Sensitive Instruction Detection Based on the Context of IoT Sensors

With the development of Internet of Things (IoT) technology and the increasing popularity of smart devices, smart homes' IoT devices are becoming more and more intelligent. However, with the access of sensor technology and the gradual increase in the number of devices, some security issues will inevitably occur in smart homes, such as ensuring that the instructions currently executed by the device are legal instructions issued by legitimate users. To solve such challenges, we propose a framework for detecting sensitive instructions. At the same time, we collected a large amount of automation strategy data in smart homes. We used machine learning to extract the sensor cooperative work context characteristics during the execution of sensitive instructions. Finally, We used the test set data to verify and evaluate the performance of our framework. Our evaluation results show that for some sensitive instructions, the goal of being able to intercept high-threat instructions actively has been achieved. The accuracy of the attack detection framework model has also reached more than 89.23%, and some devices even exceed 95%.