Research On Differential Privacy Protection Of Anonymous Network Data

The advancement of informatization construction and the popularization of Internet applications have brought convenience to people's daily communication,and also brought new challenges to the security of personal information.Anonymous communication systems can protect user's identity by layered encryption,message mixing,and traffic masquerading.At present,more and more users use anonymous communication systems.Tor is the most popular anonymous communication system,which has more than 3 million users.However,due to the anonymity and privacy of the anonymous communication system,it is difficult for researcher to collect anonymous network data.What's more,we cannot directly apply the typical network monitoring methods to anonymous communication systems.Therefore,we will carry out the collection and analysis of anonymous communication network data under the protection of differential privacy in this paper.Firstly,this paper introduces the related concepts of anonymous communication,the development and principles of the second generation onion routing Tor.Then it introduces some theoretical foundations of differential privacy.The research on differential privacy histogram publishing algorithm is investigated,summarized and summarized.Secondly,this paper proposes a Tor anonymous network data collection scheme Tor-NSDP under differential privacy protection,and elaborates on the module design and core process of the Tor-NSDP in detail.Tor-NSDP adopts secure multi-party calculation method to perform decentralized data statistics,which can avoid the risk of privacy leakage generated by the data statistics center.In the data differential privacy processing module,the Tor-NSDP sets up three data processing nodes and uses a secret sharing mechanism.It can effectively avoid a single malicious node from affecting the result of data release by adding noise.This paper analyzes the security of the Tor-NSDP scheme,and the results of the Tor-NSDP program data release are shown.Finally,when the query range is increased,the availability of data statistics result will decrease due to too much noise,so this paper proposes a differential privacy histogram publishing optimization algorithm KMHP.By continuously adjusting the cluster center,the buckets with similar frequencies in the histogram are combined to obtain the optimal histogram structure and reduce the injection of noise.In this paper,the privacy of the KMHP algorithm is proved theoretically,and by the comparison experiment with the traditional differential privacy histogram publishing algorithm,it proves that the KMHP algorithm can effectively improve the availability of published data.