Secure multiparty computation is becoming a concrete option in protocol design. Indeed, due to the astonishing improvement in computing power and refined optimization techniques, developed during the past years, the general solutions for secure multiparty computation are almost practical in several applications. The circuit-based solutions can be executed in reasonable times, and homomorphic-encryption-based solutions, for certain input sizes and suitable choices of methods and parameters, are doable too. In this paper our goal is to understand more the current costs of implementing such solutions, by using off-the-shelf tools for secure multiparty computation. Precisely, we focus on the wellknown private set-intersection problem, and on private computation of statistics on the set-intersection. Our contribution is the following:- we implement circuit-based solutions and homomorphic-encryption based solutions, and evaluate the corresponding performances and limits- we implement hybrid solutions, with a garbled-circuit part built on top of an homomorphic encryption-based part, estimate the cost of each part and evaluate the overall performance- we compare several optimized versions, in order to figure out the current ranges of concrete applicability, pointing out efficiency gains and bottlenecksWe conclude the paper discussing open issues and providing some hints for future researches.
Hybrid Solutions for 2PC: How Much do they Cost?
D'Arco P.
2025
Abstract
Secure multiparty computation is becoming a concrete option in protocol design. Indeed, due to the astonishing improvement in computing power and refined optimization techniques, developed during the past years, the general solutions for secure multiparty computation are almost practical in several applications. The circuit-based solutions can be executed in reasonable times, and homomorphic-encryption-based solutions, for certain input sizes and suitable choices of methods and parameters, are doable too. In this paper our goal is to understand more the current costs of implementing such solutions, by using off-the-shelf tools for secure multiparty computation. Precisely, we focus on the wellknown private set-intersection problem, and on private computation of statistics on the set-intersection. Our contribution is the following:- we implement circuit-based solutions and homomorphic-encryption based solutions, and evaluate the corresponding performances and limits- we implement hybrid solutions, with a garbled-circuit part built on top of an homomorphic encryption-based part, estimate the cost of each part and evaluate the overall performance- we compare several optimized versions, in order to figure out the current ranges of concrete applicability, pointing out efficiency gains and bottlenecksWe conclude the paper discussing open issues and providing some hints for future researches.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


