Managing Context Information in Mobile Devices

M obile device users want to be able to access and manipulate information and services specific to their location, time, and environment. Context information gathered from sensors, networks, device status, user profiles, and other sources can enhance mobile applications’ usability by letting them adapt to conditions that directly affect their operations. To achieve true context awareness, however, mobile systems must produce reliable information in the presence of uncertain, rapidly changing, partially true data from multiple heterogeneous sources. Mobile devices equipped with low-cost sensing elements can recognize some aspects of context. However, extracting relevant context information by fusing data from several sensors proves challenging because noise, faulty connections, drift, miscalibration, wear and tear, humidity, and other factors degrade data acquisition. Extracted contexts overlap, change with time, and yield only partially reliable approximations. Furthermore, mobile devices’ dynamic environments require that we learn context descriptions from multidimensional data. Learning systems can’t easily generalize beyond training data, however. Using even sufficiently reliable derived contexts directly to control mobile applications poses problems because users with different ideas of “context” might find application behavior irritating. To address these challenges, we present a uniform mobile terminal software framework that provides systematic methods for acquiring and processing useful context information from a user’s surroundings and giving it to applications.