| Summary: | The Internet of Things (IoT) and Smart Cities are nowadays a big trend, but with the proliferation of these systems several challenges start to appear and put in jeopardy the acceptance by the population, mainly in terms of sustainability and environmental issues. This Thesis introduces a new system composed by a modular IoT smart node that is self-configurable and sustainable with the support of machine learning techniques, as well as the research and development to achieve a innovative solution considering data analysis, wireless communications and hardware and software development. For all these, concepts are introduced, research methodologies, tests and results are presented and discussed as well as the development and implementation. The developed research and methodology shows that Random Forest was the best choice for the data analysis in the self-configuration of the hardware and communication systems and that Edge Computing has an advantage in terms of energy efficiency and latency. The autonomous communication system was able to create a 65% more sustainable node, in terms of energy consumption, with only a 13% decrease in quality of service. The modular approach for the smart node presented advantages in the integration, scalability and implementation of smart cities projects when facing traditional implementations, reducing up to 45% the energy consumption of the overall system and 60% of messages exchanged, without compromising the system performance. The deployment of this new system will help Smart Cities, in a worldwide fashion, to decrease their environmental issues and comply with rules and regulations to reduce CO2 emission.
|
|---|