Energy Footprint of Blockchain Consensus Mechanisms Beyond Proof-of-Work

Second generation consensus mechanisms, such as Proof-of-Stake, promise to provide more favourable energy consumption characteristics than those of their predecessors, such as Proof-of-Work. In this paper, we quantify and compare the energy demand of four archetypal modalities of second-generation systems: Algorand, Ethereum 2.0, Hedera Hashgraph, and Polkadot. While numerous studies that analyse the energy demands of individual distributed ledger systems have been undertaken previously, little work has been done to compare different systems that operate based on distinct technological assumptions. We approach this research question by formalising a basic mathematical consumption model for validatorbased Sybil attack resistance schemes. This model allows quantifying the energy consumption per transaction based on common input variables, such as the number of validators and the throughput characteristics of the system analysed. We find that, when applying contemporary throughput and validator counts, Hedera Hashgraph, by operating as a permissioned system, has the most favourable energy consumption characteristics with 20.95 mW h/tx. This stands in contrast to the permissionless systems Algorand with 4.427 W h/tx, and Polkadot with 115.6 W h/tx. A very broad projection for Ethereum 2.0 suggests an energy consumption of 2.862 W h/tx to 557.5 W h/tx. The present findings support the intuition that the complexity of Sybil attack resistance mechanisms, and therefore the energy needs of the overarching consensus protocols, is largely dependent on the number of active validators. Consequently, a permissioned setting in which a can control the number of validators can be beneficial to minimise energy consumption.