A Traffic Flow Prediction Method Based on the Fusion of Blockchain and Federated Learning

In the field of intelligent transportation, real-time and accurate traffic flow prediction has always been the top priority in urban development, which plays a crucial role in improving the operation efficiency of the road network. Most of the existing traffic flow prediction methods are based on machine learning, ignoring cases where the client is unwilling to participate in the prediction task or lies in order to obtain high rewards, resulting in a decline in the accuracy of traffic flow prediction when the model is aggregated. This paper proposes a Traffic Flow Prediction Method Based on Blockchain and Federated Learning (TFPM-BFL) to solve this problem. In this method, the client uses the Long Short-Term Memory (LSTM) model with attention mechanism to make local prediction and improve the prediction accuracy. An incentive mechanism based on credit rating is designed. Local and local credit values are obtained by evaluating the quality of the model uploaded by the client, and rewards are distributed according to the credit rating results, so as to encourage the client to participate in federal learning. Edge Server (ES) uses the model aggregation method based on credit value and compression rate to improve the model aggregation quality. The simulation results show that TFPM-BFL can achieve accurate and timely traffic flow prediction, effectively motivate clients to participate in Federated Learning (FL) tasks while ensuring the privacy of underlying data, and realize high-quality model aggregation.