VANET Protocol Based On Traffic Information Forecasting

With the increasement of domestic car ownership,more and more traffic problems are gradually emerged,such as traffic congestion,frequent traffic accidents and vehicle emissions.Intelligent transportation has been recognized as the next generation technology of transportation field.It is proposed for traffic problems along with the raising 5G,Internet of things and self-driving technology.Big data becomes the key point of trouble solving through the improvement of network transmission rates and the stability of the sensor acquisitions.Data features and hidden rules captured by the historical are used to avoid obstacles,plan paths,optimize routes.This paper proposes a deep learning-based traffic flow prediction model based on the traffic data collected by traffic sensors.By analyzing the temporal and spatial characteristics of traffic flow,the model combined with LSTM and spatial GCN is used to extract the spatial and temporal features,train traffic forecasting model and predict the future traffic with high accuracy.On the basis,this paper proposes an VANET routing protocol based on traffic prediction,which establishes the simulation environment based on the prediction data,establishes global delay model based on the delay of the packet passing through the road segment.and employs the heuristic ant colony algorithm to find the transmission path with the minimum time delay.The use of predictive data can make the routing path adapt to the changes of future traffic flow,getting rid of the existing routing protocol that only uses the current data or the assumed arrival distribution to calculate the routing path and cannot adapt to the accidents in traffic.The specific research contents of this paper are as follows:(1)This paper proposes an area prediction algorithm based on deep learning algorithms to extract the temporal and spatial characteristics of traffic flows.The LSTM model is used to extract the temporal characteristics of traffic flows and the spatial domain GCN algorithm to extract the spatial characteristics of traffic flows.The data is preprocessed according to the periodicity of the traffic flow,the weekly periodicity and the weekend periodicity are divided to train different models.Finally,these are merged into LG + models.The simulation is verified by the existing traffic data set of the Los Angeles Expressway ring to 15-minute predictions,30-minute predictions,45-minute predictions and 1-hour predictions.Finally,better forecasting performance are obtained in the certainty coefficient and the resolvability variance Performance.The algorithm proposed in this paper can perform more accurate regional prediction,rather than individual road section prediction.(2)Based on the predicted traffic flow data at different times above,in response to the problem that the current routing protocol cannot be adaptive to changes in traffic status,this paper proposes a VANET routing protocol based on traffic prediction.By plugging prediction flow data into delay model,the delay of the packet passing through a road segment in the future will be obtained.Therefore we can obtain the current and future delay of the packet passing through every road segment in the area.This issue turns into a minimum cost problem and uses the ant colony algorithm to solve the optimal transmission path.(3)This paper combines OMNe T ++,SUMO,Veins to build the simulation scenario,and uses real traffic data to build the network environment.As far as we know,it is the first time to use the real data set to network simulation.Considering the situation of wireless propagation loss,the simulation proves that the routing protocol proposed in this paper can indeed improve network transmission performance in sparse scenarios on the simulation platform.