基于深度学习的位置大数据统计发布与隐私保护方法

doi:10.11959/j.issn.1000-436x.2022006

通信学报 ›› 2022, Vol. 43 ›› Issue (1): 203-216.doi: 10.11959/j.issn.1000-436x.2022006

基于深度学习的位置大数据统计发布与隐私保护方法

晏燕¹^,², 丛一鸣¹, Adnan Mahmood², 盛权政²

¹ 兰州理工大学计算机与通信学院，甘肃兰州 730050
² 麦考瑞大学科学与工程学院，新南威尔士 2109

修回日期:2021-12-22 出版日期:2022-01-25 发布日期:2022-01-01
作者简介:晏燕（1980- ），女，甘肃兰州人，博士，兰州理工大学副教授、硕士生导师，麦考瑞大学访问学者，主要研究方向为数据发布隐私保护、位置隐私、多媒体信息安全等
丛一鸣（1993- ），男，黑龙江绥化人，兰州理工大学硕士生，主要研究方向为隐私保护技术、深度学习等
Adnan Mahmood（1985- ），男，博士，麦考瑞大学在站博士后，主要研究方向为信任管理、车联网安全、位置隐私等
盛权政（1971- ），男，博士，麦考瑞大学教授、博士生导师，主要研究方向为大数据分析、普适计算等
基金资助:
国家自然科学基金资助项目(61762059);国家自然科学基金资助项目(61862040)

Statistics release and privacy protection method of location big data based on deep learning

Yan YAN¹^,², Yiming CONG¹, Mahmood Adnan², Quanzheng SHENG²

¹ School of Computer and Communication, Lanzhou University of Technology, Lanzhou 730050, China
² Faculty of Science and Engineering, Macquarie University, NSW 2109, Australia

Revised:2021-12-22 Online:2022-01-25 Published:2022-01-01
Supported by:
The National Natural Science Foundation of China(61762059);The National Natural Science Foundation of China(61862040)

摘要/Abstract

摘要：

针对传统位置大数据统计划分发布结构不合理、划分发布方法效率低下的问题，提出一种基于深度学习的位置大数据统计划分结构预测方法和差分隐私发布方法，以提高位置大数据统计划分发布数据的可用性和执行效率。首先对二维空间进行细致划分和自底向上合并，从而构建合理的空间划分结构。然后将划分结构矩阵组织为三维时空序列，借助深度学习模型提取时空特征，实现对划分发布结构的预测。最后结合预测划分发布结构进行差分隐私预算分配和 Laplace 噪声添加，实现位置大数据统计划分发布信息的隐私保护。通过实际位置大数据集的实验，证明了所提方法在提高发布数据查询精度和运行效率方面的优势。

关键词: 数据发布隐私保护, 位置隐私, 隐私空间分解, 差分隐私, 深度学习

Abstract:

Aiming at the problems of the unreasonable structure and the low efficiency of the traditional statistical partition and publishing of location big data, a deep learning-based statistical partition structure prediction method and a differential publishing method were proposed to enhance the efficacy of the partition algorithm and improve the availability of the published location big data.Firstly, the two-dimensional space was intelligently partitioned and merged from the bottom to the top to construct a reasonable partition structure.Subsequently, the partition structure matrices were organized as a three-dimensional spatio-temporal sequence, and the spatio-temporal characteristics were extracted via the deep learning model in a bid to realize the prediction of the partition structure.Finally, the differential privacy budget allocation and Laplace noise addition were implemented on the prediction partition structure to realize the privacy protection of the statistical partition and publishing of location big data.Experimental comparison of the real location big data sets proves the advantages of the proposed method in improving the querying accuracy of the published location big data and the execution efficiency of the publishing algorithm.

Key words: privacy protection data publishing, location privacy, private spatial decomposition, differential privacy, deep learning

中图分类号:

TP311

晏燕, 丛一鸣, Adnan Mahmood, 盛权政. 基于深度学习的位置大数据统计发布与隐私保护方法[J]. 通信学报, 2022, 43(1): 203-216.

