An empirical study of morphing on network traffic classification

摘要: Network morphing aims at masking traffic to degrade the performance of identification and classification. Several strategies have been proposed as promising approaches, very few works, however, investigated their impact on actual classification performance. This work sets out fulfill this gap from an empirical study point view. It takes into account different exerted packet size and/or inter-arrival time. The results show that not all can effectively obfuscate Different perform distinctively, among which integration inter arrival time is best, based method worst. three classifiers also exhibit distinct robustness morphing, with C4.5 being most robust Naive Bayes weakest. In addition, our shows learn nontrivial information merely direction patterns, partially explains weakness methods.