- Rating
- 3(+1)
- Confidence
- 3(+1)
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
This paper proposes a cross-domain backdoor attack method against semantic communication systems, named XComBo, which superimposes perturbations that require matched activation in both the source domain and the channel domain to achieve a covert backdoor attack on semantic communication systems. The authors tested the effectiveness of the proposed method on three common semantic communication system models across four datasets.
Critical strengths
-
The authors claim that the proposed method is the first semantic communication attack method that distributes backdoor trigger conditions across two heterogeneous attack surfaces, the source and the wireless channel, and it requires no substitute joint optimization framework based on prior knowledge of the victim model.
-
The FDM-inspired multi-sinusoid signal trigger with channel pre-compensation, the FPR energy constraint, and the surrogate-based joint optimization are technically sensible and carefully derived. The energy constraints are explicitly tied to the goal of preserving benign behavior under incomplete trigger conditions.
Critical weaknesses
-
Section 3.1 cites one reference for timing synchronization, yet any symbol timing or carrier frequency offset would rotate the optimized phases of the signal sub-trigger (Eq. 7) and corrupt the injection (Eq. 11). The authors carefully analyzed imperfect CSI in Appendix D.1, not the arguably more fragile synchronization assumption.
-
The baseline set is limited, and the comparisons are inherently approximate: Only three adapted baselines are considered, and none is a source-domain backdoor attack on SemCom, although such attacks exist and are cited in the related work (e.g., Sagduyu et al. 2023; Xu et al. 2024; Zhou et al. 2024; Yang et al. 2026).
Questions for authors
1.Training-time wireless injection: Please clarify the assumed access for wireless injection during training-time poisoning: under what conditions can the adversary observe and interfere with the victim's training transmissions in a real system, and does the attack still succeed if the injection channel differs between training and inference?
2.Real-world evaluation: All experiments use a simulated Rician channel with idealized injection (y + λφs). Do the authors expect the attack to hold under real over-the-air conditions and would they consider validating it with a hardware or over-the-air test?
7 comments
Q1: Please clarify the assumed access for wireless injection during training-time poisoning: under what conditions can the adversary observe and interfere with the victim's training transmissions in a real system? (2/2)
To evaluate the remaining timing uncertainty, we study random timing offsets during poisoning and evaluate the models temporal misalignment:
Table R1. Random timing offsets results on WITT/GTSRB at SNR = 18 dB and FPR = 15 dB. Training and testing independently sample an shift for each image from the listed distribution, measured under PSNR (dB):
| Offset distribution | Strategy | NM | AO | BO | BD | MS |
|---|---|---|---|---|---|---|
| (-4)--(+4) | Joint | 38.64 | 37.39 | 38.18 | 41.07 | 37.05 |
| (-4)--(+4) | Random | 38.13 | 34.17 | 37.09 | 38.69 | 34.57 |
| (0)--(L-1) | Joint | 38.22 | 36.70 | 37.88 | 41.63 | 36.59 |
| (0)--(L-1) | Random | 37.92 | 29.48 | 36.89 | 28.49 | 30.06 |
XComBo retains BD PSNRs of 41.07/41.63 dB under local-ranged or full-ranged random shifts, while NM/AO/BO/MS preserve source reconstruction at 36.59–38.64 dB. These results show that the random timing offsets can be accommodated through training, without requiring victim-model access at inference.
References
[A1] Physical Layer Key Generation in 5G Wireless Networks. IEEE Wireless Communications, 2019.
[A2] Exploring Practical Vulnerabilities of Machine Learning-based Wireless Systems. USENIX Symposium on Networked Systems Design and Implementation (NSDI), 2023.
[A3] RFReact: A Real-time Capable and Channel-aware Jamming Platform. ACM Conference on Wireless Network Security (WiSec), demo, 2011.
[A4] A Real-Time and Protocol-Aware Reactive Jamming Framework Built on Software-Defined Radios. ACM Workshop on Software Radio Implementation Forum (SRIF), 2014.
[A5] Deep Joint Source-Channel Coding for Wireless Image Transmission. IEEE Transactions on Cognitive Communications and Networking, 2019.
