Research And Implementation Of Route Leak Detection Technology Based On Adaboost Algorithm

With the widespread application of BGP network protocols,route leak as a common type of BGP anomalies,poses huge threat to network stability.How to effectively detect route leak events is of great significance for maintaining the stability of the network.Machine learning can be well used to detect route leak anomalies.However,the existing machine learning detection algorithm lacks targeted abnormal feature research when performing BGP anomaly detection,which results in the detection result is not clearly typed for the anomaly event type.So,it is inconvenient for solving the abnormal problem.Based on above problems,this paper proposes a detection algorithm for route leak.The algorithm proposes three features that are different from other anomalies.At the same time,AdaBoost algorithm is introduced to improve the existing detection algorithm.We implement the algorithm based on real data,which has high detection accuracy.The main work of the thesis includes three aspects.First of all,we find three characteristics of route leak events.At the same time,we have added some common features of BGP anomaly events,and combined the two to get the final feature set.Secondly,on the basis of real data,this paper calculates the relevant features and proves their validity.Finally,we design and implement a route leak detection algorithm based on AdaBoost algorithm.The result shows that the route leak detection algorithm proposed in this paper can improve the accuracy of the algorithm and achieve targeted detection for route leak.