Research map / Adversarial robustness and security / Attacks on text and graphs
Securing Billion Bluetooth Devices Leveraging Learning-Based Techniques
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes BLEGuard, a hybrid BLE spoofing-attack detector combining cyber-physical features (UCN, INT, RSSI, CFO) with a TCN reconstruction model and text-CNN plus classical classifiers, evaluated on a self-collected testbed dataset.
Problem
Detecting spoofing attacks against Bluetooth Low Energy devices, which lack secure pairing and firmware updates, while balancing accuracy, false alarm rate and detection cost.
Approach
A hybrid pipeline in which a lightweight cyber-physical pre-detector (RSSI/CFO/UCN/INT abnormality checks) flags suspicious batches, a TCN autoencoder reconstruction model scores anomalies, and text-CNN features with SVM/KNN/RF/Naive Bayes classify individual packets.
Key results
- Average accuracy of 98.01% on the imbalanced BLE dataset
- False alarm rate of 0.09%
- Undetection rate of 1.28%
Limitations
- Preliminary results only; evaluated on a self-built testbed rather than real-world spoofing attacks
- Dataset is imbalanced (14.8% malicious) and the amount of attack variety is limited to four attacker platforms
- Hardware and compute details are not reported
Experimental setup
Datasets: self-collected BLE advertising packet dataset (1,010,526 packets)
Benchmarks: self-collected BLE testbed dataset
Metrics: accuracy false alarm rate undetection rate
Baselines: Wu et al. 2020 Lahmadi et al. 2020
Builds on: Wu et al. 2020 cyber-physical BLE detection Lahmadi et al. 2020 supervised-unsupervised learning framework text-CNN
Closest papers by meaning
- BridgePure: Limited Protection Leakage Can Break Black-Box Data Protection (NeurIPS 2025 · similarity 0.75)
- BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks (ICLR 2025 · similarity 0.75)
- Intersecting-Boundary-Sensitive Fingerprinting for Tampering Detection of DNN Models (ICML 2024 · similarity 0.75)
- CSI-Bench: A Large-Scale In-the-Wild Dataset for Multi-task WiFi Sensing (NeurIPS 2025 · similarity 0.74)
- TinyTrain: Resource-Aware Task-Adaptive Sparse Training of DNNs at the Data-Scarce Edge (ICML 2024 · similarity 0.74)
- RPP: A Certified Poisoned-Sample Detection Framework for Backdoor Attacks under Dataset Imbalance (NeurIPS 2026 · similarity 0.74)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.