Research Article
Differentially Private Autocorrelation Time-Series Data Publishing Based on Sliding Window
| Input: original time series X | | Output: time series to be published after adding noise | (1) | Read the original time series X and divide X into n subsequences using the sliding window length L, where . | (2) | for i = 1 to n: | (3) | Calculate the autocorrelation function of the subsequence . | (4) | According to the query function q, calculate the periodic sensitivity of the time-series data X, where is computed by equation (5). | (5) | Generate four IID Gauss white noise series , which have the same length as . In addition, , where . | (6) | Calculate , , , and , where . | (7) | . | (8) | Splice at the end of Z. | (9) | end for | (10) | | (11) | Return |
|