Arithmetic consistency attack-resistant integrity verification for secure outsourced computing

Abstract

Secure outsourced computing (SOC) enables clients to offload data and computational tasks to cloud servers, thereby liberating them from the constraints of storage and computational resources. As a novel computational paradigm, it has garnered widespread attention for its ability to offer significant advantages while ensuring data confidentiality and integrity. To address the issues of high computational cost, lack of bidirectional verification capability, and vulnerability to attacks in existing integrity verification schemes based on homomorphic encryption (HE), this paper proposes an arithmetic consistency attack-resistant integrity verification framework (ACIV). Firstly, we design two verification protocols combining HE and arithmetic secret sharing (ASS), which support integrity verification for both outsourced data and computational results. The incorporation of lightweight ASS significantly reduces computational cost during verification. Secondly, we construct a novel arithmetic consistency attack method, exposing the security vulnerabilities in this type of verification protocols. The malicious adversary can easily exploit the homomorphic properties of HE to tamper with data. On the basis, we propose corresponding countermeasures. Under the assumption of pre-generated immutable random numbers, the optimized protocols effectively defend against arithmetic consistency attack while ensuring data confidentiality and integrity during transmission. Theoretical analysis and performance evaluation demonstrate that the optimized version achieves stronger security and better efficiency than the baseline protocols. This work provides a feasible solution for integrity verification in SOC environments.