Bajes Zeyad Aljunaeidi; Mouiad Fadeil Alawneh; Mohammed Tawfik

ByzaSpline-Fed: Gram-Weighted Functional Aggregation for Byzantine-Robust Federated Kolmogorov–Arnold Networks in Industrial IoT

PDF (1953KB), PP.33-54

Views: 0 Downloads: 0

Author(s)

Bajes Zeyad Aljunaeidi ^1,* Mouiad Fadeil Alawneh ¹ Mohammed Tawfik ²

1. Department of Computer Science, Faculty of Information Technology, Ajloun National University, P.O. Box 43, Ajloun 26810, Jordan

2. Department of Cyber Security, Faculty of Information Technology, Ajloun National University, P.O. Box 43, Ajloun 26810, Jordan

* Corresponding author.

DOI: https://doi.org/10.5815/ijigsp.2026.05.03

Received: 4 Jun. 2026 / Revised: 28 Jun. 2026 / Accepted: 8 Jul. 2026 / Published: 8 Oct. 2026

Index Terms

Federated Learning, Byzantine Robustness, Kolmogorov–Arnold Networks, Mamba, Fault Diagnosis, Anomaly Detection, Remaining Useful Life, Industrial IoT.

Abstract

Federated learning enables industrial sites to train shared condition-monitoring models without exchanging raw vibration, acoustic, or sensor data, but its openness admits Byzantine clients that submit corrupted updates. Existing robust aggregation rules—Krum, trimmed mean, median, and their variants—operate in Euclidean parameter space and implicitly assume that a malicious update must be far from honest updates to do harm. We show that this assumption fails for models with Kolmogorov–Arnold (KAN) spline heads: an adversary can hold the spline coefficients close to the honest consensus in the L2 sense while distorting the functional shape the spline encodes, an attack we term functional spline-poisoning. We propose ByzaSpline-Fed, a Byzantine-robust aggregation rule that scores clients by Gram-weighted functional B-spline distance on a consensus knot grid and rejects functional outliers. The rule reduces exactly to spline-aware federated averaging under honest participation, admits a breakdown bound of f < K/2 under the standard honest-majority assumption, and supports optional differential privacy. We evaluate across four industrial benchmarks spanning classification, unsupervised acoustic anomaly detection, and remaining-useful-life regression—CWRU, Paderborn, MIMII, and C-MAPSS—reporting task-correct metrics under clean and adversarial conditions. Under thirty percent functional poisoning, Euclidean defenses lose up to thirty-four accuracy points while ByzaSpline-Fed remains within roughly three points of its clean performance. All reported results are averaged over eight independent random seeds, and the improvements over the strongest competing defense are statistically significant under the Wilcoxon signed-rank test (p < 0.05).

Cite This Paper

Bajes Zeyad Aljunaeidi, Mouiad Fadeil Alawneh, Mohammed Tawfik, "ByzaSpline-Fed: Gram-Weighted Functional Aggregation for Byzantine-Robust Federated Kolmogorov–Arnold Networks in Industrial IoT", International Journal of Image, Graphics and Signal Processing(IJIGSP), Vol.18, No.5, pp. 33-54, 2026. DOI:10.5815/ijigsp.2026.05.03

