Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM

A Groth16-based confidential settlement protocol on Hyperledger Besu that keeps the payer accountable while hiding the payee, amount, and payload

Pietro Tiberi; Gabriele Marcelli; Vitangelo Lasorella· 2026· DOI 10.48550/arXiv.2608.28529

The core problem

Central Bank Digital Currency (CBDC) interbank settlement systems built on Distributed Ledger Technology (DLT) inherit a structural tension from their underlying blockchain: the same transparency that enables trustless verification also exposes commercially sensitive bilateral transaction flows to every network participant. In a permissioned network of commercial banks, a central bank operator, and a securities depository, this means that settlement amounts, counterparties, and business payloads become visible to competitors and observers, undermining the confidentiality that interbank markets require. The authors, Pietro Tiberi, Gabriele Marcelli, and Vitangelo Lasorella, frame the problem as a trade-off between regulatory accountability and commercial privacy. Full anonymity would conflict with Anti-Money Laundering and Counter-Financing of Terrorism (AML/CFT) obligations, which demand that the initiating institution be identifiable. The paper therefore introduces a **relaxed sender anonymity model**: the initiating institution remains publicly identifiable on-chain for accountability and compliance, while the receiving institution, the transfer amount, and the business payload

Innovation

The experimental evaluation reports end-to-end settlement in **8-16 seconds** across the five-node network. Proof verification overhead is approximately **1 ms**, consuming around **220k gas** via EVM precompiles. Client-side proof generation takes **4-12 seconds** on commodity ARM hardware. These figures indicate that the cryptographic overhead is compatible with interbank settlement workflows, where latency budgets are typically measured in seconds to minutes rather than milliseconds. The results also show that receiver confidentiality is established at the protocol level: the receiving institution, transfer amount, and business payload are not revealed on-chain. However, the authors note a limitation in the current proof-of-concept: the `NoteRegistry` uses owner-indexed events, which can leak metadata about which owner is associated with a note. This is described as a trade-off addressable in production via uniform event broadcasting, where events are emitted in a way that does not correlate them to specific owners.

Central Bank Digital Currency (CBDC) interbank settlement systems built on Distributed Ledger Technology (DLT) inherit a structural tension from their underlying blockchain: the same transparency that enables trustless verification also exposes commercially sensitive bilateral transaction flows to every network participant. In a permissioned network of commercial banks, a central bank operator, and a securities depository, this means that settlement amounts, counterparties, and business payloads become visible to competitors and observers, undermining the confidentiality that interbank markets require.

The authors, Pietro Tiberi, Gabriele Marcelli, and Vitangelo Lasorella, frame the problem as a trade-off between regulatory accountability and commercial privacy. Full anonymity would conflict with Anti-Money Laundering and Counter-Financing of Terrorism (AML/CFT) obligations, which demand that the initiating institution be identifiable. The paper therefore introduces a **relaxed sender anonymity model**: the initiating institution remains publicly identifiable on-chain for accountability and compliance, while the receiving institution, the transfer amount, and the business payload are cryptographically obfuscated. This design targets permissioned Ethereum-compatible networks and is realized on Hyperledger Besu using QBFT consensus, with zero-knowledge proofs, hash commitments, and multi-recipient encryption as its core cryptographic primitives.

Why it matters

The paper's central contribution is the relaxed sender anonymity model, which aligns privacy with AML/CFT requirements by keeping the initiating institution publicly identifiable while hiding the receiver, amount, and payload. This is a pragmatic middle ground between full transparency and full anonymity, and it reflects the reality that regulated interbank settlement cannot operate without accountability for the payer. The use of Groth16 over BN254, Poseidon commitments in an incremental Merkle tree, and multi-recipient ECIES encryption provides a coherent cryptographic stack that avoids trusted off-chain custody servers. The on-chain `NoteRegistry` as an append-only ledger log is a notable design choice: it keeps encrypted notes on-chain, reducing trust assumptions, but introduces the owner-indexed event trade-off. The authors acknowledge this and propose uniform event broadcasting as a production mitigation. From a performance perspective, the 8-16 second settlement time and 1 ms verification overhead suggest that the protocol is viable for permissioned CBDC environments. The 4-12 second client proof generation on commodity ARM hardware is the dominant cost, but it remains within acceptable bounds for interbank settlement. The five-node network topology—three commercial banks, a central bank operator, and a securities depository—is representative of a realistic CBDC settlement ecosystem, which strengthens the external validity of the results. Overall, the work demonstrates that zero-knowledge proofs can be integrated into permissioned EVM-based CBDC settlement to achieve confidential transfers without sacrificing regulatory accountability. The main open challenge is the metadata leakage from owner-indexed events, which the authors flag as a production-level concern rather than a fundamental flaw.

Who should read this

CS practitioners and researchers

Opening member content…