The invention relates to the technical field of Internet of Vehicles
data privacy protection, and discloses an Internet of Vehicles
data privacy protection method based on mobile
perception federated learning, comprising the following steps: S1, a vehicle-mounted unit obtains vehicle operation data,
power consumption behavior data and
wireless communication data through mobile
perception, generates a
data set, and stores the
data set in a
database; and S2, the vehicle-mounted unit is based on the power
utilization behavior data in the
data set in the S1. By combining basic
encryption strength calculation of a vehicle state and final
encryption strength optimization of link stability, a two-dimensional dynamic
encryption mechanism is constructed, battery remaining capacity and running speed in the vehicle state provide a basis for basic encryption,
wireless communication link quality parameters support encryption strength secondary optimization, so that the encryption strength is adjusted as required, and the encryption efficiency is improved. And when the link is unstable, the strength is improved to guarantee privacy, when the link is stable, the complexity is reduced, the
delay is reduced, finally, the sensitive parameter
leakage rate is controlled at an extremely low level, the communication
delay is maintained within 200 milliseconds, and the vehicle-gauge-level real-time requirement is met.