VIMA: A Privacy-Preserving Integrity Measurement Architecture for Containerized Environments

Integrity verification and attestation are critical in containerized environments, where traditional Linux Integrity Measurement Architecture (IMA) falls short due to its lack of container-specific contextualization. These gaps undermine container autonomy, escalate privacy risks, and impede granular integrity checks. Addressing these challenges, this paper introduces the Virtual IMA (VIMA), a novel framework that refines Linux IMA’s principles to support containerized settings. Using nested Merkle trees, VIMA’s Two-Tree Architecture (2TA) enables detailed integrity assessments across system-wide monolithic trees and individual container trees. Integrating Merkle and zero-knowledge (ZK) proofs establishes VIMA as a secure, privacy-preserving verification and attestation solution. Our comparative analysis and initial prototype testing reveal that VIMA significantly improves upon traditional IMA with minimal performance overhead, offering substantial scope for optimization.