Application of Machine Learning for Link Adaptation in VANETs

Numerous advantages brought by Wireless Access in Vehicular Environment (WAVE) technology in the field of transportation have created a lot of attractions, expectations, and interests for a diverse population group (Engineers, researchers, scientists, etc.) all over the world. Under the WAVE standard, vehicles communicate among themselves and exchange critical safety messages in order to protect both humans and properties on public roads. However, in order to maintain dependable communication between the relevant nodes, an efficient Link Adaptation (LA) mechanism must be implemented for the Vehicular Ad Hoc Networks (VANETs) to succeed. Moreover, due to the dynamic nature of such an environment, LA implementation is a very challenging process. The current study uses a Machine Learning (ML) approach to investigate the issue, analyze the problem, and provide a model that might be utilized to address the challenging LA in VANETs. The analysis response obtained from the generated model demonstrated its ability to accurately predict the expected MCS as a function of the mobile's relative speed and available SNR. It was also interesting to see that contrary to the ML general theory, which insists on the need for a large amount of data in order to build an acceptable model, that with a few quantities of data, just like in our case, and a clear understanding of the application domain as well as all related characteristics of the specific area of interest, an effective model can also be obtained.