一种基于区块链和联邦学习融合的交通流预测方法

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

智能交通领域中实时准确地交通流预测一直是城市发展中的重中之重,这对提高路网运行效率起着至关重要的作用。现有的交通流预测方法大多是基于机器学习的,忽略了客户端不愿意参与预测任务或者为获得高奖励而撒谎的情况,从而在模型聚合时导致交通流预测的准确率下降。该文提出一种基于区块链和联邦学习融合的交通流预测方法(TFPM-BFL)来解决这一问题。在该方法中,利用加入了注意机制的长短期记忆网络(LSTM)模型进行本地预测,提高预测准确率;设计了基于信誉评定的激励机制,通过评估客户端上传的模型质量得到本地和局部信誉值,根据信誉值评定结果进行奖励分配,从而激励客户端参与联邦学习(FL);边缘服务器(ES)采用基于信誉值和压缩率的模型聚合方法来提高模型聚合质量。仿真结果表明,TFPM-BFL能够实现准确、及时地交通流预测,在保证底层数据私密的同时可以有效地激励客户端参与联邦学习任务,而且可以实现高质量的模型聚合。

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 TFPMBFL 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.