Privacy-Preserving Device Counting Using Wi-Fi Channel State Information and Deep Learning
Bibliographic record
Abstract
As smartphones and other Wi-Fi-Enabled devices become increasingly common, they offer a practical way to estimate crowd size by collecting the probe request frames these devices periodically transmit. While this passive approach avoids relying on cameras or other intrusive sensors, the adoption of MAC address randomization—designed to protect user privacy—makes it difficult to reliably count how many unique devices are present. This paper presents a machine learning–based approach for device counting that leverages Channel State Information (CSI) features extracted from probe frames. Our method enables accurate device population estimation while preserving MAC-level privacy and avoiding persistent tracking or behavioral profiling. A Siamese neural network is trained to learn a discriminative similarity function between packet pairs, allowing the grouping of temporally co-occurring packets likely originating from the same device. To address the challenge of the limited number of packets available from each randomized MAC address, we use a lightweight augmentation strategy that interpolates between CSI samples to increase training density. We evaluate our approach in indoor and outdoor settings with varying device probing behaviors under static conditions. By leveraging packet-level voting and device-counting strategies, our model leverages packet-level voting and device-counting strategies to consistently achieve high counting accuracy—averaging over 98% in indoor environments and reaching 100% outdoors—while preserving device privacy by avoiding behavioral tracking and re-identification.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".