Information Protection Model for Smart Homes Based on Federated Learning

Smart home devices collect data from a wide range of users.If malicious attackers exploit this, they can pose a serious threat to personal safety.In particular, as artificial intelligence (AI) technology becomes more widespread, the risk increases even more if malicious attackers utilize AI technology.This study proposes a privacy-preserving information protection model for smart homes based on Federated Learning (FL) as a method for protecting personal information in a smart home environment.The goal of this model is to combine FL with differential privacy, enhance privacy protection through the local training of devices, and add noise during data transmission.The experimental results demonstrate that this model provides comprehensive data protection across all stages, including device, transmission, and server processes.Additionally, the model is suitable for smart home devices with low computational power and can be extended to multidevice applications.The goal of this research is to enhance and improve information security when using smart home devices.