Cloud services and Cloud learning: autonomous reacting algorithms applied to Wireless Sensor Networks

Abstract

Internet of Things (IoT), Wireless Sensor Network (WSN), Cloud Services and Artificial Intelligence (AI) are having their breakthrough for the past few years. The markets are booming and growing at high speed, the number of solutions offered by enterprises is increasing and money is being invested in research and development. These fields complement each other in distinct ways. Therefore, there are thousands of projects where these technologies are merged to fulfil the necessities each one has. WSN allows the real-time enviroment monitoring, providing the inputs necessary for AI to learn. AI solves the problem of requiring a real person analysing large amounts of data, in short periods of time. Additionally, the Cloud gives all the necessary computational power to store data and analyse it. Furthermore, it provides a scalable and flexible way for the system to grow, that traditional computers do not. However, it is still not possible to find a generic solution, or a standard way, to implement an IoT solution with all these technologies. There is still some divergence in terms of data collection, communication methods and algorithms used. These facts, are a challenge merging multiple IoT solutions together. The current Master Thesis aims to study the best solutions and to provide a way to implement a generic solution, that helps to reduce this fragmentation and to solve some of the problems these fields are facing. The results obtained from this Master Thesis proves how the use of msgpack can make data serialization faster and reduce the message length. Moreover, it was proven the use of multiple security layers was able to reduce and avoid most of the security issues found nowadays. Additionally, this Master Thesis provides insights in how the creation of multiple microservices increases scalability and security. Furthermore, it shows how CLARA may be a good option to have an AI service that easily learns from any source.