World Sleep Industry ENCYCLOPEDIA

UWB radar respiration and heartbeat modules

6.1.1

Definition, technical boundaries, and industry role

An UWB radar respiration and heartbeat module is an embedded component that uses short ultra-wideband radio pulses, or an equivalent wideband waveform, to sense human range, minute displacement, and periodic motion and provide radar data or respiration- and heartbeat-related outputs to a host. A typical module contains pulse generation and RF transmit-receive circuitry, transmit and receive antennas, a precision clock, sampling and accumulation, processing, memory, power management, and serial interfaces. Wide bandwidth supports fine range gates that help separate a person from the background before chest or abdominal micromotion is extracted.

“UWB” describes bandwidth characteristics and a corresponding regulatory framework, not one band, protocol, or function. A UWB chip intended for tag ranging and communication may expose timestamps, distance, or a channel impulse response without providing continuous radar frames or a respiration algorithm. Conversely, an impulse-radar module may not implement standardized data communications. Procurement must establish whether the part is a communication-location device, a general radar front end, a presence module, or an application module with a fixed respiration profile.

In sleep products, the module may support bed-zone presence, bed exit, movement, resting respiratory trends, and, under limited conditions, heartbeat-related micromotion estimates. It does not contact the skin, works in darkness, and can sense through some light nonmetallic materials. It does not directly measure airflow, oxygen saturation, cardiac electrical activity, or brain activity, however, and its name does not make it a vital-signs monitor. Claims to diagnose sleep apnea, identify arrhythmia, issue life-safety alarms, or guide treatment require medical-device-level risk, performance, clinical, and regulatory evidence for the finished system.

Pulse echoes, range gates, and micromotion

Impulse UWB radar transmits brief, spectrally wide, low-energy pulses and records reflections at different delays. Delay corresponds to propagation range. Samples within a scan form a fast-time range profile, while repeated frames form slow time and show how each range gate changes. Walls, furniture, and bed structures often produce relatively stable echoes; a person entering, leaving, or moving alters amplitude and phase. Baseline subtraction, clutter modeling, and target selection identify the spatial interval for further analysis.

Quiet breathing creates periodic chest and abdominal displacement that modulates the slow-time signal. Detrending, filtering, autocorrelation, spectral analysis, or peak-interval methods in selected gates can produce a respiratory waveform and rate. Heartbeat-related displacement is weaker and more easily obscured by breathing and its harmonics, mattress transmission, posture, bedding friction, and gross motion. Reliable micromotion estimation is commonly unavailable during large movement. A defensible system flags these intervals and reacquires after stability rather than filling a continuous number.

Theoretical range resolution relates to effective bandwidth, but pulse shape, receive bandwidth, equivalent-time sampling, antenna response, accumulation, and signal-to-noise ratio determine actual performance. More pulse repetition or accumulation can strengthen a weak return while increasing energy use, data, and latency. A range gate that is too narrow loses the chest after a posture change; one that is too broad incorporates a partner, fan, or bed echo. Low spectral density enables spectrum sharing but does not permit arbitrary transmit-power increases to compensate for poor placement.

Module architecture, specifications, and host interfaces

Architectures range from RF front ends exposing raw frames, through modules offering baseband or range data and range-Doppler products, to application modules outputting presence, respiratory rate, and quality states. An integrated antenna reduces host RF-layout work but fixes the pattern and mounting orientation. A chip or external-antenna approach offers flexibility while moving high-frequency and compliance risk to the product manufacturer. On-module processing reduces host traffic and raw-data exposure; external processing enables custom algorithms and research capture.

Relevant specifications include center frequency and occupied bandwidth, radiated spectral density, pulse repetition, pulses or accumulations per frame, range-step size, minimum and maximum sensing range, dynamic range, antenna pattern, update rate, startup, and adaptive-background time. Respiration and heartbeat applications also need validity, error, delay, invalidity criteria, and reacquisition performance for varied posture, bedding, multiple people, adjacent targets, and environmental motion. A maximum-distance statement or ideal laboratory mean error does not characterize bedside behavior.

Host connections may use SPI, I²C, UART, USB, or a proprietary protocol. Outputs may include raw frames, baseband amplitude, range-Doppler matrices, target range, motion state, respiratory waveforms, rate, and confidence. Selection also covers supply and peak current, standby and duty-cycled operation, clocks, temperature, thermal behavior, boot firmware, memory, debug access, drivers, and software licensing. Interface specifications should pin byte order, scaling, timestamps, dropped-frame behavior, error codes, and version compatibility so an invalid code cannot be interpreted as a physiological value.

Respiration and heartbeat algorithms and failure modes

A processing chain typically handles DC and static clutter, then uses frame differences or other slow-time features to detect a person and choose one or more range gates. Motion gating, drift correction, respiration and heartbeat band processing, periodicity estimation, harmonic selection, and result fusion follow. Target range, activity, signal-to-noise ratio, spectral prominence, and data integrity can contribute to a quality score. The product must distinguish no person, excessive movement, multi-person ambiguity, and inadequate quality from a true zero respiratory rate, and record its algorithm and parameter versions.

