A machine learning approach for indirect human presence detection using IoT devices

• Autor(a) principal: Madeira, Rui Nuno Neves
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/13012
• topic: Human presence detection; Ambient intelligence; Internet of things; Engenharia da programação; Machine learning; Inteligência artificial; Processamento de imagens; Vigilância electrónica; Dispositivo de controlo; Sensor; Natureza humana; Internet

The recent increased democratization of technology led to the appearance of new devices dedicated to the improvement of our daily living and working spaces, capable of being remotely controlled through the internet and interoperability with other systems. In this context, human presence detection is fundamental for several purposes, such has: further automization, usage pattern learning, problem detection (illness, or intruder), etc. Current intrusion detection devices usually have flaws depending on type and many times are not coordinated for better performance. Coordinating the devices for higher level operation however requires a device, or software, that is able communicate and control them. Muzzley is a company that tries to solve this issue by creating a mobile application where the user can register all its devices and control them from there. In this dissertation we propose an approach to human presence detection using metrics based on messages between devices and the Muzzley platform. The idea is that the detection does not rely on information from specific presence detectors, but that it is able to achieve its purpose by analyzing the patterns of interactions with the devices. For this, anonimyzed datasets created by the Muzzley platform are submitted to an extensive processing in order to create meaningful features that will then be used with a machine learning algorithm for training and testing. The main contributions of this study is the processing done to create meaningful information for the task, the demonstration of the capabilities of the interactions between these devices and platforms for human presence detection, and the methods used to improve the performance of the approach.