IACR News item: 20 February 2025
Offir Friedman, Avichai Marmor, Dolev Mutzari, Yehonatan Cohen Scaly, Yuval Spiizer
ePrint Report
Threshold signatures have become a critical tool in cryptocurrency systems, offering enhanced security by distributing the signing process among multiple signers. In this work, we distribute this process between a client and a permissionless decentralized blockchain, and present novel protocols for ECDSA and EdDSA/Schnorr signatures in this setting. Typical threshold access architectures used by trusted custodians suffer from the honeypot problem, wherein the more assets the custodian holds, the greater the incentive of compromising it.
Implementing threshold signatures over permissionless blockchains poses a few challenges. First, existing networks typically work over an asynchronous reliable broadcast communication channel. Accordingly, our protocol is implemented over such a channel. As a result, it also benefits from identifiable abort, public verifiability, and guaranteed output delivery, and the client benefits from censorship resistance of blockchain systems. Second, upon signing each block, the participating quorum may dynamically change and is post-determined. Therefore, we design a fluid protocol, that supports a post-determined dynamic quorum in each communication round, thereby complying with existing broadcast channel implementations. Third, in permissionless networks, parties may join, leave, and change their stake. Therefore, we offer protocols for network reconfiguration, with complexity independent of the number of clients in the system, and our protocol efficiently supports a weighted threshold access structure for the network. Specifically, the complexity of distributed key generation and presign only depends on the number of parties and not on the overall weight, and the amortized cost of sign only depends on the individual weight.
Furthermore, our protocol introduces key improvements, including the removal of zero-knowledge proofs towards the client, and presigns with a non-interactive client. For Schnorr, the presigns are client-independent, and can be collected by the blockchain in a common pool, available for all clients in the system. These optimizations reduce communication overhead and improve the system's ability to handle traffic spikes during high-demand periods.
Our protocol is UC-secure, and is therefore natively designed for multiple clients to use the system in parallel. Notably, we propose a novel assumption, Slightly-Enhanced ECDSA Unforgeability, offering concrete security for 256-bit elliptic curves for threshold ECDSA with support for parallel execution of presigns.
In addition to securing cryptocurrency wallets, we demonstrate how our protocol enables various cross-chain applications, such as decentralized bridges, future transactions, andwallet transfer. Our system is designed for interoperability across multiple blockchains, enhancing security, scalability, and flexibility for decentralized finance (DeFi) ecosystems.
Implementing threshold signatures over permissionless blockchains poses a few challenges. First, existing networks typically work over an asynchronous reliable broadcast communication channel. Accordingly, our protocol is implemented over such a channel. As a result, it also benefits from identifiable abort, public verifiability, and guaranteed output delivery, and the client benefits from censorship resistance of blockchain systems. Second, upon signing each block, the participating quorum may dynamically change and is post-determined. Therefore, we design a fluid protocol, that supports a post-determined dynamic quorum in each communication round, thereby complying with existing broadcast channel implementations. Third, in permissionless networks, parties may join, leave, and change their stake. Therefore, we offer protocols for network reconfiguration, with complexity independent of the number of clients in the system, and our protocol efficiently supports a weighted threshold access structure for the network. Specifically, the complexity of distributed key generation and presign only depends on the number of parties and not on the overall weight, and the amortized cost of sign only depends on the individual weight.
Furthermore, our protocol introduces key improvements, including the removal of zero-knowledge proofs towards the client, and presigns with a non-interactive client. For Schnorr, the presigns are client-independent, and can be collected by the blockchain in a common pool, available for all clients in the system. These optimizations reduce communication overhead and improve the system's ability to handle traffic spikes during high-demand periods.
Our protocol is UC-secure, and is therefore natively designed for multiple clients to use the system in parallel. Notably, we propose a novel assumption, Slightly-Enhanced ECDSA Unforgeability, offering concrete security for 256-bit elliptic curves for threshold ECDSA with support for parallel execution of presigns.
In addition to securing cryptocurrency wallets, we demonstrate how our protocol enables various cross-chain applications, such as decentralized bridges, future transactions, andwallet transfer. Our system is designed for interoperability across multiple blockchains, enhancing security, scalability, and flexibility for decentralized finance (DeFi) ecosystems.
Additional news items may be found on the IACR news page.