Diffusion Fluid Antenna Systems for Resilient ISAC

Noor Waqar, Kai-Kit Wong, Fellow, IEEE , Chan-Byoung Chae, Fellow, IEEE , and Ross Murch, Fellow, IEEE · 2026

Read Diffusion Fluid Antenna Systems for Resilient ISAC by Noor Waqar, Kai-Kit Wong, Fellow, IEEE , Chan-Byoung Chae, Fellow, IEEE , and Ross Murch, Fellow, IEEE (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-23649.txt.

Introduction

Waqar and colleagues devise diffusion fluid antenna systems for resilient integrated sensing and communication, tackling the ignored object side. One object may want to stay hard to detect while a neighbor's strong echoes confuse the base station. Fluid antenna hardware adds a reconfigurable spatial degree of freedom, shaping sensing signatures through port choice beyond signal and power control. A conditional denoising diffusion model rebuilds latent spatial correlation from few seen ports. It enables two modes: generative spatial stealth, cutting visibility up to a hundredfold, and target isolation via spatial nulls.

Catalog record

CREATORNoor Waqar, Kai-Kit Wong, Fellow, IEEE , Chan-Byoung Chae, Fellow, IEEE , and Ross Murch, Fellow, IEEE
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56890
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.23649

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Diffusion Fluid Antenna Systems for Resilient ISAC The work of K. K. Wong is supported by the Engineering and Physical Sciences Research Council (EPSRC) under grant EP/W026813/1. The work of C.-B. Chae was supported by the Korean Government under Grant IITP-2025-RS-2024-00428780 and Grant IITP-RS-2026-25489110. The work of R. Murch was supported by the Hong Kong Research Grants Council Area of Excellence Grant AoE/E-601/22-R. N. Waqar and K. K. Wong are with the Department of Electronic and Electrical Engineering, University College London, London WC1E 7JE, United Kingdom. K. K. Wong is also affiliated with Yonsei Frontier Laboratory, Yonsei University, Seoul, 03722, Republic of Korea. C. B. Chae is with School of Integrated Technology, Yonsei University, Seoul, 03722, Republic of Korea. R. Murch is with the Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong SAR, China. Corresponding author: Kai-Kit Wong (e-mail: kai ​ - ​ kit . wong ​ @ ​ ucl . ac . uk \rm kai\text{-}[email protected] ). Noor Waqar, Kai-Kit Wong, Fellow, IEEE , Chan-Byoung Chae, Fellow, IEEE , and Ross Murch, Fellow, IEEE Abstract Most existing integrated sensing and communication (ISAC) research focuses on enabling a base station (BS) to support sensing and communication over shared resources through advanced waveform design and power allocation. Yet, the object-side perspective remains underexplored. For example, it may be required to make an object difficult to detect for security reasons. Moreover, an object in close proximity to a target may generate dominant reflections that confuse the BS and impair sensing reliability for the intended target. These challenges motivate the fluid antenna system (FAS) paradigm which introduces a reconfigurable spatial degree of freedom (DoF) that can reshape object sensing signatures via port selection, beyond what waveform and power control alone can provide. This paper proposes diffusion FAS, a generative artificial intelligence (AI)-driven framework that takes advantage of spatial agility to steer ISAC operation over the electromagnetic fading manifold. Instead of optimizing ISAC solely in the power domain, diffusion FAS casts ISAC as a dynamic spatial selection problem in which antenna states (i.e., ports) are selected to shape sensing signatures while maintaining communication objectives. To handle sparse measurements, we will employ a conditional denoising diffusion probabilistic model (DDPM) to reconstruct the latent spatial correlation structure from a small set of observed ports, enabling efficient exploration of the reconfigurable aperture. We demonstrate two FAS-enabled ISAC modes: (1) generative spatial stealth , which identifies localized deep fades to suppress a target’s sensing visibility by up to two orders of magnitude, and (2) target isolation , which creates spatial nulls that reject interference from adjacent objects. Our results show that combining generative priors with reconfigurable apertures offers a viable solution for secure and resilient ISAC . I Introduction I-A Background The pursuit toward sixth-generation (6G) wireless networks heralds an unprecedented unification of extreme-high rate access, super-massive connectivity, and