Digital signature algorithms
Understand how TSS and MPC work together to produce standard digital signatures
Digital signatures are a cryptographic primitive that underpins the security of blockchains and of io.sign. They provide three properties:
- Authentication verifies the identity of the signer.
- Non-repudiation prevents the signer from denying they signed the message.
- Integrity detects whether the message has changed since signing.
How TSS and MPC work together
A Threshold Signature Scheme (TSS) lets a private key be held as separate key shares across signers. To produce a signature, participating signers must collectively meet a defined threshold of signing power. For example, with three equally weighted signers and a threshold of two, any two can sign together. See Vaults, signers, and thresholds for how signer weights determine which combinations can sign.
Multi-Party Computation (MPC) is how those signers perform the cryptographic work together while keeping their individual key shares secret. They exchange protocol messages to generate a key pair and later produce signatures, without assembling the complete private key.
TSS defines the threshold needed to sign. MPC provides the computation that makes threshold signing possible. The result is a standard digital signature that anyone can verify with the public key.
In io.sign, the backend coordinates this process, while mobile or server-based signers perform the cryptographic operations. The complete private key is reconstructed only through the separate backup and recovery process.
The signature algorithm determines how the resulting signature is created and verified. io.sign supports two: the Elliptic Curve Digital Signature Algorithm (ECDSA) and the Edwards-curve Digital Signature Algorithm (EdDSA).
Curve and signature support
Every io.sign vault has an ECDSA key and an EdDSA key. Choose the algorithm for each signature with coseAlgorithm.
| ECDSA | EdDSA | |
|---|---|---|
| Curve | secp256k1 | Ed25519 |
coseAlgorithm | es256k (SHA-256), eskec256 (Keccak-256) | eddsa |
| Threshold protocol | CGGMP21 | Threshold Schnorr |
| Public key size | 33 bytes (compressed), 65 bytes (uncompressed) | 32 bytes |
| HD derivation | Unhardened | Unhardened |
| XPub format | Standard BIP32 | BIP32-like, with a 32-byte key |
| Example chains | Bitcoin (SegWit), Ethereum and EVM networks, Tron | Solana, XRP Ledger |
Elliptic Curve Digital Signature Algorithm (ECDSA)
ECDSA is the signature scheme used in Bitcoin, Ethereum, and many other blockchains. It relies on the algebraic structure of elliptic curves over finite fields.
Key facts about ECDSA:
- It relies on the assumed difficulty of the Elliptic Curve Discrete Logarithm Problem (ECDLP).
- It uses a curve and base point defined in a standard, for example secp256k1 in Bitcoin.
- The private key is a random integer; the public key is a point on the curve.
- Signing involves generating a random nonce and computing a signature (r, s).
- Verification checks a mathematical relationship between the message, public key, and signature.
ECDSA is fast and produces short signatures, which suits blockchain use.
Edwards-curve Digital Signature Algorithm (EdDSA)
EdDSA is a more recent signature scheme based on twisted Edwards curves. It's designed to be faster and more secure than ECDSA.
Notable differences from ECDSA:
- Uses a different curve form for better performance and implementation security.
- Deterministic nonce generation prevents common ECDSA pitfalls.
- Simpler and more efficient to implement.
EdDSA is increasingly common, particularly on newer blockchains. The most common variant is Ed25519, which uses Curve25519 and the hash function SHA-512.
io.sign signs with either algorithm under the same threshold rules.
Hierarchical deterministic (HD) wallets
HD wallets (defined in BIP32) let a single seed deterministically derive a structured tree of key pairs. io.sign supports HD derivation on both curves:
- Derivation is unhardened, so you can derive public keys and addresses without any private key. Pass the path as
derivationPathon a signature request. - BIP44 path structure (for example
m/44/60/0/0/0) follows chain-specific conventions. - XPub/XPrv formats are standardised:
- For ECDSA: Standard BIP32 format with version bytes, such as
xpubandypub - For EdDSA: BIP32-like format with a fixed prefix and 32-byte keys
- For ECDSA: Standard BIP32 format with version bytes, such as
- No complete seed or private key exists in io.sign. Each signer holds only its own key share.
- A single HD wallet can derive keys across multiple blockchains and signature schemes, enabling multi-asset custody from one root.
HD derivation matters for custody because:
- One seed deterministically derives many key pairs, so you can create new addresses without a new key generation.
- Derivation paths let you separate keys by user, purpose, or chain, and trace each key back to the root.
- Derivation runs under threshold control, so no single signer sees the root or derived private keys.
Lattice-based signatures (post-quantum)
"Q-Day" is the arrival of quantum computers capable of breaking today's public-key cryptography. ECDSA and EdDSA rely on the discrete logarithm problem. Lattice-based algorithms rely on the hardness of lattice problems instead, and have become the standard for quantum-resistant digital signatures.
What is coming
The National Institute of Standards and Technology (NIST) finalised these standards in 2024. We're tracking the transition to the following schemes:
- ML-DSA (Module-Lattice-Based Digital Signature Algorithm): Formerly known as Dilithium. The primary NIST standard for digital signatures, balancing security and performance.
- SLH-DSA (Stateless Hash-Based Digital Signature Algorithm): Formerly known as SPHINCS+. A backup standard based on hash functions rather than lattices.
- FN-DSA (FFT over NTRU-Lattice-Based Digital Signature Algorithm): Formerly known as Falcon. Designed for cases that need small signatures.
The hybrid transition
We expect the industry to adopt hybrid schemes in the near term. In this model, a transaction is signed by both a classical key (ECDSA) and a post-quantum key (ML-DSA). This protects against future quantum attacks without breaking compatibility with existing blockchain infrastructure.
The state of academic research in 2026
Single-party lattice signatures are now standardised, but threshold lattice signatures remain an active area of academic research. Running them under MPC is harder than threshold ECDSA.
- The rejection sampling problem. Lattice signatures such as ML-DSA require "rejection sampling", a process where the signer retries signature generation until the output satisfies a statistical distribution (preventing key leakage). In a distributed MPC setting, coordinating these retries across multiple parties without leaking private data is computationally expensive.
- Recent developments. The field is moving quickly:
- In May 2024, researchers introduced TRaccoon, a signature scheme designed to be threshold-friendly. It removes the need for rejection sampling, which makes it a strong candidate for MPC.
- In June 2025, a paper from PQShield proposed a threshold scheme that achieves signature sizes comparable to standard Dilithium for small committees.
- In August 2025, a paper named Unmasking TRaccoon explored adding identifiable abort to the original TRaccoon protocol.
- Standardisation timeline. NIST is evaluating these approaches under the Multi-Party Threshold Cryptography (MPTC) project (see NIST IR 8214C).
Until threshold lattice protocols are standardised and independently audited, io.sign uses the established threshold schemes listed in Curve and signature support.
Updated 17 days ago