Optimizing Security and Efficiency in Edge-Cloud Collaborative Environments: A Service-Aware Privacy Protection Approach

Abstract

As intelligent home services increasingly adopt IoT-Edge-Cloud architectures, protecting personal data during third-party data sharing has become an important challenge. This paper presents a Service-Aware Trust Chain framework that applies Privacy-Enhancing Technologies (PETs) according to data sensitivity. To support this approach, we introduce a 7-level security grading scheme based on Korea’s Personal Information Protection Act (PIPA) and practical deployment requirements in edge environments.A proof-of-concept system was implemented on a Raspberry Pi 5, integrating five privacy protection engines: ARX-based deidentification, Clusion searchable encryption, Microsoft SEAL homomorphic encryption, video face blurring, and sound masking. Performance measurements show that ARX and searchable encryption can be executed with low latency at the edge, whereas video and audio protection mechanisms are more suitable for batch-oriented processing. Homomorphic encryption incurs substantially higher overhead and is therefore better suited to a cloud-assisted deployment model. In addition, sound masking exhibited limited performance degradation as audio length increased, indicating its practicality for voice privacy protection in smart-home environments.To support data integrity verification, only cryptographic hashes are recorded on a cloud-hosted Trust Chain, avoiding the exposure of raw data. The proposed framework also outlines future extensions including sLLM-based data classification and DID-based access control.