Workload Prediction in BTC Blockchain and Application to the Confirmation Time Estimation

Blockchains are distributed ledgers storing data and procedures in an immutable way. The validation of the information stored therein as well as the guarantee of its immutability can be achieved without the need of a central authority. Proof-of-work is the maximum expression of the distributed nature of such systems, and requires miners to spend a large amount of energy to secure the blockchain. The cost is mostly paid by the end-users that offer fees to support the validation of their transactions. In general, higher fees correspond to shorter validation delays. However, given the limited throughput of the system and variability of the workload, the fee one needs to offer to satisfy a certain requirement on the validation delay strongly depends on the intensity of the workload that, in turns, is subject to high variability.In this work, we propose a time series analysis of the workload of Bitcoin blockchain and compare the accuracy of Facebook Prophet model with a ARIMA model. We take into account the periodicity of the workload and show by simulations how these predictions, accompanied with their confidence intervals, can be used to estimate the confirmation delays of the transactions given the offered fees.