Research On Dynamic Data Publishing Of V2G Network Based On Differential Privacy

As an important component of smart grid,V2G network shows great potential in providing valuable ancillary services.More and more electric vehicles are added to V2G networks,which generate a lot of personal data after charging or discharging.These data can provide beneficial information for enterprises or society,but when these data are obtained by malicious users,they can obtain a large amount of private information from them,causing user privacy leakage.In order to protect V2G network data security,the paper focuses on the method of V2G network data publishing under differential privacy.Firstly,an adaptive differential privacy data publishing method is proposed,which can learn the temporal and spatial correlations of data streams for multi-dimensional data.Then,a distributed w-event-level differential privacy data publishing method is proposed,which realizes the dynamic publishing of data under the untrusted central server.The specific work is as follows:(1)The scenario of dynamic release of V2G network data is analyzed,and the possible privacy leakage risk is studied for the user’s electric vehicle charging and discharging data.It analyzes and summarizes the current mainstream data release technologies,and provides technical support for dynamic data release under V2G network.(2)Aiming at the large amount of charging and discharging data in V2G network,an adaptive data publishing method based on differential privacy is proposed.The dynamic change trend is used to add noise,and then the privacy budget and hash algorithm grouping are allocated through a sliding window to ensure that the published data satisfies w-event differential privacy and saves the privacy budget.(3)For the case of untrusted servers,a distributed w-event-level differential privacy data publishing method is further proposed.By performing data prediction on multiple agents,the central server can access the data prediction value instead of the original value.In order to make the data prediction suitable for the agent,An agentbased clustering algorithm is proposed,then the data is sampled and the privacy budget is allocated on the central server,and finally noise is added locally according to the grouping results and weights.This method not only guarantees data privacy but also saves the privacy budget.(4)Based on the above algorithm,a data release privacy protection system based on differential privacy is constructed,and the functional test of the system is completed.Users can register and log in to the system,upload data in the system and call the above two algorithms for data release,providing users with a data sharing and publishing platform based on differential privacy.