[A6] Deep Learning Enabled Semantic Communication Systems. IEEE Transactions on Signal Processing, 2021.
[A7] Deep Learning-Based End-to-End Wireless Communication Systems With Conditional GANs as Unknown Channels. IEEE Transactions on Wireless Communications, 2020.
[A8] BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain. arXiv:1708.06733, 2017.
[A9] Adversarial Attacks Against Shared Knowledge Interpretation in Semantic Communications. IEEE Transactions on Cognitive Communications and Networking, 2025.
[A10] Channel-Aware Adversarial Attacks Against Deep Learning-Based Wireless Signal Classifiers. IEEE Transactions on Wireless Communications, 2022.
[A11] Adversarial Attacks on Deep-Learning Based Radio Signal Classification. IEEE Wireless Communications Letters, 2019.
\
Q1: Does the attack still succeed if the injection channel differs between training and inference?
We additionally evaluate changes in channel distribution and Rician fading parameters without reoptimizing the model.
Table R2. Reconstruction PSNR of XComBo under different training and inference channels on WITT/GTSRB at SNR=18 dB and FPR=15 dB, measured under PSNR(dB).
| Training channel | Inference channel | NM | AO | BO | BD | MS |
|---|---|---|---|---|---|---|
| Rician | Rician | 38.25 | 37.00 | 37.89 | 44.65 | 36.91 |
| Rician | Rician | 38.88 | 37.52 | 38.33 | 45.18 | 37.27 |
| Rician | Rician | 40.98 | 39.59 | 40.44 | 46.08 | 39.28 |
| Rayleigh | Rician | 37.91 | 36.52 | 37.41 | 44.27 | 36.27 |
| Rayleigh | Rician | 38.48 | 36.95 | 37.82 | 44.64 | 36.66 |
| Rayleigh | Rician | 40.49 | 38.88 | 39.83 | 45.38 | 38.43 |
| Rayleigh | Rayleigh | 38.02 | 36.60 | 37.39 | 44.49 | 36.29 |
| Rayleigh | AWGN | 40.76 | 39.13 | 40.04 | 45.36 | 38.66 |
| AWGN | Rician | 29.54 | 28.73 | 29.46 | 33.75 | 29.27 |
| AWGN | Rician | 31.81 | 30.26 | 31.37 | 35.94 | 30.48 |
| AWGN | Rician | 43.37 | 41.79 | 42.68 | 44.15 | 41.30 |
| AWGN | Rayleigh | 30.08 | 29.11 | 29.40 | 34.48 | 29.30 |
| AWGN | AWGN | 43.84 | 42.20 | 43.07 | 44.19 | 41.77 |
The Rayleigh-trained model maintains BD PSNR of 44.27–45.38 dB across the tested Rician, Rayleigh, and AWGN channels. The Rician-trained model maintains 44.65–46.08 dB across the tested values. The AWGN-trained model has lower target fidelity under stronger fading, but still obtains BD PSNR of 33.75–35.94 dB in the reported low- Rician and Rayleigh settings. Its NM PSNR also decreases under these fading channels, suggesting that the lower BD fidelity partly reflects degradation of the underlying communication model. The results show that target reconstruction can be retained despite differences between training and inference channels.
Q2: Real-world evaluation: All experiments use a simulated Rician channel with idealized injection (y + λφs). Do the authors expect the attack to hold under real over-the-air conditions and would they consider validating it with a hardware or over-the-air test?
We thank the reviewer for this suggestion. We conducted over-the-air validation using three USRP B210, configured as the legitimate transmitter, receiver, and adversarial transmitter, respectively. Experiments were conducted at a carrier frequency of 2.2GHz, without timing synchronization between the legitimate and adversarial transmitters.
Within the available time and equipment constraints, we implemented an OTA attack without channel estimation, as is also adopted in SemCom systems [H1-H2], using WITT/GTSRB model trained with random timing offset. This implementation enables direct evaluation of the cross-domain threat over a real wireless transmittion.
