WirelessTPP is a development toolkit for Mixture Density Networks (MDN) and Temporal Point Process (TPP) for temporal performance prediction in Openairinterface5G.
Model List [Back to Top]
We implement 2 state-of-the-art temporal point process (TPP) papers:
| No | Publication | Model | Paper |
|---|---|---|---|
| 1 | ICML'20 | THP | Transformer Hawkes process |
| 2 | ICLR'20 | IntensityFree | Intensity-Free Learning of Temporal Point Processes |
Dataset [Back to Top]
We use EDAF and Openairinterface 5G for creation of the dataset.
Quick Start [Back to Top]
This code is tested with Python 3.9. To create a Python 3.9 environment with Conda, you can use the following command:
conda create --name wireless_tpp python=3.9This command will create a new Conda environment with Python 3.9 installed.
conda activate wireless_tppWe provide an end-to-end example for users to run a standard TPP model with WirelessTPP.
First of all, we can install the package either by using pip or from the source code on Github.
To install the latest stable version:
pip install wireless-tppTo install the latest on GitHub:
git clone https://github.com/samiemostafavi/wireless-tpp.git
cd wireless-tpp
python setup.py installPreprocess edaf files and create database file:
python main.py -t preprocess_edaf -s data/s63_results
Plot the processed data (packet_arrival)
python main.py -t packet_arrival -u plot_data -s data/s63_results -c config/dataset_config.json -g s63 -n test0
Plot the processed data (link_quality)
python main.py -t link_quality -u plot_data -s data/s63_results -c config/dataset_config.json -g s63 -n test0
python main.py -t link_quality -u plot_data -f -s data/s63_results -c config/dataset_config.json -g s63 -n test0
Plot the processed data (scheduling)
python main.py -t scheduling -u plot_data -s data/s63_results -c config/dataset_config.json -g s63 -n test0
python main.py -t scheduling -u plot_data -v -s data/s63_results -c config/dataset_config.json -g s63 -n test0
Create the dataset (packet arrival)
python main.py -t packet_arrival -u create_training_dataset -s data/s63_results -c config/dataset_config.json -g s63_arrival -n test0
Create the dataset (link quality retransmissions)
python main.py -t link_quality -u create_training_dataset -s data/s63_results -c config/dataset_config.json -g s63_link_retx -n test0
Create the dataset (link quality mcs)
python main.py -t link_quality -u create_training_dataset -s data/s61-64_results -c config/dataset_config.json -g s61-64_link_mcs -n test0
Create the dataset (scheduling)
python main.py -t scheduling -u create_training_dataset -s data/s63_results -c config/dataset_config.json -g s63_scheduling -n test0
First modify the yaml file config/training_config.yaml with the datasets information and the configuration of the model you would like to train.
Train a model (packet arrival)
python main.py -t packet_arrival -u train_model -f -c config/training_config.yaml -i IF_train_s63_packetarrival_0
Train a model (link quality)
python main.py -t link_quality -u train_model -f -c config/training_config.yaml -i THP_train_s63_linkquality_0
Train a model (scheduling)
python main.py -t scheduling -u train_model -f -c config/training_config.yaml -i IF_train_s63_scheduling_0
By running the following commands you can check visually if the model is performing or not. It will take samples from the test dataset, runs predictions on them, and plots the results.
Validate packet arrival model (probabilistic)
python main.py -t packet_arrival -u generate_predictions -s data/s63_results -p probabilistic -c config/prediction_config.json -g s63_arrival -n test0 -i 1105474_140407072232064_241202-120400
python main.py -t packet_arrival -u plot_predictions -s data/s63_results -n test0 -i 1117955_140163946500736_241203-075703
Validate packet arrival model (sampling)
python main.py -t packet_arrival -u generate_predictions -s data/s63_results -p sampling -c config/prediction_config.json -g s63_arrival -n test0 -i 1105474_140407072232064_241202-120400
python main.py -t packet_arrival -u plot_predictions -s data/s63_results -n test0 -i 1119416_139811246588544_241203-084053
Validate retx link quality model (probabilistic)
python main.py -t link_quality -u generate_predictions -s data/s63_results -p probabilistic -c config/prediction_config.json -g s63_linkquality -n test0 -i 1106489_139985643180672_241202-123840
python main.py -t link_quality -u plot_predictions -s data/s63_results -n test0 -i 1121178_140591209673344_241203-091730
Validate retx link quality model (sampling)
python main.py -t link_quality -u generate_predictions -s data/s63_results -p sampling -c config/prediction_config.json -g s63_linkquality -n test0 -i 1105474_140407072232064_241202-120400
python main.py -t link_quality -u plot_predictions -s data/s63_results -n test0 -i 1122926_140662824829568_241203-102211
Validate the scheduling model (probabilistic)
python main.py -t scheduling -u generate_predictions -s data/s63_results -p probabilistic -c config/prediction_config.json -g s63_scheduling -n test0 -i 1112063_140079729623680_241202-162330
python main.py -t scheduling -u plot_predictions -s data/s63_results -n test0 -i 1124079_140705547641472_241203-105910 -m 1
python main.py -t scheduling -u plot_predictions -s data/s63_results -n test0 -i 1124079_140705547641472_241203-105910 -m 2
python main.py -t scheduling -u plot_predictions -s data/s63_results -n test0 -i 1124079_140705547641472_241203-105910 -m 3
Validate the scheduling model (sampling)
python main.py -t scheduling -u generate_predictions -s data/s63_results -p sampling -c config/prediction_config.json -g s63_scheduling -n test0 -i 1112063_140079729623680_241202-162330
python main.py -t scheduling -u plot_predictions -s data/s63_results -n test0 -i 1126371_140091332403840_241203-121033 -m 1
python main.py -t scheduling -u plot_predictions -s data/s63_results -n test0 -i 1126371_140091332403840_241203-121033 -m 2
python main.py -t scheduling -u plot_predictions -s data/s63_results -n test0 -i 1126371_140091332403840_241203-121033 -m 3
Benchmark [Back to Top]
License [Back to Top]
This project is licensed under the Apache License (Version 2.0). This toolkit also contains some code modified from other repos under other open-source licenses. See the NOTICE file for more information.
Todo List [Back to Top]
Citation [Back to Top]
Acknowledgment [Back to Top]
The following repositories are used in WirelessTPP, either in close to original form or as an inspiration: