bitcoin-dev

Altruistic Rebroadcasting - A Partial Replacement Cycling Mitigation

Altruistic Rebroadcasting - A Partial Replacement Cycling Mitigation

Original Postby Antoine Riard

Posted on: December 21, 2023 21:59 UTC

In a recent discourse on the security model for time-sensitive second-layer networks, it was noted that introducing an altruistic party to rebroadcast transactions could parallel dynamics observed with BIP157 filter servers, where few nodes serve many clients.

Concerns were raised over how these second-layer nodes would determine the global mempool minimum fee and the potential for adversaries to exploit the system by inflating the block template's top percentage to prevent the rebroadcast of malicious transactions.

The conversation also delved into the issue of an adversary repeatedly broadcasting the same UTXO amount under different transaction IDs, thereby consuming all the available altruistic bandwidth. It was highlighted that current economic estimations based on per-block metrics such as 3.125 BTC every 10 minutes might not account for the possibility that adversaries could reuse this value continuously during inter-block intervals. The notion of introducing rate-limitation per UTXO as a solution was discussed, but concerns were raised about its effectiveness and the risk of creating new vulnerabilities for legitimate transactions.

Furthermore, the concept of package relay and its design considerations were examined, especially in relation to the challenge of preventing an adversary from using the same UTXO liquidity across numerous concurrent replacement transactions, effectively exhausting altruistic bandwidth. It was argued that the financial resources assumed for the adversary might be underestimated.

As a proposed long-term solution, attention was drawn to efforts directed towards fixing replacement cycling, which involve eliminating package malleability and relying on self-contained fee-bumping reserve UTXOs—a method detailed in a December 2023 post on the Linux Foundation mailing list (https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2023-December/022191.html). This approach suggests a shift away from per-transaction fees to bundled transactions with built-in fee mechanisms.