# US 11,443,137 B2 · Method and apparatus for detecting signal features

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

- Number: US 11,443,137 B2
- Document ID: US11443137B2
- Status: Granted
- Kind: B2
- Application number: 16/528,317
- Filed: 2019-07-31
- Published: 2021-02-04
- Granted: 2022-09-13
- Earliest claimed: 2019-07-31
- Assignees: Rohde & Schwarz GmbH & Co. KG
- Applicants: Rohde & Schwarz GmbH & Co. KG
- Inventors: Andrew Schaefer; Baris Guezelarslan; Benjamin Fischbach
- Web page: https://patentgrep.com/patent/US11443137B2

## CPC

- G06N 3/088: Non-supervised learning, e.g. competitive learning
- G06F 18/2133: based on naturality criteria, e.g. with non-negative factorisation or negative correlation
- G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0499: Feedforward networks
- G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning
- G06N 3/094: Adversarial learning
- G06V 10/454: {Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]}
- G06V 10/82: using neural networks
- H03M 1/1071: {Measuring or testing}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- H03M: CODING; DECODING; CODE CONVERSION IN GENERAL (using fluidic means F15C4/00; optical analogue/digital converters G02F7/00; coding, decoding or code conversion, specially adapted for particular applications, see the relevant subclasses, e.g. G01D, G01R, G06F, G06T, G09G, G10L, G11B, G11C, H04B, H04L, H04M, H04N; ciphering or deciphering for cryptography or other purposes involving the need for secrecy G09C)

## Abstract

A measurement apparatus comprising an acquisition memory adapted to store data sections of at least one acquired measurement signal; a processor adapted to calculate a measurement parameter vector, v, for each data section of the acquired measurement signal; and a trained autoencoder neural network adapted to process the measurement parameter vectors, v, applied as input data to the trained autoencoder neural network to provide at a middle layer of said autoencoder neural network an encoded vector, h, with characteristic signal features of the acquired measurement signal.

## Claims

### Claim 1

A method for detecting signal features of a measurement signal comprising the steps of:
providing at least two separate data sections of at least one measurement signal;
determining a measurement parameter vector, v, for each provided data section of the measurement signal; and
processing the measurement parameter vectors, v, as input data by a trained autoencoder neural network to detect signal features of said measurement signal.

### Claim 2 (depends on 1)

The method according to claim 1 wherein an encoded vector, h, including a number of characteristic signal features of the measurement signal is derived from a middle layer of said trained autoencoder neural network.

### Claim 3 (depends on 1)

The method according to claim 1 wherein a dimension of an output encoded vector, h, of said autoencoder neural network comprising the number of characteristic signal features derived from the middle layer of said autoencoder neural network is reduced to provide a vector, h′, in a feature space with lower dimension.

### Claim 4 (depends on 1)

The method according to claim 1 wherein the measurement signal is derived by a measurement apparatus from a device under test, DUT.

### Claim 5 (depends on 4)

The method according to claim 4 wherein characteristics of data sections of the measurement signal represented in the low dimensional feature space are displayed on a screen of a user interface of said measurement apparatus.

### Claim 6 (depends on 5)

The method according to claim 5 wherein data sections of the measurement signal comprising similar signal features in the low dimensional feature space are clustered to determine feature areas of similar data sections displayed on the screen of the user interface of said measurement apparatus.

### Claim 7 (depends on 6)

The method according to claim 6 wherein a first kind of label is assigned to each determined feature area.

### Claim 8 (depends on 6)

The method according to claim 6 wherein a second kind of label is assigned to the data section of the measurement signal based on the at least one determined feature area.

### Claim 9 (depends on 1)

The method according to claim 1 wherein the measurement signal comprises an analog measurement signal acquired by a probe of a measurement apparatus digitized and stored in an acquisition memory of the measurement apparatus.

### Claim 10 (depends on 9)

The method according to claim 9 wherein the digital measurement signal stored in the acquisition memory of said measurement apparatus is segmented into data sections processed to determine a measurement parameter vector, v, for each data section.

### Claim 11 (depends on 1)

The method according to claim 1 wherein the autoencoder neural network is trained in an unsupervised machine learning process on the basis of applied measurement parameter vectors, v.

### Claim 12

A measurement apparatus comprising:
an acquisition memory adapted to store data sections of at least one acquired measurement signal;
a processor adapted to calculate a measurement parameter vector, v, for each data section of the acquired measurement signal; and
a trained autoencoder neural network adapted to process the measurement parameter vectors, v, applied as input data to the trained autoencoder neural network to provide at a middle layer of said autoencoder neural network an encoded vector, h, with characteristic signal features of the acquired measurement signal.

### Claim 13 (depends on 12)

The measurement apparatus according to claim 12 comprising at least one probe to derive the at least one analog measurement signal from a device under test, DUT, and an analog to digital converter adapted to convert the analog measurement signal into a digital signal comprising data sections stored in the acquisition memory of said measurement apparatus.

### Claim 14 (depends on 12)

The measurement apparatus according to claim 12 wherein the trained autoencoder neural network comprises a variational autoencoder neural network or an adversarial autoencoder neural network.

### Claim 15 (depends on 12)

The measurement apparatus according to claim 12 wherein the measurement apparatus comprises a signal analyzer.

### Claim 16

A computer-implemented method for signal feature detection within at least one measurement signal,
wherein measurement parameter vectors, v, representing separate data sections of the measurement signal are processed by a trained autoencoder neural network to detect automatically characteristic signal features of the measurement signal displayed on a display of a user interface.

### Claim 17

A computer-implemented software tool for signal feature detection in one or more measurement signals by processing separate data sections of each measurement signal to calculate associated measurement parameter vectors, v, applied as input data to a trained autoencoder neural network to extract characteristic signal features of the measurement signals.