Dynamic Merkle B-tree with Efficient Proofs

TL;DR

This paper introduces a self-balancing B-Merkle tree that uses concise intranode commitments, enabling smaller proofs and lower tree heights based on element order, improving efficiency for authenticated data structures.

Contribution

It presents a novel dynamic B-Merkle tree structure that reduces proof sizes and tree height by incorporating concise intranode commitments within a self-balancing tree.

Findings

Smaller proof sizes compared to traditional q-ary Merkle trees.

Lower tree height driven by element order rather than key length.

Efficient subset and non-membership proofs with similar communication costs.

Abstract

We propose and define a recursive Merkle structure with q-mercurial commitments, in order to create a concise B-Merkle tree. This Merkle B-Tree builds on previous work of q-ary Merkle trees which use concise, constant size, q-mercurial commitments for intranode proofs. Although these q-ary trees reduce the branching factor and height, they still have their heights based on the key length, and are forced to fixed heights. Instead of basing nodes on q-ary prefix trees, the B Merkle Tree incorporates concise intranode commitments within a self-balancing tree. The tree is based on the ordering of elements, which requires extra information to determine element placement, but it enables significantly smaller proof sizes. This allows for much lower tree heights (directed by the order of elements, not the size of the key), and therefore creates smaller and more efficient proofs and operations. Additionally, the B Merkle Tree is defined with subset queries that feature similar communication costs to non-membership proofs. Our scheme has the potential to benefit outsourced database models, like blockchain, which use authenticated data structures and database indices to ensure immutability and integrity of data. We present potential applications in key-value stores, relational databases, and, in part, Merkle forests.