Two people in the same bed or adjacent beds may occupy the same range gate. One person's gross movement can also contaminate the other's return through mattress coupling and multipath. Pets, fans, curtains, HVAC vanes, rocking cribs, caregivers, and cleaning robots create periodic or transient changes. Metal frames, springs, electric blankets, and reflective walls can form strong echoes or blind areas, while heavy bedding, body direction, and mounting angle change chest visibility. An algorithm should disclose unsupported and confused conditions rather than hide them with smoothing.

Respiration and heart-rate estimates are susceptible to harmonic, double-rate, and half-rate errors. Respiratory harmonics may enter a nominal heartbeat band, and filter ringing after motion can create false peaks. Very slow or irregular breathing, weak cardiac displacement, body-size variation, prone posture, thick covers, and pediatric rate ranges require specific validation. Consumer radar cannot be the sole monitor for an infant, medically vulnerable older adult, or patient. Any no-breathing alarm needs analysis of misses, false alarms, system outages, and the human response pathway.

Antenna placement, material propagation, and coexistence

The module's main beam should cover the intended target region while the carrier respects antenna keep-outs, ground, spacing, and board-size guidance. Headboard-facing, bedside, ceiling, and under-mattress positions create different range changes and occlusion and need tests with the actual bed, mattress, and postures. Metal, conductive textiles, batteries, large connectors, and lossy absorbers should not obstruct an integrated antenna. Nearby screws, brackets, and wiring can also distort the pattern.

UWB signals can pass through some plastics, wood, gypsum board, clothing, and bedding, but “penetration” is not unconditional through-wall capability. Thickness, moisture, dielectric properties, metallic coatings, layered construction, and incidence angle cause attenuation, reflection, and dispersion. A changed enclosure, furniture panel, or mattress lot calls for renewed range calibration, blind-zone, and noise checks. Range gating and spatial design should deliberately prevent sensing in neighboring rooms, corridors, or unrelated beds.

Other UWB devices, Wi‑Fi systems, switching supplies, displays, and digital clocks may create interference or spurs. A multi-module deployment needs testing of pulse timing, duty cycle, randomization, or coordination rather than assuming low power prevents interaction. Power integrity, clock jitter, grounding, shielding, electrostatic immunity, and electromagnetic compatibility all affect weak micromotion. Pre-compliance tests should use the final cover, cable set, active host modes, and worst-case transmission while confirming that filters or coexistence settings do not destroy sensing performance.

Data privacy, cybersecurity, and radio compliance

UWB radar generally produces no recognizable face image, yet range profiles, bed occupancy, entry and exit timing, respiratory waveforms, and heartbeat trends reveal behavior and possible health information. Privacy starts with the minimum sensing volume, edge processing, and data tiers. When the module can produce a quality-controlled event, raw frames need not be uploaded by default. The product should state purpose, retention, access roles, and deletion and establish a separate basis for secondary model training or product improvement.

Firmware and configuration affect both legal transmission and result integrity. Production systems can use secure boot, signed updates, authenticated or disabled debug, device identities, encrypted links, and least privilege. Remote changes to center frequency, pulse timing, duty cycle, gain, range gates, and alarm thresholds need authorization, version records, and rollback. Cloud systems, mobile clients, and third-party algorithms belong in the threat model. Resale, account deletion, and vendor shutdown require key revocation and treatment of local and cloud data.

UWB is not a globally uniform license-free permission. Jurisdictions regulate device categories, bands, uses, indoor or outdoor conditions, operator eligibility, peak and average emissions, spectral density, and mitigation. U.S. rules distinguish imaging, surveillance, indoor, handheld, and other UWB categories, with use or operator limits for some categories. European harmonized standards and implementing decisions also evolve. A certified module remains subject to host integration, labeling, antenna, firmware, and installation conditions; departure may require a permissive change or complete end-product authorization.

Validation, production quality, and lifecycle control

Validation starts from intended use and decomposes respiration, heartbeat, presence, or bed-exit claims by range, angle, posture, people, bedding, bed type, environmental motion, startup, and output delay. Bench tests examine clocks, pulses, noise, dynamic range, and interfaces. Controlled targets verify range and microdisplacement response. Real sleep scenarios cover a single person, a partner, pets, turns, exits, and moved equipment. Training, tuning, and independent test datasets should be split by participant so adjacent segments from one person cannot inflate results.

The reference must fit the claim. Presence and bed exit may use synchronized annotation or an independent sensor; respiration can be compared with a validated respiratory belt, airflow, or other appropriate reference; heart rate and medical sleep events require corresponding ECG, polysomnography, or clinical reference. Reports should provide valid-data yield, error distribution, confidence intervals, harmonic failures, post-motion reacquisition, multi-person performance, missed and false events, and outages, not only correlation or mean error. Display and handling of invalid outputs are also test subjects.

Production controls include component and antenna-material traceability, solder and visual inspection, power and interface tests, RF transmit-receive checks, calibration programming, firmware hashes, and sampled end-to-end tests. Golden units, limits, and measurement-system analysis reveal lot drift. Changes in RF silicon, oscillator, package, antenna substrate, enclosure resin, algorithm library, or compiler can affect sensing and compliance and require supplier notification, impact assessment, and revalidation. Procurement contracts should anticipate end-of-life replacement, long-term firmware, vulnerability response, and transfer of certification evidence.