search

hust-diangroup / ns3-ai

Enable the interaction between ns-3 and popular frameworks using Python, which mean you can train and test your AI algorithms in ns-3 without changing any frameworks you are using now!
GNU General Public License v2.0
233 stars 83 forks source link

feat: add multi agent environment #126

Open Kakadus opened 4 months ago

Kakadus commented 4 months ago

This enhances ns3-ai by adding a multi-agent environment. Its api is inspired by its single agent counterpart but has the advantage of a single function NotifyCurrentState. For extensibility, a delay can be configured for the result of the python agent.

We developed it to showcast several use-cases in context of LTE reinforcement learning. We will soon showcast our work at https://github.com/DEFIANCE-project

OliverZim commented 3 months ago

I am currently working on adding documentation and an example that utilizes the Ns3MultiAgentEnv.

Asterohobit commented 3 months ago

This looks promising. Will use this in the future.

OliverZim commented 2 months ago

I added documentation and examples for the multi-agent interface #dc35b6a. @Asterohobit How was your experience with the new feature so far?

Asterohobit commented 2 months ago

@Asterohobit How was your experience with the new feature so far?

Currently, I train single agents. I still use the multi-agent environment because of the fixes.