Research On Identification Of Active/Passive Device Fingerprint In Smart Fire

The number of Io T devices is huge and has certain network attack ability.Most of the Io T devices also involve the user’s privacy and associate with the sensitive devices of users.With the rapid development of Io T industry and the large-scale use of Io T devices,in the trend of deep integration of Smart City and Io T,the security problems caused by Io T devices have caused widespread Attention.In order to meet the high requirements of equipment consistency for emergency management of Smart Fire System,based on the active and passive device fingerprint construction methods,this thesis divides the IOT devices in Smart Fire System into perceptron and actuator for relevant experimental research,and verifies the recognition performance of the two methods for the two devices.The main work of this thesis is summarized as following three aspects:(1)In this thesis,the active device fingerprint is constructed based on the implicit identifier of the device,and the implicit identifier is selected by using the information theory of filtering so as to improve the relavant performance of the device fingerprint,and the effect evaluation algorithm is given based on the identification performance index of the device fingerprint.The passive device fingerprint is constructed based on the communication traffic characteristics of the device,and the feature selection of the passive fingerprint is optimized to make the authenticiation ability of the passive fingerprint better.(2)This thesis discussed the identification of equipment from the perspective of classification.For active device fingerprint,this thesis adopted the Naive Bayesian classification algorithm of device fingerprint,and designed the classifier threshold determination algorithm based on the weighted average of recognition accuracy and recall rate to optimize the initial algorithm,and compared the algorithm.The experimental results show that the improved Naive Bayesian algorithm reaches the best recognition accuracy,94.4%,and the recognition performance is improved by8.1%.The fingerprint technology of active device has good performance in recognition of perceptron,and the recognition accuracy is 95.3%,but the recognition performance of actuator is poor,and the recognition accuracy is only 78.6%.(3)Aiming at the poor performance of active device fingerprint technology in actuator recognition,this thesis constructed the passive fingerprint,and used Random Forest algorithm to complete the relevant experiments.The experimental results showed that the accuracy and precision of passive fingerprint are both excellent,but it is easy to confuse the devices with similar models.Specifically,the accuracy of traffic fingerprint to perceptron is about 90%,the recognition accuracy of actuator is 92.8%,and the recognition performance of actuator is improved by 14.2%.Finally,this thesis compared and verified that the experimental method of improved traffic fingerprint features have good recognition stability.