| Input: : the sensors that need to be sampled at |
| Output: The noised data of sensors in |
| for each sensors do |
| Predict by using the LSTM model |
| if then |
| Let the sensor itself as a group |
| Add the group to |
| else |
| Add the sensor into |
| Employ K-means algorithm to cluster sensors in according to . |
| Add each group into based on the cluster results. |
| Introduce Laplacian noise into each group . |
| Allocate the group perturbed statistic to each sensor according to . |
| return The noised data of sensors in |
