# Incentive Mechanism Design for Privacy-Preserving Decentralized Blockchain Relayers

Crypto.world

> Public blockchains expose every smart-contract interaction, so an observer can link users to their on-chain activity. Relayers — intermediaries that submit transactions on a user's behalf — hide the…

**arXiv**: 2601.06699

**Published**: 2026-01-10

**Authors**: Boutaina Jebari, Khalil Ibrahimi, Hamidou Tembine, Mounir Ghogho

## AI summary

## Overview

**Research area:** Blockchain privacy, mechanism design, and algorithmic game theory — specifically incentive design for decentralized transaction relayers.

**Technical level:** Advanced. The paper relies on non-cooperative game theory (mixed-strategy Nash equilibrium, potential games, strong equilibria) and evolutionary game dynamics (replicator dynamics, invasion barriers). The numerical section is accessible, but the analytical core requires comfort with equilibrium proofs.

**Scope (one sentence):** The paper models a set of staked blockchain relayers as players in a Volunteer's-Dilemma-style game and derives an incentive mechanism whose unique mixed Nash equilibrium makes all relayers behave indistinguishably, then evaluates the resulting privacy–reliability–robustness–cost trade-off numerically.

## What This Paper Is About

Public blockchains expose every smart-contract interaction, so an observer can link users to their on-chain activity. Relayers — intermediaries that submit transactions on a user's behalf — hide the user's identity behind the relayer's address, but today most systems depend on a *single* relayer, which is both a single point of failure and a deanonymization risk (timing, traffic correlation, propagation clues). The paper asks how to spread this role across many relayers so that no observer can tell which one acted, while ensuring that at least one relayer actually uploads, given that uploading costs money and relayers are self-interested.

## Key Contributions

1. **A game-theoretic framework for decentralized relayers** in which the incentive mechanism enforces *indistinguishable probabilistic actions*: at equilibrium every node uses the same upload strategy, which the authors argue is required for privacy against an on-chain observer.
2. **A new variant of the Volunteer's Dilemma (VD)** with symmetric rewards, a *probabilistic* (outcome-dependent) cost reflecting whether a smart-contract submission is accepted or reverted, and a staking-enforced penalty applied to all relayers when nobody uploads.
3. **Analytical results:** proof of existence, uniqueness, and evolutionary stability of the symmetric mixed-strategy Nash equilibrium; a potential-function characterization; identification of the set of N-strong equilibria; and an invasion-barrier result against mutant strategies.
4. **A numerical study of the privacy–reliability–robustness–cost trade-off**, showing how equilibrium upload probability, outage probability, and expected reward vary with the number of relayers, upload costs, and the penalty.

## Main Findings

- **Upload probability falls as the relayer set grows.** With b = 100, p = 100, c_l = 1 and c_f ∈ {25, 50, 75}, the equilibrium upload probability q_u* decreases with N, consistent with the free-riding effect in Volunteer's Dilemma games. Higher c_f also lowers q_u*.
- **Outage probability rises with N and c_f, then stabilizes.** For N ≤ 15 there is a clear privacy–reliability trade-off; beyond a threshold, outage variations become very small (on the order of 0.001) and continue to decrease as N grows, so extra relayers add anonymity-set size without further degrading reliability. With c_f = 25 the system achieves outage probability lower than 0.05.
- **Lower upload cost means much higher equilibrium payoff.** Relayers obtain up to 91.65% of the benefit b when c_f = 25, versus 79.49% when c_f = 50 and 63% when c_f = 75. Expected reward R* varies only slightly with N, so an individual player is roughly indifferent to the size of the relayer set in payoff terms.
- **Reverted-transaction cost c_l erodes reliability.** Raising c_l from 1 up to c_f = 50 (with b = p = 100, N ∈ {5, 10, 30}) lowers q_u*. For N = 5, when c_f = c_l = 50 and b = 100, outage probability rises from approximately 0.06 to nearly 0.17.
- **Penalties repair reliability under high costs.** With b = 100, c_f = 50, c_l = 25, increasing p raises q_u* and R* and sharply lowers outage probability. When p exceeds five times the upload reward, outage probability falls below 0.05 for all tested N while expected reward stays positive.
- **The equilibrium is globally asymptotically stable in simulation.** Under replicator dynamics with b = 100, c_f = 25, c_l = 1, p = 100 and adaptation rate μ = 0.1, trajectories from different initial conditions all converge to q_u*. Convergence is slower when the initial probability is far from equilibrium and for larger populations.
- **Strong-equilibrium structure.** The N-strong equilibria of the game are exactly the profiles in which one player chooses Upload and all others choose Not Upload; every pure-strategy Nash equilibrium is a global maximizer of the sum of utilities.
- **No empirical or on-chain data is reported.** The numerical values are explicitly described as normalized rather than literal gas measurements; only the ratios among b, c_f, c_l and p matter. No dataset sizes, benchmark comparisons, or deployment results appear in the provided content.

