A Trust Model Based on Service Classification in Mobile Services

Yang Liu, Zhikui Chen, Feng Xia, Xiaoning Lv, Fanyu Bu

Abstract

Internet of Things (IoT) and B3G/4G communication are promoting pervasive mobile services with advanced features. However, security problems continue to hinder development. This paper proposes a trust model to protect user security. A billing or trust operator acts as an agent to provide trust authentication for all service providers. Services are classified via sensitive value calculations, allowing for the determination of a user's trustiness for a corresponding service. For decision-making, three trust regions are defined: high, medium, and low. These regions dictate the authentication methods required for access based on the calculated trust value. Authentication history and penalties for failure are also integrated into the model.

Introduction

In the burgeoning world of the Internet of Things (IoT) and 4G/5G mobile services, security is a double-edged sword. Traditional security measures can be overly complex, creating friction that degrades the user experience. For instance, requiring a fingerprint scan to check a weather app is excessive, while relying on a simple cookie for a banking transaction is dangerous. This research identifies the need for a balanced approach—a system that understands the context of a service request and dynamically adjusts authentication requirements.

The Proposed Trust Model

Service Classification via Fuzzy Mathematics

Using the Analytic Hierarchy Process (AHP), the model creates a comparison matrix to rank services. By comparing targets (e.g., Governmental/Military vs. Entertainment), the system assigns a "sensitive value" ($S_i$) to each service. This process converts subjective importance into a numerical weight used for real-time decision-making.

Tiered Trust Regions

The model defines three distinct ranks that map to specific authentication requirements:

  • High Rank: No extra authentication needed (seamless access).
  • Medium Rank: Requires a PIN.
  • Low Rank: Requires biometrics (face, fingerprint, or iris scan).

Dynamic Threshold Adjustment

The system evaluates a user’s Authentication History to adjust trust regions. If a user has a consistent history of successful high-rank logins ($T_1$) or PIN entries ($T_2$), the trust region expands, facilitating easier access. Conversely, a poor history or lack of data makes the thresholds stricter.

Adjustment value as a function of authentication history percentage ($T_i$).

Penalty Coefficient and Brute-Force Protection

To prevent brute-force attacks, the authors introduce a penalty formula ($Y' = YP^n$). Each failed attempt exponentially reduces the user's trust value. This ensures that even a previously "trusted" user is forced into the biometric verification tier after repeated failures, effectively neutralizing stolen credentials like PINs.

Key Results

Simulations demonstrated that the model effectively balances convenience and security:

  1. History-Dependent Security: When authentication history is considered, the "High Trust" region for moderately sensitive services shrinks if the user lacks a long-term history of successful logins, making the system more conservative.
  2. Rapid Failure Response: Simulations showed that after repeated failed PIN attempts, a user’s trust value drops significantly from the "Medium" rank to the "Low" rank, requiring biometric proof of identity.
Trust value degradation and region shift after the fifth authentication failure.

Conclusion

This trust model provides a mathematically consistent way to map service sensitivity to authentication friction. By incorporating history and exponential penalties, it offers robust protection against identity compromise while maintaining user convenience for low-sensitivity tasks. Future work will focus on refining the penalty coefficients and testing the model in more complex, real-world mobile environments.

Summary updated

Abstract

Internet of Things (IoT) and B3G/4G communication are promoting the pervasive mobile services with its advanced features. However, security problems are also baffled the development. This paper proposes a trust model to protect the user's security. The billing or trust operator works as an agent to provide a trust authentication for all the service providers. The services are classified by sensitive value calculation. With the value, the user's trustiness for corresponding service can be obtained. For decision, three trust regions are divided, which is referred to three ranks: high, medium and low. The trust region tells the customer, with his calculated trust value, which rank he has got and which authentication methods should be used for access. Authentication history and penalty are also involved with reasons.
BibTeX
@inproceedings{liu2010trust,
  author = {Yang Liu and Zhikui Chen and Feng Xia and Xiaoning Lv and Fanyu Bu},
  title = {{A Trust Model Based on Service Classification in Mobile Services}},
  booktitle = {GreenCom/CPSCom},
  year = {2010},
  url = {https://openreview.net/forum?id=y2VHHLBLtz},
}
Figure