Secure Entropy Gatherer
Daniel Mahu, Vlad Dumitrel, Florin Pop · 2015
High quality entropy data is a very useful asset for a large variety of applications, including computer security, simulations, and gaming. Not only is it important for this type of data to be indistinguishable from true randomness, it must also be possible to generate it in an efficient and scalable manner. Although entropy for local use can be easily obtained from various hardware sources, it does not always suffice, either in terms of quantity or quality. In this paper we explore the benefits of obtaining entropy using a distributed system. The rate at which entropy data is produced can be improved by generating it in parallel on multiple sites. Additionally, because the bits of entropy come from multiple and possibly heterogeneous sources, the quality of the entropy can also be improved. We propose a centralized, distributed system that is able to generate entropy on multiple sites across the network, and aggregate the generated data locally. We show that the system is secure, and that it offers good performance and high quality entropy.