Research on LSTM Based Traffic Flow Prediction Adaptive Beacon Transmission Period and Power Joint Control

Abstract

In vehicle-to-vehicle communication, each vehicle regularly broadcasts its own status information in its beacon broadcast to create awareness of the surrounding vehicles. Moreover, most security applications in vehicular ad hoc networks are based on periodic security information. However, vehicular ad hoc networks are characterized by large node density changes, fast network topology changes, diverse wireless channel quality changes and so on, which makes it difficult to guarantee real-time and reliable beacon message transmission. Therefore, we want to predict the future traffic flow to improve the beacon transmission performance. Among many prediction models, we choose the LSTM prediction model with higher stability and accuracy. In this paper, based on the short-time traffic flow prediction model of the improved LSTM network, the traffic flow time data is taken as the input sample, and the LSTM network is used to predict the traffic flow. Then, the combined control algorithm of the adaptive beacon transmission period and transmission power is realized through the predicted parameters. Experimental simulation and analysis show that this algorithm can significantly reduce the distribution delay and improve the packet delivery rate, which proves that our method can effectively reduce channel congestion and improve the performance of vehicle network.