Dynamic Fraud Proof

TL;DR

The paper tackles achieving fast finality for off-chain computation in Ephemeral Rollups on the Solana VM by introducing Dynamic Fraud Proofs, an adaptive assert/challenge mechanism that shortens the challenge window under ideal conditions. It combines a configurable set of randomly selected verifier nodes with economic incentives to ensure rapid fraud detection and settlement, while providing safe extensions when challenges arise. The authors formalize a probabilistic trust model, present an extension mechanism for the challenge window, and discuss bonding and slashing to encourage honest behavior, aiming to deliver near real-time finality without compromising base-layer security. This work offers a scalable, flexible approach to cross-layer state management that mitigates censorship risks and state fragmentation, with practical implications for real-time transaction processing on Solana via ERs.

Abstract

In this paper, we present a novel fraud-proof mechanism that achieves fast finality and, when combined with optimistic execution, enables real-time transaction processing. State-of-the-art optimistic rollups typically adopt a 7-day challenge window, during which any honest party can raise a challenge in case of fraud. We propose a new assert/challenge construction called "Dynamic Fraud Proofs" that achieves sub-second finality in ideal scenarios, while dynamically delaying settlement in the event of fraud detection and challenge resolution. The system relies on 1) a dynamic challenge period and 2) a configurable number of randomly selected verifier nodes who must interactively approve a state commitment without raising a challenge. If these conditions are not met, the state is not finalized, and the challenge period and approval criteria are dynamically adjusted. We provide a detailed analysis of the system's design, explaining how it maintains the assumption of a single honest node and addresses censorship attacks by inverting the traditional challenge process. Additionally, we formalize the system's probabilistic security model and discuss how bonding, incentives, and slashing mechanisms can encourage honest behavior, thereby increasing the likelihood of fast settlement in ideal scenarios.

Figures (2)

• Figure 1: The system is comprised of several key components: The Delegation Program manages account locking/unlocking, state commitments, and finalization on the base SVM. The Operator is a node running an Ephemeral SVM session, while the Provisioner listens for delegation events, selects operators, and manages runtime provisioning. The Security Committee verifies execution and data availability, supported by the Data Availability Layer that stores transactions from the ephemeral session. An RPC Router offers a convenient interface for clients to access and modify state across both the base layer and Ephemeral Rollup sessions.
• Figure 2: The process of state finalization in an ER involves multiple steps to ensure integrity and security. The ER node first submits a commitment of the state difference it intends to finalize. A challenge window is then initiated, allowing participants to raise challenges against any invalid state commitment. If a challenge is raised, a dispute resolution process begins, preventing the state from being finalized until the dispute is resolved. Depending on the outcome of the challenge period—whether enough challengers sign off on the state diff—the state is either settled or the challenge window is extended.