Detecting Wireless Jammers and Hackers with JADE

on YouTube ↗28m

" summary

Claims an automated, data-driven jammer detector called JADE that uses hourly operator cell-level KPIs, TensorFlow LSTM autoencoders plus RSSI prediction and transfer learning to produce a frequency-agnostic model. Training uses semi-supervised “learn normal” data and three labelled jammers; evaluation hits ~0.95 F1 versus ~0.7 for random forest. Deployed in-field, it detects discontinuous and military jammers and localizes jammers within a few hundred meters.

AI-written summary · about

" chapters (10)

1. 0:00Introduction
2. 2:59Network complexity
3. 7:05The impact of jammers
4. 10:47Manual detection limitations
5. 12:57Anomaly detection approach
6. 15:30Jade framework
7. 19:03Evaluation and baselines
8. 22:16Jade results and application
9. 24:16Conclusion
10. 25:06Q&A

" related talks

• 11mGotta Detect 'Em All: Fake Base Station and Multi-Step Attack Detection...Kazi Samin Mubasshir, Imtiaz Karim, Elisa Bertino · USENIX Security '25
• 40mMe, My Adversary & AI: Investigating & Hunting with Machine LearningJess Garcia · RSAC 2021
• 17mCellDAM: User-Space, Rootless Detection and Mitigation for 5G Data PlaneSongwu Lu, Zhaowei Tan, Jinghao Zhao, Boyan Ding · NSDI 2023
• 25mMonitoring AI with AIStepan Pushkarev · Scale By The Bay 2018
• 28mIs Neural Network Better Than Ash at Detecting Team RocketJuan De Dios Santos · PyData Global 2020
• 32mMan or Machine?: Developing a "Turing Test" for Radio IntelligenceAdam Anderson · GRCon 2019
• 29mSara Hooker & Sean McPherson, Delta Analytics at MLconf Seattle 2017Sara Hooker, Sean McPherson · MLconf Seattle 2017
• 12mEarFisher: Detecting Wireless Eavesdroppers by Stimulating and Sensing Memory EMRCheng Shen, Jun Huang · NSDI ’21