Table 4.
Combining our preprocessing strategy with generalized compression algorithms.
| Preprocessing Applied | |||
|---|---|---|---|
| Algorithm | No (Ratio) | Yes (Ratio) | Time (ms) |
| None | 1.00 | 2.99 | - |
| Gzip [47] | 1.84 | 4.82 | 38.08 |
| bzip2 [48] | 1.87 | 5.05 | 21.71 |
| LZMA [49] | 2.68 | 5.36 | 121.32 |
| LZ4 [50] | 1.43 | 4.05 | 82.00 |
| Brotli [51] | 2.60 | 5.59 | 762.41 |
| zstd [52] | 1.97 | 5.01 | 377.04 |
| zlib [53] | 1.84 | 4.83 | 38.92 |
The compression ratios are related to the average size of a chunk (182 kB). The underlying dataset was acquired from a real vehicle. It contains 589,904 samples from 39 different sensors.