ultra-reliable communication under stringent latency constraints [ 1 , 2 , 3 ] . A further recent trend has seen 6G discussion dominated by integrated sensing and communication (ISAC), a paradigm shift that recognizes the rising importance of sensing beyond the primary communication-oriented mobile networks [ 4 , 5 , 6 ] . To support high-precision sensing while meeting stringent communication requirements, a substantial body of work has been reported [ 7 , 8 ] , which focuses on developing advanced waveform and power allocation to enable a base station (BS) to support both sensing and communication tasks over shared resources. Moreover, multiple-input multiple-output (MIMO) architectures play a central role in enabling ISAC by offering the spatial degrees of freedom (DoF) needed to jointly shape communication and sensing signals [ 9 ] . The benefits of MIMO are well established, and the evolution toward extra-large MIMO appears compelling [ 10 ] . Yet, its scalability limitations are increasingly evident. First, even in communications without sensing requirements, fifth generation (5G) New Radio (NR) precoding is far from trivial: it relies on the maximization of signal-to-interference plus noise ratio (SINR) based on quantized channel state information (CSI), which imposes substantial feedback and overhead [ 11 ] . More worryingly, the increase in the peak-to-average power ratio as a result of precoding greatly degrades the efficiency of power amplifiers at each antenna, causing the power consumption to explode at the BS when the antenna count is very large [ 12 ] . These issues are expected to worsen under ISAC settings. On the other hand, despite the exciting prospect of ISAC, it raises security and privacy concerns [ 13 ] . What if we require an object to be hidden? How can we suppress an object’s reflections and reduce its sensing visibility? To respond to these concerns, the authors of [ 14 ] introduced a security and privacy-preserving network (SPPN) for information handling and collaborative efforts among trusted parties. In the physical layer, jamming was considered. In [ 15 ] , these problems were tackled using advanced strategies such as null-space projection and the use of artificial noise. Recently, [ 16 ] provided an architecture defined into three layers: the hardware layer, the omniscient layer, and the application layer. The idea is to consider new design principles for artificial intelligence (AI)-empowered networks that support security in ISAC. Although AI can help a receiver detect and interpret sensing outcomes, it has limited leverage in addressing object-side objectives such as reducing sensing visibility. In particular, once an external transmitter illuminates the environment, the object’s reflections are governed primarily by electromagnetic scattering physics and the geometry/material properties of the object, which cannot be altered at the sensing node. As a result, purely AI-driven approaches are confined to post-processing, e.g., improving detection, classification, or interference suppression, rather than actively preventing an object from being sensed. Sending a jamming signal from the sensing object can be effective in reducing its sensing visibility but the use of jamming is controversial as it preemptively pollutes the entire radio environment. Achieving true sensing suppression hence requires additional controllable DoFs at the physical layer. In this regard, the fluid antenna system (FAS) that considers the utilization of reconfigurable apertures capable of reshaping the propagation/scattering process, can be the solution. I-B The FAS Paradigm First investigated in [ 17 , 18 ] , FAS is a hardware-agnostic system concept that considers the antenna as a reconfigurable physical-layer resource to broaden system design and network optimization [ 19 , 20 , 21 , 22 , 23 ] . In practice, it can be implemented using a range of antenna technologies such as movable elements [ 24 ] , liquid antennas [ 25 , 26 ] , metamaterials [ 27 , 28 ] , reconfigurable pixels [ 29 , 30 , 31 ] . The theoretical premise is profound: even within a relatively small physical footprint, the multipath fading observed across FAS positions exhibit rich spatial fluctuations, and FAS yields extraordinary diversity and reliability gains [ 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 ] . CSI estimation for FAS has been recently addressed in [ 42 , 43 , 44 ] . Beyond point-to-point single-user links, this spatial diversity has also catalyzed the

Read the complete text