Enhancing Efficiency of Byzantine-Tolerant Coordination Protocols via Hash Functions

摘要

Distributed protocols resilient to Byzantine failures are notorious to be costly from the computational and communication point of view. In this paper we discuss the role that collision-resistant hash functions can have in enhancing the efficiency of Byzantine-tolerant coordination protocols. In particular, we show two settings in which their use leads to a remarkable improvement of the system performance in case of large data or large populations. More precisely, we show how they can be applied to the implementation of atomic shared objects, and propose a technique that combines randomization and hash functions. We discuss also the earnings of these approaches and compute their complexity.