MobiCrowd: A Collaborative Location-Privacy Preserving Mobile Proxy

Reza Shokri, Panos Papadimitratos, Jean‐Pierre HubauxInfoscience 2010

Abstract

Location-Based Services (LBS) offer significant utility but often at the cost of user privacy, as they require sharing precise coordinates with potentially untrustworthy servers. This paper presents MobiCrowd, a collaborative LBS client that mitigates these risks by leveraging nearby peers. By fetching Point of Interest (POI) data from local buffers or neighboring devices via ad-hoc connections, MobiCrowd reduces the frequency of direct server queries. This "hiding in the crowd" approach creates gaps in tracking history, protecting users from continuous surveillance and risks like absence disclosure. The system features tunable parameters, allowing users to balance privacy, performance, and battery consumption.

Introduction

The rise of GPS-enabled smartphones has made LBS ubiquitous. However, the centralized nature of these services creates a privacy paradox: to find a nearby coffee shop, a user must reveal their exact location to a service provider. This data can be harvested to build detailed movement profiles. A particularly sensitive risk is absence disclosure—the ability for a provider to infer when a user is away from home, potentially facilitating physical security breaches. While obfuscation and centralized proxies exist, they often require changes to the LBS business model or fail to prevent long-term tracking. MobiCrowd proposes a user-centric alternative that requires no infrastructure changes.

The MobiCrowd Architecture

MobiCrowd operates as a local proxy on the mobile device, intercepting LBS queries and attempting to resolve them through a three-tiered hierarchy:

  1. Local Buffer: The device first checks its own cache (POI buffer) for previously retrieved data.
  2. Collaborative P2P Search: If the data is missing, the device uses an ad-hoc WiFi connection to broadcast the query to nearby peers. If a neighbor has the information in their buffer, they provide it directly.
  3. LBS Fallback: Only when local and peer resources are exhausted does the device contact the LBS server via cellular or infrastructure WiFi.

By satisfying queries locally or via peers, the user creates "blackouts" in their tracking history, making it much harder for a server to reconstruct their path or identify them through continuous monitoring.

Implementation and User Agency

The system was implemented on real-world hardware, specifically Nokia N810 and N900 devices, integrated with the Maemo Mapper application. A key innovation of MobiCrowd is the inclusion of tunable parameters that empower the user to manage their own privacy-utility trade-offs:

  • Privacy Metric: A real-time display of how "hidden" the user is from the LBS server.
  • Response Timeout: A setting to determine how long the device should wait for a peer response before giving up and querying the server.
  • Collaboration Level: A slider that allows users to control how often their own device helps others, balancing their desire for privacy against their device's battery life and resource costs.

Key Results and Insights

Simulations and real-world testing demonstrate that even moderate levels of peer collaboration can significantly disrupt a server's ability to track a user. The "power of omission" proves to be a highly effective privacy strategy; the most reliable way to protect data is to avoid sending it to the server in the first place.

Furthermore, the system leverages the latent power of ad-hoc networking. Modern smartphones are capable of direct communication, and MobiCrowd turns a crowd of strangers into a distributed, private database. The energy overhead for this collaboration remains low, as the cost of servicing peer queries is distributed across the network, and the system does not require LBS providers to change their APIs or infrastructure.

Limitations and Future Work

While promising, MobiCrowd faces several challenges:

  • Density Dependency: The system’s effectiveness relies on "crowd" density. In rural areas or sparsely populated times, the system falls back to the LBS server more frequently, offering less protection.
  • Data Freshness: The current implementation must address how the POI buffer handles rapidly changing data, such as live traffic or temporary events.
  • Peer Security: While the system protects users from the central server, it creates a transient window where peers can see what a user is searching for. While arguably safer than a permanent server log, malicious peers remain a potential research area.
  • Incentive Structures: Future work is needed to explore the game theory of collaboration, ensuring users are incentivized to help others even when their own battery is low.

Conclusion

MobiCrowd demonstrates that privacy does not have to be a zero-sum game between users and service providers. By leveraging collaborative, peer-to-peer proxies, users can reclaim control over their location data without sacrificing the benefits of modern LBS. This user-centric approach provides a viable path forward for privacy in an increasingly tracked world.

Summary updated

Abstract

We demonstrate the first, to the best of our knowledge, LBS client that safeguards user privacy against non-trustworthy LBS servers. Our collaborative approach allows the client to obtain responses from its near-by peers rather than submitting the query to the LBS. Our demo makes the user aware of her privacy level and offers tunable user parameters. The demo contributes a novel user-centric approach that enhances privacy without altering the business model of location-based services. Introduction. Smart phone users are now offered localization devices (integrated GPS receivers) or positioning services (e.g., based on near-by infrastructure). This enables a range of location-based services (LBS), allowing users to submit queries on their points of interest (POI) when they
BibTeX
@inproceedings{shokri2010mobicrowd,
  author = {Reza Shokri and Panos Papadimitratos and Jean‐Pierre Hubaux},
  title = {{MobiCrowd: A Collaborative Location-Privacy Preserving Mobile Proxy}},
  booktitle = {Infoscience},
  year = {2010},
}
Figure