Proactive Secret Sharing
Proactive secret sharing periodically refreshes key shares to limit the damage from gradual compromise over time.
Key Takeaways
- Proactive secret sharing defends against mobile adversaries: attackers who compromise different participants over time. By refreshing key shares at regular intervals, previously stolen shares become useless.
- The secret and public key never change during a refresh: each participant adds values derived from random polynomials with a zero constant term, so the underlying secret stays intact while every share is replaced.
- Critical for long-lived signing setups: systems like threshold signature schemes and distributed operator sets rely on proactive refresh to remain secure over months or years without rotating the on-chain address.
What Is Proactive Secret Sharing?
Proactive secret sharing (PSS) is a cryptographic technique that periodically refreshes the shares in a secret sharing scheme so that an attacker cannot accumulate enough shares over time to reconstruct the secret. It was introduced by Herzberg, Jarecki, Krawczyk, and Yung at CRYPTO 1995 in their paper "Proactive Secret Sharing Or: How to Cope with Perpetual Leakage."
In a standard threshold scheme like Shamir's Secret Sharing, a secret is split into n shares such that any t of them can reconstruct it. This works well if compromises happen all at once: as long as fewer than t shares leak, the secret is safe. But what if an attacker slowly compromises participants one by one? Over the lifetime of a system, they might eventually collect t shares from different time periods and reconstruct the secret. This is the mobile adversary problem, and proactive secret sharing solves it.
After each refresh, the old shares are incompatible with the new ones. An attacker who stole a share last month and steals another share this month cannot combine them. They would need to compromise t participants within a single refresh period to succeed.
How It Works
The refresh protocol preserves the secret while replacing every share. The core insight is that adding a polynomial with a zero constant term to the existing sharing polynomial changes all share values without changing the secret (which is the constant term).
The Refresh Protocol
In a distributed setting without a trusted dealer, the refresh works as follows:
- Each participant generates a random polynomial of degree t−1 with its constant term set to zero
- Each participant evaluates their polynomial at every other participant's index and sends them the result over a secure channel
- Each participant adds all received values to their current share
- Each participant securely deletes their old share
The combined effect of all these random polynomials is equivalent to adding a single random polynomial with a zero constant term to the original sharing polynomial. Since the constant term is zero, interpolating the new shares at x = 0 still recovers the original secret.
Mathematical Illustration
Consider a (2, 3) threshold scheme where the secret s is shared among three participants using a degree-1 polynomial f(x) = s + ax:
# Original sharing polynomial
f(x) = s + a*x
# Shares: f(1), f(2), f(3)
# Any 2 shares reconstruct s = f(0)
# --- Refresh Phase ---
# Each participant i generates a random degree-1
# polynomial q_i(x) with q_i(0) = 0:
q_1(x) = r_1 * x
q_2(x) = r_2 * x
q_3(x) = r_3 * x
# Combined update polynomial:
Q(x) = q_1(x) + q_2(x) + q_3(x) = (r_1 + r_2 + r_3) * x
# New sharing polynomial:
f'(x) = f(x) + Q(x) = s + (a + r_1 + r_2 + r_3) * x
# f'(0) = s (the secret is unchanged)
# But f'(1), f'(2), f'(3) are all different from beforeThe new shares are completely different from the old ones, but they still encode the same secret. An attacker holding old share f(1) and new share f'(2) cannot reconstruct s because those values lie on different polynomials.
The Mobile Adversary Model
Traditional security models assume a static adversary: the set of corrupted parties is fixed at the start. The mobile adversary model, formalized by Ostrovsky and Yung, allows the adversary to corrupt different parties in different time periods. As long as fewer than t parties are corrupted in any single period, the secret remains safe.
This is realistic for long-running systems. Over years of operation, any given server might be breached through software vulnerabilities, insider threats, or physical access. Proactive refresh limits the damage: the adversary must breach enough parties within one refresh window, not across the entire lifetime of the system.
Integration with FROST
FROST (Flexible Round-Optimized Schnorr Threshold signatures) is a threshold signing protocol that uses distributed key generation to create key shares among a group of signers. FROST supports proactive share refresh through a protocol similar to its DKG phase: each signer generates a zero-constant polynomial and distributes evaluations, then adds received values to their signing share.