As shown in Table R3, BD reconstructed target images at 38.09--41.37 dB in PSNR, while individual triggers and mismatched pairs retained source-image reconstruction, consistent with simulation. Some degradationm of reconstruction quality is expected compared with transmission using channel equalization. The higher BD PSNR relative to simulation may reflect less severe timing misalignment during OTA testing (compared with full random misalignment in simulation) although we started legitimate and adversarial transmitter individually and manually without synchronization.
Table R3. Simulation OTA results on WITT/GTSRB. Target A–C denote the three attack targets. OTA values are mean standard deviation across repeated recordings, calculated from each recording’s mean per-image PSNR.
| Pattern trigger | Signal trigger | Configuration | Simulation PSNR (dB) | OTA PSNR (dB) |
|---|---|---|---|---|
| -- | -- | NM | 33.60 | 30.65 ± 2.94 |
| Target A | -- | AO | 31.14 | 29.78 ± 0.77 |
| Target B | -- | AO | 31.39 | 28.73 ± 1.84 |
| Target C | -- | AO | 31.49 | 28.71 ± 2.23 |
| -- | Target A | BO | 30.78 | 31.69 ± 3.86 |
| -- | Target B | BO | 31.48 | 31.38 ± 0.45 |
| -- | Target C | BO | 32.29 | 29.59 ± 3.02 |
| Target A | Target A | BD | 30.15 | 38.09 ± 0.51 |
| Target B | Target B | BD | 30.57 | 41.37 ± 1.58 |
| Target C | Target C | BD | 29.41 | 41.05 ± 2.96 |
| Target A | Target B | MS | 30.36 (averaged under all mismatch configurations) | 28.32 ± 1.47 |
| Target B | Target C | MS | 30.36 (averaged under all mismatch configurations) | 28.88 ± 2.05 |
references:
[H1] Deep Joint Source-Channel Coding for Wireless Image Transmission. IEEE Transactions on Cognitive Communications and Networking, 2019.
[H2] Backdoor Attacks and Defenses on Semantic-Symbol Reconstruction in Semantic Communications. IEEE International Conference on Communications (ICC), 2024.
We thank the reviewer for their constructive feedback on practical wireless conditions and hardware validation. We appreciate their time and patience in considering our further clarification, additional simulation and real over-the-air experiments.
Q1: Please clarify the assumed access for wireless injection during training-time poisoning: under what conditions can the adversary observe and interfere with the victim's training transmissions in a real system? (1/2)
Training-time injection access. Training-time poisoning does not require observing or interfering with physical wireless transmissions. For backpropagation-based end-to-end training, physical wireless channels generally cannot directly propagate the required channel gradients. SemCom systems therefore use differentiable simulated channel modules [A5–A7]. The assumed poisoning access covers selected training samples, reconstruction targets, and the simulated channel module, allowing the attacker to add the signal sub-trigger througth the training pipeline. Channel conditions and timing are specified within the simulation, so physical channel estimation and real-time synchronization are not required during poisoning. Adversary can obtain these capabilities in practical scenarios such as outsourced training [A8].
Inference-time capabilities. Inference stage does not require access to or control over the deployed victim model.
Channel estimation. The adversary can exploit co-location with the legitimate receiver to obtain highly correlated estimates of the transmitter-to-receiver channel under spatially correlated channel conditions [A1]. For the adversary-to-receiver channel, observations of receiver-originated transmissions support estimation through channel reciprocity [A2].
Timing synchronization. Timing synchronization is a relatively strong assumption in practical deployment, although commonly assumed in wireless simulations [A9–A11]. So we discuss two possible implementation approaches, and discusses XComBo's effectiveness under out-of-sync conditions: (a) Preamble detection and reactive transmission. The adversary detects the beginning of a transmitted packet’s preamble and responds by transmitting the backdoor signal. RFReact reports microsecond-scale operation [A3], while an FPGA-based USRP N210 framework reports a response of approximately 80 ns after signal detection [A4]. (b) Repeated signal transmission. The adversary can transmit multiple copies of the backdoor signal to accommodate potentially large temporal misalignment [A2]. This approach may leave a random relative offset δ between the legitimate signal and the backdoor signal. Neither approach requires modifying the receiver’s model or inference pipeline.