Reference

[1]N. M. Al-Zidi, M. Tawfik, A. M. Al-Hejri, I. Fathail, T. A. Aldhaheri, and Q. Al-Tashi, “Smart system for real-time remote patient monitoring based on Internet of Things,” in Proc. 2nd Int. Conf. Computational Methods in Science & Technology (ICCMST), 2021, pp. 1–6.
[2]P. Kairouz, H. B. McMahan, et al., “Advances and open problems in federated learning,” Found. Trends Mach. Learn., vol. 14, no. 1–2, pp. 1–210, 2021.
[3]B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” in Proc. AISTATS, 2017, pp. 1273–1282.
[4]W. Zhang, X. Li, H. Ma, Z. Luo, and X. Li, “Federated learning for machinery fault diagnosis with dynamic validation and self-supervision,” Knowl.-Based Syst., vol. 213, art. 106679, 2021, doi: 10.1016/j.knosys.2020.106679.
[5]J. Du, N. Qin, D. Huang, Y. Zhang, and X. Jia, “An efficient federated learning framework for machinery fault diagnosis,” IEEE Trans. Neural Netw. Learn. Syst., vol. 35, no. 7, pp. 10086–10097, 2024.
[6]S. Lu, Z. Gao, Q. Xu, C. Jiang, A. Zhang, and X. Wang, “Class-imbalance privacy-preserving federated learning for decentralized fault diagnosis with biometric authentication,” IEEE Trans. Ind. Informat., vol. 18, no. 12, pp. 9101–9111, 2022.
[7]Y. Xiao, H. Shao, J. Lin, Z. Huo, and B. Liu, “BCE-FL: a secure and privacy-preserving federated learning system for device fault diagnosis under non-IID condition in IIoT,” IEEE Internet Things J., vol. 11, no. 8, pp. 14241–14252, 2024, doi: 10.1109/JIOT.2023.3340745.
[8]M. Tawfik, A. A. Abu-Ein, H. M. Noaman, A. H. Abdelhaliem, and I. S. Fathi, “FedMedSecure: federated few-shot learning with cross-attention mechanisms and explainable AI for collaborative healthcare cybersecurity,” Sci. Rep., vol. 15, art. 40050, 2025, doi: 10.1038/s41598-025-25107-z.
[9]A. Gu and T. Dao, “Mamba: linear-time sequence modeling with selective state spaces,” arXiv:2312.00752, 2023.
[10]H. Yi, D. Li, Z. Lu, Y. Jin, H. Duan, L. Hou, F. Z. Duraihem, E. Awwad, and N. A. Saeed, “VibrMamba: a lightweight Mamba-based fault diagnosis of rotating machinery using vibration signal,” Measurement, vol. 249, art. 116881, 2025, doi: 10.1016/j.measurement.2025.116881.
[11]M. Li, L. Zhu, M. Luo, and T. Ke, “Remaining useful life prediction of airplane engine based on bidirectional Mamba and causal discovery,” Sensors, vol. 25, no. 11, art. 3429, 2025, doi: 10.3390/s25113429.
[12]Z. Liu, Y. Wang, S. Vaidya, F. Ruehle, J. Halverson, M. Soljačić, T. Y. Hou, and M. Tegmark, “KAN: Kolmogorov–Arnold networks,” in Proc. ICLR, 2025.
[13]Z. Tang, X. Hou, X. Wang, and J. Zou, “A cross-working condition-bearing diagnosis method based on image fusion and a residual network incorporating the Kolmogorov–Arnold representation theorem,” Appl. Sci., vol. 14, no. 16, art. 7254, 2024, doi: 10.3390/app14167254.
[14]M. Tawfik, “MambaResp-KAN: a state space model with Kolmogorov–Arnold networks and diffusion-based augmentation for explainable respiratory disease classification,” Int. J. Image Graph. Signal Process. (IJIGSP), vol. 18, no. 3, pp. 189–208, 2026, doi: 10.5815/ijigsp.2026.03.10.
[15]M. Tawfik, I. S. Fathi, S. S. Nimbhore, I. M. Alsmadi, and M. S. Sawah, “E-RespiNet: an LLM-ELECTRA driven triple-stream CNN with feature fusion for asthma classification,” PLoS One, vol. 20, no. 11, art. e0334528, 2025, doi: 10.1371/journal.pone.0334528.
[16]P. Blanchard, E. M. El Mhamdi, R. Guerraoui, and J. Stainer, “Machine learning with adversaries: Byzantine tolerant gradient descent,” in Proc. NeurIPS, vol. 30, 2017, pp. 119–129.
[17]D. Yin, Y. Chen, R. Kannan, and P. Bartlett, “Byzantine-robust distributed learning: towards optimal statistical rates,” in Proc. ICML, 2018, pp. 5650–5659.
[18]E. M. El Mhamdi, R. Guerraoui, and S. Rouault, “The hidden vulnerability of distributed learning in Byzantium,” in Proc. ICML, 2018, pp. 3521–3530.
[19]X. Cao, M. Fang, J. Liu, and N. Z. Gong, “FLTrust: Byzantine-robust federated learning via trust bootstrapping,” in Proc. NDSS, 2021.
[20]M. Tawfik, N. M. Al-Zidi, B. Alsellami, A. M. Al-Hejri, and S. Nimbhore, “Internet of things-based middleware against cyber-attacks on smart homes using software-defined networking and deep learning,” in Proc. 2nd Int. Conf. Comput. Methods Sci. Technol. (ICCMST), 2021, pp. 7–13, doi: 10.1109/ICCMST54943.2021.00014.
