Enrichment of raw sensor data to enable high-level queries

Abstract

Sensor networks are increasingly used across various application domains. Their usage has the advantage of automated, often continuous, monitoring of activities and events. Ubiquitous sensor networks detect location of people and objects and their movement. In our research, we employ a ubiquitous sensor network to track the movement of players in a tennis match. By doing so, our goal is to create a detailed analysis of how the match progressed, recording points scored, games and sets, and in doing so, greatly reduce the eort of coaches and players who are required to study matches afterwards. The sensor network is highly efficient as it eliminates the need for manual recording of the match. However, it generates raw data that is unusable by domain experts as it contains no frame of reference or context and cannot be analyzed or queried. In this work, we present the UbiQuSE system of data transformers which bridges the gap between raw sensor data and the high-level requirements of domain specialists such as the tennis coach.