Monkey King: An Enhanced Asynchronous Consensus Protocol with Provable Security

The asynchronous consensus protocol, as one of the core technologies of the blockchain, can ensure that the nodes in the system can agree on a certain value or status without assuming network capabilities. However, existing asynchronous consensus protocols face problems such as low efficiency and limited scalability. To address these problems, we propose an asynchronous consensus algorithm based on sharding, i.e., Monkey King, which innovatively introduces sharding into the field of asynchronous consensus. Our protocol divides the network into fixed-size shards, runs asynchronous common subset protocol (ACS) in each shard, selects a node from each shard to form the mainchain, then runs ACS once on the mainchain, and finally broadcasts the consensus result to All nodes. We implement the MK protocol and test the performance of different shard numbers under 100 nodes. Finally, we conduct comparative experiments between our protocol and Dumbo1/Dumbo2 under different numbers of nodes. Experimental results show that under 100-node scenario, the latency of our protocol is reduced by 94.51% and 79.04% respectively compared to Dumbo1 and Dumbo2. At the same time, under the condition of 100 nodes and a batch size of 10000, the throughput of our protocol increases by 520.1% and 12.7% respectively compared to Dumbo1 and Dumbo2, which reflects that our protocol has better performance in multi-node scenarios.