Efficient federal learning method in Internet of Vehicles scene

Abstract

The invention provides an efficient federal learning method in an Internet of Vehicles scene. The method comprises the following steps of 1, enabling a road side unit to obtain an alternative learning task set, 2, selecting a training task, 3, establishing an initial model parameter, and sending the training task and a network address thereof to vehicles in a coverage range, 4, analyzing task information of the training task by each vehicle, and then determining whether to participate in the training process or not, if so, establishing communication connection with the road side unit through the network address, 5, sending the initial model parameters to each established vehicle; 6, enabling each vehicle to use local data to carry out local training on the current model parameters and uploading the current model parameters to a road side unit, 7, once a local training model uploaded by a certain vehicle is received, calculating the weight of the local training model in real time, weighing and aggreging the weight of the local training model into the global model in real time, and generating current model parameters and returning to all the vehicles in real time. and 8, iteratively executing the step 6 to the step 7 until the set number of iterations is met.

Description

technical field [0001] The present invention relates to the technical field of information security, in particular to the data privacy protection technology in the Internet of Vehicles scenario, and in particular to an efficient federated learning method in the Internet of Vehicles scenario. Background technique [0002] With the continuous development of Internet of Vehicles technology, the circulation of data between vehicles is essential. Using the road condition information data generated by the on-board equipment, combined with its own environmental perception capabilities, vehicle users can have a more accurate comprehensive judgment. Circulating data between vehicles is one of the effective means to improve the efficiency of data utilization and fully tap the value of data. However, car owners are reluctant to upload data to the data center due to concerns about personal privacy data leakage, which hinders the circulation of data and affects the development of the In...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the technical problem of long time-consuming training in the traditional federated learning method, the present invention provides an efficient federated learning method in the Internet of Vehicles scenario, so that it is suitable for scenarios with high requirements for training efficiency and meets the requirements of huge data data volume, keep data out of the local area, protect privacy, and share efficiently

Images