Optimal Multi-Tier Clustering of Permissioned Blockchain Systems for IoT

Abstract

In this work we describe and analyze a novel multi-tier architecture for recording Internet of Things (IoT) data. Each tier in the architecture consists of vertically interconnected node clusters that may be operated by different service providers. Blocks of IoT data are accepted by a cluster tier by tier, until they reach the top tier which adds them to the global blockchain and makes them available to client applications. To allow for large geographical coverage and reduce the dependency on any individual node in the cluster, clusters run a multiple entry version of the Practical Byzantine Fault Tolerance (PBFT) consensus algorithm. We describe an analytical model of the system and find the optimal allocation of clusters and nodes within clusters for a given number of tiers and system coverage area. Our results show that the total number of orderers for given system coverage and total load are main predictors of the block linking time, and indicate that minimizing block linking time is easier to achieve with a smaller number of tiers and large number of clusters with a smaller number of orderers per cluster. These observations can be used as guidelines for planning and dimensioning of multi-tier cluster architectures that create and maintain blockchain structures in a given coverage area.