Optimal scheduling in sensor networks

Abstract

We aim to address, some of the fundamentals on how to model the optimal control of data problems arising in this big data era accurately and effectively. In this thesis, we have introduced a framework for formulating and solving the optimal scheduling problem of both communication and computation in a sensor network, with minimising the latency, i.e. the overall task completion time, and the total energy consumption as the objectives. We are able to schedule both the communication and computation by formulating the proposed optimal scheduling problem as a nonlinear programming (NLP) problem. We present two novel optimal scheduling problems in sensor networks: a single objective scheduling problem, minimising the overall task completion time, and a bi-objective scheduling problem, minimising both the overall task completion time and the total energy consumption. We propose the design of a decentralised discrete processing and transmission protocol, effectively turning the continuous-time, uncountable speed set solution into a discrete-time, countable speed set implementation. This significantly reduces the computation complexity compared to solving a mixed integer nonlinear programming (MINLP) instead. By implementing the normal constraint (NC) method, we are able to generate an evenly-distributed point-wise approximation of the Pareto curve to show the energy and latency trade-offs of our proposed bi-objective optimal scheduling problem. The Pareto curve can provide a guideline and reference for parameter design and selection. We also demonstrate how to modify and implement our framework by studying the optimal communication setup of smart meters in a smart building. The modified formulation and several case studies on optimal communication topology and transmission rates setups in various smart meter networks are presented. Numerical results show that the overall energy consumption can be reduced by implementing the optimal communication architecture and transmission rate setup, rather than implementing a straightforward communication architecture with uniform channel bandwidth.