Abstract:
<p>Localization is projected to play a critical role in mobile computing applications. Nevertheless, the state of the art is inadequate especially when operating on mobile devices. More specifically, the on-phone GPS sensor has an unacceptable energy consumption and does not operate indoors. Alternate localization techniques, based on WiFi or GSM, alleviate some of the GPS limitations but provide degraded accuracy and assume pre-installed infrastructure. As a result, these solutions need extensive war-driving for collecting location fingerprints and, in many instances, limit services to regions close to drivable paths. Moreover, when infrastructure is scarce or missing, the localization accuracy is poor. Lastly, relying on hardware deployments is costly and raises scalability concerns when targeting wide regions.</p><p>To address the shortcomings of current solutions, we propose four new localization systems: (1) CompAcc enables energy-efficient, war-driving-free localization using the phone inertial sensors and digital maps, (2) Escort provides indoor localization by exploiting the phone inertial sensors and social environments where people are mobile, (3) SurroundSense enables indoor logical localization (e.g., inside Target) by sensing the user ambiance through the phone sensors, and (4) EnLoc proposes energy-efficient localization via personalized mobility profiling and predictions.</p>
