Deep Learning-Driven Friendly Jamming for Secure Multicarrier ISAC Under Channel Uncertainty

Bui Minh Tuan; Van-Dinh Nguyen; Diep N. Nguyen; Nguyen Linh Trung; Nguyen Van Huynh; Dinh Thai Hoang; Marwan Krunz; Eryk Dutkiewicz

cs.LGMar 5, 2026

Deep Learning-Driven Friendly Jamming for Secure Multicarrier ISAC Under Channel Uncertainty

Authors: Bui Minh Tuan, Van-Dinh Nguyen, Diep N. Nguyen, Nguyen Linh Trung, Nguyen Van Huynh, Dinh Thai Hoang, Marwan Krunz, Eryk Dutkiewicz

Organizations: the School of Electrical and Data Engineering, University of Technology Sydney, NSW 2007, Australia · the School of Computer Science and Statistics, Trinity College Dublin, Dublin 2, D02PN40, Ireland · Smart Green Transformation Center (GREEN-X), VinUniversity, Hanoi, Vietnam · the Faculty of Electronics and Telecommunications, University of Engineering and Technology, Vietnam National University, Hanoi, Vietnam ·

Abstract

Integrated sensing and communication (ISAC) systems promise efficient spectrum utilization by jointly supporting radar sensing and wireless communication. This paper presents a deep learning-driven framework for enhancing physical-layer security in multicarrier ISAC systems under imperfect channel state information (CSI) and in the presence of unknown eavesdropper (Eve) locations. Unlike conventional ISAC-based friendly jamming (FJ) approaches that require Eve's CSI or precise angle-of-arrival (AoA) estimates, our method exploits radar echo feedback to guide directional jamming without explicit Eve's information. To enhance robustness to radar sensing uncertainty, we propose a radar-aware neural network that jointly optimizes beamforming and jamming by integrating a novel nonparametric Fisher Information Matrix (FIM) estimator based on f-divergence. The jamming design satisfies the Cramer-Rao lower bound (CRLB) constraints even in the presence of noisy AoA. For efficient implementation, we introduce a quantized tensor train-based encoder that reduces the model size by more than 100 times with negligible performance loss. We also integrate a non-overlapping secure scheme into the proposed framework, in which specific sub-bands can be dedicated solely to communication. Extensive simulations demonstrate that the proposed solution achieves significant improvements in secrecy rate, reduced block error rate (BLER), and strong robustness against CSI uncertainty and angular estimation errors, underscoring the effectiveness of the proposed deep learning-driven friendly jamming framework under practical ISAC impairments.

Figures & tables

Work	ISAC	Eve CSI	Sensing Role	Uncertainty Handling	MC	Method
[ 3 , 4 , 12 ]	Yes	No (AoA only)	Region/angle-aided design	Sensitive to AoA errors; limited robustness	No	Optimization
[ 10 , 18 , 19 ]	Yes	Yes / bounded	Joint waveform/beamforming	Typically assumes perfect or bounded CSI; limited angular uncertainty treatment	Limited	SDR/FP/AO
[ [ 14 , 15 , 16 ]	No	Bounded/statistical	Not considered	Robust to user CSI errors; not ISAC sensing-guided	Yes	Robust opt.
[ 22 , 23 ]	Partial	Imperfect CSI	Learning/prediction	Data-driven CSI robustness; limited CRLB-guaranteed sensing constraints	Yes	Deep learning
This work	Yes	No	Radar-echo-guided jamming	CSI & angular uncertainty with CRLB constraints	Yes	Learning + TT-Q

TABLE I : Comparison between representative secure ISAC approaches and the proposed method.

Fig. 1 : Two multicarrier ISAC architectures considered in this work. In the overlapping scheme, all subcarriers are jointly used for communication, sensing, and friendly jamming. In the non-overlapping scheme, the subcarriers are partitioned into disjoint sets, with one set allocated to communication and the remaining set used for sensing and friendly jamming.

