Boson Consensus - A Scalable Blockchain Consensus Algorithm

[qizhou]

@plotozhu

“A cross-shard receipt will be incentivized to be eventually processed at destination shard.” shard-to-shard communication should be described and discussed to prove that your scheme can be practical.

Please check the following assumptions when I discussed the shard-to-shard communications:

@qizhou

Assumptions:

A1. A BP node of shard j is a light-client of shard i and full node of the root chain (similar to beacon chain in ETH2.0), and thus the node is able to obtain and verify the receipts of all blocks of shard i via a Merkle root hash (Note that “all blocks” constraint may be relaxed to “recent blocks”)

A2. The Merkle root hashes of cross-shard receipts from all shards are included in the root chain in a deterministic order.

A3. A shard block contains a hash pointer to root block and a cursor :

• The root block hash pointer tells which root block the BP of the shard block observed and included. According to A1, the BP could fully recover all receipts from shards by the Merkle root hashes of the blocks of root chain since the genesis root block. By ordering the receipts according to their position in the Merkle root and the position in the root chain, we can obtain a deterministic receipt queue, which will be processed by the shard in sequence.
• The cursor tells which receipt has been processed by the shard block in the queue, and thus the rest of the queue is called post-block unprocessed receipt queue , which can be very long due to DoS (however, the queue can be constructed by any BP of the shard on-demand and asynchronously according to A1 and A2, and thus all BP of the shard to be attacked won’t be overwhelmed by the DoS attack)