| Summary: | In smart homes devices take partial control of the house and make decisions that increase its safety and functionality. Due to the high cost of cable installation, wireless sensor networks are considered to be a good choice for smart home systems. In smart homes, to reliably detect events such as intrusions, gas leakages, or accidents, is an essential functionality. A correct control action, such as alarming or shutting gas pipes relies entirely on reliable event detection. Given that reliability is the major concern in these devices, detection solutions should be found that make decisions with the smallest possible cost. A way to achieve this is by using detection and data fusion techniques so measurements from multiple sensors make the final decision whether the event has happened or not. Furthermore, estimation aided detection in every sensor is essential so as to provide a noiseless environment for the decision concerning the happening of an event in each node of the network. In this thesis, several distributed detection techniques are reviewed and their suitability for low power sensor networks is investigated. A wireless sensor network is designed and completely implemented in order to test the functionality and the reliability of the methods. In the presence of a detected event the sensor network sends a Twitter notification to the user and, meanwhile, actuates a control decision that could solve the detected problem. The experiments show that the studied detected methods are able to offer reliable performance even in the presence of high noises in the measurements. It is concluded that wireless sensor networks can be effectively used in smart home applications, provided that detection methods of low complexity and reliability are implemented.
|
|---|
