Confidentiality and practicality of trusted execution environments for cross-organizational multi-model consistency checking

abstract

Confidentiality-preserving consistency checking is essential in cross-organizational engineering projects, where model data must be jointly processed although participating parties may not trust one another or want to disclose their data. Such processing can create confidentiality risks when data is accessed, transferred, or temporarily stored outside its owner’s controlled environment. Trusted Execution Environments (TEEs) may reduce these risks through isolated computation and remote attestation, but their security and practicality for this domain remain insufficiently understood. This article does not propose a new TEE security mechanism or architecture. Instead, based on an explicit system and threat model, we systematically identify vulnerabilities across the multi-model consistency-checking workflow and map them to established TEE guarantees. The analysis shows that TEEs mitigate critical threats, particularly during runtime processing, but do not fully address leakage and trust-establishment risks during setup, interaction, and result disclosure. Some remaining vulnerabilities require complementary safeguards, including attestation-based provisioning, distributed trust, or controlled result disclosure. A prototype implementation evaluates the practical overhead and scalability of TEE-protected consistency checking and demonstrates its feasibility for representative workloads. The resulting domain-specific assessment clarifies the security guarantees, limitations, and practical applicability of TEEs for confidential multi-model consistency checking.