The group's public key remains unchanged after refresh, which means the on-chain address derived from that key also stays the same. This is critical for Bitcoin-based systems where moving funds to a new address requires an on-chain transaction and associated fees.
Research such as Dynamic-FROST (D-FROST) extends this further by combining FROST with proactive share refresh to support both share renewal and committee membership changes without altering the group public key.
Use Cases
Long-Lived Signing Infrastructure
Any system that holds signing keys for extended periods benefits from proactive refresh. Bitcoin bridges, custody platforms, and Layer 2 networks with distributed operator sets all face the mobile adversary threat. Without refresh, security degrades over time as the probability of cumulative compromise increases.
Spark, for example, uses a distributed set of operators who collectively hold one side of a 2-of-2 arrangement for each virtual UTXO. The operators manage their signing shares using FROST signing, and share refresh ensures that compromising an operator at one point in time does not permanently compromise the system. The on-chain UTXO never moves during refresh because the public key stays the same.
Custodial Key Management
Institutional custody solutions that use threshold signatures across multiple hardware security modules or geographic locations can run periodic refresh to limit exposure. Even if an HSM is breached and its share extracted, that share becomes worthless after the next refresh cycle.
Validator and Committee Rotation
Blockchain systems with rotating validator committees use proactive secret sharing to hand off signing authority. When a validator leaves and a new one joins, a refresh can incorporate the new participant while revoking the departing one's share. This avoids the need to generate a new key and migrate all references to it.
Cross-Chain Bridges
Cross-chain bridges that hold locked assets under a threshold key are high-value targets. Proactive refresh reduces the risk that a slow, persistent attacker can collect enough shares to drain the bridge. Combined with key rotation policies, PSS significantly raises the cost of attack.
Comparison with Static Secret Sharing
Static secret sharing and proactive secret sharing differ fundamentally in how they handle time:
| Property | Static Secret Sharing | Proactive Secret Sharing |
|---|---|---|
| Share lifetime | Permanent: shares never change | Ephemeral: shares are refreshed each period |
| Adversary model | Static: fixed set of corrupted parties | Mobile: adversary can corrupt different parties over time |
| Compromise accumulation | Shares from different times can be combined | Old and new shares are incompatible |
| Key change required | Must generate new key to revoke old shares | Same key and public address persist |
| Operational overhead | None after initial setup | Periodic refresh rounds require coordination |
In a static scheme, once a share is stolen, it remains valid forever. An attacker with patience can collect shares across years. With PSS, the attacker faces a strict deadline: they must compromise the threshold number of participants within a single refresh period.
Risks and Considerations
Secure Deletion Assumption
Proactive secret sharing assumes that participants securely delete old shares after each refresh. If an attacker can recover deleted data (through disk forensics, memory dumps, or backup systems), the refresh provides no benefit. In practice, secure deletion on modern storage hardware (SSDs with wear leveling, cloud storage with replication) is difficult to guarantee.
Synchronization Requirements
All participants must complete the refresh protocol within the same time window. If some participants are offline, the protocol cannot proceed. Asynchronous variants of PSS exist but typically require stronger assumptions or tolerate fewer corruptions. A 2026 ePrint paper formalizes a hierarchy of mobile adversary models for fully asynchronous networks and shows that optimal resilience is achievable but with added complexity.
Secure Channel Requirement
During refresh, participants exchange polynomial evaluations over secure channels. Under a mobile adversary, establishing these channels is non-trivial: if the adversary controls a participant during the refresh, it can observe and potentially manipulate the exchanged values. Research continues on minimizing the assumptions needed for these channels.
Communication Overhead
Each refresh requires O(n²) messages where n is the number of participants, since every participant sends a value to every other participant. For large committees, this communication cost can be significant. The refresh frequency must balance security (shorter periods are safer) against operational cost (each refresh consumes bandwidth and computation).
Does Not Prevent Real-Time Attacks
Proactive secret sharing protects against gradual compromise across time periods, but it does not help if an adversary compromises t or more participants simultaneously. It complements other defenses like multi-party computation, geographic distribution, and hardware security, but does not replace them.
This glossary entry is for informational purposes only and does not constitute financial or investment advice. Always do your own research before using any protocol or technology.