Reseaech And Implementation Of Tor Network Path Selection Strategy Against Sybil Attacks

Tor is the most popular and widely used anonymous network.Tor allows volunteers to build relay nodes anonymously and quickly.And the threshold to build a relay node is quite low.This open attitude,while making Tor develop rapidly,also makes it vulnerable to Sybil attacks.Optimizing the Tor network path selection strategy for Sybil attack nodes is one of the effective strategies to resist Sybil attacks.However,the existing Tor network path selection strategies are designed to defend against specific forms of attacks,such as end-to-end attacks,website fingerprint analysis attacks,etc.,but they cannot comprehensively resist multiple forms of Sybil attacks.In view of the above problems,this thesis proposes a machine learning method to determine Sybil nodes,and based on this,designs a Tor network path selection strategy that can resist multiple Sybil attacks.Finally,this thesis implements a Tor network path selection strategy switching system that can resist multiple Sybil attacks.The specific work is as follows:1.This thesis proposes a method forextracting Sybil Family node pairs and high-risk Sybil nodes based on machine learning.For ease of explanation,this thesis refers to the set of multiple nodes configured by the same volunteer as Family,and refer to the Family that has been labeled according to regulations as Explicit Family,otherwise,it is called Sybil Family.This thesis analyzes the similarity of node features in Explicit Family in detail,and uses machine learning to determine whether "two nodes belong to the same Family",and then extracts all Sybil Family node pair sets.Using the extraction results of Sybil Family node pair sets,this thesis analyzes and categorizes the potential attack intentions of the nodes in the set,and extracts several high-risk Sybil nodes,such as "high-risk Sybil Exit node","high-risk large Sybil nodes",and "high-risk Sybil Guard nodes".2.Based on the extracted set of Sybil Family node pairs and high-risk Sybil nodes,this thesis designs a path selection strategy to resist multiple Sybil attacks.This strategy can resist various forms of Sybil attacks in the Tor network,such as end-to-end attacks,exit traffic tampering attacks,website fingerprint analysis attacks,and bridge relay node blocking attacks.Moreover,this path selection strategy will not affect the Tor network bandwidth seriously.3.Based on the above strategy,this thesis designs and implements a Tor network path selection strategies switching system that can resist Sybil attacks.This system can generate and update path selection strategies against Sybil attacks in real time based on the current Tor network relay nodes information.In addition,the system also supports users to add custom path selection strategies.It also supports multi-strategy overlay usage and free switching of different strategies.