Secure Multi-Party Computation
Secure Multi-Party Computation (SMPC) · Also known as: MPC, Multi-Party Computation, Privacy-Preserving Computation, Güvenli Çok Taraflı Hesaplama
Secure Multi-Party Computation (SMPC) is a cryptographic paradigm that enables two or more parties to jointly compute a function over their private inputs without revealing those inputs to one another. Introduced by Andrew Yao in 1982 through his seminal garbled-circuit construction, SMPC provides provable privacy guarantees grounded in computational hardness assumptions. It underpins modern privacy-preserving data analysis, enabling collaborative computation on sensitive datasets in finance, healthcare, and machine learning.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use SMPC when two or more parties must jointly compute a function on sensitive data and no single trusted third party is acceptable. It is suited for private set intersection, joint statistical analysis, privacy-preserving machine learning, and secure auctions. Assumptions include the existence of computational hardness primitives (e.g., one-way functions). Limitations include high communication overhead and latency, especially for large circuits. For scenarios tolerating statistical disclosure, Differential Privacy may be a lighter alternative.
Strengths & limitations
- Provable privacy: input confidentiality is guaranteed under well-defined cryptographic assumptions
- No trusted third party required: parties retain full control of their private data throughout
- Correctness guarantee: the computed output is verifiably consistent with all parties' inputs
- Flexible function support: arbitrary functions expressible as circuits can be computed securely
- High communication complexity: multiple rounds of cryptographic message exchange scale poorly with circuit depth and party count
- Computational overhead: garbling and evaluating large circuits is orders of magnitude slower than plaintext computation
- Setup complexity: implementing secure OT and garbling correctly requires deep cryptographic expertise
- Collusion vulnerability: privacy guarantees break down if a threshold number of parties collude, depending on the security model
Frequently asked
Is SMPC the same as homomorphic encryption?
No. Homomorphic Encryption (HE) allows a single party to compute on encrypted data held by another, whereas SMPC involves multiple parties jointly computing on their respective private inputs. HE is often used as a building block within SMPC protocols, but the two address distinct threat models and use cases.
How many parties can SMPC support?
SMPC protocols generalize to any number of parties n. Two-party protocols (2PC), such as Yao's garbled circuits, are the most efficient. General n-party protocols exist but incur communication costs that scale with n. Threshold variants allow privacy to hold as long as fewer than t colluding parties fall below a defined corruption bound.
Can SMPC guarantee correctness if some parties behave maliciously?
Yes, under the malicious adversary model. Protocols add zero-knowledge proofs or message-authentication codes so that any deviation from the protocol is detected and the computation aborted. This robustness comes at significant additional cost compared to the semi-honest (honest-but-curious) model.
Sources
- Yao, A. C. (1982). Protocols for secure computations. 23rd Annual Symposium on Foundations of Computer Science, 160–164. DOI: 10.1109/SFCS.1982.38 ↗
How to cite this page
ScholarGate. (2026, June 2). Secure Multi-Party Computation (SMPC). ScholarGate. https://scholargate.app/en/privacy/secure-multiparty-computation
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Differential PrivacyPrivacy↔ compare
- Federated LearningPrivacy↔ compare
- k-AnonymityPrivacy↔ compare