Research On Network Flow Watermark Technology Based On Interval Characteristic

As Internet applications penetrate all aspects of society,the situation of network security and privacy protection is becoming increasingly severe.Anonymous network technology can better hide personal privacy information such as identity,but it is also easily exploited by attackers to cover their illegal activities and hinder monitoring and tracking.Active network flow watermarking technology is one of the main methods to solve illegal intrusion tracking.The current watermarking methods are difficult to obtain the best balance in watermark robustness and concealment,and it is difficult to meet the requirements of real-time requirements on the real-time problem of watermarking.This thesis has studied and analyzed the existing research methods.Aiming at the shortcomings of the current time interval characteristic class watermarking method,a dynamic turbo code watermarking method based on interval centroid is proposed.On the one hand,the robustness of watermarking is improved by the error correction ability of the watermarking method,on the other hand,the concealment of watermarking is enhanced by dynamic matching.Based on this,in order to improve the real-time performance of the application,a geometric Huffman-based dual-encoding watermarking method is proposed.The main research work of this thesis is as follows:(1)By comparing various watermarking methods and combining the powerful error correction ability of Turbo codes,an Interval Centroid-Based Dynamic Turbo Watermarking(ICBDTW)method based on slot characteristics is proposed to reduce the original stream.Interference,the watermark sequence is implemented based on the similarity comparison.The embedded end first initializes the system parameters,and generates a set of Turbo codes,captures the data stream for time slot division,compares the original centroid sequence with the Turbo code set,records and stores the most similar Turbo code number and distinct code bits.The subscript of the element modulates the original stream according to the security modulation rules.The receiving end replaces the corresponding position of the detected watermark sequence according to the subscript and number in the shared database,and dynamically corrects the detected sequence by using Turbo,and calculates the similarity between the corrected watermark sequence and the Turbo code sequence corresponding to the number,and passes the threshold matching.Determine whether the stream is associated or not.The effectiveness of this method in improving watermark robustness and concealment is verified by theoretical proof and experiments.(2)Aiming at the shortcoming of the time-slot watermarking method,a dual-encoding watermarking method based on Geometric Huffman Coding(GHC)is proposed.The watermark matching device is generated based on the GHC,and the original watermark sequence is mapped to a random value in the corresponding centroid range by the matcher and the PRNG(pseudo random number generator),thereby obtaining the target centroid size of the modulation,and the mapping is performed.The method compresses multiple watermark bits into a slot centroid range,reduces the consumption of buffer time,and improves real-time performance.Finally,the experimental analysis of the watermark realtime and other aspects is carried out to verify the effectiveness of this method.