AoI-Aware Resource Allocation for Platoon-Based C-V2X Networks via Multi-Agent Multi-Task Reinforcement Learning

Citation Author(s):
Mohammad
Parvini
Submitted by:
Mohammad Parvini
Last updated:
Mon, 05/10/2021 - 04:27
DOI:
10.21227/3kfr-ct25
License:
5
2 ratings - Please login to submit your rating.

Abstract 

The simulation code for the paper:

"AoI-Aware Resource Allocation for Platoon-Based C-V2X Networks via Multi-Agent Multi-Task Reinforcement Learning"

 

The overall architecture of the proposed MARL framework is shown in the figure.

 

Modified MADDPG: This algorithm trains two critics (different from legacy MADDPG) with the following functionalities:

The global critic which estimates the global expected reward and motivates the agents toward a cooperating behavior and an exclusive local critic for each agent that estimates the local individual reward.

 

Modified MADDPG with Task decomposition: This algorithm is similar to the Modified MADDPG; however, in this algorithm, the local holistic reward function of each agent is further decomposed into multiple sub-reward functions based on the tasks each agent has to accomplish, and the task-wise value functions are learned separately.

 

"For both algorithms, the global critic is built upon the twin delayed policy gradient (TD3)."

Instructions: 

Please make sure that the following prerequisites are met:

 

python 3.7 or higher

PyTorch 1.7 or higher + CUDA

It is recommended that the latest drivers be installed for the GPU.

In order to run the code:

 

***

Please make sure that you have created the following directories:

    1) ...\Classes\tmp\ddpg

    2) ...\model\marl_model

The final results and the network weights will be saved in these directories.

 

***