Error Correction Code Based Proof-of-Work on Ethereum

Abstract

These days, because of the centralization problem of proof-of-work (PoW), many researchers propose application-specific integrated circuit (ASIC)-resistant PoW and alternative consensus algorithm (e.g., proof-of-stake, delegated proof-of-stake, and Byzantium fault tolerance). However, networks of these alternative consensus algorithms present less decentralization than ASIC-resistant PoW. Specifically, in alternative consensus algorithms, a limited participant can generate blocks; but in ASIC-resistant PoW, anyone can join to generate a block. Thus, ASIC-resistant PoW presents a better-decentralized network than alternative algorithms. In this work, we utilize error-correction code based proof-of-work (ECCPoW) as known as ASIC-resistant PoW. The ECCPoW utilizes a low-density parity-check (LDPC) code that has flexible parameters: variable length code, continuously changed parity check matrix (PCM). Thus, ECCPoW is possible to impair ASIC by changing the parameter of LDPC. Previous researches on ECCPoW algorithms present its theory and the implementation on Bitcoin. However, they do not discuss the stability of its block generation time. Block generation time (BGT) must follow a distribution that has a finite mean to achieve the stability that can ensure confirmation of transactions. In ECCPoW algorithms, BGT follows a long-tailed distribution due to varying cryptographic puzzles. If this long-tailed distribution has a none finite mean, such as the heavy-tailed distribution, the confirmation of transactions is not guaranteed. Thus, validating the distribution of BGT is necessary to determine if consensus algorithms can guarantee the confirmation of transactions. In this work, we present the implementation, simulation, and stability validation of ECCPoW Ethereum. In the implementation, we demonstrate how Ethereum applies ECCPoW algorithm as a consensus algorithm. Moreover, in simulation, we perform a multinode simulation to show how ECCPoW algorithm works on Ethereum with difficulty change. In stability validation, to present moderate evidence if the BGT has a finite mean, we obtain a goodness-of-fit result using the Anderson-Daring test. Our implementation is at GitHub.