Secure Pick Up: Implicit Authentication When You Start Using the Smartphone

Wei-Han Lee, Xiaochen Liu, Yilin Shen, Hongxia Jin, Ruby B. Lee

TL;DR

Secure Pick Up (SPU) is a lightweight, in-device authentication system that identifies smartphone users by the unique way they move their arms when picking up their phones. By using a weighted multi-dimensional Dynamic Time Warping (DTW) algorithm, SPU achieves 96.3% accuracy with negligible battery drain, reducing the need for manual PIN or fingerprint entries by nearly a third.

What Problem Does This Paper Address?

We unlock our phones dozens of times a day, yet the friction of traditional authentication (PINs, patterns, and even fingerprints) is high enough that roughly 64% of users choose to leave their devices unprotected. While biometric and behavioral authentication (like gait analysis or keystroke dynamics) exist, they often suffer from significant drawbacks: they require massive amounts of training data, rely on privacy-invasive cloud processing, or demand specific user actions (like walking in a straight line).

The authors of this paper address the "usability-security" gap by looking for a behavioral signature that happens naturally every time you use your phone: the act of picking it up. The goal was to create a system that is implicit (happens in the background), in-device (no cloud required), and adaptive (learns as your habits change).

Methods

The core of SPU is Weighted Multi-dimensional Dynamic Time Warping (DTW).

  1. Signal Extraction: The system monitors the accelerometer and gyroscope in the background. When a "trigger-action" occurs (like pressing the power or home button), SPU backtracks through the sensor data to find the "pick-up" segment—the transition from a stable state to the moment the screen wakes up.
  2. The Algorithm: DTW is a well-known method for measuring similarity between two temporal sequences that may vary in speed. SPU improves on standard DTW by applying different weights to different sensor axes. Through empirical testing, the researchers found that the accelerometer is more "informative" than the gyroscope for this specific movement, leading to a 0.6 to 0.4 weight ratio.
  3. Adaptive Updating: To handle "behavioral drift" (the fact that you might pick up your phone differently when tired or in a rush), the system updates the user's profile. If a user fails the implicit check but passes a backup explicit check (like a PIN), SPU averages the new movement into the stored profile.

Datasets

The researchers evaluated SPU using three distinct datasets collected from 24 participants (14 male, 10 female, average age 34.3):

  • Lab Dataset: 2,880 samples collected under controlled conditions (6 different locations, sitting vs. standing).
  • Real-World Dataset: 3,115 samples collected over a week of "natural" use on the participants' own devices.
  • Security/Attack Dataset: To test robustness, they recruited 12 "adversaries" to attempt three types of attacks: Random (brute force), Context-Aware (knowing where the phone was picked up), and Educated (mimicking the victim after watching a video of their movement).

Key Insights and Ideas

  • The "Stable State" Constraint: SPU specifically targets movements starting from a stable state (e.g., a table or nightstand). While this only covers about 35.6% of all unlocks (excluding picking up from a pocket or bag), it provides the most reliable behavioral signature.
  • More Sensors $\neq$ Better Performance: In a surprising "aha" moment, the researchers found that adding a magnetometer actually degraded accuracy. Because magnetic fields are highly sensitive to the phone's orientation and environment, the data was too noisy to be useful for authentication.
  • Asymmetric Weights: Not all movement data is equal. The study proved that translational movement (accelerometer) is a stronger identifier for "picking up" than rotational movement (gyroscope).
  • Privacy by Design: Because the system runs entirely on-device and doesn't require "negative" training data (data from other users), it avoids the privacy pitfalls of many machine-learning-based biometric systems.

Key Results

  • High Accuracy: SPU achieved a peak authentication accuracy of 96.3%.
  • Ultra-Low Latency: The decision-making process takes only 2.4 milliseconds, making it virtually instantaneous to the user.
  • Friction Reduction: In real-world testing, the system reduced the number of times users had to manually unlock their phones by 32.9%.
  • Security: Against "Educated" mimicry attacks (the toughest scenario), SPU maintained a 0% False Acceptance Rate (FAR) at certain thresholds, though this came at the cost of a higher False Rejection Rate (18%).
  • Efficiency: The system adds only 2% to the total daily power consumption of the smartphone.

Limitations and Future Work

The primary limitation is that SPU currently only functions when the phone is picked up from a stable surface. It cannot yet implicitly authenticate a user pulling a phone out of a pocket or a moving bag, as those signals are too inconsistent.

The authors suggest that future work could involve:

  • Multi-user profiles: Allowing the system to distinguish between family members or guests.
  • Cross-device application: Extending the model to tablets and smartwatches.
  • Context-Awareness: Integrating more sophisticated environment detection to handle a wider variety of "start" states beyond just the stable surface.

Summary updated

Abstract

We propose Secure Pick Up (SPU), a convenient, lightweight, in-device, non-intrusive and automatic-learning system for smartphone user authentication. Operating in the background, our system implicitly observes users' phone pick-up movements, the way they bend their arms when they pick up a smartphone to interact with the device, to authenticate the users. Our SPU outperforms the state-of-the-art implicit authentication mechanisms in three main aspects: 1) SPU automatically learns the user's behavioral pattern without requiring a large amount of training data (especially those of other users) as previous methods did, making it more deployable. Towards this end, we propose a weighted multi-dimensional Dynamic Time Warping (DTW) algorithm to effectively quantify similarities between users' pick-up movements; 2) SPU does not rely on a remote server for providing further computational power, making SPU efficient and usable even without network access; and 3) our system can adaptively update a user's authentication model to accommodate user's behavioral drift over time with negligible overhead. Through extensive experiments on real world datasets, we demonstrate that SPU can achieve authentication accuracy up to 96.3% with a very low latency of 2.4% milliseconds. It reduces the number of times a user has to do explicit authentication by 32.9%, while effectively defending against various attacks.
BibTeX
@inproceedings{lee2017secure,
  author = {Wei-Han Lee and Xiaochen Liu and Yilin Shen and Hongxia Jin and Ruby B. Lee},
  title = {{Secure Pick Up: Implicit Authentication When You Start Using the Smartphone}},
  booktitle = {SACMAT},
  year = {2017},
  url = {https://openreview.net/forum?id=4zP_VlGkF0},
}

Loading related content...

Figure