## Methodology in Plain English

The authors strip the problem down to what an on-chain observer can see: transaction contents, the relayer address that submitted, and inclusion time. Network-level traffic analysis and cryptography are deliberately excluded from the model.

Each user message is broadcast to all N ≥ 3 staked relayers before a shared upload window that is derived from blockchain state (block height or slot number), so no relayer's local clock gives it away. Every relayer independently decides Upload or Not Upload. If several upload, the first valid transaction is accepted and the rest revert.

Money flows are simple: the successful uploader pays c_f, unsuccessful uploaders pay a smaller c_l (with 0 < c_l < c_f), everyone gets reward b if any upload succeeds (b > c_f), and if nobody uploads, everyone is penalized p through the staking contract. From these payoffs the authors write the expected utility of uploading versus not uploading, define the gain function g(q_u) as the difference, and derive a polynomial h(q_u) whose root in (0,1) is the equilibrium upload probability q_u*. A derivative argument shows the polynomial crosses zero exactly once, giving uniqueness. They then show the game is an exact potential game with potential φ(q_u) = ∫₀^{q_u} g(x)dx + C, which yields asymptotic stability under payoff-improving evolutionary dynamics, and they prove an invasion barrier: a small enough fraction ε ≤ α of mutants cannot out-earn the resident strategy. Finally, they solve the equilibrium condition numerically and simulate the replicator dynamic to check convergence.

## Why This Matters

**Impact on research.** Most relayer literature (zkRelay, account-abstraction bundlers, Tornado Cash, OpenZeppelin Defender) treats relayer behavior as static and trusted. Nym is described as the closest incentive-driven privacy system, but it operates at the network layer. This paper's distinctive move is to make *indistinguishability itself* the object of mechanism design: privacy no longer rests on honest operation but on equilibrium incentives. It also extends the Volunteer's Dilemma literature to a setting the authors say prior VD work has not covered — blockchain relayers and smart-contract submission — with two non-standard features: outcome-dependent (accepted vs. reverted) costs and staking-based penalties for total inaction.

**Real-world applications.**

- **Privacy-preserving incentivized data collection:** the paper's own example, where contributors must submit an on-chain commitment proving they provided data; routing it through a relayer set makes them indistinguishable from other relayers.
- **Smart-contract interaction privacy generally:** decoupling a user's on-chain identity from the transaction origin for DeFi, voting, supply-chain, and open-data-market applications.
- **Account abstraction and bundler infrastructure:** Ethereum bundlers that submit transactions for users could adopt the probabilistic, symmetric upload rule instead of first-come-first-served.
- **Mixer and interoperability relayer sets:** deposit/withdrawal unlinkability (Tornado Cash-style) and cross-chain relay designs (zkRelay-style) that currently rely on trusted relayer behavior.

**Industry relevance.** Any protocol that pays intermediaries to act on behalf of users — relayers, bundlers, mixnet nodes, oracle networks — faces the same tension: rewarding the *first* actor leaks propagation information and invites racing, while pre-selecting an uploader via public randomness needs unpredictable randomness (RANDAO is described as vulnerable to prediction; verifiable delay functions and threshold beacons add operational overhead). The mechanism here avoids that dilemma by making everyone upload with the same probability and paying the reward to all participants, with penalties as the reliability lever.

## Future Directions

1. **Repeated interactions and dynamic fees.** The analysis covers a single upload window; the authors explicitly place the dynamic fees arising from repeated interactions outside the model's scope.
2. **Practical randomness and selection.** The paper argues pre-selected uploaders would need decentralized unpredictable randomness that is currently hard to achieve; a workable beacon or VDF-lite design remains open.
3. **End-to-end privacy accounting.** The model assumes the observer is confined to blockchain-level data; combining this incentive layer with network-level anonymity (Dandelion++, Loopix-style mixnets) and cryptographic primitives is left untested.
4. **Empirical calibration.** All numerical values are normalized ratios rather than literal gas measurements, so parameterizing b, c_f, c_l and p from real gas markets and testing an implementation are natural next steps. The paper's own concluding section on future research directions is not included in the provided text.

## Target Audience

Researchers and graduate students in blockchain security, mechanism design, and game theory; protocol designers building relayer, bundler, or mixnet incentive systems; and engineers at organizations running relayer infrastructure who want a principled alternative to first-uploader-wins reward rules. Readers need a working knowledge of Nash equilibrium and basic probability; the evolutionary-dynamics proofs and the potential-game argument will be heavy going for beginners.

## Authors’ abstract

Public blockchains, though renowned for their transparency and immutability, suffer from significant privacy concerns. Network-level analysis and long-term observation of publicly available transactions can often be used to infer user identities. To mitigate this, several blockchain applications rely on relayers, which serve as intermediary nodes between users and smart contracts deployed on the blockchain. However, dependence on a single relayer not only creates a single…

[The full abstract on arXiv](https://arxiv.org/abs/2601.06699)

[Read the original paper](https://arxiv.org/abs/2601.06699)

---

Canonical: https://crypto.world/paper/incentive-mechanism-design-for-privacy-preserving-decentralized-2601.06699
