Commodity Tri-Band Wi-Fi 6E Spectroscopy
for Material Classification
Classify building materials using off-the-shelf Wi-Fi 6E hardware. Tri-band frequency-differential attenuation across 2.4, 5, and 6 GHz delivers richer material fingerprints than dual-band approaches.
Different building materials attenuate 2.4 GHz, 5 GHz, and 6 GHz signals at distinctly different rates. WiSpec exploits this physical phenomenon across all three Wi-Fi 6E bands to create unique material fingerprints using nothing more than commodity hardware.
Higher frequencies attenuate faster through materials. Three bands (2.4, 5, 6 GHz) yield 3 pairwise differentials and a spectral curvature metric for richer fingerprints.
No specialized RF equipment needed. Works with off-the-shelf routers and Wi-Fi cards costing under $200 total.
Ablation study comparing single-band, dual-band, and tri-band configurations with McNemar’s test and paired t-tests.
Simultaneously transmit on 2.4 GHz, 5 GHz, and 6 GHz (Wi-Fi 6E) bands. The router sends beacon frames while the receiver measures signal characteristics across all three frequencies.
Compute 3 pairwise differentials: ΔA5−2.4, ΔA6−2.4, ΔA6−5, plus a spectral curvature metric. Each material produces a unique tri-band fingerprint.
Feed tri-band features into ensemble classifiers (Random Forest, SVM, XGBoost). Stratified 5-fold cross-validation with ablation across single-, dual-, and tri-band configurations.
Three pairwise differentials plus Scurv (spectral curvature) — the non-linearity of attenuation across the three frequency points — provide a richer material fingerprint than any dual-band subset.
Each building material attenuates 2.4 GHz, 5 GHz, and 6 GHz at distinctly different rates, creating a unique tri-band electromagnetic fingerprint.
Rigorous comparison across single-band, dual-band, and tri-band configurations.
| Classifier | Accuracy | Precision | Recall | F1 Score |
|---|---|---|---|---|
| Random Forest | 90.9% | 0.91 | 0.90 | 0.90 |
| SVM (RBF) | 88.4% | 0.89 | 0.88 | 0.88 |
| XGBoost | 91.2% | 0.91 | 0.91 | 0.91 |
| k-NN | 84.7% | 0.85 | 0.84 | 0.84 |
| Gradient Boost | 89.6% | 0.90 | 0.89 | 0.89 |
Characterize interior layouts, wall composition, and structural elements using Wi-Fi signals — without visual access. Map room boundaries and identify construction materials remotely.
Rapid structural assessment during emergencies. Identify floor materials, wall thickness, and potential hazards before entry.
Material-aware automation adapting HVAC, lighting, and acoustic settings based on real-time wall and floor composition data.
Non-destructive verification of building materials during and after construction. Detect substitutions or structural deficiencies.
WiSpec runs on commodity hardware. No specialized RF equipment required.
End-to-end modular pipeline: ~5,000 lines of Python covering data collection through publication-quality output.
@software{ranish_wispec_2026,
author = {Ranish, Abhinav},
title = {{WiSpec}: Commodity Tri-Band
Wi-Fi 6E Spectroscopy for Material
Classification},
year = {2026},
note = {Student research, Arizona State
University},
url = {https://github.com/aranish/wispec}
}Ranish, A. (2026). WiSpec: Commodity
Tri-Band Wi-Fi 6E Spectroscopy for Material
Classification and Structural
Reconnaissance [Software]. Student research
conducted at Arizona State University.Free for academic research, personal learning, and student thesis (with citation). Commercial use requires a paid license. Contact chatgpt@asu.edu for licensing inquiries.