Modelling Packet Error Rate using Machine Learning in VANETs

Vehicular Ad Hoc Networks (VANETs) cannot achieve their full potential if the involved nodes are unable to transmit and receive road state information effectively. Due to the nature of this environment and based on the IEEE802.11p standard, all involved mobiles must broadcast data regularly (CAMs) or when the need arises (DENMs). Unfortunately, this type of communication protocol is the primary source of the broadcast storm, which depletes the available channel bandwidth and renders further communication difficult. To alleviate this problem, the current work proposes modeling the Packet Error Rate (PER) using Machine Learning (ML) to help each involved node to locally compute its own PER based on Signal-to-Noise Ratio (SNR) and Modulation Code Scheme (MCS). The estimated PER will assist the transmitter in deciding whether communication is necessary at any given time. The results of the developed model prove to be effective and efficient in estimating the PER with up to $\mathbf{9 9 \%}$ accuracy.