Uma abordagem de privacidade de dados em serviços baseados em localização

• Ano de defesa: 2019
• Autor(a) principal: Duarte Neto, Eduardo Rodrigues
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Privacidade de dados; Ofuscação; Localizações falsas; Serviços de localizações; k-anonimato
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/47042

The increasing popularity of mobile devices with Internet connectivity and the presence of global positioning systems (GPS) as a standard component have provided the increasing emergence of location-based services (LBS). As a consequence, a large amount of location data has been generated. Such data can be used for a variety of purposes, such as traffic flow analysis, infrastructure planning, understanding of human behavior, etc. However, accessing personal location data can pose severe privacy issues for most users, especially when the service provider itself is a potential malicious agent. Location data by their nature are highly correlated to other types of information, which combined can be used to infer sensitive data from individuals associated with the data. Many of the existing approaches involving location anonymization in location-based services consider the service provider as a trusted entity, responsible for performing the process of anonymizing user data. In this work we propose the PrivLBS, an obfuscation algorithm that uses the dummy location technique, whose objective is to protect the user’s location among other locations present in the request, without loss of service’s quality. We also propose an attack algorithm that seeks to identify the actual location of the user in requests sent to the service provider. Experimental results demonstrate that our attack algorithm obtained a high rate of identification of the real location of the user when applied to anonymized requests by strategies used by related works, whereas under equal conditions PrivLBS was able to protect the user’s requests.