Towards Causally Interpretable Wi-Fi CSI-Based Human Activity Recognition with Discrete Latent Compression and LTL Rule Extraction
Published: 24 April 2026
Authors: Luca Cotti, Luca Lavazza, Marco Cominelli, Liying Han, Gaofeng Dong, Francesco Gringoli, Mani B. Srivastava, Trevor Bihl, Erik P. Blasch, Daniel O. Brigham, Kara Combs, Lance M. Kaplan, Federico Cerutti
The paper proposes a pipeline that compresses Wi-Fi CSI signals into discrete representations and extracts Linear Temporal Logic (LTL) rules for human activity recognition. Rather than relying solely on a black-box classifier, it produces deterministic rules that make temporal and causal dependencies explicit.
Its cybersecurity relevance lies in making wireless sensing more inspectable and governable. Wi-Fi CSI can enable sensitive inferences about presence and activity; understanding the rules behind a model helps assess its reliability, limitations, and privacy implications, while supporting more controllable countermeasures and multi-antenna fusion.