An Intelligent Congestion Avoidance Mechanism Based on Generalized Regression Neural Network for Heterogeneous Vehicular Networks

The information generated by safety and traffic efficiency applications needs strict communication requirements to be smoothly exchanged in intelligent transportation system. Unfortunately, data congestion is still a challenge that negatively affects network performance. In this paper, we propose an Intelligent Congestion Avoidance Mechanism (ICAM) to prevent congestion in Heterogeneous Vehicular Network (HetVNET) that adapts the Dedicated Short Range Communication (DSRC) transmission power using a Generalized Regression Neural Network (GRNN) to predict data congestion. We compare the performance of the proposed GRNN congestion prediction model to other well-known models such as Multiple Linear Regression (MLR), Support Vector Machine (SVM) for regression, Decision Tree Regression (DTR), and Multi-layer Perceptron for Regression (MLPR). Numerical results show that the proposed GRNN congestion prediction model outperforms those other models in terms of accuracy, reliability and stability. Furthermore, simulation results show a substantial network performance improvement compared to other congestion control methods in terms of packet delivery ratio, average delay, and packet loss ratio.