Time-Modulated Intelligent Reflecting Surfaces for Integrated Sensing, Communication and Security: A Generative AI Design Framework

Zhihao Tao, Athina Petropulu, H. Vincent Poor ∗ · 2025

Read Time-Modulated Intelligent Reflecting Surfaces for Integrated Sensing, Communication and Security: A Generative AI Design Framework by Zhihao Tao, Athina Petropulu, H. Vincent Poor ∗ (2025) 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-2509-05565.txt.

Introduction

Walls that reflect radio waves in a timed pattern can protect wireless links while sensing their surroundings, Zhihao Tao, Athina Petropulu, and H. Vincent Poor propose in a 2025 wireless engineering paper. Their targets may themselves be eavesdroppers, so sensing and secrecy must be designed together.

The time-modulated reflecting surface is tuned so legitimate users receive clean data while signals in every other direction appear scrambled. A generative flow network learns a sampling strategy over the huge space of surface settings, guided by rewards mixing data rate and sensing beam strength. Against exhaustive search and earlier security methods, their design sampled far more efficiently and held up more robustly in tests.

Catalog record

CREATORZhihao Tao, Athina Petropulu, H. Vincent Poor ∗
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-05565

Browse the full library · About the Tower of Babel

Full text

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

Time-Modulated Intelligent Reflecting Surfaces for Integrated Sensing, Communication and Security: A Generative AI Design Framework Zhihao Tao, Athina Petropulu, H. Vincent Poor ∗ Abstract We propose a novel approach to achieve physical layer security for integrated sensing and communication (ISAC) systems operating in the presence of targets that may be eavesdroppers. The system is aided by a time-modulated intelligent reflecting surface (TM-IRS), which is configured to preserve the integrity of the transmitted data at one or more legitimate communication users (CUs) while making them appear scrambled in all other directions. The TM-IRS design leverages a generative flow network (GFlowNet) framework to learn a stochastic policy that samples high-performing TM-IRS configurations from a vast discrete parameter space. Specifically, we begin by formulating the achievable sum rate for the legitimate CUs and the beampattern gain toward the target direction, based on which we construct reward functions for GFlowNets that jointly capture both communication and sensing performance. The TM-IRS design is modeled as a deterministic Markov decision process (MDP), where each terminal state corresponds to a complete configuration of TM-IRS parameters. GFlowNets, parametrized by deep neural networks are employed to learn a stochastic policy that samples TM-IRS parameter sets with probability proportional to their associated reward. Experimental results demonstrate the effectiveness of the proposed GFlowNet-based method in integrating sensing, communication and security simultaneously, and also exhibit significant sampling efficiency as compared to the exhaustive combinatorial search and enhanced robustness against the rule-based TM-IRS design method. Keywords: Dual-Function Radar-Communication (DFRC), intelligent reflecting surface (IRS), time modulation, physical layer security (PLS), generative AI (GenAI), GFlowNets. 1 Introduction The explosive growth of wireless devices and the increasing demand for both high data rates and ubiquitous environmental awareness have propelled integrated sensing and communication (ISAC) to the forefront of 6G research and standardization. ISAC systems aim to jointly perform communication and sensing tasks using the same waveform, hardware, or spectrum, thereby reducing cost, improving spectral efficiency, and enabling tighter coordination between devices and their environments [ 1 , 2 , 3 ] . By fusing these traditionally separate functionalities, ISAC paves the way for transformative applications such as autonomous driving, smart cities/factories, human-device interaction, and surveillance systems [ 4 , 5 ] . Within the broader ISAC paradigm, dual-function radar-communication (DFRC) systems have emerged as a compelling architecture that uses a shared transmit waveform to simultaneously probe the physical environment and convey data to communication users (CUs). DFRC designs benefit from streamlined hardware, coherent integration, and real-time synchronization between radar and communication operations [ 6 , 1 , 7 ] . Orthogonal frequency-division multiplexing (OFDM)-based DFRC systems [ 8 ] , in particular, offer high flexibility, wide bandwidth, and compatibility with existing communication standards, making them a natural candidate for ISAC implementations. Despite