A proposed study aims to analyze the impact of different tx insertion and confirmation rates on a network, with an emphasis on Replace-By-Fee (RBF) transactions and their patterns on the current mainnet.

To conduct this analysis, a simulation setup involving a network of 100 nodes is suggested, where 90 are listening and 10 are non-listening. Researchers plan to manipulate three parameters: the size of the network, the transaction creation rate, ranging from 5 transactions per second (tx/s) to 70 tx/s, and the transaction confirmation rate with three potential speeds—3.75 tx/s, 7.5 tx/s, or 15 tx/s. The Bitcoin Core version 26.0 will be used consistently across all nodes, which will each operate with their own wallet. Adjustments such as reducing the MAX_OUTBOUND_FULL_RELAY_CONNECTIONS parameter may be necessary to better emulate the conditions of larger networks.

For the mining process, blocks are to be mined every ten minutes by selecting a random peer, and transactions created from block rewards will be split into smaller amounts. In addition to these transactions, researchers will generate small transactions at the specified rate using either Taproot or P2WSH formats. The fee rate for these transactions will be dynamically calculated using the estimatesmartfee 5 command adjusted by a random factor.

Performance metrics will primarily focus on compact block relay efficiency, assessed by enabling debug mode for compact blocks and analyzing log entries that indicate successful block reconstructions. Expectations include the possibility of inconsistent mempools and increased 'txn requested' numbers when sustained tx creation rates exceed 18 tx/s. Such inconsistencies are likely to surface when conducting RBF transactions or dealing with chains of unconfirmed transactions, especially those involving Child Pays For Parent (CPFP) scenarios.

This systematic approach to studying transaction handling in blockchain networks is essential for identifying potential bottlenecks and optimizing network configurations to handle real-world usage patterns effectively.