|
| Type | Method | Feature description |
|
| Independent division of spatial data based on differential privacy | Quad-heu [21] | Division based on QuadTree structure with heuristic judgment strategy for node region adjustment and merging |
| PrivTree [22] | Introduced controllable deviation noise to decide whether to perform subtree partitioning, eliminating the requirement of predefined tree depth |
| HQP [23] | Adaptive sampling mechanism to select data samples based on proportional-integral-derivative (PID) controllers |
| lnLN_DPSD [24] | Add noise only at leaf nodes to prevent noise cancellation |
| Aba [25] | Designed and implemented a new arithmetic privacy budget allocation strategy |
| UG [26] | Division based on a single homogeneous grid, adding homogeneous noise to each of the divided grids |
| DPIH [27] | Divide the original data set into a fixed grid and inject noise and then perform IH-tree partitioning |
|
| Relevant division of spatial data based on differential privacy | AG [26] | Selfadaptive two-layer grid division structure, based on the first-level grid division, determine the second-level division granularity |
| Kd-PPDP [28] | Adding similarity judgment for proximity grids on the top of AG |
| STAG [29] | On the basis of random sampling, a three-layer grid structure is realized |
| UBQP-gra [30] | Unbalanced QuadTree division of regions according to the density of location points, stopping the division when the region meets the set conditions, reducing the large number of errors caused by blank nodes |
| HDPHD [32] | Determine density areas horizontally and vertically and further classify area units into different types by setting low- and high-density thresholds |
| DP-HDAQT [33] | The target area is divided into two classes: dense and sparse areas, and then different strategies are set for different areas |
|
