Research On Activity Monitoring Based On Context Temporal Relation In Ubiquitous Computing Environment

Pervasive computing is a new computing paradigm following desktop computing, and it emphasizes the integration of physical space and information space. With the development of pervasive computing, to provide computing services transparently taking advantage of context-aware technology for users under working status has become one of research issues in pervasive computing area. Studies show that activities to be accomplished by several standardized operations, often can’t reach their goals, and even cause safety accidents due to execution conditions not met, user’s abnormal behaviors or improper operations, such as missing some operations or inaccurate operating sequence, et al. To avoid the occurrence of situation above and standardize user’s operations, obviously, it is necessary to monitor the user’s activity real-time, and provide auxiliary services suitable for current activity state and location according to the monitoring results. In view of analysis above, the paper focuses on how to detect the abnormal cases according to the contexts related to activity.The activity considered in this paper is accomplished by several actions or standardized operations in some objective conditions, including much temporal information, and the temporal information also provides evidence and cues for abnormal cases detecting. Therefore, it is necessary to describe the temporal character to determine activity state according to the contexts and their changes accessed from sensors and other computing devices embedded into physical space.This paper firstly describes the framework of activity monitoring system based on context-aware technology. Considering contexts change dynamically during the execution of activity, context pattern is defined in this paper, and the temporal relationship between different context patterns is described to express the temporal information included in activity. We summarize abnormal patterns to detect abnormalities that context patterns do not satisfy the temporal relationship corresponding to normal situation, and the algorithm is described to judge abnormality. Finally, we prescribe activity monitoring model based on context temporal relation about microstructure analysis experiment. The abnormalities in inlay operation are detected according to context change in different moments, and the method is verified according to context inputs in different moments, proving that context-aware computing into activity monitoring will result in considerable benefits.