Efficient Multilevel Federated Compressed Reinforcement Learning of Smart Homes Using Deep Learning Methods

Internet of Things connects all real time devices using the wireless nature for collecting, sharing and processing of data. These data are analyzed using machine learning models based on the structure of data. Reinforcement learning is a type of learning method which performs with past experience of data. Traditional algorithms use data with a specific environment with a learning process for prediction. Federated Learning (FL) is achieved through the integration of the various learning models for achieving accuracy. The proposed learning algorithm uses multilevel FL over the smart homes with two house data for analysis of the user behavior. Various kinds of sensors are used for analyzing the behavior, namely local and global. The data is shared with agents and servers with the use of communication networks. It suffers the bandwidth issues because of heterogeneity in the data, so this is overcome by using FL compression method. Multilevel FL compression method achieves reduced latency with efficient interaction. The proposed technique achieves better accuracy when compared to existing RL method with maximum performance and reliability.