Activity Recognition In Operational Process Under Ubiquitous Computing Environment

With the rapid development of computer technology and wireless network, wide use in sensors and mobile devices, computing model has developed from the traditional desktop computing model to the ubiquitous computing model. In ubiquitous computing environment, the user could transparently make use of computing power and service resources, therefore, the ubiquitous computing environment should have the ability to access and deal with the surrounding environment and infer contextual information related to user activity. Studies have shown that if it can be accurately identified the behavior and the action of the user's action or activities engaged in, then provide users with accurate, timely and appropriate service information based on the identification the activities which the user conducts, it is especially useful to help the user to operate and complete the activities. Consequently, activity recognition has become an important topic in ubiquitous computing research field in recent years.During the activity recognition process, it is a challenging task to infer the user's activities based on the surrounding environment changes and the user's action and behavior. The activities to recognize in this paper is in a process of specification and it is composed by a number of complex operations in order to achieve a certain purpose or to complete certain tasks of complex organic activity entity, according to context changes over time, if we could identify the activities that the user operates and the abnormal movements, the tips and services are provided to the user.In this paper, we refer to the previous research of activity recognition and summarize the general steps of the activity recognition. We put forward the hierarchical system framework for activity recognition:context acquisition layer, context processing layer and context-aware services layer. During the context processing layer which includes activity modeling and activity recognition, the context reasoning is adopted to recognize the activity. In order to formalize the contextual information describing the processes and activities of entities, we analyze constraint on the flow structure and activities and activity flow model are established based on the logical relationship between activities in the process. The context pattern and the context evolving pattern are proposed to describe the activity and the different status of activities in order to accurately identify a user's activity and abnormal operating activities. We propose several rules which are used to distinguish the activity status. When the activity model to meet the above rules, we could use the activity recognition algorithm to recognize the corresponding activity based on the activity model, the known activity and its status and the context values at some point. For the activities does not meet the active rules, we achieve the goal of activity recognition through the activity model and the context valued during a period of time and monitor the abnormal movements when the user conducts the activity.In this paper, we model the specific activities of the application process and design the prototype system of the activity recognition. We realize the prototype system and the proposed activities recognition algorithm, then verify the proposed activities recognition method, which is more effective to realize activity recognition and monitor the user's abnormal operation.