Langheinrich, Marc Giordano, Silvia Papandrea, Michela 2015-03-17 In recent years we are witnessing a noticeable increment in the usage of new generation smartphones, as well as the growth of mobile application development. Today, there is an app for almost everything we need. We are surrounded by a huge number of proactive applications, which automatically provide relevant information and services when and where we need them. This switch from the previous generation of passive applications to the new one of proactive applications has been enabled by the exploitation of context information. One of the most important and most widely used pieces of context information is location data. For this reason, new generation devices include a localization engine that exploits various embedded technologies (e.g., GPS, WiFi, GSM) to retrieve location information. Consequently, the key issue in localization is now the efficient use of the mobile localization engine, where efficient means lightweight on device resource consumption, responsive, accurate and safe in terms of privacy. In fact, since the device resources are limited, all the services running on it have to manage their trade-off between consumption and reliability to prevent a premature depletion of the phone’s battery. In turn, localization is one of the most demanding services in terms of resource consumption. In this dissertation I present an efficient localization solution that includes, in addition to the standard location tracking techniques, the support of other technologies already available on smartphones (e.g., embedded sensors), as well as the integration of both Human Mobility Modelling (HMM) and Machine Learning (ML) techniques. The main goal of the proposed solution is the provision of a continuous tracking service while achieving a sizeable reduction of the energy impact of the localization with respect to standard solutions, as well as the preservation of user privacy by avoiding the use of a back-end server. This results in a Smart Localization Service (SLS), which outperforms current solutions implemented on smartphones in terms of energy consumption (and, therefore, mobile device lifetime), availability of location information, and network traffic volume. application/pdf https://n2t.net/ark:/12658/srd1318735 https://susi.usi.ch/global/documents/318735 https://susi.usi.ch/documents/318735/files/2015INFO014.pdf eng info:eu-repo/semantics/altIdentifier/urn/urn:nbn:ch:rero-006-114977 info:eu-repo/semantics/altIdentifier/ark/12658/srd1318735 info:eu-repo/semantics/openAccess License undefined Localization Mobile computing Mobile-context based applications Mobile resources usage optimization Mobility prediction Mobility modeling Activity inference Points of interest Point of interest classification User behaviour inference info:eu-repo/classification/udc/004 SLS: Smart localization service : human mobility models and machine learning enhancements for mobile phone’s localization http://purl.org/coar/resource_type/c_db06