these advantages, DFRC systems are increasingly recognized to suffer from critical security vulnerabilities at the physical layer. Since the same signal is used for both radar and communication purposes, radar targets may inadvertently or maliciously intercept communication data. Consequently, conventional DFRC designs are vulnerable to eavesdropping attacks by the targets [ 9 , 10 , 11 , 12 ] . Therefore, developing physical layer security (PLS) mechanisms that can safeguard communication while enabling effective target sensing is crucial. PLS exploits the physical characteristics of the wireless medium, such as channel fading, noise, interference, and spatial diversity, to complement, or in some circumstances, replace higher-layer cryptographic techniques [ 13 , 14 , 15 , 16 ] . Among the many PLS mechanisms proposed for securing DFRC systems, directional modulation (DM) has attracted particular interest because it embeds information in the spatial signature of the transmitted waveform: a receiver aligned with the intended steering direction observes an undistorted constellation, whereas other directions see a scrambled one [ 17 , 18 , 19 ] . Compared with other PLS approaches such as secrecy rate maximization [ 20 , 21 ] or artificial-noise injection [ 22 , 23 ] , DM can offer comparable secrecy in a more cost-effective and energy-efficient manner [ 24 ] . DM implementations have been proposed for fully digital or hybrid beamforming architectures with multiple radio-frequency (RF) chains and fine-grained phase control at each antenna element or each transmitted symbol [ 25 , 26 , 27 , 10 , 24 ] . Also, many methods [ 28 , 24 , 29 ] necessitate full channel state information (CSI) on the eavesdropper as well as the legitimate users. Alternatively, a time-modulated array (TMA) driven by OFDM signals provides DM without the need for CSI [ 30 , 19 , 31 ] . By using single-pole-single-throw (SPST), the TMA periodically connects and disconnects antennas to the RF chain, generating controllable harmonics whose periods are aligned with the OFDM symbol duration. As a result, each subcarrier of the transmitted OFDM signal carries a weighted mixture of all original symbols, where the mixing coefficients depend on the TMA parameters, i.e., connection times, or on states, and on state durations. The subcarrier-induced mixing represents scrambling of the transmitted symbols in all directions except the intended direction and operates independently of any CSI. In the absence of noise, the scrambling towards an intended direction can be eliminated by a rule-based design [ 30 ] , and this can be achieved with low complexity. DM implemented via OFDM-based TMAs offers a hardware cost-efficient solution for securing DFRC systems, while also enabling significantly higher data rates through OFDM [ 9 ] . However, the periodic deactivation of antenna elements degrades the system’s energy utilization efficiency [ 32 ] . To address this issue, recent research [ 33 ] shifts time modulation to an intelligent reflecting surface (IRS). IRS is a passive metasurface composed of programmable elements that dynamically adjust the phase of incident electromagnetic waves to realize beamforming [ 34 , 35 ] . By exerting the periodic TM on each IRS element, the system in [ 33 ] is designed to implement a 3D directional modulation. Also, the large aperture of an IRS delivers substantial beamforming gain that compensates for power lost of TMA during element deactivation. In [ 33 ] , the TM-IRS parameters are still obtained using simple, closed-form rules. Although the rule-based TMA approach is straightforward to implement, it does not account for noise and guarantees undistorted signal reception in only a single CU direction. Extending it to support communication with multiple users is challenging. This limitation is significant, as multi-user scenarios are common in modern wireless systems, particularly in ISAC settings. In this paper, we formulate a TM-IRS-assisted DFRC system and propose a time modulation approach that is not rule-based, as such can handle noise and multiple CUs. We assume that the target/eavesdropper’s location lies within a region of the 3D space. During the target tracking stage, this region is determined based on previous detections and predicted target positions. We define a secrecy rate based on the difference between the CU sum rate and the potential eavesdropper rate, and maximize the minimal secrecy rate across all possible locations within the suspected target region. The novelty of our approach lies in the method used to design the TM-IRS system. In particular, we propose a generative AI (GenAI)-based framework for TM-IRS-assisted DFRC systems that simultaneously support secure multi-user

Read the complete text