Development of a wireless sensor device and data evaluation with a learning algorithm (Masterthesis)
18.03.2015, 16:00, Room 3945
The development of wireless body sensor network systems has received increasing attention during the last years. The main application and motivation therefore is the healthcare sector with its rising costs and increasingly ageing patients. But there is also an increasing demand for wearable bio sensors for entertainment and sports. Advances in electronics, computer science and wireless technologies made it able to process larger data in miniaturized hardware, which led to an increased mobility and connectivity of electronic devices and opened new possibilities. The first goal of this thesis is to develop and build a small, low-power and wireless sensor node to detect and record the human movement. The appropriate sensors and wireless technologies for this purpose are presented and integrated in a sensor node design concept. The developed hardware and chosen components, e.g. parts for radio communication, microcontroller, etc., resulting from this concept are shown and explained, as well as the functionality of the developed software. The second part of this thesis is the evaluation of the sensor data. Therefore some different features of the recorded human movement data were extracted to train a support vector machine learning model, which should be able to classify further input movement data. The results are evaluated for the amount of correct outputs and the needed computation time. In conclusion it can be stated, that the built sensor node works as intended and supplies
sufficient data to evaluate human movement. It is possible to predict the tested actions: walking, running, jumping and doing nothing, with a learning machine.