ANOMALY DETECTION IN WIRELESS SENSOR NETWORKS OF «SMART HOME» SYSTEM

Subject of Research.The paper reviews the problem of anomaly detection in home automation systems. The authors define present security networks specificities and highlight the need of informational and physical impact detection on sensors aimed at information security. Method. Artificial neural network is proposed for anomaly detection. This method processes the data on characteristics of security network devices for anomalous behavior detection. The artificial neural network should be preliminarily trained on the data of that type. The implementation tools for the proposed method of anomaly detection are described. Main Results. The scenario has been created for the experiment so that the model of «Smart home» system produces the data of network information streams and the artificial neural network makes decisions based on this data. As a result, the training and testing sets have been created. The anomaly has been considered to be a state with the artificial neural network result less than 0.9. Based on the test results the artificial neural network determines the network node state with 91% precision. Practical Relevance. The proposed method can be used in information and security systems when connected devices should be monitored. Anomaly detection technology excludes inconspicuous violation of information confidentiality and integrity.