Federated Learning-based Privacy-Preserving Traffic Flow Prediction Scheme for VANETs

Abstract

Vehicular ad hoc networks (VANETs) bring us a lot of convenience. However, the credibility of the messages transmitted among vehicles and the security of VANETs still face risks and challenges. In this paper, we propose a privacy-preserving traffic flow prediction scheme based on federated learning and long short-term memory recurrent neural network (LSTM RNN), which can perform traffic flow prediction for VANETs. We use LSTM RNN as the deep learning neural network for predicting traffic flow. Federated learning framework is used for model interaction and data privacy protection. Security and performance analysis show that the proposed scheme can not only predict the traffic flow, but also protect the privacy of users and model.