Multi-Party Computation beyond Custody Applications

Current applicability of MPC in the crypto space is primarily oriented towards custody in the case of providers such as Fireblocks who use it primarily for Threshold Signing capabilities. On the other hand, SMPC (Secure Multi Party Computation) is a cryptographic technique that allows multiple parties to get a function outputs over their private inputs. Ability to have quantum-safe privacy-preserving computation opens many use cases enabling data ownership and regulatory compliance.

1. MPC beyond Custody Applications Dejan Radic - BlockSplit 5 - May 2024

2. What is the value if data is not processed? If everyone keeps the data, nobody benefits! If everyone gives the data, what is the value for data owners?

3. That’s where MPC comes into play!

4. MPC - Multi Party Computation MPC - protocol that for data processing while not disclosing the underlying data. It enables collaborative computation of multiple parties, where result end at designated location. Computation of function F Arithmetic (+, -, x, /) Logic (AND, OR, XOR) Comparison (=, >, <) Statistics Text!?

5. Yao's Millionaires' problem Andrew Yao, 1982 Two millionaires, Alice and Bob, who are interested in knowing which of them is richer without revealing their actual wealth.

6. Crypto Use Case - Threshold Signing

7. Additive Secret Sharing Example

8. 7091 - 3894 + 13362 + 7751 = 24400 24400 / 4 = € 6100

9. Communication Throughput

10. Example with moderator

11. Example with moderator

12. What if a party is dishonest? Verifiability!

13. Verifiability Interactive challenging with half-results Proof of Stake style tokenization Trusted 3rd party Reputation scheme Commitment scheme

14. Associated Technologies PKI for authentication Symmetric encryption for inter-party confidentiality ZKP for correctness of computations Shamir Secret Sharing Homomorphic Encryption Federated Learning

15. Conclusion Not just custody applications through threshold signing Collaborative computation by keeping the data private Combining arithmetic & logic operations for analytics Verifiability to detect dishonest parties Engineering trade-off: throughput, decentralization, precision Data Ownership enabled through tokenization Model creation by using Federated learning

16. Questions?

17. Thank you!