Fig. 2 : ISAC FJ workflow: (1) message encoding, (2) beamforming generation, (3) power-normalized transmission, (4) radar echo acquisition, (5) Fisher information estimation, (6) friendly jamming optimization, and (7) message decoding. The dashed arrow highlights the sensing feedback loop, which enables adaptive jamming without requiring explicit eavesdropper CSI.

\begin{split}R_{k}^{(n)}&=\bar{\tau}\log_{2}\Bigg(1+\frac{\xi_{k}^{(n)}\left|(\mathbf{h}_{k}^{(n)})^{H}\mathbf{f}_{k}^{(n)}\right|^{2}}{\sum\limits_{j\neq k}\xi_{j}^{(n)}\left|(\mathbf{h}_{k}^{(n)})^{H}\mathbf{f}_{j}^{(n)}\right|^{2}+\zeta^{(n)}\left|(\mathbf{h}_{k}^{(n)})^{H}\mathbf{v}^{(n)}\right|^{2}+\sigma_{c}^{2}}\Bigg),\\ R_{E,k}^{(n)}&=\bar{\tau}\log_{2}\det\Big(\mathbf{I}+\left(\zeta^{(n)}\mathbf{H}_{E}^{(n)H}\mathbf{v}^{(n)}\mathbf{v}^{(n)H}\mathbf{H}_{E}^{(n)}+\sigma_{E}^{2}\mathbf{I}\right)^{-1}\mathbf{H}_{E}^{(n)H}\mathbf{f}_{k}^{(n)}\mathbf{f}_{k}^{(n)H}\mathbf{H}_{E}^{(n)}\Big).\end{split}

(28)

Table 4

\begin{split}R_{k}^{(n)}&=\bar{\tau}\log_{2}\Bigg(1+\frac{\gamma_{k}^{(n)}\left|(\mathbf{h}_{k}^{(n)})^{H}\mathbf{f}_{k}^{(n)}\right|^{2}}{\sum\limits_{j\neq k}\gamma_{j}^{(n)}\left|(\mathbf{h}_{k}^{(n)})^{H}\mathbf{f}_{j}^{(n)}\right|^{2}+\zeta^{(n)}\left|(\mathbf{h}_{k}^{(n)})^{H}\mathbf{v}^{(n)}\right|^{2}+\sigma_{c}^{2}}\Bigg),\\ R_{E,k}^{(n)}&=\bar{\tau}\log_{2}\det\Bigg(\mathbf{I}+\Big(\zeta^{(n)}\mathbf{H}_{E}^{(n)H}\mathbf{v}^{(n)}\mathbf{v}^{(n)H}\mathbf{H}_{E}^{(n)}+\sigma_{E}^{2}\mathbf{I}\Big)^{-1}\mathbf{H}_{E}^{(n)H}\mathbf{f}_{k}^{(n)}\mathbf{f}_{k}^{(n)H}\mathbf{H}_{E}^{(n)}\Bigg).\end{split}

(41)

Table 5

Parameter	Value
$N_{t}$ , $N_{r}$ , $N_{e}$	16, 4, 2
Number of users, $K$	2
Number of subcarriers, $N$	64
Set of comm subcarriers, $\mathcal{N}_{c}$	First 32 subcarriers (1–32)
Set of sensing/jamming, $\mathcal{N}_{s}$	Remaining 32 subcarriers (33–64)
Modulation scheme	QPSK ( $M=4$ )

TABLE II : Simulation Parameters

Fig. 3 : Sum secrecy rate comparison with perfect AoA estimation.

Fig. 4 : BLER on the legitimate channel.

Fig. 5 : Convergence of training process.

Fig. 6 : Average secrecy rate versus CRLB.

Fig. 7 : Average secrecy rate in worst case per user’s channel.

Fig. 8 : Average secrecy rate versus SNR with different perturbation levels of estimation.

Fig. 9 : Comprehensive comparison of the average secrecy rates between FC encoder and quantised TT.

Fig. 10 : Comparison of average secrecy rates and training loss between FC and TT method.

