Towards private and accurate IoT applications
Over the past years, the fast-growing trend of Internet of Things (IoT) is bringing millions of new smart devices and sensors into homes, office buildings and industries. These smart devices and sensors enable smart IoT applications (e.g., energy prediction, activity recognition, etc.) to increase the quality and efficiency of our lives. To achieve promising performance for smart IoT applications, it requires massive data from different users and sensors to guarantee the performance due to machine learning and deep learning purposes. However, the edge devices of IoT applications often collect and store only limited data, which is insufficient for training modern learning models. Collaboratively training sets steps to achieve better application performance among different devices, while introducing the concern of data privacy. On the other hand, directly applying privacy-preserving techniques such as differential privacy can dramatically degrade the performance of IoT applications.
In this research, we aim to achieve privacy-first smart IoT applications while ensuring their accurate performance for multi-user and multi-sensor scenarios. First, we propose Personalized Federated Deep Reinforcement Learning~(PFDRL), a system that helps local users to achieve private and accurate energy management. PFDRL replaces the central server with decentralized federated learning (DFL) framework and enables a personalized federated reinforcement learning to tackle the standby energy reduction in residential buildings. Next, we propose PFed-LDP, a personalized federated local differential privacy framework for global users. The PFed-LDP design includes a weight-enhanced local differential privacy (LDP) with a dynamic layer sharing mechanism in the federated learning framework to accomplish privacy-preserving, personalized and communication efficient IoT applications. Finally, we introduce PrivateGNN, a system that utilizes a privacy-enhanced self-attention graph neural network in multi-sensor scenarios. PrivateGNN helps to prevent the latent activities identification from multi-sensor environment while ensuring the accurate performance of the original purpose of the sensors.
- Yangfeng Ji, Chair, CS/SEAS/UVA
- Dr. Bradford Campbell, Advisor, CS/SEAS/UVA
- Dr. Lu Feng, CS/SEAS/UVA
- Dr. Cong Shen, ECE/SEAS/UVA
- Dr. Jorge Ortiz, ECE/UVA/Rutgers