Yan YAN, Yiming CONG, Mahmood Adnan, Quanzheng SHENG. Statistics release and privacy protection method of location big data based on deep learning[J]. Journal on Communications, 2022, 43(1): 203-216.

图/表 15

图1

图2

表1

表2

表3

表4

图3

图4

图5

表5

表6

图6

图7

图8

图9

参考文献 39

[1]	ZHU L , YU F R , WANG Y G ,et al. Big data analytics in intelligent transportation systems:a survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2019,20(1): 383-398.
[2]	GE M Z , BANGUI H , BUHNOVA B . Big data for Internet of things:a survey[J]. Future Generation Computer Systems, 2018,87: 601-614.
[3]	TIAN Z H , WANG Y H , SUN Y B ,et al. Location privacy challenges in mobile edge computing:classification and exploration[J]. IEEE Network, 2020,34(2): 52-56.
[4]	SOWMIYA B , ABHIJITH V S , SUDERSAN S ,et al. A survey on security and privacy issues in contact tracing application of covid-19[J]. SN Computer Science, 2021,2(3): 136.
[5]	JIANG H B , LI J , ZHAO P ,et al. Location privacy-preserving mechanisms in location-based services[J]. ACM Computing Surveys, 2022,54(1): 1-36.
[6]	ALI A , ZHU Y M , ZAKARYA M . A data aggregation based approach to exploit dynamic spatio-temporal correlations for citywide crowd flows prediction in fog computing[J]. Multimedia Tools and Applications, 2021,80(20): 31401-31433.
[7]	YANG L , WANG L Z . Mining traffic congestion propagation patterns based on spatio-temporal co-location patterns[J]. Evolutionary Intelligence, 2020,13(2): 221-233.
[8]	DWORK C , . Differential privacy[C]// Proceedings of the 33rd International Colloquium on Automata,Languages and Programming. Berlin:Springer, 2006: 1-12.
[9]	DWORK C , . Differential privacy:a survey of results[C]// International Conference on Theory and Applications of Models of Computation. 2008: 1-19.
[10]	QARDAJI W , YANG W N , LI N H . Differentially private grids for geospatial data[C]// Proceedings of 2013 IEEE 29th International Conference on Data Engineering. Piscataway:IEEE Press, 2013: 757-768.
[11]	XIONG P , ZHANG L F , ZHU T Q . Reward-based spatial crowdsourcing with differential privacy preservation[J]. Enterprise Information Systems, 2017,11(10): 1500-1517.
[12]	WANG J , ZHU R B , LIU S B ,et al. Node location privacy protection based on differentially private grids in industrial wireless sensor networks[J]. Sensors, 2018,18(2): 410.
[13]	张啸剑, 金凯忠, 孟小峰 . 基于自适应网格的隐私空间分割方法[J]. 计算机研究与发展, 2018,55(6): 1143-1156.
	ZHANG X J , JIN K Z , MENG X F . Private spatial decomposition with adaptive grid[J]. Journal of Computer Research and Development, 2018,55(6): 1143-1156.
[14]	FANAEEPOUR M , RUBINSTEIN B I P . Differentially private counting of users’ spatial regions[J]. Knowledge and Information Systems, 2018,54(1): 5-32.
[15]	WEI J H , LIN Y P , YAO X ,et al. Differential privacy-based location protection in spatial crowdsourcing[J]. IEEE Transactions on Services Computing,2019,doi:10.1109/TSC.2019.2920643.
[16]	RODRíGUEZ K M , BOSSY M , MAFTEI R ,et al. New spatial decomposition method for accurate,mesh-independent agglomeration predictions in particle-laden flows[J]. Applied Mathematical Modelling, 2021,90: 582-614.
[17]	CORMODE G , PROCOPIUC C , SRIVASTAVA D ,et al. Differentially private spatial decompositions[C]// Proceedings of 2012 IEEE 28th International Conference on Data Engineering. Piscataway:IEEE Press, 2012: 20-31.
[18]	吴英杰, 卢清, 蔡剑平 ,等. 基于四分树的差分隐私二维数据划分发布方法[J]. 华中科技大学学报(自然科学版), 2016,44(3): 99-104.
	WU Y J , LU Q , CAI J P ,et al. Differential privacy two-dimensional data partitioning publication algorithm based on quad-tree[J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2016,44(3): 99-104.
[19]	ZHANG J , XIAO X K , XIE X . PrivTree:a differentially private algorithm for hierarchical decompositions[C]// Proceedings of the 2016 International Conference on Management of Data.[S.l.:s.n.], 2016: 155-170.