Metric	FC Encoder	DTTE	Pure TT Encoder
Architecture	Multi-layer FC network	TT core + FC head	Single TT-matrix layer
Compression	None	Hybrid TT + FC	Full TT decomposition
Parameters ( $N_{t}{=}16$ , $K{=}2$ )	62,563	28,371	2,800
FLOPs (Inference)	$\approx$ 125,000	$\approx$ 57,000	$\approx$ 5,600
Secrecy Rate	Highest (baseline)	Near-baseline	Lower due to overcompression
Inference Time	Slowest	Moderate	Fastest

TABLE III: Comparison of Encoder Architectures: FC, DTTE, and Pure TT

Fig. 11 : Average secrecy rate versus SNR in different levels of non-overlapping.

Fig. 12: Secrecy rate degradation, defined as the loss relative to the ideal hardware baseline.

Wireless Communications

Wireless communications studies how information is sent over radio links and how networks manage signals, spectrum, and resources reliably.

See all 399 papers

Short-Length Code Designs for Integrated Sensing and Communications: A Deep Learning Approach
Sep 27, 2026Muah Kim, Shuangyang Li, Tayyebeh Jahani-Nezhad +2Integrated Sensing and Communication Wireless Communications
Faculty of Electrical and Computer Engineering, Dresden University of Technology, Dresden 01069, Germany · Faculty of Electrical Engineering and Computer Science, Technical University of Berlin, Berlin 10587, Germany
Task-Oriented Candidate-Latent Feedback for Coarse-to-Fine Sensing in Distributed OFDM-ISAC Networks
Aug 4, 2026Shiv Shankar, Radha Krishna Ganti, J Klutto MillethIntegrated Sensing and Communication
Indian Institute of Technology (IIT), Madras, Chennai India · Centre of Excellence in Wireless Technology (CEWiT), Chennai India
Adaptive Pilot Selection for Unified Semantic Communication and Semantic Sensing in ISAC
Sep 25, 2026Muhammad Abubakar Rashid, Muhammad Hannan Akram, Haejoon Jung +1Integrated Sensing and Communication Task-Oriented Semantic Communication
School of Electrical Engineering and Computer Science, National University of Sciences and Technology, Islamabad, Pakistan · Department of Electronic Engineering, Kyung Hee University, Yongin 17104, Republic of Korea

Sep 27, 2026J/K move · Enter open · S save

Short-Length Code Designs for Integrated Sensing and Communications: A Deep Learning Approach

Muah Kim, Shuangyang Li, Tayyebeh Jahani-Nezhad, Rafael F. Schaefer+1

Faculty of Electrical and Computer Engineering, Dresden University of Technology, Dresden 01069, Germany · Faculty of Electrical Engineering and Computer Science, Technical University of Berlin, Berlin 10587, Germany

Integrated Sensing and Communication Wireless Communications

Open paper AskPDF arXiv

Integrated sensing and communication (ISAC) enables joint communication and sensing using a shared waveform, but its signal design is challenging due to the inherent trade-off between the two objectives, particularly in the short blocklength regime. This paper proposes an autoencoder (AE)-based framework for ISAC waveform design in noncoherent settings. We derive a modified Cramér-Rao bound for multi-target delay estimation and analyze the maximum-likelihood decoding rule for noncoherent communication under correlated fading. These results reveal structural connections and trade-offs between communication and sensing objectives in waveform design. Based on this analysis, the AE learns waveform representations that jointly optimize both functionalities, with a tunable parameter controlling the trade-off. Simulation results show that the proposed design outperforms conventional schemes in both communication reliability and sensing accuracy, especially under short blocklength and fading conditions.

Abstract

Wireless Communications

Explore similar work

Short-Length Code Designs for Integrated Sensing and Communications: A Deep Learning Approach

Task-Oriented Candidate-Latent Feedback for Coarse-to-Fine Sensing in Distributed OFDM-ISAC Networks

Adaptive Pilot Selection for Unified Semantic Communication and Semantic Sensing in ISAC