MobChain: Three-Way Collusion Resistance in Witness-Oriented Location Proof Systems Using Distributed Consensus.

Faheem Zafar, Abid Khan, Saif Ur Rehman Malik, Mansoor Ahmed, Carsten Maple, Adeel AnjumSensors 2021 · OpenReview

TL;DR

This paper introduces "MobChain," a novel decentralized system for generating secure location proofs that specifically addresses the "three-way collusion" problem where the prover, witness, and location authority all conspire. By leveraging blockchain and distributed consensus for participant selection, MobChain enhances security and trustworthiness in location-based services without significantly compromising performance.

What Problem Does This Paper Address?

Location-based services (LBS) are increasingly prevalent, relying on users' geographical locations. To access these services, users often need to provide "location proofs" – digital certificates verifying their presence at a specific place and time. However, these proofs are vulnerable to tampering and collusion. Existing systems have struggled with collusion attacks, particularly the "three-way collusion" where the prover, location authority (LA), and witness (if present) all act maliciously to generate a fake proof. This is a critical gap because if these three key participants collude, existing witness-oriented schemes become insecure. The paper aims to fill this gap by proposing a system that actively resists this specific type of collusion.

Methods

The core of MobChain's innovation lies in its decentralized approach to participant selection and its use of blockchain. Instead of relying on a centralized authority or allowing the prover to choose participants, MobChain employs a distributed consensus mechanism among "supervisor nodes" in an "admin layer."

Here's a breakdown:

  • Decentralized Participant Selection: When a prover requests a location proof, supervisor nodes in the admin layer engage in a distributed consensus process to select a suitable LA and witness from the available co-located entities. This selection is based on factors like uptime and the number of previously entertained requests, aiming to favor honest participants.
  • Blockchain for Provenance and Decisions: Two blockchains are utilized:
  • Decision Blockchain: Records the consensus decisions made by supervisor nodes for participant selection. This ensures transparency and immutability of the selection process.
  • Location Provenance Blockchain: Stores the generated location proofs chronologically, creating a tamper-evident history of a user's movements.
  • Witness-Oriented Protocol: The system builds upon witness-oriented LPS, where a witness verifies the prover's location and the LA's assertion. However, MobChain enhances this by decentralizing the selection of these crucial roles.
  • Separation of Concerns: The system separates the control of participant selection from the actual location proof generation, making it harder for malicious parties to orchestrate a collusion.

Datasets

Not discussed in the paper. The paper focuses on a theoretical framework and a proof-of-concept implementation.

Key Insights and Ideas

  • Three-Way Collusion is the Achilles' Heel: The paper clearly identifies three-way collusion as the most significant vulnerability in existing witness-oriented LPS, and this is the primary problem MobChain aims to solve.
  • Decentralized Selection is Key to Collusion Resistance: By removing the control of participant selection from the prover or a single LA, MobChain significantly hinders the ability of malicious parties to orchestrate a collusion. The consensus mechanism among supervisor nodes makes it computationally infeasible for a prover to compromise enough nodes to select colluding participants.
  • Blockchain as a Foundation for Trust: The use of blockchain for both decision-making and provenance provides a robust, tamper-evident audit trail, enhancing the overall trustworthiness of location proofs.
  • Performance Trade-offs: While MobChain enhances security, the decentralized consensus mechanism introduces some overhead, particularly in decentralized decision time. However, the paper demonstrates that this overhead is manageable and competitive with existing solutions.

Key Results

The paper presents a proof-of-concept (POC) implementation and experimental evaluation. Key findings include:

  • Three-Way Collusion Resistance: MobChain is claimed to be the first witness-oriented LPS that is resistant to three-way collusion.
  • Competitive Performance: The POC application, developed using Java AKKA, shows that MobChain's performance, even with the overhead of decentralized decision-making, competes with state-of-the-art solutions like WORAL.
  • Decentralized Decision Time (DDT): The DDT increases with the number of active workers and the consensus threshold, but remains within practical limits. For instance, with 1600 active workers, DDT is around 12365 ms.
  • Proof Generation Time (PGT): PGT is also affected by active workers and consensus threshold, but the impact of secure localization is noted as a significant component.
  • Storage Requirements: Location proof size is largely independent of the consensus threshold, while decision block size increases with the consensus threshold and key size. MobChain's storage requirements are noted as being high, especially for supervisor nodes.
  • Comparison with Existing Techniques: Table 4 shows MobChain has a comparable proof generation time to WORAL (<1 sec) and a similar number of entities involved (3), but a larger proof size (≥ 4.7 KB vs. <2 KB for WORAL).

Limitations and Future Work

  • Storage Requirements: The paper acknowledges that MobChain has high storage requirements, particularly for supervisor nodes, which is a limitation for smart devices.
  • Performance under High Concurrent Requests: While the POC shows performance, the paper notes that a high number of concurrent requests to a single supervisor node can degrade performance, suggesting a need for further optimization.
  • Real-World Deployment: The current implementation is a POC, and transforming it into a full application for real-world testing is a future goal.
  • Optimization of Participant Selection: Further work is needed to optimize the witness and LA selection algorithms to reduce the impact of active workers on the network.
  • Storage Optimization: Efforts are planned to optimize location proof and decision block sizes to reduce storage requirements, especially on mobile devices.
  • Parallelization and Load Balancing: Techniques like parallelization and load balancing are suggested to mitigate performance degradation under high concurrent requests.

Summary updated

BibTeX
@inproceedings{zafar2021mobchain,
  author = {Faheem Zafar and Abid Khan and Saif Ur Rehman Malik and Mansoor Ahmed and Carsten Maple and Adeel Anjum},
  title = {{MobChain: Three-Way Collusion Resistance in Witness-Oriented Location Proof Systems Using Distributed Consensus.}},
  booktitle = {Sensors},
  year = {2021},
  url = {https://openreview.net/forum?id=gsBTmUbi1O},
}
Figure