Research And Implementation Of Tor Path Selection Algorithm Based On Node Information

Tor is currently the most well-known anonymous communication network,but as it continues to expand,it has also attracted a large number of attackers.To improve the security and performance of Tor network,researchers are exploring location-aware routing algorithms.However,recent studies have shown that existing work in this area poses a high risk of leaking user locations,which severely compromises user anonymity.Therefore,this paper conducts research into this issue with the aim of improving user anonymity on Tor.First,a path dataset is constructed and a path location leakage evaluation model PLE is proposed.In order to better evaluate the leakage of path location,this paper defines threat assumption,sender anonymity and guard anonymity as the basis for model design.The path data is collected by TorPS simulation.It contains the node information of the Tor historical consensus data,and expands the location information and network topology information.On this basis,the model uses LightGBM and GCN to learn data features,and evaluates the degree of location leakage through path data.Experiments show that PLE can better evaluate the position leakage degree of path selection algorithm.Second,this paper proposes a path selection algorithm PLLA,which incorporates dynamic awareness of location leakage.The PLLA algorithm establishes a danger threshold and screens out paths with better anonymity based on the location leakage calculated by PLE.This technique can be applied to existing location-aware path selection algorithms.Experimental results demonstrate that compared to Vanilla Tor and Counter-RAPTOR,PLLA can improve anonymity by approximately 21%and 43%.Moreover,the anonymity of PLLA gradually increases with the number of destinations visited by users.In addition,the PLLA algorithm suffers less loss in other safety and performance evaluations.making it more suitable for real-world scenarios.Compared to other methods,PLLA is expected to have significant practical applications.Finally,this paper introduces the design and implementation of the Tor path selection evaluation system ANPE.In order to facilitate a convenient and comprehensive analysis of path selection algorithm designs,ANPE includes four functional modules:path evaluation,evaluation result analysis,user management and data management.These modules enable the application of the PLE evaluation model,which focuses on evaluating path selection algorithms from perspectives such as anonymity,hijacking resilience,and location leakage.