Research And Implementation Of Abnormal Detection System For Webcam Traffic

In recent years,computers and communications have promoted rapid development.At the same time,it also promotes the popularity of mobile phones,computers,webcams and other network equipments.At present,the webcams are distributed in schools,roads,factories,banks,hospitals and other places,being inseparable from people's lives.The botnet is currently the biggest threat to cybersecurity,and the webcam can become an important part of the botnet.Abnormal camera traffic includes cyberattack traffic from webcams in botnets.At present,there is no research on the classification of abnormal camera traffic.Therefore,the purpose of this thesis is to study and analyze the abnormal traffic of the camera.By summing up camera traffic,this thesis summarizes several features that distinguish camera traffic from non-camera traffic and distinguish malicious camera traffic from normal camera traffic.And then the thesis applies an abnormal detection model to classify abnormal camera traffic.After that,the thesis has improved the multi-pattern matching algorithm to distinguish the attack types of abnormal traffic,improving the efficiency of matching.Finally,On the basis of the previous research,a detection system which has high detection and can distinguish the attack types of abnormal webcam flow can be designed and realized.The main achievements of this thesis are as follows.(1)Analyzing the differences between camera traffic and other network traffic,several new features are summarized to distinguish camera traffic from non-camera traffic and distinguish malicious camera traffic from normal camera traffic.Then this thesis applies a combination of supervised learning and semi-supervised learning model to distinguish the camera traffic.The experiments are classifying on three different types of camera traffic,and the results show that the features summarized in this thesis can improve the accuracy of camera traffic on average by 13%,while malicious camera traffic on average by 5.5%.(2)For the low efficiency of single-pattern matching in classifying malicious traffic,this thesis improves the existing multi-pattern matching algorithm to match the types of attacks.This thesis improves on the time complexity and through experimental verification,the improved algorithm can identify attacks more quickly and has practical value.(3)To achieve the above research,this thesis designs and implements the camera traffic anomaly detection system.The system includes data acquisition module,data packet decoding module,classification model module,detection engine module and other auxiliary modules.And then the system is tested under the simulated environment.By analyzing the results of the test,the system can effectively distinguish the abnormal traffic packets of the camera and divide the abnormal traffic into the corresponding attack types.