On the Computation of Multi-Scalar Multiplication for Pairing-Based zkSNARKs

Abstract

Multi-scalar multiplication refers to the operation of computing multiple scalar multiplications in an elliptic curve group and then adding them together. It is an essential operation for proof generation and verification in pairing-based trusted setup zero-knowledge succinct non-interactive argument of knowledge (zkSNARK) schemes, which enable privacy-preserving features in many blockchain applications. Pairing-based trusted setup zkSNARKs usually follow a common paradigm. A public string composed of a list of fixed points in an elliptic curve group called common reference string is generated in a trusted setup and accessible to all parties involved. The prover generates a zkSNARK proof by computing multi-scalar multiplications over the points in the common reference string and performing other operations. The verifier verifies the proof by computing multi-scalar multiplications and elliptic curve bilinear pairings.

Multi-scalar multiplication in pairing-based trusted setup zkSNARKs has two characteristics. First, all the points are fixed once the common reference string is generated. Second, the number of points n is typically large, with the thesis targeting at n = 2^e (10 ≤ e ≤ 21). Our goal in this thesis is to propose and implement efficient algorithms for computing multi-scalar multiplication in order to enable efficient zkSNARKs.

This thesis primarily includes three aspects. First, the background knowledge is introduced and the classical multi-scalar multiplication algorithms are reviewed. Second, two frameworks for computing multi-scalar multiplications over fixed points and five corresponding auxiliary set pairs are proposed. Finally, the theoretical analysis, software implementation, and experimental tests on the representative instantiations of the proposed frameworks are presented.