基于时空数据联邦的安全k近邻查询分析
Secure k-Nearest Neighbor query analysis based on Spatio-temporal data federation
王腾 1左军瑞1
作者信息
- 1. 中国电子科技集团公司网络通信研究院,河北石家庄 050050
- 折叠
摘要
[目的/意义]时空数据联邦指的是基于联邦学习框架,在不共享具体数据的情况下,协同多个时空数据持有者进行查询和计算.这种方式可以实现较为精确的查询结果,而且能保证数据的安全.[方法/过程]提出了一种轻量级的联邦kNN查询方法FedkNN,允许不同数据持有者协作完成kNN查询,而无需共享位置数据,从而避免将数据持有者的信息暴露给不受信任的聚合服务器,并且易于部署和实施.[结果/结论]通过在真实世界数据集和合成数据集上实验,FedkNN算法在运行时间和通信成本方面优于现有算法,同时实现了与现有算法相当的准确度.
Abstract
[Purpose/Significance]Spatio-temporal data federation refers to the collaboration of multiple spatio-temporal data owners in queries and calculations through a federated learning framework,without sharing specific data,to achieve both accurate query results and data privacy and security protection.[Method/Process]This paper proposes a lightweight privacy-preserving federated kNN query method.This method allows federated kNN queries to be completed among different data holders without the need to share location data,thereby avoiding the exposure of data holder's information to untrusted aggregation servers,and it is easy to deploy and implement.[Results/Conclusion]We conducted experiments on both real-world datasets and synthetic datasets.The experimental results demonstrate that our proposed algorithm outperforms existing algorithms in terms of runtime and communication cost,while achieving comparable accuracy to existing algorithms.
关键词
时空数据联邦/kNN查询/保序加密/空间查询/网络空间安全Key words
Spatio-temprol data federation/k-Nearest Neighbor query/Oder-preserving Encryption/spatial query/cybersecurity引用本文复制引用
基金项目
河北省重点研发计划支持项目(23376201D)
出版年
2024
NETL
NSTL
万方数据