# US 2018/0324595 A1 · Spectral sensing and allocation using deep machine learning

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

- Number: US 2018/0324595 A1
- Document ID: US20180324595A1
- Status: Published application
- Kind: A1
- Application number: 15/973,003
- Filed: 2018-05-07
- Published: 2018-11-08
- Earliest claimed: 2017-05-05
- Assignees: Ball Aerospace & Technologies Corp.
- Applicants: Ball Aerospace & Technologies Corp.
- Inventors: James Michael Shima
- Web page: https://patentgrep.com/patent/US20180324595A1

## CPC

- H04W 16/10: Dynamic resource partitioning
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/06: Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- H04B 7/0413: MIMO systems
- H04W 16/14: Spectrum sharing arrangements {between different networks}
- H04W 24/02: Arrangements for optimising operational condition
- H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04B: TRANSMISSION

## Abstract

Methods and systems for identifying occupied areas of a radio frequency (RF) spectrum, identifying areas within that RF spectrum that are unusable for further transmissions, and identifying areas within that RF spectrum that are occupied but that may nonetheless be available for additional RF transmissions are provided. Implementation of the method then systems can include the use of multiple deep neural networks (DNNs), such as convolutional neural networks (CNN's), that are provided with inputs in the form of RF spectrograms. Embodiments of the present disclosure can be applied to cognitive radios or other configurable communication devices, including but not limited to multiple inputs multiple output (MIMO) devices and 5G communication system devices.

## Claims

### Claim 1

A method for analyzing spectrum, comprising:
generating an RF spectrogram; and
analyzing the RF spectrogram using a first deep neural network (DNN) to identify occupied portions of the RF spectrum.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
determining whether a portion of the RF spectrum is occupied;
in response to determining that the portion of the RF spectrum is occupied, analyzing the RF spectrum using a second DNN to identify at least one of occupied but nonetheless usable portions of the RF spectrum, and identifying unusable portions of the RF spectrum.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
determining whether a portion of the RF spectrum is occupied:
in response to determining that the portion of the RF spectrum is analyzed, analyzing the RF spectrum using a second DNN to identify at least one of occupied but nonetheless usable portions of the RF spectrum, and analyzing the RF spectrum using a third DNN to identify unusable portions of the RF spectrum.

### Claim 4 (depends on 3)

The method of claim 3, wherein the result of analyzing the RF spectrum using a second DNN to identify at least one of occupied but nonetheless usable portions of the RF spectrum is a heat map.

### Claim 5 (depends on 4)

The method of claim 4, wherein the result of analyzing the RF spectrum using a third DNN to identify unusable portions of the RF spectrum is a mask.

### Claim 6 (depends on 5)

The method of claim 5, wherein analyzing the RF spectrum using a second DNN to identify at least one of occupied but nonetheless usable portions of the RF and analyzing the RF spectrum using a third DNN to identify unusable portions of the RF spectrum are performed in parallel.

### Claim 7 (depends on 6)

The method of claim 6, further comprising executing an application for implementing the DNN that analyzes the RF spectrum on a configurable RF device.

### Claim 8 (depends on 7)

The method of claim 7, further comprising:
configuring parameters of the RF device using the results of the analyses.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
outputting an RF signal using the parameters.

### Claim 10 (depends on 9)

The method of claim 9, wherein the configurable RF device is a cognitive radio.

### Claim 11 (depends on 9)

The method of claim 9, wherein the configurable RF device is a mobile telephone.

### Claim 12

A radio frequency (RF) device, comprising:
an antenna;
a processor;
memory;
application programming stored in the memory and executed by the processor, wherein the application programming implements a first deep neural network operable to perform spectral sensing of an RF spectrum received by the antenna of the RF device.

### Claim 13 (depends on 12)

The RF device of claim 12, wherein the application programming implements an additional deep neural network operable to determine at least one of channel quality metrics of the RF spectrum and occupied channel identification within the RF spectrum.

### Claim 14 (depends on 13)

The RF device of claim 13, wherein the application programming implements a second deep neural network operable to determine at least one of channel quality metrics of the RF spectrum, and wherein the application programming implements a third deep neural network operable to identify occupied channels within the RF spectrum.

### Claim 15 (depends on 14)

The RF device of claim 14, wherein the device is a configurable RF device.

### Claim 16 (depends on 15)

The RF device of claim 15, wherein the device has a plurality of configurable radio parameters, and wherein at least some of the parameters are set in response to outputs of at least one of the configurable neural networks executed on the device.

### Claim 17 (depends on 16)

The RF device of claim 16, wherein the device includes a radio transceiver connected to a plurality of antennas.

### Claim 18 (depends on 17)

The RF device of claim 17, wherein the device includes a plurality of antennas, and wherein at least one of the configurable parameters is selecting antenna for use in transmitting a signal.

### Claim 19 (depends on 18)

The RF device of claim 18, wherein the device is a mobile telephone.

### Claim 20

A method of analyzing a radio frequency (RF) spectrum, comprising:
receiving RF signals;
transforming the received RF signals into a spectrogram;
analyzing the spectrogram using a first deep neural network (DNN) to determine the presence of signals;
in response to determining that signals are present, analyzing the spectrogram:
in a second DNN to determine an occupancy of the spectrum; and
in a third DNN to determine unusable areas of the spectrum;
setting a radio parameter in response to the output of at least one of the DNNs; and
transmitting a signal using the radio parameter.