Method, apparatus, and system for wireless sensing based on deep learning

Patent metadata

Field	Value
Publication number	US-12352889-B2
Application number	US-202418401681-A
Country	US
Kind code	B2
Filing date	Jan 1, 2024
Priority date	Feb 13, 2020
Publication date	Jul 8, 2025
Grant date	Jul 8, 2025

Abstract

Methods, apparatus and systems for wireless sensing based on deep learning are described. For example, a described method comprises: transmitting a wireless signal through a wireless multipath channel of a venue, wherein the wireless multipath channel is impacted by a motion of an object in the venue; receiving the wireless signal through the wireless multipath channel of the venue, wherein the received wireless signal differs from the transmitted wireless signal due to the wireless multipath channel and the motion of the object; obtaining a time series of channel information (TSCI) of the wireless multipath channel based on the received wireless signal; computing a plurality of autocorrelation functions based on the TSCI, each autocorrelation function (ACF) computed based on CI of the TSCI in a respective sliding time window; constructing at least one ACF vector, wherein each respective ACF vector is a vector associated with a respective ACF comprising multiple vector elements each associated with a respective time lag, each vector element being a value of the respective ACF evaluated at the respective time lag; rearranging the at least one ACF vector into rearranged ACF data, wherein each ACF vector is a one-dimensional (1D) ACF-block; and performing a wireless sensing task based on a task engine to do a processing using the rearranged ACF data as an input.

First claim

We claim:

1. A system for wireless sensing, comprising:
at least one transmitter-receiver pair (TX-RX pair), each TX-RX pair comprising a respective transmitter and a respective receiver, wherein:
the respective transmitter of each TX-RX pair is configured to transmit a respective wireless signal through a respective wireless multipath channel of a venue,
the respective wireless multipath channel is impacted by a motion of an object in the venue,
the respective receiver of each TX-RX pair is configured to receive the respective wireless signal from the respective transmitter of the TX-RX pair through the respective wireless multipath channel of the venue and generate at least one respective time series of channel information (TSCI) of the respective wireless multipath channel based on the respective received wireless signal, each respective TSCI associated with a respective transmit antenna of the respective transmitter and a respective receive antenna of the respective receiver,
the respective received wireless signal differs from the respective transmitted wireless signal due to the respective wireless multipath channel and the motion of the object; and

a processor configured to:
obtain a plurality of autocorrelation functions (ACF's) computed based on all the TSCI generated by the receivers of the at least one TX-RX pair, wherein each autocorrelation function (ACF) is computed based on CI of a respective TSCI in a respective sliding time window,
construct a plurality of ACF vectors based on the plurality of ACF's, wherein each respective ACF vector is a vector associated with a respective ACF comprising multiple vector elements each associated with a respective time lag, each vector element being a value of the respective ACF evaluated at the respective time lag,
construct a (k+1)-dimensional ((k+1)-D) ACF-block based on the plurality of ACF vectors, wherein the (k+1)-D ACF-block is a (k+1)-D matrix formed by assembling and organizing a plurality of 1-) matrices associated with the plurality of ACF vectors, each 1-D matrix being a respective ACF vector, wherein k is an integer,
reorganize the (k+1)-D ACF-block with a first data structure into rearranged ACF data with a second data structure by initializing a series of current matrices comprising the (k+1)-D ACF block as a single initial current matrix, and performing the following iteratively:
partitioning each current matrix of the series of current matrices into a respective plurality of sub-matrices,
scanning the respective plurality of sub-matrices of the respective current matrix in a respective scanning order,
constructing a respective series of ordered sub-matrices that are the respective plurality of sub-matrices ordered according to the respective scanning order, and
replacing the respective current matrix in the series of current matrices by the series of ordered sub-matrices to increase a length of the series of current matrices,

obtain the rearranged ACF data based on the series of current matrices, and
perform a wireless sensing task based on a task engine to do a processing using the rearranged ACF data with the second data structure as an input.

2. The system of claim 1 , wherein:
the wireless sensing task is performed based on the task engine to do the processing using a neural network with the rearranged ACF data as an input to the neural network.

3. The system of claim 2 , wherein the processor is further configured to:
compute a feature of each CI, wherein the feature comprises one of: magnitude, magnitude square, function of magnitude, phase, magnitude of a CI component (component magnitude), component magnitude square, function of component magnitude, or phase of the CI component (component phase), of the CI, and
compute the value of any ACF based on the feature of each CI of the TSCI in the respective sliding time window associated with the ACF,
wherein, in each ACF vector, the values of the respective ACF evaluated at the respective time lags are arranged in an ascending order or a descending order of the respective time lags among the vector elements of the ACF vector.

