Abstract
Event detection has been studied and researched for many years and it has been applied in real world applications with the aim of characterising a situation in the real world. In order to capture a situation, Wireless Sensor Networks (WSNs) are deployed and sensor nodes are used to sense the entities of interest for the real world application; sensing the environment results in the production of a large and often continuous production of raw data. In this context, event detection is used in order to extract the most relevant and useful information from this large set of data. The constraints of nodes have to be taken into account such as energy, computation, and memory. The environment is observed from a program that is hosted on a sensor node. Machine learning and data mining techniques are embedded in the program to learn from the environment and detect events. A collaborative sensing is a technology to process an event from distrusted nodes which can enhance an accuracy result that can be fault or event. This research studied processing sensor data to detect events using multiple sensor nodes. A model and/or rules are defined in order to detect an outlier from data matching between sensor data and the model and/or rules. An outlier is analysed and processed to detect an event. The main contributions of this work have been on collaborative sensing in different sensors including clustering analysis for data labelling, classification analysis in order to process an outlier for an event detection.