search

mawongh / wirelessRL

Reinforcement Learning paradigms in the domain of Self-Organizing Networks
3 stars 4 forks source link

Would you write a more detail README file? #1

Open zhuwenxing opened 6 years ago

zhuwenxing commented 6 years ago

I think this project is the most complete work that implements reinforcement learning in wireless communication.But the whole project is so huge, i hope that you can tell something more details in README kindly!I am a beginner in the filed combines reinforcement learning and wireless communication .Looking for your reply!

mawongh commented 6 years ago

Hello Zhu Wenxing, based on your request, I will include more detailed information in the repository, however, since as you said the scope is big, it would take some time to include specifically what you're interested in. In the meantime, please feel free to ask me any question about the repo.

Regards,

Manuel Wong

zhuwenxing commented 6 years ago

Thank you for your reply!
I wanna say that i haveinvestigated in this field for a long time. I am sure that reinforcement learning will have a good application prospect in the field of wireless communication. The relevant papers found in IEEE don’t share their code on github unfortunately. I sent an email to the author but did not get a reply so far. One of the difficulties I encountered in applying reinforcement learning to wireless communication is to create a reinforcement learning environment for wireless communication. When we apply reinforcement learning to practical problems. , we need to build our own specific environment instead of using openai’ gym. In the environment.py of your repository, my understanding is that you use Python to call the NS3 emulator instead of building an environment from scratch with Python. (If you can build a wireless communication environment from scratch in Python, this is certainly better, but using Python to call NS3 with the command line is also a good idea.) Back to the topic, if possible, I hope you can add Environment, Prerequisites, How to use, Basic usage and other parts in the readme.I think the readme of this repository (https://github.com/NeymarL/ChineseChess-AlphaZero) is very standard and can be used as a reference.To be honest, I have no idea how to run this repository successfully, which file should be run firstly and what the function of some files is. I'm very sorry for troubleing you. Thank you again sincerely! By the way, may I have your email,so the commmunication will be more effective!

mawongh commented 6 years ago

Hello, sorry for the delayed response. My email is mawongh@gmail.com. Now, the first step is surely to install the NS3 simulator, which for me was one of the most complicated tasks. Then there is was I call the "glue code" that connects the python classes to the NS3, this is the lena-simple_env.cc file which will read a the configuration for the network from a .csv file (check line 41) and run the simulator, the simulator will output a .rem file (radio environment map).

mawongh commented 6 years ago

Then the only file you might need to focus is the environment.py which contains the network class, this class has several methods that perform the respective calls to the NS3 simulator, all communication with the NS3 simulator and the "glue code" are made through the configuration .csv file and the output .rem file.