Auth+Track: Enabling Authentication Free Interaction on Smartphone by Continuous User Tracking

Title of the Paper

Auth+Track: Enabling Authentication-Free Interaction on Smartphones by Continuous User Tracking

Paper Information

• Domain: User Authentication, Mobile Computing, Human-Computer Interaction
• Keywords: User Authentication, Continuous Tracking, Smartphones, User Experience, Seamless Interaction, Biometrics, Multimodal

Research Background and Problem

• Identified Problems or Challenges:

  1. Current user authentication on smartphones is a cumbersome and time-consuming process.
  2. Despite the widespread adoption of password and biometric authentication technologies, frequent unlocking still causes inconvenience for users.
  3. Existing implicit authentication methods lack accuracy, reliability, and the ability to capture user states effectively, often failing to eliminate redundant authentication.
  4. Most methods lack continuous tracking of user-device interaction history, leading to fragmented sessions and frequent authentication.

• Significance:

  1. Studies show that users spend approximately 2.6 minutes daily on authentication processes, with 24.1% of authentications being redundant.
  2. Unnecessary operations significantly impact user experience, especially in scenarios requiring frequent unlocking.

• Research Motivation and Related Work:

  • Motivation: To develop an intelligent interaction model that significantly reduces redundant authentication when users are consistently holding or near their smartphones.
  • Related Work: Previous studies have focused on biometric authentication (e.g., facial recognition, fingerprint) or implicit authentication (e.g., behavioral pattern tracking, device location recognition), but these approaches struggle to balance security and user convenience.

Solution

• Proposed Method or Solution:

  1. Introduced Auth+Track, a novel authentication model combining "gateway authentication" with "continuous user tracking."
  2. Extended the traditional authentication states by proposing a "User Around" state, where the device eliminates repeated authentication when it senses the user nearby.

• Innovations:

  1. Introduced the "User Master Device" state, offering a smarter and smoother lock-unlock logic.
  2. Proposed the PanoTrack prototype system for user tracking, leveraging panoramic vision technology to achieve real-time tracking of user body and hand states.
  3. Utilized both hand and body features, relying on a fisheye camera to detect user behavior within a range of ≤2 meters.

• Implementation Steps and Key Technologies:

  1. Used a fisheye camera to capture panoramic images while detecting key body points and hand states.
  2. Developed a continuous user tracking algorithm based on machine learning to accurately identify the relationship between hand, body, and device.
  3. Enhanced the user tracking model with hybrid strategies to handle scenarios involving "body-only tracking" and "hand-only tracking."

Research Outcomes

• Specific Results:

  1. The PanoTrack system achieved 99.5% accuracy and 94.7% recall in user tracking within a ≤2m range.
  2. The system significantly reduced redundant authentication in fragmented sessions, cutting the average smartphone unlocking time by more than half (1.45 seconds compared to 2.98 seconds).
  3. User feedback indicated high approval of the new authentication model, deeming it more efficient and reliable.

• Technical Advantages:

  1. Compared to existing implicit authentication technologies, PanoTrack offers a wider detection range, supports more user behaviors, and achieves higher accuracy.
  2. The hybrid strategy combining hand and body cues enhances tracking robustness.

• Experiments or Evaluation Results:

  • Experiment 1: Component Evaluation:
    • Body key point detection accuracy within ≤2m range approached 100%.
    • Hand state detection: 4-class classification accuracy reached 97.5%.
    • User identity assignment accuracy: 97% in desktop scenarios, 100% in handheld scenarios.
  • Experiment 2: Multi-Scenario Performance Testing:
    • In real-world scenarios such as laboratories, streets, and cafes, the system achieved a recall rate of 94.7% and a precision rate of 99.5% for user tracking.
  • Experiment 3: User Experience Evaluation:
    • User feedback showed positive results in terms of efficiency, usability, and user preference.
    • Compared to traditional authentication models, Auth+Track received higher approval ratings from users.

• Limitations and Future Directions:

  1. Limitations:
     • The current PanoTrack system relies on an external fisheye camera, and hardware integration needs improvement.
     • The system's robustness to lighting, occlusion, and usability in complex outdoor environments requires further testing.
     • Experiments were relatively idealized and did not fully cover diverse user behaviors and environments.
  2. Future Directions:
     • Further optimize algorithm efficiency and energy consumption to enable real-time deployment on mobile devices.
     • Integrate additional sensors, such as motion sensors (IMU), to address edge cases like pocket scenarios.
     • Incorporate privacy-preserving designs, such as edge computing, to ensure data is processed locally.