A study of neural network based IoT device information security system

Abstract

The development of ubiquitous computing environment has increased interest in IoT (Internet of Things) technology. As IoT has flexible and open characteristics applicable in various fields of industry, increased external accessibility has raised the possibility of external threats. As the existing IoT network was small on scale, there was less risk of security attack. However, continuous IoT development brought the large scale network environment by combining various networks, therefore causing higher risk of security attack compared to small scale network environment. By figuring out the response and implementation time while operating IoT devices to detect intrusion through virus or hacking, the artificial neural network learns various responses using a wide variety of mobile devices. This process may help to deal with current virus and hacking. In addition, by detecting virus and malware in real time, this process may also help to prevent future intrusion. As IoT security risk sharply arises, we suggested an intrusion detection system using artificial neural network model in this study. The system method which is developed in this study can be adjusted to fit various fields and situations of IoT by facilitating flexible modification of critical values. Considering limitations of IoT, the research method which detects anomaly through learning the response and implementation time is expected to be widely used for information security system of various fields which utilize IoT in the future from various angles.