[Paper Review] Location Privacy in Mobile Edge Clouds: A Chaff-based Approach

Ting He, Ertuğrul Necdet Çiftçioğlu+2 more|arXiv (Cornell University)|Sep 10, 2017

Privacy-Preserving Technologies in Data18 references3 citations

TL;DR

This paper proposes a chaff-based defense mechanism to protect user location privacy in mobile edge clouds (MECs), where eavesdroppers can track users by monitoring service migrations between MECs. By strategically deploying chaff services that mimic legitimate user mobility, the optimal strategy reduces the eavesdropper’s tracking accuracy to zero when user mobility is sufficiently random, even against advanced eavesdroppers using maximum likelihood detection.

ABSTRACT

In this paper, we consider user location privacy in mobile edge clouds (MECs). MECs are small clouds deployed at the network edge to offer cloud services close to mobile users, and many solutions have been proposed to maximize service locality by migrating services to follow their users. Co-location of a user and his service, however, implies that a cyber eavesdropper observing service migrations between MECs can localize the user up to one MEC coverage area, which can be fairly small (e.g., a femtocell). We consider using chaff services to defend against such an eavesdropper, with focus on strategies to control the chaffs. Assuming the eavesdropper performs maximum likelihood (ML) detection, we consider both heuristic strategies that mimic the user's mobility and optimized strategies designed to minimize the detection or tracking accuracy. We show that a single chaff controlled by the optimal strategy or its online variation can drive the eavesdropper's tracking accuracy to zero when the user's mobility is sufficiently random. We further propose extended strategies that utilize randomization to defend against an advanced eavesdropper aware of the strategy. The efficacy of our solutions is verified through both synthetic and trace-driven simulations.

Motivation & Objective

• To address the threat of cyber eavesdropping in mobile edge clouds, where service migrations reveal user locations through side channels.
• To protect user location privacy when the MEC provider is untrusted and may monitor service migrations.
• To design chaff service strategies that minimize the eavesdropper’s ability to track the real user using maximum likelihood detection.
• To enhance robustness against advanced eavesdroppers who are aware of the chaff strategy and can adapt their detection.

Proposed method

• The authors model user mobility as a Markov chain and use it to generate chaff service trajectories that mimic real user movement patterns.
• They propose heuristic strategies (IM, ML, OO, MO) and an optimal strategy based on solving a Markov decision process (MDP) to minimize detection accuracy.
• The optimal strategy is derived using dynamic programming to minimize the eavesdropper’s likelihood of correctly identifying the real user.
• An online variant of the optimal strategy is introduced to handle real-time deployment without prior knowledge of future mobility.
• Robust strategies (RML, ROO) are designed using randomization to defend against eavesdroppers aware of the chaff strategy.
• Performance is evaluated using both synthetic mobility models and real-world taxi cab traces with 959 Voronoi cells and 174 user trajectories.

Experimental results

Research questions

• RQ1 Can chaff services effectively obscure a user’s true location from an eavesdropper monitoring service migrations in MECs?
• RQ2 Under what conditions can the optimal chaff strategy reduce the eavesdropper’s tracking accuracy to zero?
• RQ3 How does the performance of chaff strategies vary under different user mobility models, including spatially and temporally skewed patterns?
• RQ4 Can robust chaff strategies defend against an advanced eavesdropper who knows the chaff deployment strategy?
• RQ5 How does the number of chaff services affect the trade-off between privacy and system cost?

Key findings

• The optimal chaff strategy and its online variant can reduce the eavesdropper’s tracking accuracy to zero when the user’s mobility has sufficient entropy.
• Under a basic eavesdropper using maximum likelihood detection, the ML and OO strategies significantly reduce tracking accuracy compared to baseline, even with only one chaff.
• The MO strategy performs poorly when the user’s trajectory dominates the likelihood of the myopic path, as it fails to deviate from the maximum likelihood location.
• With two chaffs, the original strategies (IM, ML, OO, MO) fail against an advanced eavesdropper, but robust strategies RML and ROO substantially reduce tracking accuracy.
• Trace-driven simulations using real taxi cab mobility data confirm that the proposed strategies maintain high privacy protection even in spatially- and temporally-skewed mobility environments.