communication, sensing, control, and computation to meet the demand of the fusion techniques such as unmanned aerial vehicles (UAVs), autonomous driving, virtual and augmented reality, Internet of vehicles, and smart cities. In ISAC, the sensing and communication are not regarded as independent functions but are jointly designed toward the goals and mutual benefits of both, which introduce two gains to the wireless network, namely integration gain and coordination gain. Integration gain is from the dual usage of the signal, spectrum, and device in the ISAC for both functions, thereby reducing the cost. Coordination gain is from the deeper level integration in ISAC to achieve mutual assistance between the two functions, such as the communication-assisted sensing and the sensing-assisted communication. This project will explore the interplay between ISAC and non-orthogonal multiple access (NOMA) in the following aspect: 1) use NOMA to boost the load capability of ISAC [Paper]; 2) follow the concept of NOMA to manage or exploit the interference in ISAC; 3) investigate the potential of NOMA as the enabler of more sophisticated ISAC frameworks [Paper].

b. UAV Communications and Sensing

With the rapid development of the low-altitude economy, there is an increasing demand for intelligent, scalable, and spectrum-efficient UAV enabled low-altitude networks to support emerging applications such as aerial logistics, urban air mobility, and real-time sensing. We propose a novel framework for UAV networks that leverages non-orthogonal multiple access (NOMA) and integrated sensing and communication (ISAC) to be critical enablers for low-altitude networks. To evaluate the real-world potential of this framework, we begin by evaluating the performance of NOMA-enabled UAV networks in dense urban environments. Using stochastic geometry to model the spatial distribution of UAVs and ground users, we analyze the system's connectivity and interference characteristics, which directly support applications such as aerial base stations as a component of urban digital infrastructure [Paper] [Paper] [Paper] [Paper] [Paper] [Paper]. We then investigate the joint trajectory planning and power allocation for UAVs serving fixed-position ground users, such as those at parcel drop-off stations or critical infrastructure sites. Based on a simplified two-dimensional (2D) model with UAVs flying at a fixed altitude, this study provides practical insights into real-time decision-making for intelligent UAVs’ deployments [Paper] [Paper] [Paper]. As a further step, we proposed several machine learning techniques to against the challengs posed by dynamic environments mobile users, and three dimensions (3D) trajectories. [Paper] [Paper] [Paper] [Paper] [Paper]. Lastly, we introduce sensing capabilities into UAV networks, allowing UAVs to simultaneously perform sensing and communication tasks using shared spectral and hardware resources [Paper] [Paper] [Paper]. This integration is crucial for real-time environmental monitoring, cargo delivery, and intelligent infrastructure inspection. These studies demonstrate how advanced communication, sensing, and optimization technologies can empower UAV systems to become a foundational infrastructure in the low-altitude economy.

a. Pinching-Antenna Systems (PASS)

This project investigates the novel Pinching Antenna Systems (PASS) and advocates simple yet efficient wireless communications over the “last meter”. A PASS consists of a specialized transmission line known as a dielectric waveguide. Similar to leaky-wave antennas, electromagnetic waves within the waveguide are emitted by pinching small separate dielectric particles at specific points along its length. These dielectrics are typically attached to the tips of plastic clips or clothespins, forming structures known as pinching antennas (PAs). By leveraging the waveguide’s ability to transmit or receive radio waves at any pinched point, PASS enables wireless communication precisely where needed via the pinching beamforming. The full details of the proposed concept can be found in the presentation slides [Slides].

b. Continuous Aperture Arrays for 6G

The holy grail of multiple-antenna technology is to increase the number of antenna elements, thereby enhancing spatial degrees-of-freedom (DoFs) and achieving significant capacity gains in wireless networks. In recent years, several novel antenna array architectures have emerged, including reconfigurable intelligent surfaces, holographic MIMO, and dynamic metasurface antennas. Despite their diverse architectures, these systems share a common evolutionary trend: larger aperture sizes, denser antenna configurations, higher operating frequencies, and more flexible structures. Naturally, this evolutionary trend leads to the creation of an (approximately) continuous electromagnetic (EM) aperture, i.e., a continuous aperture array (CAPA). Generally speaking, a CAPA operates as a single large aperture antenna with a continuous current distribution, which represents a significant shift from conventional spatially discrete arrays (SPDAs). It comprises a (virtually) infinite number of radiating elements coupled with electronic circuits and fed by a limited number of radio-frequency chains. A CAPA serves as an ideal transmission aperture by enabling continuous control over the amplitude and phase of the current over the surface of the CAPA where RF signals are modulated. It supports fully analog beamforming across the entire communication bandwidth, allowing precise alignment of the radiation pattern.

