Abstract: Aiming at the problem that it is difficult for existing environmental appliances to improve human sleep environment in real time and personalized, a new technology for detecting human sleep signs based on ultra-wideband radar was proposed. The technology uses ultra-wideband radar to capture physiological signals during sleep. The classification model is constructed by deep neural network to realize real-time and accurate recognition of sleep stages. Further, according to the sleep staging results, the system automatically executes the corresponding environmental control logic to optimize the sleep environment. In the verification stage, a set of air conditioning automatic control system was established. Through the effect test, the accuracy of the system in detecting sleep stages and the actual operation performance of the functional logic were verified, which proved that the system has certain potential to improve sleep.

Key words: Ultra-wideband radar, Sleep stages, Neural network, Air conditioning control