IEEE recognition highlights growing interest in joint communications and sensing research
A recent study on joint communications and sensing has drawn sustained attention within the signal processing community, placing it among the 25 most downloaded articles in the IEEE Journal of Selected Topics in Signal Processing, a highly selective and top-tier journal in the field, during 2024 and 2025.
The paper, Joint Communications and Sensing Hybrid Beamforming Design via Deep Unfolding, examines how model-based learning can address one of the central challenges in future wireless systems: designing hybrid beamforming schemes that balance data transmission and sensing accuracy within tight hardware and latency constraints. The work focuses on massive MIMO systems, where analogue and digital precoders are tightly coupled, and conventional optimisation methods often struggle to converge fast enough.
Instead of using a black-box neural network, the authors apply a deep unfolding approach to hybrid beamforming, reformulating a projected gradient ascent algorithm as a trainable architecture. This model-based learning method preserves the structure and interpretability of the original optimisation while using data to tune its step sizes, enabling faster convergence to an effective trade-off between communications rate and sensing performance in joint communications and sensing systems.
“By unfolding the gradient-based design and learning only a small set of parameters, we can improve convergence speed and system performance without sacrificing interpretability.”
Instead of using a black-box neural network, the authors apply a deep unfolding approach to hybrid beamforming, reformulating a projected gradient ascent algorithm as a trainable architecture. This model-based learning method preserves the structure and interpretability of the original optimisation while using data to tune its step sizes, enabling faster convergence to an effective trade-off between communications rate and sensing performance in joint communications and sensing systems.
“The aim was to preserve the structure of the original optimisation while enabling rapid convergence to a high objective value suitable for practical joint communications and sensing,” said Nhan Nguyen, Assistant Professor and Academy Research Fellow at the Centre for Wireless Communications, University of Oulu. “By unfolding the gradient-based design and learning only a small set of parameters, we can improve convergence speed and system performance without sacrificing interpretability.”
Simulation results show that the proposed deep unfolding approach to hybrid beamforming improves joint communications and sensing performance, achieving up to a 33.5% increase in communications sum rate and a 2.5 dB reduction in sensing beam pattern error compared to conventional hybrid beamforming designs. The method also reduces run time and computational complexity by up to 65%, addressing key efficiency constraints in massive MIMO and 6G-oriented wireless systems.
The level of readership has led the IEEE Signal Processing Society to invite the authors to present their work in an upcoming webinar and to contribute a dedicated blog post for the Society’s channels. Both invitations point to growing interest in practical design methods for systems that integrate sensing and communications rather than treating them as separate functions.
The study brings together researchers from several institutions, with contributions from Van Ly Nguyen, Nir Shlezinger, Yonina Eldar, Lee Swindlehurst, and Markku Juntti.
For the research programme, the recognition reflects a broader shift in how joint communications and sensing is being approached. As future wireless systems move towards higher frequencies and larger antenna arrays, the challenge increasingly lies in translating joint design concepts into deployable solutions that respect computational and hardware limits.
The forthcoming IEEE webinar will provide an opportunity to discuss these design questions with an international audience. The IEEE Signal Processing Society will announce the webinar on its LinkedIn channel.