# US 10,871,551 B2 · Least squares fit classifier for improved sensor performance

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 10,871,551 B2
- Document ID: US10871551B2
- Status: Granted
- Kind: B2
- Application number: 15/396,678
- Filed: 2017-01-01
- Published: 2017-07-06
- Granted: 2020-12-22
- Earliest claimed: 2015-12-31
- Applicants: Herbert U Fluhler
- Inventors: Herbert U Fluhler
- Web page: https://patentgrep.com/patent/US10871551B2

## CPC

- G01S 7/2922: {by using a controlled threshold}
- G01S 7/2886: {using I/Q processing}
- G01S: RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES

## Abstract

An improvement to sensor detection performance is described through use of novel least squares based (or other) sensor measurements fitting, which can significantly improve or increase the sensor Probability of Detection (Pd) while simultaneously improving or decreasing the sensor Probability of False Alarm (Pfa). Instead of just thresholding a scalar magnitude as is done in prior art signal detection methods, the new method seeks to classify time (or space, or spatio-temporal) sequenced Signal measurements from time (or space, or spatio-temporal) sequenced Noise measurements through unique features characteristic of each. The proposed method ma be implemented on most modern Radar Signal Data Processors (SDP) and therefore affords a near term, low cost opportunity to both significantly increase detection and tracking performance and/or enable ready adoption of new additional auxiliary missions currently not possible with resource constrained sensors.

## Claims

### Claim 1

A method for processing Doppler radar data to improve differentiation between a true signal and noise, said method comprising: accepting received timed data into a Doppler filter; fitting the timed data to at least one of: a model of a feature characteristic of a signal to produce a first output comprising a measure of goodness of fit to the feature characteristic of a signal and fitting the timed data to a model of a feature characteristic of noise to produce a second output comprising a measure of goodness of fit to the feature characteristic of noise; entering the at least one of the first and second outputs as a measurement feature into a classifier algorithm; and based on predetermined or dynamically computed classification parameters, producing an output classifying the timed data as comprising only noise with a predetermined level of confidence and/or as comprising noise and a signal with a predetermined level of confidence wherein the timed data comprises In Phase and Quadrature (I&Q) data and said method comprises mapping In Phase and Quadrature (I&Q) vectors, determining a helix spin rate for the mapped I&Q vectors, and comparing the determined helix spin rate to an expected helix spin rate for a true signal associated with the Doppler filter.

### Claim 2 (depends on 1)

The method of claim 1, wherein the timed data is selected from the group consisting of: amplitude and phase data, In Phase and Quadrature (I&Q) data, spatial sequenced measurement data, spatio-temporal measurement data, pulsed Doppler waveform data comprising a sequence of pulses, and combinations of these.

### Claim 3 (depends on 1)

The method of claim 1, wherein the timed data is fitted to both of the model of a feature characteristic of a signal to produce a first output comprising a measure of goodness of fit to the feature characteristic of a signal and the model of a feature characteristic of noise to produce a second output comprising a measure of goodness of fit to the feature characteristic of noise.

### Claim 4 (depends on 1)

The method of claim 1, wherein the method additionally comprises performing a target detection thresholding algorithm before or after said accepting timed data into a Doppler filter.

### Claim 5 (depends on 1)

The method of claim 1, wherein the timed data comprises I&Q data and said method comprises determining a variability in amplitudes of the I&Q data.