3rd TBSI Workshop on Learning Theory, Oral Presentation

Presentation Topic: Markov Decision Process based Resource Allocation in V2X Communication Networks with Poisson Traffic

In this work, the sharing of the spectrum and transmission power in vehicular networks is modeled as a resource allocation problem. In summary, the main contribution of this work are as follows:

1. A Poisson traffic is proposed for testing the efficiency of V2X communication network.
2. A reinforcement learning algorithm based on Markov Chain is designed to maximize the capacity for high bandwidth content delivery and improve the payload delivery reliability.
3. A neural network named Deep Q-Network (DQN) with experience replay is established to improve the performance of Q-learning in reinforcement learning.