As the demand for privacy-preserving computation grows across industries—from healthcare to finance to AI—two leading cryptographic technologies have emerged: Secure Multiparty Computation (MPC) and Fully Homomorphic Encryption (FHE). While they are often framed as competitors, the reality is more nuanced. These tools offer distinct strengths and trade-offs and, increasingly, researchers are exploring how they can work together for more robust and flexible secure computation systems.
This article explores the relationship between MPC and FHE, drawing from recent research and practical perspectives.
What Are MPC and FHE?
- MPC (Secure Multiparty Computation):
Enables multiple parties to compute a function over their private inputs without revealing them to each other. Privacy is maintained through distributed computation and cryptographic protocols, often relying on secret sharing. - FHE (Fully Homomorphic Encryption):
Allows computation to be performed directly on encrypted data, without needing to decrypt it first. The result, once decrypted, matches the output of the computation on the original plaintext.
MPC vs. FHE: A Technical Trade-Off
As noted in A Pragmatic Introduction to Secure Multiparty Computation:
“Homomorphic encryption offers an asymptotic communication improvement compared to MPC, but at the expense of computational efficiency.”
This quote captures the key trade-off:
| Feature | MPC | FHE |
|---|---|---|
| Computation cost | Lower | Higher (especially bootstrapping) |
| Communication overhead | Higher (many rounds, multi-party) | Lower (can be single party) |
| Trust model | Multiple parties, no trusted third party | Single party can compute securely |
| Interactivity | Interactive | Non-interactive |
| Latency | Higher due to rounds | Potentially lower (in batch mode) |
So, FHE improves communication but increases computational load, while MPC excels in computational efficiency but requires more network interaction.
Hybrid Use: MPC with FHE
Rather than choosing one over the other, modern cryptographic systems increasingly combine MPC and FHE for the best of both worlds.
Secure Function Evaluation via FHE and MPC
A basic hybrid model involves:
- Each party encrypts their input using an FHE scheme.
- The computation is performed independently on ciphertexts.
- A distributed decryption (MPC) reveals the final result.
This allows:
- Secure input protection via FHE
- Joint output reconstruction via MPC
- One-shot, non-interactive computations
However, this model is limited to single-function evaluation—you can’t change the function midway, nor run dynamic computations interactively.
Towards Fully Secure and Reactive Integration
Recent research aims to create robust protocols that combine MPC and FHE more effectively. To achieve this, the following criteria must be met:
✅ Malicious Security (Robustness):
MPC should remain secure even if some parties are dishonest or deviate from the protocol. Honest parties should always receive correct results.
✅ Guaranteed Output Delivery:
Even in the presence of adversaries, honest parties should always get the output—no one should be able to block results.
✅ Reactive Computation Support:
Rather than computing a fixed function once, the system should support dynamic, stateful, multi-step computations—essential for applications like secure AI inference or encrypted database queries.
✅ Communication Independence from FHE Ciphertext Size:
FHE ciphertexts are large and heavy. Efficient integration with MPC requires communication protocols that don’t scale with ciphertext size, ensuring the system stays performant.
Practical Implications
Combined FHE-MPC systems could enable:
- Secure cloud computing where encrypted data is processed by untrusted servers, with result verification by multiple stakeholders.
- Federated machine learning with cryptographic guarantees for both privacy and integrity.
- Cross-border health data analysis without revealing patient-level data to centralized authorities.
Conclusion: Not Competitors—Complementary Tools
MPC and FHE are not mutually exclusive. Each excels under different conditions, and together they can offer enhanced functionality, especially in environments requiring:
- Low trust
- High interactivity
- Rigorous privacy and integrity guarantees
As real-world secure data processing systems mature, hybrid architectures combining FHE and MPC are likely to become the norm.
We love to share our knowledge on current technologies. Our motto is ‘Do our best so that we can’t blame ourselves for anything“.