DeepRP: Bottleneck Theory Guided Relay Placement for 6G Mesh Backhaul Augmentation

This repository provides simplified implementation for a paper that is in submission.

More specifically, some files are highlighted:

• In ./utils/bottleneck.py, the algorithms of bottleneck structure construction (BSC) and clique gradient computation (CliqueGrad) are implemented in details based on the the data structure of heap queue.
• In ./ppo_agent.py and ./drl_train.py, an actor-critic agent is trained by proximal policy optimization.
• In ./heuristic_main.py and ./random_main.py, two baseline relay placement methods are implemented.
• In ./config.py, a toy example on network configuration is provided.

A more complete and detailed implementation accompanied with real-world datasets will be released upon the acceptance of the paper.