Resource Sharing Based Spectrum Allocation In Dense C-V2X System Using Reinforcement Learning

The current extensive research being done in solving resource allocation problems in vehicular communication systems has led to many reinforcement learning-based solutions that target the dense network scenarios. This paper explores a dense environment and proposes a machine learning-based solution for resource and power distribution amongst multiple cellular links (CL). We take into consideration that more than one Non-Cellular Vehicular Link (NCVL) can use a single CL assigned to a single Cellular Vehicular Link (CVL), improving the spectral efficiency. The same CL is assigned to multiple NCVLs using a reinforcement learning algorithm based on the min-max channel gain concept. Priority in resource allocation is given to CVL with the minimum channel gain to minimize the interference in the network such that the spectrum efficiency is improved using reinforcement learning. The power is then distributed individually to each user equipment (UE) again prioritizing the throughput transmission throughout the network. Further, it can be observed from the results that the proposed system outperforms the other algorithms, especially in terms of the spectral efficiency.