Research on ALSTM-SVR Based Traffic Flow Prediction Adaptive Beacon Message Joint Control

In Vehicular Ad Hoc Networks (VANETs), vehicles rely on periodic broadcasting of beacon messages to perceive their surrounding environment. The interaction between vehicles’ Basic Safety Messages (BSMs) allows drivers to promptly gather information about the surrounding traffic conditions, enabling them to identify potential hazards and prevent accidents. However, effectively distributing beacon messages in complex and dynamic traffic environments presents a significant challenge. To address this challenge, this paper proposes an adaptive joint control scheme for beacon message dissemination based on an improved attention mechanism, specifically, the combination of Long Short-Term Memory (LSTM) and Support Vector Regression (SVR), for traffic flow prediction. The scheme leverages historical traffic flow data as input and applies various data preprocessing techniques, including reconstruction and normalization. By employing an LSTM model with an improved attention mechanism and denormalization, accurate predictions of traffic flow can be achieved. The predicted parameters are then utilized for the joint control of beacon message transmission. Simulation results demonstrate that the proposed joint control scheme significantly enhances network performance by effectively reducing channel congestion. The approach provides improved reliability and efficiency in distributing beacon messages, contributing to the overall safety and effectiveness of VANETs.