A comprehensive technical exploration of privacy-preserving trading mechanisms on Stellar, comparing zero-knowledge proofs, multi-party computation, fully homomorphic encryption, and trusted execution environments. The authors built working prototypes using MPC and TEE approaches, with TEEs emerging as the most practical solution for production systems.
This article examines how to prevent front-running and MEV extraction in decentralized exchanges by implementing dark pool-style privacy mechanisms on Stellar. The authors survey cryptographic approaches including ZK proofs (which lack inherent confidentiality), MPC (which scales linearly with participants), FHE (which is orders of magnitude too slow for practical trading), and TEEs (which offer native performance with hardware trust tradeoffs). They built experimental prototypes using MP-SPDZ for MPC and Phala's TEE Cloud for a production-ready dark pool, demonstrating order submission, matching, and atomic settlement via Soroban smart contracts. The TEE-based system uses Intel TDX attestation to bind cryptographic keys and container digests, enabling clients to verify they're communicating with genuine, audited code. The article concludes that TEEs represent the pragmatic choice for production systems today, though remaining work includes contract-level attestation verification and OS kernel verification against reproducible builds.
