Fairness and Efficiency in DAG-Based Cryptocurrencies

Bitcoin is a decentralised digital currency that serves as an alternative to existing transaction systems based on an external central authority for security. Although Bitcoin has many desirable properties, one of its fundamental shortcomings is its inability to process transactions at high rates. To address this challenge, many subsequent protocols either modify the rules of block acceptance (longest chain rule) and reward, or alter the graphical structure of the public ledger from a tree to a directed acyclic graph (DAG). Motivated by these approaches, we introduce a new general framework that captures ledger growth for a large class of DAG-based implementations. With this in hand, and by assuming honest miner behaviour, we (experimentally) explore how different DAG-based protocols perform in terms of fairness, as well as efficiency. To do so, we isolate different parameters of the network (such as k, the number of pointers to previous blocks) and study their effect on those performance metrics. Our results demonstrate how the DAG-based ledger protocols described by our framework cope with a high transaction load. More specifically, we show that even in a scenario where every miner on the system is honest in terms of when they publish blocks, what they point to, and what transactions each block contains, fairness and efficiency of this kind of ledgers can break down at specific hash rates if miners have differing levels of connectivity to the P2P network sustaining the protocol. (The full version of this paper can be found in[2])