Systems, apparatuses and methods for adaptive noise reduction
Inventors
CENCI, IVAN • TOGNETTI, SIMONE
Assignees
Publication Number
US-10134378-B1
Publication Date
2018-11-20
Expiration Date
2035-04-16
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
An apparatus includes a sensor module configured for receiving sensed information indicative of a sensed signal. The sensed signal includes a source signal component and a source noise component. The apparatus also includes a reference module configured for reference information indicative of a reference signal. The reference signal also includes a reference noise component. The apparatus also includes a filter module configured as a fixed lag Kalman smoother. The filter module is configured for adaptively filtering the reference signal to generate an estimate of the source noise component. The apparatus also includes a processing module configured for calculating an output signal based on the sensed signal and the estimate of the source noise component. The apparatus also includes an interface module configured for transmitting an indication of the output signal. The filter module is further configured for, based on the output signal, tuning the Kalman smoother.
Core Innovation
The invention relates to systems, apparatuses, and methods for adaptive noise reduction using a fixed lag Kalman smoother. The apparatus receives a sensed signal composed of a source signal component and a source noise component, as well as a reference signal containing noise associated with the source noise. The filter module adaptively filters the reference signal to generate an estimate of the source noise component, specifically by setting the state noise covariance matrix equal to the measurement noise, setting the observation matrix to match the reference signal, and setting the oblivion coefficient to about one.
The processing module calculates an output signal associated with the source signal by combining the sensed signal and the estimate of the source noise component. The filter module further tunes the Kalman smoother based on the output signal by adjusting parameters such as the state noise covariance, oblivion coefficient, or the order of the Kalman smoother. This approach addresses the problem in the prior art wherein existing adaptive noise canceling techniques, including Kalman filtering, have limited adaptivity and lag errors, failing to efficiently handle non-stationary signals such as physiological signals with sharp spikes and discontinuities.
Claims Coverage
The patent includes two independent claims covering an apparatus and a method for adaptive noise reduction, outlining key inventive features of the system configuration and operation.
Adaptive filtering with fixed lag Kalman smoother
A filter module configured as a fixed lag Kalman smoother that adaptively filters the reference signal to generate an estimate of the source noise component by setting observation data equal to the source signal, setting the observation matrix equal to the reference signal, and setting the oblivion coefficient to about one.
Output signal calculation based on sensed and estimated noise signals
A processing module configured to calculate an output signal associated with the source signal component, based on the sensed signal and the estimate of the source noise component, typically by subtracting the estimate of the source noise from the sensed signal.
Dynamic tuning of Kalman smoother parameters
The filter module is configured to tune the Kalman smoother based on the output signal by modifying the state noise covariance, oblivion coefficient, or the order of the Kalman smoother. Specifically, the state noise covariance can be incremented or decremented based on whether the sensed or reference signals exceed defined thresholds, with parameters such as thresholds and increments being user-modifiable.
Signal sensing and referencing with physiological sensors
The apparatus includes sensor modules configured for receiving sensed signals including photoplethysmographic information and reference signals including accelerometer information, involving corresponding sensors such as photoplethysmographic sensors and accelerometers.
Delay introduction between source and output signals
The method introduces a delay of half the order of the Kalman smoother between the source signal and the output signal, and similarly between the reference signal and the output signal, to implement fixed lag smoothing effectively.
The independent claims collectively cover an adaptive noise reduction system and method employing a fixed lag Kalman smoother configured to process sensed and reference signals for estimating and removing noise, with dynamic tuning capabilities to enhance adaptivity and reduce lag errors.
Stated Advantages
Enhanced adaptivity for adaptive noise reduction, effectively handling signals with sharp spikes, discontinuities, and non-stationary behavior.
Reduced or eliminated lag error by using a fixed lag Kalman smoother with centered moving average windows and delay compensation.
Capability to dynamically tune key parameters like state noise covariance, oblivion coefficient, and order of the Kalman smoother based on output error and signal conditions, improving noise estimation accuracy.
Documented Applications
Removal of motion artifacts from photoplethysmographic signals.
Extraction of fetal heartbeat signal from combined maternal and fetal ECG signals by removing maternal ECG as noise.
General removal of periodic interference such as 60 Hz noise in electrocardiography, noise in speech signals, antenna sidelobe interference, and background hum in music recordings.
Interested in licensing this patent?