Precision Time Protocol (PTP) Optimization framework aims to optimize the time synchronization using the linuxptp suite by optimizing the Proportional-Integral servo’s parameters. The proposed framework automatically searches for the best variables using a Genetic Algorithm. Different evaluation functions can be used to fulfill optimization goal.
The software is copyrighted by the authors and is licensed under the GNU General Public License. See the file, COPYING, for details of the license terms.
PTP is implemented in PTP-aware networking hardware, that implements the IEEE 1588 standard. To succesfully run the code, an adequate Networking adapter and a proper driver must be installed in the system.
As this script uses linuxptp suite it is targeted against the linux operating system. Python3 must be installed.
Use the pip to install required python packages.
pip install numpy
pip install sklearnTo succesfully run the script with dedault settings run following code:
python3 main.py --i EnpXfYwhere EpnXfY stnads for the interface name. If you want to check the interface name please run following command to print all interfaces name:
ip aProvided script accepts a set of parameters:
| Argument | Description | Default |
|---|---|---|
| --pop_size | Initial population size | 8 |
| --pop_size | Number of epochs | 10 |
| --max_kp | Max value of Kp | 15 |
| --max_kp | Max value of Ki | 15 |
| --num_random | Number of random parents added to each epoch | 2 |
| --num_inherited | Number of the best parents that are crossed to create a new generation | 5 |
| --num_replicated | Number of the best parents that are replicated to create a new generation | 4 |
| --mutation_coef | Mutation coefficient | 1 |
| --debug_level | Determines level of debug prints | 1 |
| --i | Interface | - |
| --t | Time of a single test | 120 |
| --metric | Evaluation metric (1 - MSE, 2 - RMSE, 3 - MAE) | 1 |
| --elite_size | Number of elite chromosomes | 1 |
All contributions will be considered for acceptance trough pull requests.