Fairness-based Data Rate Control For Channel Congestion Mitigation in Vehicular Communications

Rate control is one of the promising approaches to mitigate channel congestion in wireless vehicular networks once resource allocation-based techniques are overloaded due to too many requests. Recently, state-of-the-art rate control methods have been proposed to mitigate channel congestion by reducing the data rate from low-risk vehicles. However, there is a lack of fairness guarantees during the risk assessment. As a result, if many vehicles are at the same risk level, there are possibilities that several of them may not be able to access the wireless channels to transmit messages on time due to the long queue of channel occupation requests. In this work, we present a novel fairness risk-based transmission control model, namely F-RTC, to automatically adjust the message broadcasting rate of a group of vehicles under channel congestion. F-RTC works based on Double Deep Q-learning Networks with a centralized control center to measure Jain's fairness index for each vehicle. F-RTC can suggest cutting down or increasing the data rate of the vehicles based on their score results. The evaluation results show that F-RTC outperforms the state-of-the-art risk-based rate control methods in terms of congestion rate reduction and channel busy ratio while having no significant impact on the packet delivery rate. We release the source code of the learning model for interesting readers at https://github.com/lanvernon/FRTC.