This study introduces a novel privacy-preserving sleep quality assessment algorithm leveraging millimeter-wave radar technology, designed to enhance the accuracy of sleep stage classification through non-contact monitoring of heart rate, respiratory rate, and body movement signals. While traditional contact-based monitoring methods, such as polysomnography (PSG), deliver reliable results in clinical settings, their application in long-term use is limited due to discomfort and complexity, particularly when monitoring special populations in home environments. The innovation of this study lies in the application of a dual sliding window algorithm to detect the stability of physiological signals, combined with real-time privacy-enhancing mechanisms to ensure data security. By dynamically adjusting sliding window sizes and stability thresholds, the algorithm optimizes feature extraction, achieving high accuracy in identifying periods of significant heart rate fluctuations and steady states. Although this research focuses primarily on heart rate stability monitoring, the algorithm demonstrates extensibility to respiratory and other physiological signals. Experimental results indicate that the proposed privacy-aware non-contact monitoring system achieves high accuracy, with practical applications in real-time home monitoring and elderly care, while ensuring user data protection.