Reverse Analysis Of Communication Protocol For Mini-UAV System

UAV has the advantages of low cost,high mobility,conveniently,pilotless,and high cost-effectiveness,making it rapidly developed and applied to military and civilian fields.But if "unconstrained" rapid development,it would bring serious challenges to national and social security.Therefore,while speeding up the regulation of laws and regulations,the majority of researchers have gradually turned their attentions to solving this potential safety problem technically.Inferring the format and meaning of the UAV system communication protocol by protocol reverse analysis technology can effectively strengthen the supervision of UAV.As far as the current research is concerned,there are lots of difficulties in reverse analysis of communication protocol for UAV system.First,the communication protocol data obtained by the software or technology is disorderly and large.Secondly,non-standard private protocol is used in the UAV system communication protocol,so that there is no effective reference standard in the process of protocol reverse analysis.This thesis aimed at the off-line network trace sequence,focuses on the keyword extraction algorithm,sequence clustering algorithm and protocol format parsing algorithm,and according to the key and difficult problems in the algorithm,put forward the corresponding solutions.1)A keyword extraction method based on LDA model is proposed.First,the N-gram is used to preprocess protocol data set to form a sub-sequence set.Then,the LDA model is used to mine sub-sequences with the highest frequency of each keyword,and finally the sub-sequences obtained after de-duplication of the top L sub-sequences of all keywords were represented as the protocol data set keywords.2)In view of sequence clustering,this thesis proposes a protocol sequence clustering method based on K-means++ algorithm.In order to find the optimal number of clusters,a method combining the Density Peak Clustering(DPC)with the Silhouette coefficient was proposed.Firstly,the decision graph of DPC was used to determine the range of cluster number,and then use the Silhouette coefficient to finally determine the optimal number of clusters.3)In order to overcome the shortcomings of progressive multi-sequence alignment algorithm,this thesis proposes a multi-sequence alignment algorithm based on Di Align algorithm.This algorithm is based on segment-segment similarity and performs multi-sequence alignment based on relatively conservative segments,and automatically generate the number and location of gaps,effectively avoiding the effect of gap penalties on alignment accuracy.The test results of both Dobby drone communication protocol and public protocol including HTTP and SMTP were presented.Experiment results show that higher accuracy in protocol sequence clustering and format inference could be obtained.