Blockchain-Based Mobile Edge Computing for Autonomous Vehicle Model Sharing

The patent segments the centralized cloud-based model training system into multiple distributed autonomous vehicle nodes that each perform local model training independently. This segmentation eliminates the single point of failure in centralized cloud systems, as each vehicle maintains its own model training capability. The blockchain network then coordinates these segmented nodes to share training data and models, ensuring both model accuracy through collective learning and system reliability through distributed architecture.

The patent implements a feedback mechanism through blockchain technology where each autonomous vehicle's training data and model updates are recorded on an immutable distributed ledger. This feedback system allows the network to verify the authenticity of training data from multiple sources, detect and reject falsified data through consensus validation, and continuously improve model accuracy while maintaining data security. The blockchain provides transparent audit trails that enable the system to identify and correct data quality issues.

The patent extracts the model training function from the centralized cloud-based service center and relocates it to the edge devices (autonomous vehicles) themselves. Each vehicle performs local model training using its own sensors and computing resources, extracting only essential model parameters and training insights for sharing with the network via blockchain. This extraction eliminates the need for continuous bulk data uploads to the cloud, significantly reducing network burden while maintaining high model training quality through distributed edge computing.