Adaptive Meek Technology for Anti-Traffic Analysis
Autor: | Haoyao Xie, Hao Shentu, Hui Zhao, Shangnan Yin, Liangmin Wang |
---|---|
Rok vydání: | 2020 |
Předmět: |
Traffic analysis
Computer science business.industry Reliability (computer networking) ComputerSystemsOrganization_COMPUTER-COMMUNICATIONNETWORKS Cloud computing 02 engineering and technology Communications system Communications security Computer security computer.software_genre Obfuscation (software) 020401 chemical engineering Obfuscation 0202 electrical engineering electronic engineering information engineering Overhead (computing) 020201 artificial intelligence & image processing 0204 chemical engineering business computer |
Zdroj: | NaNA |
Popis: | With the rapid increase in the demand of users for communication security, Tor is widely used as the most reliable anonymous communication system. Meek is used as a pluggable transport for Tor, which protects Tor users from traffic analysis by disguising the traffic to the Tor network as the traffic to the cloud server. However, in the face of machine-learning attacks that use side-channel information of traffic, Meek still exposes its lack of obfuscation ability. In this research, we propose an adaptive Meek technology (AM) that resists statistical analysis of traffic to disguise Tor-based Meek traffic as the target traffic to meet the minimum cost of improving indistinguishability. First, this technology selects the target traffic, which can be constantly updated to minimize the overhead of obfuscation. Then, it provides a reliable obfuscation solution to maximize the mimicry effect by focusing on side-channel information. Throughout the obfuscation process, our technology can make obfuscation of Meek traffic achieve dynamics and diversity. We use the traffic dataset of the real scenario, then conduct an experimental evaluation through our well-trained classifiers. After our defense, the effect of the classifier was successfully reduced while the overhead is within an acceptable range. |
Databáze: | OpenAIRE |
Externí odkaz: |