An Improved Trust-Free Approach for Preserving Privacy in Internet of Things

Sangwoo Park, Soowoong Lee, JooSeok SongInternational Journal on Cybernetics & Informatics 2022 · DOI

In the rapidly expanding world of the Internet of Things (IoT), our devices are constantly whispering our secrets—location data, health metrics, and financial habits—to service providers. Traditionally, protecting this data required us to trust either the service provider itself or a "Trusted Third Party" (TTP). But what if we didn't have to trust anyone?

TL;DR

The Enhanced Blind Peer Approach (EBLP) is a trust-free privacy framework for IoT that allows users to anonymize their data through untrusted peers. By introducing a "feasible moving area" selection policy, EBLP fixes a critical flaw in previous models: vulnerability to trajectory attacks, where service providers could track a user's path even after an identity swap.

What Problem Does This Paper Address?

Privacy in IoT is a balancing act. Current methods like data obfuscation or dummy queries often result in inaccurate service results, while homomorphic encryption is often too computationally expensive for tiny IoT sensors.

The most common solution is relying on a Trusted Third Party (TTP) to anonymize data. However, this creates a single point of failure; if the TTP is compromised or malicious, your privacy is gone. A newer concept, the Blind Peer Approach (BLP), attempted to remove the TTP by having users swap identities with "blind peers" (other users who don't need to be trusted).

The problem? BLP was a half-baked concept. It lacked a formal protocol and was wide open to trajectory attacks. If a service provider sees User A at Point 1 and then suddenly sees "User B" at Point 2 (which is exactly where User A was headed), it’s trivial to deduce that User A simply changed their name.

Methods

The authors propose EBLP, which formalizes the blind peer process and introduces a clever selection strategy to thwart tracking.

The "Feasible Moving Area" (FMA)

The core innovation is the FMA. This is the physical area a user could realistically reach within a specific timeframe. Instead of picking any random peer to swap identities with, EBLP only selects peers located within this FMA.

The Two-Phase Protocol:

  1. Blind Peer Selection Phase: The user broadcasts their position to nearby peers. Only peers within the FMA respond. The user then randomly selects one of these "feasible" peers to act as their proxy.
  2. Masquerading Phase:
    • The user encrypts their query with a session key and the Service Provider’s (SP) public key.
  3. The selected blind peer forwards this encrypted blob to the SP.
  4. The SP decrypts the query, processes it, and encrypts the result using the session key provided in the query.
  5. The result is sent back to the peer, who forwards it to the user.

Because the peer is "blind," they cannot see the contents of the query or the result. Because the peer was chosen from the FMA, the SP sees a "confused trace" that looks like natural movement, making it impossible to distinguish the original user from the peer.

Datasets

The authors do not use empirical datasets or real-world sensor logs in this paper. Instead, the evaluation is based on theoretical modeling and comparative analysis of communication and computational overhead against existing privacy methods (Anonymity, Dummy, Clocking Area, and Cooperation).

Key Insights and Ideas

  • Trust-Free is the Future: By removing the need for a TTP, EBLP decentralizes privacy. You don't need to find a "good" actor; you just need a "blind" one.
  • Physicality Matters in Digital Privacy: The "aha moment" here is realizing that identity anonymization fails if it ignores the laws of physics. By ensuring that identity swaps only happen between users who could plausibly be each other, the "trajectory" of the data remains consistent with real-world movement.
  • One-Time Setup, Continuous Privacy: While the selection phase adds a bit of initial overhead, it only needs to happen once per session, making the long-term cost of continuous queries very low.

Key Results

The paper evaluates EBLP across four major attack vectors:

  • Resilience: EBLP is shown to be resilient against Semantic Context, Trajectory, Historical, and Inversion attacks. Notably, it is the only trust-free method listed that successfully defeats trajectory attacks.
  • Communication Overhead: EBLP requires $2n + a$ communications (where $n$ is the number of queries and $a$ is the selection phase). While slightly higher than the original BLP ($2n$), the authors argue the added security is worth the negligible cost of the selection phase.
  • Computational Cost: The number of calculations is higher than simpler methods like "dummies" because it involves encryption and peer selection logic, but it remains well within the "acceptable" range for modern IoT hardware.

Limitations and Future Work

The authors acknowledge that the definition of "Feasible Moving Area" is currently a bit abstract. The FMA for a pedestrian is very different from the FMA for a car or a drone.

Future directions include: * Developing specialized FMAs for specific applications like M-Health or Smart Cities. * Analyzing how the system performs in high-density environments where hundreds of peers might be in the FMA. * Refining the protocols for specific IoT communication standards.

Reference: Park, S., Lee, S., & Song, J. (2022). An Improved Trust-Free Approach for Preserving Privacy in Internet of Things. International Journal on Cybernetics & Informatics (IJCI), 11(4).

Summary updated

Abstract

As the researches for Internet of things(IoT) are very active, those for preventing privacy leakage in IoT is also getting important. Various methods have been proposed, but each has its own drawbacks such as reliance on the service provider or a trusted third party. Recently a new method called Blind Peer Approach (BLP) has been proposed to resolve the drawbacks by cooperating untrusted blind peer instead of trusted third party. Blind Peer Approach, however, is still in the early stage, therefore the detailed procedure is missing. In addition, the method still has a critical issue, which is the fact that BLP is vulnerable to trajectory attack. In this paper, we propose Enhanced Blind Peer Approach (EBLP) and its detailed procedures. This method is resilient to the trajectory attack by carefully selecting the blind peer in the feasible moving area. We analyse the resiliency for some privacy attacks for EBLP. Our analysis shows that EBLP has acceptable communication and computational costs.
BibTeX
@article{park2022improved,
  author = {Sangwoo Park and Soowoong Lee and JooSeok Song},
  title = {{An Improved Trust-Free Approach for Preserving Privacy in Internet of Things}},
  journal = {International Journal on Cybernetics \& Informatics},
  year = {2022},
  doi = {10.5121/ijci.2022.110412},
}
Figure