[20]	ZHANG J , CORMODE G , PROCOPIUC C M ,et al. PrivBayes[J]. ACM Transactions on Database Systems, 2017,42(4): 1-41.
[21]	YANG M M , ZHU T Q , XIANG Y ,et al. Density-based location preservation for mobile crowdsensing with differential privacy[J]. IEEE Access, 2018,6: 14779-14789.
[22]	黄泗勇, 陈婷婷, 卢清 ,等. 基于kd-树的差分隐私二维空间数据划分发布方法[J]. 山东大学学报(工学版), 2015,45(1): 24-29,36.
	HUANG S Y , CHEN T T , LU Q ,et al. Differentially privacy two-dimensional dataset partitioning publication algorithm based on kd-tree[J]. Journal of Shandong University (Engineering Science), 2015,45(1): 24-29,36.
[23]	YAN Y , HAO X H , ZHANG L X . Hierarchical differential privacy hybrid decomposition algorithm for location big data[J]. Cluster Computing, 2019,22(4): 9269-9280.
[24]	CAI S , XIN L , DUOHAN B . Spatial statistic data release based on differential privacy[J]. Transactions on Internet and Information Systems, 2019,13(10): 5244-5259.
[25]	张啸剑, 付楠, 孟小峰 . 基于本地差分隐私的空间范围查询方法[J]. 计算机研究与发展, 2020,57(4): 847-858.
	ZHANG X J , FU N , MENG X F . Towards spatial range queries under local differential privacy[J]. Journal of Computer Research and Development, 2020,57(4): 847-858.
[26]	DWORK C . Calibrating noise to sensitivity in private data analysis[J]. Lecture Notes in Computer Science, 2012,3876(8): 265-284.
[27]	MCSHERRY F . Privacy integrated queries[J]. Communications of the ACM, 2010,53(9): 89-97.
[28]	YU Y , SI X S , HU C H ,et al. A review of recurrent neural networks:LSTM cells and network architectures[J]. Neural Computation, 2019,31(7): 1235-1270.
[29]	DO?AN E . LSTM training set analysis and clustering model development for short-term traffic flow prediction[J]. Neural Computing and Applications, 2021,33(17): 11175-11188.
[30]	GUO S N , LIN Y F , LI S J ,et al. Deep spatial–temporal 3D convolutional neural networks for traffic data forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2019,20(10): 3913-3926.
[31]	ZHAO J C , DENG F , CAI Y Y ,et al. Long short-term memory - fully connected (LSTM-FC) neural network for PM 2.5 concentration prediction[J]. Chemosphere, 2019,220: 486-492.
[32]	SHI X J , CHEN Z R , WANG H ,et al. Convolutional LSTM network:a machine learning approach for precipitation nowcasting[C]// Proceedings of Conference and Workshop on Neural Information Processing Systems (NIPS).[S.l.:s.n.], 2015: 802-810.
[33]	WANG Y B , LONG M S , WANG J M ,et al. PredRNN:recurrent neural networks for predictive learning using spatiotemporal LSTMs[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems.[S.l.:s.n.], 2017: 879-888.
[34]	陈思, 付安民, 苏铓 ,等. 基于差分隐私的轨迹隐私保护方案[J]. 通信学报, 2021,42(9): 54-64.
	CHEN S , FU A M , SU M ,et al. Trajectory privacy protection scheme based on differential privacy[J]. Journal on Communications, 2021,42(9): 54-64.
[35]	李洪涛, 任晓宇, 王洁 ,等. 基于差分隐私的连续位置隐私保护机制[J]. 通信学报, 2021,42(8): 164-175.
	LI H T , REN X Y , WANG J ,et al. Continuous location privacy protection mechanism based on differential privacy[J]. Journal on Communications, 2021,42(8): 164-175.
[36]	付钰, 俞艺涵, 吴晓平 . 大数据环境下差分隐私保护技术及应用[J]. 通信学报, 2019,40(10): 157-168.
	FU Y , YU Y H , WU X P . Differential privacy protection technology and its application in big data environment[J]. Journal on Communica tions, 2019,40(10): 157-168.
[37]	WANG J , LIU S B , LI Y K ,et al. Differentially private spatial decompositions for geospatial point data[J]. China Communications, 2016,13(4): 97-107.
[38]	BKAKRIA A , TASIDOU A , CUPPENS-BOULAHIA N ,et al. Optimal distribution of privacy budget in differential privacy[C]// International Conference on Risks and Security of Internet and Systems. Berlin:Springer, 2018: 222-236.
[39]	YAN Y , GAO X , MAHMOOD A ,et al. Differential private spatial decomposition and location publishing based on unbalanced quadtree partition algorithm[J]. IEEE Access, 2020,8: 104775-104787.

