Towards summarized representation of time series data in pervasive computing systems

Ubiquitous computing systems are connected with a number of sensors and devices immersed in the environment, spread throughout providing proactive context aware services to users. These systems continuously receive tremendous amount of information about their environment, users and devices. Such a huge amount of information deserves special techniques for efficient modeling, storage and retrieval. In this paper we propose the modeling of context information as time series and applying the time series approximation techniques to reduce the storage space requirements and for faster query processing. We applied an algorithm based on non-linear interpolation to approximate such data and evaluated the approximation error, storage space requirements and query processing time.