Pervasive computing in swimming: A model describing acceleration data of body worn sensors in crawl swimming

We envision a pervasive, wearable computing system that can support swimmers in achieving their desired exercise goals by constantly monitoring their swim performance and providing the necessary feedback. In this paper we describe the underlying SwimModel to get a full understanding of the recorded acceleration data during swimming. This SwimModel consists of a body motion model based on classical mechanics. The model inputs are the arm and leg forces as well as the body torques. Based on the input forces and torques the model calculates the swimmer's acceleration, velocity and position as well as the angular accelerations, angular velocities and angles. Internally, the model considers the buoyancy force and the drag forces. The transformation of the body motion into the sensor coordinate system and its sensing modalities is done by a sensor model. We validate our SwimModel by comparing the simulated sensor data with real sensor data recorded during swimming.