Skip to main content

Advances, Systems and Applications

Table 4 Privacy protection average performance of the total differential privacy budget

From: Real-time trajectory privacy protection based on improved differential privacy method and deep learning model

Dataset

ε

Query Error

JS Divergence

Kendall Coefficient

UG

AG

Top-Down

UG

AG

Top-Down

UG

AG

Top-Down

Yonsei

0.1

0.00094

0.00055

0.00046

4.07402

1.56135

0.53506

0.00156

0.0016

0.00168

0.3

0.00092

0.00031

0.00028

4.00918

1.38628

0.31496

0.00156

0.0016

0.00168

0.5

0.00096

0.00044

0.00026

3.99288

1.47073

0.44056

0.00168

0.00172

0.00181

0.7

0.00126

0.0003

0.00021

2.76302

0.88986

0.12478

0.00186

0.00191

0.002

0.9

0.00124

0.00055

0.0003

2.41934

0.89516

0.27262

0.00206

0.00211

0.0022

Citi Bike

0.1

0.00112

0.00101

0.00096

1.66402

1.50994

1.46392

0.00424

0.00432

0.00458

0.3

0.00121

0.001

0.00092

3.52368

2.77579

2.5524

0.00424

0.00496

0.00704

0.5

0.00115

0.00091

0.00082

3.54326

2.87243

2.67206

0.00424

0.0055

0.00827

0.7

0.00108

0.0009

0.00084

3.78866

2.90351

2.63912

0.00433

0.00582

0.00872

0.9

0.0011

0.00095

0.00081

3.10342

2.65329

2.51884

0.00439

0.00616

0.00902