[21]C. Dwork and A. Roth, “The algorithmic foundations of differential privacy,” Found. Trends Theor. Comput. Sci., vol. 9, no. 3–4, pp. 211–407, 2014.
[22]W. A. Smith and R. B. Randall, “Rolling element bearing diagnostics using the Case Western Reserve University data: a benchmark study,” Mech. Syst. Signal Process., vol. 64–65, pp. 100–131, 2015.
[23]C. Lessmeier, J. K. Kimotho, D. Zimmer, and W. Sextro, “Condition monitoring of bearing damage in electromechanical drive systems by using motor current signals of electric motors: a benchmark data set for data-driven classification,” in Proc. PHM Soc. Eur. Conf., vol. 3, no. 1, 2016, doi: 10.36001/phme.2016.v3i1.1577.
[24]H. Purohit, R. Tanabe, K. Ichige, T. Endo, Y. Nikaido, K. Suefusa, and Y. Kawaguchi, “MIMII dataset: sound dataset for malfunctioning industrial machine investigation and inspection,” in Proc. DCASE Workshop, 2019, pp. 209–213.
[25]A. Saxena, K. Goebel, D. Simon, and N. Eklund, “Damage propagation modeling for aircraft engine run-to-failure simulation,” in Proc. Int. Conf. PHM, 2008, pp. 1–9.
[26]C. Fung, C. J. M. Yoon, and I. Beschastnikh, “The limitations of federated learning in sybil settings,” in Proc. RAID, 2020, pp. 301–316.
[27]M. Yurochkin, M. Agarwal, S. Ghosh, K. Greenewald, T. N. Hoang, and Y. Khazaeni, “Bayesian nonparametric federated learning of neural networks,” in Proc. ICML, 2019, pp. 7252–7261.
[28]H. Wang, M. Yurochkin, Y. Sun, D. Papailiopoulos, and Y. Khazaeni, “Federated learning with matched averaging,” in Proc. ICLR, 2020.
[29]M. Fang, X. Cao, J. Jia, and N. Z. Gong, “Local model poisoning attacks to Byzantine-robust federated learning,” in Proc. USENIX Security, 2020, pp. 1605–1622.
[30]V. Shejwalkar and A. Houmansadr, “Manipulating the Byzantine: optimizing model poisoning attacks and defenses for federated learning,” in Proc. NDSS, 2021.
[31]G.-B. Truong, T.-T. Tran, N.-L. Than, V. Q. Nguyen, T. H. Nguyen, and V.-T. Pham, “SC-MambaFew: few-shot learning based on Mamba and selective spatial-channel attention for bearing fault diagnosis,” Comput. Electr. Eng., vol. 123, art. 110004, 2025.
[32]Q. Sun, L. Lu, Y. Chen, L. Xiang, and A. Hu, “FedMAPS: a federated meta-learning framework with adaptive cross-domain contrastive learning for few-shot fault diagnosis,” Mech. Syst. Signal Process., 2026.
[33]J. Sun, F. Zhou, J. Chen, C. Wang, X. Hu, and T. Wang, “A federated adversarial fault diagnosis method driven by fault information discrepancy,” Entropy, vol. 26, no. 9, art. 718, 2024, doi: 10.3390/e26090718.
[34]M. Abadi, A. Chu, I. Goodfellow, H. B. McMahan, I. Mironov, K. Talwar, and L. Zhang, “Deep learning with differential privacy,” in Proc. ACM CCS, 2016, pp. 308–318.
[35]C. de Boor, A Practical Guide to Splines, rev. ed. New York: Springer-Verlag, 2001.
[36]E. Bagdasaryan, A. Veit, Y. Hua, D. Estrin, and V. Shmatikov, “How to backdoor federated learning,” in Proc. AISTATS, 2020, pp. 2938–2948.
[37]Divya and D. Suresha, “KanAVNet: a CNN-BiLSTM-CTC-based audio-visual speech recognition system for Kannada to assist the hearing impaired,” Int. J. Image Graph. Signal Process. (IJIGSP), vol. 18, no. 2, pp. 87–106, 2026, doi: 10.5815/ijigsp.2026.02.06.
[38]A. Paszke, S. Gross, F. Massa, et al., “PyTorch: an imperative style, high-performance deep learning library,” in Proc. NeurIPS, vol. 32, 2019, pp. 8024–8035.
[39]D. P. Kingma and J. Ba, “Adam: a method for stochastic optimization,” in Proc. ICLR, 2015.
[40]F. Wilcoxon, “Individual comparisons by ranking methods,” Biometrics Bull., vol. 1, no. 6, pp. 80–83, 1945.
[41]I. S. Fathi and M. Tawfik, “Enhancing IoT systems with bio-inspired intelligence in fog computing environments,” Stat. Optim. Inf. Comput., vol. 13, no. 5, pp. 1916–1932, 2025, doi: 10.19139/soic-2310-5070-2305.
[42]M. Tawfik, A. H. Abdelhaliem, and I. S. Fathi, “Transforming IoT security through large language models: a comprehensive systematic review and future directions,” Stat. Optim. Inf. Comput., vol. 14, no. 2, pp. 1018–1044, 2025, doi: 10.19139/soic-2310-5070-2424.

International Journal of Image, Graphics and Signal Processing (IJIGSP)