Autonomous Dominant Resource Fairness for Blockchain Ecosystems

Abstract

Blockchain systems have been a part of mainstream academic research, and a hot topic at that. It has spread to almost every subfield in the computer science literature, as well as economics and finance. Especially in a world where digital trust is much sought for, blockchains offer a rich variety of desired properties, such as immutability, public auditing, decentralised record keeping, among others. Not only has it been a research topic of its own, the integration of blockchains into other systems has been proposed as solutions in many areas, ranging from grid computing, cloud and fog computing, to internet of things, self driving vehicles , and smart cities. In many cases the primary function attributed to blockchains in these contexts is resource management. Although much attention is paid to this topic, the focus is on single resource allocation scenarios. Even the cases where multiple resource types are to be allocated, are treated as single resource type scenarios, and problems are formulated as allocating standardised bundles consisting of a fixed amount of each of them, such as virtual machines. The present study addresses the problem of allocating multiple resource types among tasks with heterogeneous resource demands with a smart contract adaptation of Precomputed Dominant Resource Fairness; an algorithm that approximates Dominant Resource Fairness, without loop iterations, which makes it preferable in the blockchain context because of the block gas limit. We present the resulting algorithm, Autonomous Dominant Resource Fairness, along with the empirical data collected from the tests run on the algorithm. The results show that Autonomous Dominant Resource Fairness is a gas-cost efficient algorithm, which can be used to manage hundreds of resource types for unlimited number of users.