Erasure-Coded Blockchain Storage for Fast Local Data Repair

The patent divides the redundant storage requirement into segments by applying erasure coding techniques. Instead of storing complete copies of transaction data in all nodes, the data is segmented into data blocks and parity blocks through RS encoding, then further optimized with LRC encoding to create locally repairable code groups. This segmentation allows Byzantine fault tolerance to be maintained with reduced storage capacity per node.

The patent changes the storage parameter from full redundant copies to erasure-coded fragments. By transforming the storage model using RS (Reed-Solomon) encoding parameters and LRC (Locally Repairable Code) encoding parameters, the system achieves the same fault tolerance level with reduced data storage capacity requirements. The parameter transformation converts k data blocks into n encoded blocks where n > k, but each node stores only a portion of the total encoded data.

The patent segments the RS-encoded data into multiple LRC code groups, where each group contains a subset of data blocks and corresponding parity blocks. This segmentation allows quick access to specific data by only querying the relevant LRC group rather than the entire dataset, thereby improving access speed while maintaining data integrity through the distributed nature of the encoding.

The patent applies local quality optimization by creating locally repairable code groups with localized parity blocks. Each LRC group is self-contained and can be accessed independently, allowing fast retrieval of specific data blocks without requiring access to the entire encoded dataset. This local organization maintains integrity through distributed verification while enabling rapid data access.

The patent segments data into different access frequency categories and applies LRC encoding selectively to frequently accessed data blocks. This segmentation strategy identifies hot data that benefits from LRC's faster access characteristics while leaving cold data in standard RS encoding, thereby improving overall access speed for critical operations without unnecessarily increasing storage complexity for all data.

The patent applies LRC encoding partially rather than universally - specifically to data blocks that require frequent access. This partial application of the more complex LRC encoding technique provides performance benefits where needed while avoiding the full storage complexity overhead across the entire dataset, achieving an optimal balance between access speed and system complexity.