# US 2023/0109658 A1 · Spectrum-analysis-isolation-synthesis machine learning-based receiver system and method for spectrum coexistence and sharing applications

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

- Number: US 2023/0109658 A1
- Document ID: US20230109658A1
- Status: Published application
- Kind: A1
- Application number: 17/493,622
- Filed: 2021-10-04
- Published: 2023-04-06
- Earliest claimed: 2021-10-04
- Assignees: Booz Allen Hamilton Inc.
- Applicants: Booz Allen Hamilton Inc.
- Inventors: Ioannis SYLLAIOS; Alex SAUNDERS; Katherine OLSEN; Michael CALABRO
- Web page: https://patentgrep.com/patent/US20230109658A1

## CPC

- H04B 1/18: Input circuits, e.g. for coupling to an antenna or a transmission line (coupling networks between antennas or lines and receivers independent of the nature of the receiver H03H)
- G06F 18/23: Clustering techniques
- H04B 1/7101: {with estimation filters}
- H04B: TRANSMISSION
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Exemplary embodiments of the present disclosure are directed to a machine learning-based receiver that operates on the basis of a wide-band RF front-end. The RF front-end outputs clean IQ data to a SPectrum-Analysis-Isolation-Synthesis (SPAIS) system in order to collect analytics of signals present within the monitored band of frequencies, including signal center frequencies, bandwidth and power. The SPAIS system pre-processes the captured wideband IQ data in order to locate frequency bands with signal energy, isolate and synthesize the located frequency bands in corresponding IQ data for individual processing by a neural network with high classification accuracy.

## Claims

### Claim 1

A signal processing method, comprising:
receiving, by an RF receiver, a continuous spectrum of RF signals;
digitizing, by a signal processor, the RF signals as in-phase (I) and quadrature (Q) data;
estimating, by the signal processor, a spectrum of the IQ data;
identifying, by the signal processor, one or more data signals in the IQ data;
estimating, by the signal processor, at least a center frequency and a bandwidth of each data signal;
reconstructing, by the signal processor, each of the one or more data signals based on at least the data signal's estimated center frequency and bandwidth;
classifying, by a neural network of the signal processor, the one or more reconstructed data signals; and
generating, by the signal processor, an output signal which includes a classification result of each of the one or more reconstructed data signals.

### Claim 2 (depends on 1)

The method of claim 1, wherein estimating the spectrum of the IQ data comprises:
performing a time-to-frequency domain transform operation on the IQ data; and
analyzing a plurality of coefficients generated by the time-to-frequency domain operation to estimate the spectral density of the one or more data signals.

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
estimating a bandwidth of each data signal.

### Claim 4 (depends on 2)

The method of claim 2, further comprising:
estimating a power of each data signal.

### Claim 5 (depends on 1)

The method of claim 1, wherein reconstructing each of the one or more data signals comprises:
performing an inverse frequency-to-time domain transform operation.

### Claim 6 (depends on 5)

The method of claim 5, wherein the inverse frequency-to-time-domain transform operation comprises reconstructing and synthesizing each of the one or more data signals at a lower data rate than the time-to-frequency-domain transform operation.

### Claim 7 (depends on 1)

The method of claim 1, comprising:
training the neural network to detect and classify RF signals using at least the one or more reconstructed data signals and associated signal analytics.

### Claim 8 (depends on 1)

The method of claim 1, wherein receiving the continuous spectrum of received RF signals comprises:
processing the received RF signals using at least a bandpass filter.

### Claim 9 (depends on 1)

The method of claim 1, wherein each of the one or more data signals is reconstructed based on at least the data signal's estimated center frequency and the estimated bandwidth of the data signal.

### Claim 10

A signal processing system, comprising:
a receiver configured to receive a continuous spectrum of RF signals;
at least one processor configured to:
digitize the continuous spectrum of RF signals as in-phase (I) and quadrature (IQ) data;
estimate a spectrum of the IQ data;
identify one or more data signals in the IQ data;
estimate at least a center frequency and a bandwidth of each data signal;
reconstruct each of the one or more data signals based on the data signal's estimated center frequency and bandwidth;
classify the one or more reconstructed data signals; and
generate an output signal which includes a classification result of each of the one or more reconstructed data signals.

### Claim 11 (depends on 10)

The system of claim 10, wherein to estimate the spectrum of the IQ data, the at least one processor is configured to:
perform a time-to-frequency-domain transform operation on the IQ data; and
analyze a plurality of coefficients generated by the time-to-frequency-domain transform operation to estimate the spectral density of the one or more data signals.

### Claim 12 (depends on 11)

The system of claim 11, wherein the at least one processor is further configured to:
estimate a bandwidth of each data signal.

### Claim 13 (depends on 11)

The system of claim 11, wherein the at least one processor is further configured to:
estimate a power of each data signal.

### Claim 14 (depends on 10)

The system of claim 10, wherein to reconstruct each of the one or more data signals, the at least one processor is configured to perform an inverse frequency-to-time-domain transform operation.

### Claim 15 (depends on 14)

The system of claim 14, wherein by performing the inverse frequency-to-time-domain transform operation the at least one processor is configured to:
reconstruct and synthesize each of the one or more data signals at a lower data rate than the time-to-frequency-domain transform operation.

### Claim 16 (depends on 10)

The system of claim 10, wherein the at least one processor is further configured to:
train the neural network to detect and classify RF signals using at least the one or more reconstructed data signals.

### Claim 17 (depends on 10)

The system of claim 10, wherein the receiver processes the received RF signals through at least a bandpass filter.

### Claim 18

A non-transitory computer readable medium storing program code for performing RF signal processing, the program code causing a signal processor to perform operations comprising:
receiving, by an RF receiver, a continuous spectrum of RF signals;
digitizing, by a signal processor, the continuous spectrum of RF signals as in-phase (I) and quadrature (IQ) data;
estimating, by the signal processor, a spectrum of the IQ data;
identifying, by the signal processor, one or more data signals in the IQ data;
estimating, by the signal processor, at least a center frequency and a bandwidth of each data signal;
reconstructing, by the signal processor, each of the one or more data signals based on at least the data signal's estimated center frequency and bandwidth;
classifying, by a neural network of the signal processor, the one or more reconstructed data signals; and
generating, by the signal processor, a data signal which includes a classification result of each of the one or more reconstructed data signals.