Radar Sleep Monitoring for Dual-User Breathing Disambiguation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing contactless sleep monitoring systems struggle to accurately distinguish between the breathing of an intended user sleeping close to the radar unit and an unintended user sleeping farther away, particularly in dual-user scenarios, without centroid information, and to determine the user's sleep intent and duration accurately.
Innovation Solution
A cloud-based sleep-monitoring system with parallel processing pipelines for sleep-stage classification, primary user identification, presence detection, and intent-to-sleep classification, using 60 GHz radar technology, employs a reconciliation module to filter out data from unintended users and generate a hypnogram and sleep score for the intended user, leveraging machine learning models and radar data processing techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If radar monitoring is performed in a dual-user scenario without centroid information, then the monitoring coverage is extended, but the ability to distinguish between intended and unintended users deteriorates
Solution Approach 1:
The monitoring area is divided into multiple zones including a first region (bed area) and a second region (adjacent area). The system segments user detection by creating distinct monitoring zones with different purposes - the first region for primary sleep monitoring and the second region for detecting unintended users or disturbances, thereby maintaining user distinction accuracy while extending overall coverage
Solution Approach 2:
Out-of-bed indicators serve as intermediary signals that mediate between raw radar data and user identification. These indicators detect motion in the second region adjacent to the bed and provide contextual information that helps the system distinguish whether detected breathing patterns belong to the intended user or an unintended user, resolving the contradiction between coverage and precision
2Measurement precision
If parallel processing pipelines are implemented for sleep-stage classification, primary user identification, presence detection, and intent-to-sleep classification, then the processing accuracy is improved, but the system complexity increases
Solution Approach 1:
The system is divided into four independent parallel processing pipelines, each responsible for a specific function (sleep-stage classification, primary user identification, presence detection, intent-to-sleep classification). This segmentation allows each pipeline to be optimized independently while working together to achieve high overall accuracy without creating excessive system complexity
Solution Approach 2:
The radar unit serves multiple functions simultaneously - it detects presence, monitors breathing patterns for sleep staging, identifies the primary user, and detects out-of-bed conditions. This multi-functionality reduces the need for separate specialized sensors and systems, thereby managing complexity while maintaining high processing accuracy across all monitoring dimensions
3Measurement precision
If out-of-bed indicators are used to detect user presence and intent, then the sleep intent detection accuracy is improved, but the processing time increases
Solution Approach 1:
The system continuously monitors the second region adjacent to the bed for out-of-bed indicators even before the user is fully detected in the first region. By performing preliminary detection in the adjacent area, the system prepares user presence and intent information in advance, reducing the processing time required when the user actually enters the monitoring zone while maintaining high detection accuracy
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate, contactless sleep monitoring up to 1.5 meters without calibration, disambiguating between intended and unintended users, and generating precise sleep scores and hypnograms by filtering and trimming data effectively.
Implementation Method 1
The monitoring device is typically positioned next to the user's bed (e.g., on a nightstand adjacent to the bed) and uses radar to detect movement within a detection zone to identify a respiratory waveform of a user. Monitoring devices can use radar sensors to detect the range, velocity, and identity of objects in motion.
Implementation Method 2
Monitoring devices can use radar sensors to detect the range, velocity, and identity of objects in motion.
Data Source
AI summary
Technologies of contactless and real-time sleep tracking are described. One method of a sleep-monitoring device includes receiving radar data from a radar unit, the radar data representing a user's breathing. The method determines, using the radar data, sleep-stage data, presence data, and user identification data associated with the user. The method determines, using the i) sleep-stage data, ii) the presence data, and iii) the user identification data, a first event corresponding to sleep associated with the user and a second event corresponding to the sleep associated with the user, the second event occurring after the first event. The method modifies, using the presence data, the user identification data, the first event, and the second event, the sleep-stage data to obtain modified sleep-stage data. The method generates, using the modified sleep-stage data, a hypnogram of the sleep associated with the user.