Unleashing the Power of the Crowd: Towards Efficient and Sustainable Mobile Crowdsensing

Mobile crowdsensing has become a popular computing paradigm. It enables ubiquitous mobile devices, including wearable and industrial equipment, to collect and share sensing data at large scales. The crowdsourced data is analyzed comprehensively to understand phenomena of common interest or to create sensor-enriched maps. Examples include the monitoring of city-scale traffic, the sharing of discount information in shopping malls, and so on. This thesis explores challenges associated with efficiently bootstrapping and continuously improving the performance of mobile crowdsensing systems with crowdsourced data. The research questions include exploring the feasibility of adopting crowdsourced data to create new services, creating algorithms for the efficient collecting, organizing and utilizing crowdsourced data, and offering reliable services during the bootstrap stage while keep improving the service quality afterwards. In particular, we try to obtain answers from the practice of designing and developing two real-life mobile crowdsensing systems: one uses cellular signal-strength traces contributed by mobile users to achieve mobile energy efficiency; another adopts crowdsourced images and inertial sensor readings to offer indoor location-based services.