Real-Time Index Authentication for Event-Oriented Surveillance Video Query using Blockchain

Seyed Yahya Nikouei, Ronghua Xu, Deeraj Nagothu, Yu Chen, Alexander J. Aved, Erik Blasch

Abstract

As smart city infrastructures expand, the volume of surveillance data poses significant challenges for centralized cloud-based processing, particularly regarding latency and security. This paper presents a decentralized framework that leverages edge computing and blockchain technology to secure event-oriented surveillance video queries. By implementing a three-tier Edge-Fog-Cloud architecture, the system performs local video indexing and secures these indexes using a private Ethereum blockchain. Experimental evaluations on a physical testbed demonstrate that the proposed approach ensures data integrity and authenticity with negligible latency, providing a scalable solution for real-time smart surveillance.

Introduction

Modern urban environments are increasingly reliant on smart surveillance systems. However, the traditional model of streaming raw high-definition video to a central cloud for analysis is becoming unsustainable due to bandwidth constraints and high latency. Furthermore, centralized architectures present a single point of failure and are vulnerable to "Man-in-the-Middle" (MitM) attacks, where adversaries can alter video metadata or inject false records. While edge computing addresses latency by processing data locally, edge devices are often resource-constrained and susceptible to compromise. This research addresses the critical need for a trustless mechanism to verify the integrity of indexing data generated at the edge.

System Architecture

The researchers propose a hierarchical Edge-Fog-Cloud model designed to distribute computational tasks and enhance security. The system is divided into three functional layers:

  1. Edge Layer: Comprised of low-power devices (e.g., Asus Tinker Boards) that perform initial object detection and tracking using lightweight Convolutional Neural Networks (CNNs).
  2. Fog Layer: Intermediate nodes that aggregate data from multiple edge devices, classify behaviors, and generate searchable key-value indexes for the video.
  3. Cloud Layer: Serves as a long-term archive and high-level policy engine for historical data analysis.
The decentralized system architecture integrates Cloud, Fog domains, and a Blockchain Network to ensure secure data synchronization and index authentication.

Security and Authentication Mechanisms

To secure the data pipeline, the framework employs two primary mechanisms: hybrid encryption and blockchain-based anchoring.

Hybrid Encryption

Data transmitted between Edge and Fog nodes is protected via a hybrid scheme. It combines RSA for secure asymmetric key exchange and AES for efficient symmetric encryption of the data payload. This ensures that even if an edge device is compromised, the communication channel remains secure against network sniffing.

The hybrid encryption protocol utilizes RSA for secure key exchange followed by AES for high-speed data transfer between Edge and Fog nodes.

Blockchain Anchoring

The core innovation lies in the use of a private Ethereum network to authenticate video indexes. Instead of storing large video files on-chain, Fog nodes generate a cryptographic hash of the video index and record it via a Smart Contract. This "anchoring" of trust allows users to verify the authenticity of any queried video segment by comparing its current index hash against the immutable record on the blockchain. Additionally, every node is assigned a Virtual Identity (VID) derived from its blockchain address, eliminating the need for a centralized certificate authority.

Experimental Evaluation

The system was validated using a physical testbed featuring Asus Tinker Boards and a private Ethereum mining cluster. Testing involved real-time pedestrian tracking in a university setting. Key performance results include:

  • Low Latency: The authentication process adds only 2.3ms of overhead, making it suitable for real-time use.
  • Query Efficiency: Index queries on fog nodes are highly optimized, taking approximately 6ms, compared to 53ms on edge nodes.
  • Robustness: The hybrid encryption scheme successfully secured the channel without introducing significant jitter or delays in data transfer.

Conclusion

This research demonstrates that blockchain technology can be effectively integrated into edge-based surveillance systems to provide decentralized trust. By storing only cryptographic hashes on the blockchain, the system achieves high security with minimal performance impact. Future work will explore more complex event-oriented indexing and the scalability of consensus mechanisms for city-wide deployments.

Summary updated

BibTeX
@inproceedings{nikouei2018real,
  author = {Seyed Yahya Nikouei and Ronghua Xu and Deeraj Nagothu and Yu Chen and Alexander J. Aved and Erik Blasch},
  title = {{Real-Time Index Authentication for Event-Oriented Surveillance Video Query using Blockchain}},
  booktitle = {ISC2},
  year = {2018},
  doi = {10.1109/isc2.2018.8656668},
}
Figure