IoT implementation and evaluation of distributed consensus algorithms

Abstract

This thesis is concerned with consensus algorithms suitable for implementation on resource-constrained devices operating in highly dynamic wireless networks – conditions that can be found in various IoT environments. Consensus is a classical problem in the field of distributed computing – we discuss both its terminating and its stabilizing variant.This thesis starts by presenting the system model and core concepts of a class of root component based algorithms – introduced in [SS21] and [WSS19] – that solve both the terminating and the stabilizing consensus problem in dynamic networks under eventually stabilizing message adversaries.Following the theoretical aspects of the root component based algorithms, we give a brief discussion of the MinMax consensus algorithm by [CM21] and put some central system model concepts of the MinMax algorithm in relation to core concepts of the root component based stabilizing algorithm.After the theoretical part, this thesis provides a detailed account of an actual implementation of the root component based stabilizing consensus algorithm and the MinMax consensus algorithm. This thesis also discusses the implementation of a simulation environment that was created to aid development. Finally we report on an experimental side-by-side evaluation of the root component based algorithm and the MinMax algorithm with regard to stabilization time.