Assisting humans to achieve optimal sleep by changing ambient temperature

Vivek Gupta, Siddhant Mittal, Sandip Bhaumik, Raj Roy

Abstract

The environment plays a critical role in human sleep mechanisms. Research indicates that the sleeping environment, waking times, and total sleep duration significantly impact health, with poor conditions potentially leading to various sleep disorders. This paper proposes a method to maximize sleep efficiency by identifying an individual's real-time sleep cycle and adjusting ambient temperature accordingly. By utilizing Fast Fourier Transform (FFT) of heart rate signals to extract heart rate variability (HRV) data—specifically the low frequency to high frequency (LF/HF) power ratio—sleep stages are detected through an automated algorithm. A feedback mechanism then modulates ambient temperature based on the detected sleep stage to facilitate optimal rest.

Introduction

Sleep quality is intrinsically linked to core body temperature ($T_{core}$), which follows a natural circadian rhythm. To initiate and maintain sleep, the body must effectively shed heat. However, most residential environments maintain a static temperature that fails to account for the varying thermoregulatory needs of different sleep phases, such as the transition from Light Sleep to Slow Wave Sleep (SWS) or REM.

Traditional sleep monitoring methods like Polysomnography (PSG) are expensive and intrusive, making them unsuitable for widespread home use. This study proposes a non-invasive, automated system that detects sleep cycles via HRV and dynamically adjusts the environment to improve sleep quality.

Methodology

The proposed system employs a four-step automated algorithm to create a closed-loop control environment.

Frequency Domain Analysis

Using FFT on RR intervals extracted from ECG signals, the system identifies three primary frequency bands: Very Low Frequency (VLF), Low Frequency (LF), and High Frequency (HF). These bands serve as proxies for autonomic nervous system activity.

Sleep Stage Detection

The system differentiates between Light Sleep, SWS, and REM based on the $LF/HF$ ratio. Typically, $LF/HF$ remains above unity at night but drops significantly during SWS. High-frequency power ($>0.65$ Hz) is specifically used to identify the REM stage.

Feedback Loop Mechanism

A Deterministic Finite Automata (DFA) monitors the "slope" of sleep transitions every five minutes. It operates across three states: State Plus (increasing temperature), State Minus (decreasing temperature), and State Neutral.

Figure 1: State machine diagram for the temperature feedback loop.

Temperature Modulation

  • Transition to Deep Sleep (SWS): The system decreases ambient temperature by a set amount (neg-delta) to facilitate the necessary drop in core body temperature.
  • Transition to REM: The system increases temperature (pos-delta). During REM, human thermoregulation is reduced, making the body more susceptible to cold-induced awakenings.

Results and Discussion

The algorithm was validated using the Physionet HRV toolkit. Simulations demonstrated that the system could accurately map the cascade of sleep stages using only heart rate data.

Figure 4: Simulated ambient temperature fluctuations tracking the sleep cycle over time.

Key findings include:

  • Autonomic Synchronization: Parasympathetic dominance during SWS and sympathetic increases during REM provide reliable triggers for environmental adjustment.
  • Thermal Facilitation: By lowering room temperature during transitions to SWS, the system actively "pulls" the user into deeper sleep stages.
  • Efficiency Gains: Since approximately 70% of sleep is spent in non-REM stages (Stages 1, 2, and SWS), maintaining a cooler environment during these specific windows is critical for preventing early-morning awakenings.
  • Dynamic Range: Simulations showed ambient temperature successfully tracking sleep stages, fluctuating between approximately 17°C and 21°C to match physiological needs.

Conclusion and Future Work

This study demonstrates the feasibility of a reactive, personalized sleep environment. By adjusting temperature in real-time based on physiological state, sleep efficiency can be significantly enhanced. Future work will focus on clinical validation with human subjects and the integration of wearable PPG sensors to replace ECG-based monitoring, making the system more accessible for consumer use.

Summary updated

Abstract

Environment plays a vital role in the sleep mechanism of a human. It has been shown from many studies that sleeping and waking environment, waking time and hours of sleep is of very significant importance which can result in sleeping disorders and variety of diseases. This paper finds the sleep cycle of an individual and according changes the ambient temperature to maximize his/her sleep efficiency. We suggest a method which will assist in increasing sleep efficiency. Using Fast-Fourier-Transformation (FFT) of heart rate signals to extract heart rate variability data such that low frequency / high frequency (LF/HF) power ratio we are detecting sleep stages using an automated algorithm and then applying feedback mechanism to alter the ambient temperature depending upon the sleep stage.
BibTeX
@inproceedings{gupta2016assisting,
  author = {Vivek Gupta and Siddhant Mittal and Sandip Bhaumik and Raj Roy},
  title = {{Assisting humans to achieve optimal sleep by changing ambient temperature}},
  booktitle = {BIBM},
  year = {2016},
  url = {https://openreview.net/forum?id=cGQxz5jKKY7},
}
Figure