Finding abnormal events in home sensor network environment using correlation graph

Anomaly detection in sensor network seems a challenge when encountering the limitation of the energy requirement and dynamics environments. It is to rapidly analyze and identify the abnormal events among the extreme volume data. Using correlation graph representation to correlate the events generated by sensor networks is capable to find the intentional dependency behavior´s insight for detecting home sensor network abnormal events. In this study, we proposed an anomaly detection mechanism based on correlation graphs of sensor networks for rapidly identifying abnormal home events. The proposed mechanism which makes the following contributions: (a) it is automatically identify the abnormal event under home sensor network environment (b) it eliminates irrelevant events for saving the computation power (c) it is easily to apply on different machine learning classifiers for enhancement. The evaluation from Intel Berkeley Research lab sensor network data set. The proposed mechanism performs well in sensor events elimination and abnormal event detection.