层号	各层模型	输出结构	参数量
1	LSTM	(None, 1,32)	528 512
2	LSTM	(None,50)	16 600
3	Dense	(None,4096)	208 896
合计			754 008

层号	各层模型	输出结构	参数量
1	Conv2D	(None,64,64,32)	320
2	MaxPooling2D	(None,32,32,32)	0
3	Conv2D	(None,32,32,64)	18 496
4	MaxPooling2D	(None,16,16,64)	0
5	Conv2D	(None,16,16,64)	36 928
6	Flatten	(None,16384)	0
7	Dense	(None,4096)	67 112 960
合计			67 168 704

层号	各层模型	输出结构	参数量
1	ConvLSTM2D	(None,None,64,64,30)	33 600
2	BatchNormalization	(None,None,64,64,30)	120
3	ConvLSTM2D	(None,None,64,64,30)	64 920
4	BatchNormalization	(None,None,64,64,30)	120
5	Conv3D	(None,None,64,64,1)	811
合计			99 571

层号	各层模型	输出结构	参数量
1	STLSTM2D	(None1,64,64,30)	76 800
2	Reshape	(None,64,64,30)	0
3	Conv2D	(None,64,64,1)	31
4	Reshape	(None,1,64,64,1)	0
合计			76 831

数据集	模型	MSE	RMSE	MAE	宏平均	微平均
	LSTM	0.032 2	0.179 5	0.091 1	0.975 9	0.976 6
BikeShare	CNN	0.329 1	0.573 7	0.244 9	0.855 5	0.859 4
	ConvLSTM	0.001 5	0.038 4	0.023 3	0.925 8	0.933 6
	ST-LSTM	0.027 4	0.165 5	0.135 7	0.988 1	0.988 3
	LSTM	0.153 3	0.391 5	0.276 6	0.608 2	0.718 8
Macquarie Park	CNN	0.308 7	0.555 6	0.283 8	0.880 7	0.859 4
	ConvLSTM	0.038 2	0.195 6	0.081 1	0.934 7	0.937 5
	ST-LSTM	0.046 9	0.216 5	0.046 9	0.947 5	0.953 1
	LSTM	0.250 4	0.490 0	0.331 5	0.662 7	0.790 1
Yellow_tripdata	CNN	0.175 7	0.402 8	0.234 3	0.725 5	0.853 2
	ConvLSTM	0.317 3	0.548 3	0.338 7	0.647 6	0.783 7
	ST-LSTM	0.089 6	0.298 8	0.230 9	0.801 0	0.911 5