c. Near-Field Communication/Sensing (NFC/NISE)

As the fifth-generation (5G) wireless network continues to be commercialized, research and development efforts are already underway on a global scale to investigate the possibilities for the sixth-generation (6G) wireless network. Compared to the previous generations of wireless networks, 6G is expected to be data-driven, instantaneous, ubiquitous, and intelligent, thereby facilitating new applications and services, such as extended reality (XR), holographic communication, pervasive intelligence, digital twin, and Metaverse. Therefore, 6G has significantly higher performance targets compared to the past generations, such as a 100 times higher peak data rate of at least 1 terabit per second (Tb/s), an air interface latency of 0.01~0.1 milliseconds, and a 10 times larger connectivity density of up to ten million devices per square kilometer. To reach these stringent targets, 6G is expected to integrate the three technical trends: 1) extremely large-scale antenna arrays, 2) tremendously high frequencies, and 3) new types of antennas. The significant increase in the size of antenna arrays, extremely high frequencies, and the emerging new metamaterial-based antennas cause a paradigm shift for the EM characteristics in 6G. Generally, the EM field radiated from antennas can be divided into two regions: near-field region and far-field region. The EM waves in these regions exhibit different propagation properties. Specifically, the wavefront of EM waves in the far-field region can be reasonably well approximated as being planar. Conversely, more complex wave models, such as spherical waves, are required to accurately depict propagation in the near-field region. From the first-generation (1G) to 5G wireless networks, the near-field was generally limited to a few meters or even centimeters because of the low-dimensional antenna arrays and low frequencies. Therefore, communication systems could be efficiently designed based on far-field approximation. However, given the large aperture of ELAAs and tremendously high frequencies, 6G networks exhibit a large near-field region on the order of hundreds of meters. Furthermore, for the investigation of metamaterial-based continuous-aperture antennas, the conventional far-field plane wave assumption is also not suitable. Therefore, in 6G networks, the near-field region is not negligible, which motivates the investigation of the new near-field communication (NFC).

d. Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RISs)

The novel concept of simultaneously transmitting and reflecting (STAR) reconfigurable intelligent surfaces (RISs) is proposed in this project, where the incident wireless signal is divided into transmitted and reflected signals passing into both sides of the space surrounding the surface, thus facilitating a full-space manipulation of signal propagation. In this project, we will investigate (1) the fundamental electromagnetic and hardware models [Paper1][Paper2]; (2) the basic communication signal model and operating protocols [Paper1][Paper4]; (3) the corresponding fundamental performance analysis, beamforming design, and channel estimation methods [Paper4][Paper5][Paper6][Paper7]; (4) the interplay between STAR-RIS and other promising techniques and advanced tools [Paper7]. The full details of the proposed concept can be found in the presentation slides [Slides].

e. Next Generation Multiple Access (NGMA)

Next generation wireless networks are expected to achieve significantly high capacity, extremely low latency, ultra high reliability, as well as massive and ubiquitous connectivity for supporting diverse disruptive applications. One of the most fundamental issues is the design of sophisticated multiple access (MA) techniques for the next generation wireless networks, namely next generation multiple access (NGMA). The key concept of NGMA is to enable a tremendous number of users/devices to be efficiently, flexibly, and intelligently connected with the network over the given wireless radio resources for not only satisfying stringent communication requirements but also realizing heterogenous functions. This project will explore the development of NGMA by exploiting NOMA and multiple antenna techniques with the aid of convex optimization and machine learning methods. In particular, a unified framework for NGMA is proposed based on multi-antenna transmission and NOMA, including both downlink and uplink multi-user communication scenarios. The aim of the proposed unified NGMA framework is to fully reap the benefits of multiple antenna and NOMA techniques and to enable highly flexible and customized communication services for future wireless networks. This opens up numerous research opportunities and directions.

The details of the proposed unified NGMA framework can be founded in the survey/tutorial paper. [Paper]