[Paper Review] A Lightweight Blockchain-based Privacy Protection for Smart Surveillance at the Edge

Alem Fitwi, Yu Chen+1 more|arXiv (Cornell University)|Sep 21, 2019

Blockchain Technology Applications and Security16 references5 citations

TL;DR

This paper proposes Lib-Pri, a lightweight blockchain-based privacy protection scheme for edge-based smart surveillance systems. It enables real-time, policy-driven privacy enforcement via on-device object detection and reversible blurring, with access control and audit logging managed through a private blockchain, ensuring privacy-preserving video analytics without network congestion.

ABSTRACT

Witnessing the increasingly pervasive deployment of security video surveillance systems(VSS), more and more individuals have become concerned with the issues of privacy violations. While the majority of the public have a favorable view of surveillance in terms of crime deterrence, individuals do not accept the invasive monitoring of their private life. To date, however, there is not a lightweight and secure privacy-preserving solution for video surveillance systems. The recent success of blockchain (BC) technologies and their applications in the Internet of Things (IoT) shed a light on this challenging issue. In this paper, we propose a Lightweight, Blockchain-based Privacy protection (Lib-Pri) scheme for surveillance cameras at the edge. It enables the VSS to perform surveillance without compromising the privacy of people captured in the videos. The Lib-Pri system transforms the deployed VSS into a system that functions as a federated blockchain network capable of carrying out integrity checking, blurring keys management, feature sharing, and video access sanctioning. The policy-based enforcement of privacy measures is carried out at the edge devices for real-time video analytics without cluttering the network.

Motivation & Objective

• To address the growing concern over privacy violations in pervasive video surveillance systems.
• To develop a lightweight, secure, and real-time privacy-preserving solution for edge-based surveillance cameras.
• To integrate blockchain technology with edge computing and object detection for decentralized, auditable privacy enforcement.
• To enable fine-grained access control and tamper-proof logging of video access while preserving data integrity.
• To support opt-out mechanisms for scenarios where privacy is less critical, such as emergency response or proctoring.

Proposed method

• Deploying smart cameras with embedded single-board computers (e.g., Jetson Nano, Raspberry Pi) for on-edge video processing.
• Using pre-trained deep neural networks for real-time object detection to identify privacy-sensitive regions like faces, windows, and private body parts.
• Applying reversible image scrambling (chaotic masking) to blur sensitive objects directly at the edge, preserving data for authorized recovery.
• Storing blurring keys and access policies on a private blockchain to ensure authenticity and access control via smart contracts.
• Logging viewer identity, access time, and video references on the blockchain to enable end-to-end auditability.
• Implementing an off-blockchain storage system for reversibly scrambled videos, accessible only to authorized users with proper privileges.

Experimental results

Research questions

• RQ1 How can privacy be preserved in real-time smart surveillance systems without overloading network bandwidth?
• RQ2 What role can lightweight blockchain play in enabling decentralized, auditable access control for surveillance video?
• RQ3 How can object detection and reversible blurring be effectively integrated at the edge to protect sensitive visual content?
• RQ4 In what scenarios can an opt-out mechanism be safely applied without compromising system integrity?
• RQ5 How can viewer identity and access to video data be securely tracked and logged without leaking sensitive information?

Key findings

• The Lib-Pri system successfully performs real-time privacy enforcement at the edge using on-device object detection and reversible blurring, minimizing network traffic.
• Smart contracts on a private blockchain manage access privileges and log all video access events, ensuring auditability and accountability.
• Blurring keys are securely shared and stored on the blockchain, enabling authorized recovery of original video frames only when legally justified.
• The system supports dynamic policy configuration per camera or zone, allowing flexible deployment across diverse surveillance environments.
• The integration of viewer-specific metadata embedding and blockchain logging reduces the risk of unauthorized video leakage.
• An opt-out mechanism is implemented to disable privacy protection when necessary, such as in emergency or high-priority monitoring scenarios.