4. The system of claim 3 , wherein:
each ACF of the plurality of ACF's is processed with a respective ACF filter to generate a respective filtered ACF, wherein both the ACF and the filtered ACF are univariate functions of time lag; and
each vector element of any ACF vector is a value of the respective filtered ACF evaluated at the respective time lag.

5. The system of claim 4 , wherein:
a common ACF filter is used to process all ACF of the plurality of ACF's;
in each ACF vector, the values of the respective filtered ACF evaluated at the respective time lags are arranged in an ascending order or a descending order of the respective time lags among the vector elements of the ACF vector.

6. The system of claim 5 , wherein:
the respective time lags associated with the vector elements are equally spaced, with a common time lag increment or a common time lag decrement.

7. The system of claim 1 , wherein the processor is further configured to:
reorganize an ACF vector in a sub-matrix individually towards the rearranged ACF data.

8. The system of claim 1 , wherein the processor is further configured to:
reorganize at least two ACF vectors in a sub-matrix jointly towards the rearranged ACF data.

9. The system of claim 1 , wherein the processor is further configured to:
reorganize multiple ACF vectors in a current matrix into the rearranged ACF data.

10. The system of claim 1 , wherein the rearranged ACF data comprises at least one of:
a 1D data structure, a sequence of 1D data structure, a time series of 11) data structure, an array of ID data structure, a collection of 1D data structure,
a 2D data structure, a sequence of 2D data structure, a time series of 21) data structure, an array of 2D data structure, a collection of 2D data structure,
a 3D data structure, a sequence of 3D data structure, a time series of 3D data structure, an array of 3D data structure, a collection of 3D data structure,
a k1-D data structure, a sequence of k1-D data structure, a time series of k1-D data structure, an array of k1-D data structure, a collection of k1-D data structure, wherein k1 is an integer,
or a combination of the above.

11. The system of claim 10 , wherein the processing comprises at least one of:
a neural network (NN) processing with the rearranged ACF data as input,
a feed-forward NN (FNN) processing with the rearranged ACF data as input in a form of 1D data structure, 2D data structure, 3D data structure, or k1-D data structure,
a convolutional NN (CNN) processing with the rearranged ACF data as input in a form of ID data structure, 2D data structure, 3D data structure, or k1-D data structure,
a recurrent NN (RNN) processing with the rearranged ACF data as input in a form of a sequence of 1D data structure, a sequence of 2D data structure, a sequence of 3D data structure, or a sequence of k1-D data structure,
a transformer NN processing with the rearranged ACF data as input in the form of a collection of ID data structure, a collection of 2D data structure, a collection of 3D data structure, or a collection of k1-D data structure.

12. The system of claim 11 , wherein the processor is further configured to:
obtain a time series of ACF (TSACF) computed by a second processor of a particular receiver based on a particular TSCI generated by the particular receiver, wherein each respective ACF is computed based on CI of the particular TSCI in a respective first sliding time window and is associated with a respective time stamp associated with the respective first sliding time window;
construct

Assignees

• Zhu Guozhen
• Wang Beibei
• Gao Weihang
• Hu Yuqian
• Wu Chenshu
• Zeng Xiaolu
• Liu K J Ray
• Au Oscar Chi Lim
• Origin Res Wireless Inc

Inventors

• Zhu Guozhen
• Wang Beibei
• Gao Weihang
• Hu Yuqian
• Wu Chenshu
• Zeng Xiaolu
• Liu K J Ray
• Au Oscar Chi-Lim

Classifications

• G01S13/56 Primary

  for presence detection {(presence detection using near field arrangements G01V3/00, e.g. G01V3/08, G01V3/12; burglar, theft or intruder alarms with electrical actuation G08B13/22 - G08B13/26)} · CPC title

• H04L5/0051

  of dedicated pilots, i.e. pilots destined for a single user or terminal · CPC title

• H04L27/2647

  Arrangements specific to the receiver only (equalisation H04L27/01) · CPC title

• H04L5/0048

  Allocation of pilot signals, i.e. of signals known to the receiver (allocation of control signalling H04L5/0053; use of control signalling H04L5/0091) · CPC title

• H04L5/0026

  Division using four or more dimensions, e.g. beam steering or quasi-co-location [QCL] · CPC title