Research Projects

The work aims to leverage the emerging cutting-edge Dynamic Metasurface Antenna (DMA) technologies and advanced RF-Digital hybrid signal processing architectures for next generation wireless systems, operating in the sub-THz and THz bands. Together with advanced machine learning (ML) models, this project aims to develop a cohesive framework that unlocks the full potential of wireless communication via integrated design. The focus of this proposal is on harnessing the unique capabilities of DMA-based multiple antenna architectures, together with state-of-the-art waveforms such as Orthogonal Time Frequency Space (OTFS), to overcome the key challenges that plague high speed wireless communication, such as severe path loss, molecular absorption, and hardware limitations, thereby enabling ultra-high data rate, reliable, and intelligent communication in future wireless networks. In order to fully exploit the high-speed communication opportunities offered by the wideband channels, it is critical to explore advanced hardware solutions capable of operating in such massive bandwidth systems. In this context, the recently proposed DMA technology offers a promising solution due to its low power consumption, reconfigurability, and highly reduced hardware complexity in comparison to conventional phased arrays. Furthermore, the use of large antenna arrays, a key requisite for beamforming, leads to a substantial reduction in the Rayleigh distance. This in turn leads to the user equipment (UE) simultaneously experience near-field and far-field propagation, collectively termed as a hybrid-field channel. Naturally, in such scenarios, the conventional far-field assumption, widely used in previous works breaks down, leading to extraordinary challenges in channel estimation and beamforming. Addressing this therefore compels the design of novel signal processing models capable of incorporating these hybrid field components. Finally, in addition to the significant challenges listed above, practical deployment of THz technology also faces severe impediments in achieving accurate and real-time channel estimates and beam steering due to stringent hardware constraints coupled with the necessity to accurately tracking the channels in a highly time-varying environment. The recently proposed delay-Doppler waveforms such as OTFS offer an excellent, cost-effective recourse in such a scenario as they can be readily overlaid on existing transceivers. The project also envisages the generation of high-quality datasets using generative modelling approaches that accurately reflect the propagation characteristics of sub-THz and THz wireless communication. This module will leverage Generative Adversarial Networks (GANs), which have emerged as an excellent alternative for existing deep learning-based techniques, to generate realistic and high-fidelity channel datasets required for learning in these systems. The proposed approaches bring us a step closer toward the realization of energy-efficient and scalable THz systems, under practical hardware and mobility constraints, suitable for deployment in 6G and beyond wireless networks. The proposal is a significant advancement over the current state-of-the-art as it develops a data-driven, model-based hybrid field framework that enables real-time, low-overhead CSI acquisition, beamforming and localization in DMA-based ultra-wide bandwidth systems. It aims to significantly push system communication speeds and capabilities beyond those envisioned by the existing research in 6G, via the effective amalgamation of generative modelling, emerging antenna technologies, advanced hybrid field modelling, delay-Doppler domain waveforms, coupled with state-of-the-art learning methodologies. The ultimate goal of our proposed research is to develop pragmatic strategies that can help commercialize high speed sub-THz technology for both short-range indoor and long-range outdoor communication.