Extending Battery Life by Employing Fog Computing in CoAP

Autor	Mo Shen
Date	8. February 2019
Degree	Bachelor
Topic	Extending Battery Life by Employing Fog Computing in CoAP
Title	Extending Battery Life by Employing Fog Computing in CoAP
Abstract	Mobile devices are one of the most important components in the vision of Internet of Things (IoT). With the rapid development of wireless communication protocols, building a network consisting of a massive number of sensors is not a great challenge any more. The Constrained Application Protocol (CoAP) is one of the emerging protocols that focusses on the efficiency and reliability of Machine-to-Machine (M2M) communications. The CoAP Option Observe allows to subscribe to a resource instead of polling it constantly. Still, Observe lacks a flexible and standardised interface to specify which data is of interest. This often results in unnecessary high levels of network traffic and power consumption. Thus, more work can be done to optimize existing approaches or explore new possibilities. This thesis is motivated by the concept of fog computing and attempts to pursue a widely applicable method in this domain. This method aims to be energy efficient in dealing with long-term resource-monitoring tasks and focuses on extending battery life of embedded devices. As a result, an Application Programming Interface (API) is introduced and implemented. Two kinds of applications of this API are assumed and simulated in the experiments. With the help of this API, CoAP devices are able to process collected data at the data source. This reduces the necessary wireless communication and is effective to reduce power consumption. Additionally, for those computation-intensive tasks, this API gives mobile devices the ability to transfer part of work to remote servers via dynamic code migration. This method avoids a shortened battery lifetime caused by the sustained high CPU load. Through analyzing the obtained results of the two experiments, the amount of power that could be saved turns out to be significant. The response time of computation-intensive tasks could be reduced up to 10% through dynamic code migration.