Lorawan network with RL gateway agent, all of them simulated by NS3 and NS3Gym

Hi everyone, I’m working on an idea about creating an RL gateway agent with the LoRaWAN module NS3, and the RL part works on NS3Gym.

I created an environment with 10 end devices and 1 network server. Gateway, like an UAV, then collects data from each end device. In this scenario, I must minimize the time difference between the data generation time on each node and the network server. But now I think, how can I add some constraints for the end device or gateway, or all parts of the environment? Please give me some idea and any advice for me. Thanks to everyone.

Note that all scenarios were simulated with NS3 (C++) and an RL agent with Python.