From Packets to Patterns: Interpreting Encrypted Network Traffic as Longitudinal Behavioral Signals

arXiv cs.LG / 5/5/2026

📰 NewsSignals & Early TrendsModels & Research

共有:

Key Points

The paper investigates whether encrypted smartphone network traffic can be used as a passive, always-on sensing signal to detect longitudinal behavioral patterns tied to sleep disturbance, stress, and loneliness.
It builds a model that uses a transformer backbone with per-user adapters to capture both typical individual behavior and deviations from that baseline over time.
To improve interpretability, the authors apply a sparse autoencoder to extract distinct behavioral features from learned traffic representations.
Using generalized estimating equations with Mundlak decomposition, the study distinguishes stable between-person differences from within-person changes and finds different temporal drivers for each outcome (stress: mostly between-person, loneliness: mostly within-person, sleep disturbance: both).
The results show that important within-person dynamics are not captured by predefined network-traffic features, highlighting the advantage of learned representations for longitudinal behavioral sensing.

Abstract

Human behavior is difficult to observe continuously at scale, yet it leaves measurable traces in everyday device use. We test whether encrypted smartphone network traffic -- a ubiquitous, always-on, passive sensing modality -- can passively capture behavioral patterns related to sleep, stress, and loneliness. We model shared behavioral structure using a transformer backbone with per-user adapters, allowing the model to represent both typical individual behavior and deviations from it. To make these representations interpretable, we apply a sparse autoencoder to extract behavioral features corresponding to distinct patterns of activity. We relate these features to sleep disturbance, stress, and loneliness using generalized estimating equations with Mundlak decomposition, separating between-person differences from within-person changes over time. We find that the three outcomes reflect distinct temporal structures: stress is primarily associated with stable between-person differences, loneliness with within-person variation, and sleep disturbance with a combination of both. Notably, these within-person dynamics are not captured by predefined network-traffic features, demonstrating the value of learned representations for longitudinal behavioral sensing. These results establish encrypted network traffic as a viable passive sensing modality, revealing interpretable behavioral dynamics -- particularly deviations from an individual's baseline -- that are not visible in raw traffic features.