Research On Tor Anonymous Network Resource Distribution Mechanism

Onion Routing Tor is an anonymous communication system.Users use Tor nodes to construct multi-hop communication circuit to achieve anonymous network access.All relays in the Tor network are publicly distributed and are vulnerable to enumeration attack,which makes the Tor network unavailable.Therefore,Tor Network introduces bridges.Tor does not distribute bridges publicly and adopts a restrictive resource distribution strategy.However,Tor does not detect the user's identity when distributing bridges.An attacker can pretend to be an honest user to enumerate resources and makes the resources unavailable,which reduces the availability of the system.Therefore,research on the resources distribution mechanism based on user detection is of great significance to the availability of the system.The cost of running bridges is high.The number of bridges in the system is small and the set of users is large.The same resource will be distributed to multiple users.Therefore,detecting attackers in the process of resource distribution faces the following challenges: high false positive rate,low honest user available rate,and high cost.Aiming at these problems,this thesis proposes a resource distribution model based on user behavior.The main work and contributions of this thesis are as follows:(1)In order to survey the distribution mechanism bridges in the Tor network when all resources cannot be obtained.We describe the resource distribution as a ticket collection process,and research the system's resistance to enumeration attacks,the number of total resources,and resource update strategy.We analyze the system's resistance to enumeration attacks from two aspects.The certain users can obtain the number of resources and the attacker wants to obtain all the resources need the number of users.We evaluate the number of resources and the update strategy through the ticket collection problem and acquired resources.And we use the Markov model to verify the evaluation results.(2)In order to resolve high false positive rate and low honest user available rate because the same resource is distributed to multiple users,this thesis proposes a detection method based on integer linear programming.We establish a bipartite graph between suspicious attackers and unavailable resources,and use the integer linear programming method to find the smallest subset of the set of suspicious attackers.Experimental results show that when the unavailable resources in the system do not exceed 50%,the precision of the detection method is more than 80%,which is about 10% higher than the detection method based on multi-stage.(3)In order to make attackers get as few resources as possible when the system distribute resources to honest users to reduce the cost,we propose a resource distribution model based on user behavior.This distribution model can prevent attackers from disguising themselves as normal users and delaying attacks,and can reduce the waste of resources in the process of attacker differentiation.This thesis sets up simulation experiments to evaluate the resource distribution model according to resource available rate,the resistance to enumeration attacks and normal user availability available rate.According to the experimental results,when attackers do not disguise,the scheme can make resource available rate and normal user available rate of the system achieve 100% in the second stage.Even when attackers disguise,the scheme can consume less resources to make available rate and normal user available rate achieve 100%.Comparing the existing resource distribution scheme of Tor,the scheme proposed in this thesis reduces the impact of attackers on system availability.Attackers obtain 30% less resources than in the scheme proposed in this thesis than the scheme based on multi-stage detection of attackers,that is,30% less resources are wasted.