基于深度学习的位置大数据统计发布与隐私保护方法

Statistics release and privacy protection method of location big data based on deep learning

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 39

相关文章 15

Metrics

推荐阅读 0

参数	数据集
参数	BikeShare	Macquarie Park	Yellow_tripdata
覆盖范围	6.59 km×7.13 km	2.43 km×2.72 km	110.59 km×110.13 km
数据规模	70万	340万	1 400万
q₁	0.11 km×0.11km	0.04 km×0.04 km	0.44 km×0.44 km
q₂	0.22 km×0.22 km	0.08 km×0.08 km	0.88 km×0.88 km
q₃	0.44 km×0.44 km	0.16 km×0.16 km	1.76 km×1.76 km
q₄	0.88 km×0.88 km	0.32 km×0.32 km	3.52 km×3.52 km
q₅	1.76 km×1.76 km	0.64 km×0.64 km	7.04 km×7.04 km
q₆	3.52 km×3.52 km	1.28 km×1.28 km	14.08 km×14.08 km

[1]	陈东昱, 陈华, 范丽敏, 付一方, 王舰. 基于深度学习的随机性检验策略研究[J]. 通信学报, 2023, 44(6): 23-33.
[2]	李荣鹏, 汪丙炎, 张宏纲, 赵志峰. 知识增强的语义通信接收端设计[J]. 通信学报, 2023, 44(6): 70-76.
[3]	马帅, 裴科, 祁华艳, 李航, 曹雯, 王洪梅, 熊海良, 李世银. 基于生成模型的地磁室内高精度定位算法研究[J]. 通信学报, 2023, 44(6): 211-222.
[4]	冯涛, 陈李秋, 方君丽, 石建明. 基于本地化差分隐私和属性基可搜索加密的区块链数据共享方案[J]. 通信学报, 2023, 44(5): 224-233.
[5]	余晟兴, 陈泽凯, 陈钟, 刘西蒙. DAGUARD：联邦学习下的分布式后门攻击防御方案[J]. 通信学报, 2023, 44(5): 110-122.
[6]	张淑芬, 董燕灵, 徐精诚, 王豪石. 基于目标扰动的AdaBoost算法[J]. 通信学报, 2023, 44(2): 198-209.
[7]	汤凌韬, 王迪, 刘盛云. 面向非独立同分布数据的联邦学习数据增强方案[J]. 通信学报, 2023, 44(1): 164-176.
[8]	袁程胜, 郭强, 付章杰. 基于差分隐私的深度伪造指纹检测模型版权保护算法[J]. 通信学报, 2022, 43(9): 181-193.
[9]	王瀚仪, 李效光, 毕文卿, 陈亚虹, 李凤华, 牛犇. 多级本地化差分隐私算法推荐框架[J]. 通信学报, 2022, 43(8): 52-64.
[10]	杨洁, 董标, 付雪, 王禹, 桂冠. 基于轻量化分布式学习的自动调制分类方法[J]. 通信学报, 2022, 43(7): 134-142.
[11]	杨秀璋, 彭国军, 李子川, 吕杨琦, 刘思德, 李晨光. 基于Bert和BiLSTM-CRF的APT攻击实体识别及对齐研究[J]. 通信学报, 2022, 43(6): 58-70.
[12]	廖勇, 王世义. 高速移动环境下基于RM-Net的大规模MIMO CSI反馈算法[J]. 通信学报, 2022, 43(5): 166-176.
[13]	廖育荣, 王海宁, 林存宝, 李阳, 方宇强, 倪淑燕. 基于深度学习的光学遥感图像目标检测研究进展[J]. 通信学报, 2022, 43(5): 190-203.
[14]	张勇, 李丹丹, 韩璐, 黄小红. 隐私保护的群体感知数据交易算法[J]. 通信学报, 2022, 43(5): 1-13.
[15]	赵增华, 童跃凡, 崔佳洋. 基于域自适应的Wi-Fi指纹设备无关室内定位模型[J]. 通信学报, 2022, 43(4): 143-153.