Bicomp: A Bilayer Scalable Nakamoto Consensus Protocol
Blockchain has received great attention in recent years and motivated innovations in different scenarios. However, many vital issues which affect its performance are still open. For example, it is widely convinced that high level of security and scalability and full decentralization are still impossible to achieve simultaneously. In this paper, we propose Bicomp, a bilayer scalable Nakamoto consensus protocol, which is an approach based on high security and pure decentralized Nakamoto consensus, and with a significant improvement on scalability. In Bicomp, two kinds of blocks are generated, i.e., microblocks for concurrent transaction packaging in network, and macroblocks for leadership competition and chain formation. A leader is elected at beginning of each round by using a macroblock header from proof-of-work. An elected leader then receives and packages multiple microblocks mined by different nodes into one macroblock during its tenure, which results in a bilayer block structure. Such design limits a leader's power and encourages as many nodes as possible to participate in the process of packaging transactions, which promotes the sharding nature of the system. Furthermore, several mechanisms are carefully designed to reduce transaction overlapping and further limit a leader's power, among which a novel transaction diversity based metric is proposed as the second level criteria besides the longest-chain-first principle on selecting a legitimate chain when fork happens. Security issues and potential attacks to Bicomp are extensively discussed and experiments for evaluation are performed. From the experimental results based on 50 nodes all over the world, Bicomp achieves significant improvement on scalability than that of Bitcoin and Ethereum, while the security and decentralization merits are still preserved.