Method, apparatus, and system for fall-down detection based on a wireless signal

Patent metadata

Field	Value
Publication number	US-2021173045-A1
Application number	US-202117180762-A
Country	US
Kind code	A1
Filing date	Feb 20, 2021
Priority date	Jul 17, 2015
Publication date	Jun 10, 2021
Grant date	—

Abstract

Methods, apparatus and systems for periodic or transient motion detection, e.g. fall event detection, based on wireless signals are described. In one example, a described system comprises: a transmitter configured for transmitting a first wireless signal through a wireless multipath channel of a venue; a receiver configured for receiving a second wireless signal through the wireless multipath channel; and a processor. The second wireless signal differs from the first wireless signal due to the wireless multipath channel that is impacted by a target motion of an object in the venue. The processor is configured for: obtaining a time series of channel information (TSCI) of the wireless multipath channel based on the second wireless signal, computing a time series of spatial-temporal information (STI) of the object based on the TSCI, and detecting the target motion of the object based on the time series of STI (TSSTI).

First claim

We claim:

1 . A system for target motion detection, comprising:
a transmitter configured for transmitting a first wireless signal through a wireless multipath channel of a venue; a receiver configured for receiving a second wireless signal through the wireless multipath channel, wherein the second wireless signal differs from the first wireless signal due to the wireless multipath channel that is impacted by a target motion of an object in the venue; and a processor configured for:
obtaining a time series of channel information (TSCI) of the wireless multipath channel based on the second wireless signal,
computing a time series of spatial-temporal information (STI) of the object based on the TSCI, and
detecting the target motion of the object based on the time series of STI (TSSTI).

2 . The system of claim 1 , wherein:
each STI is related to at least one of following features of the object: position change, distance change, speed or acceleration; and the target motion is a transient motion or a periodic motion.

3 . The system of claim 2 , wherein the processor is further configured for:
computing a time series of features (TSF) based on the TSCI, wherein the TSSTI is computed based on the TSF, wherein each feature of the TSF comprises at least one of following characteristics related to a corresponding channel information (CI): magnitude, phase, autocorrelation function (ACF), local maximum, or local minimum.

4 . The system of claim 3 , wherein:
the object is a human being; the target motion is a fall-down motion of the human being; a feature of the TSF comprises at least one of: a magnitude of a component of a CI, or an autocorrelation function of the TSCI; and each STI comprises the speed or the acceleration.

5 . The system of claim 2 , wherein the processor is further configured for:
computing, during an offline stage of the system, at least one representative STI template based on a set of training TSSTI, wherein each training TSSTI is computed based on a respective training TSCI obtained based on a respective training wireless signal from a respective training transmitter received by a respective training receiver in a respective training venue when a respective training object in the respective training venue undergoes a respective training target motion; and detecting, during an online stage of the system, the target motion of the object based on the TSSTI and the at least one representative STI template.

6 . The system of claim 5 , wherein the processor is further configured for:
cleaning the set of training TSSTI; and computing the at least one representative STI template based on the set of cleaned training TSSTI.

7 . The system of claim 6 , wherein the processor is further configured for:
cleaning each training TSSTI by segmenting the TSSTI into: a respective initial non-target segment of STI, a respective target segment of STI, and a respective trailing non-target segment of STI, wherein the target segment of STI corresponds to the respective training target motion of the respective training object.

8 . The system of claim 7 , wherein the TSSTI is segmented based on at least one of: a respective first constraint on the respective initial non-target segment, a respective second constraint on the respective target segment, or a respective third constraint on the respective trailing non-target segment.

9 . The system of claim 8 , wherein the processor is further configured for:
constraining the respective initial non-target segment to start from a beginning of the TSSTI and to have a first duration that is at least one of: less than a first threshold, or being a fraction of a duration of the TSSTI; constraining the respective target segment to have a second duration not greater than a target duration associated with the target motion of the object; or constraining the respective trailing non-target segment to end at an end of the TSSTI and to have a third duration being a fraction of the duration of the TSSTI.

10 . The system of claim 7 , wherein the processor is further configured for:
segmenting the training TSSTI based on a mapping between the training TSSTI and an additional training TSSTI with a mapping score between the two TSSTI associated with the mapping, wherein each STI of training TSSTI is mapped to at least one STI in the mapping, wherein the mapping score is a similarity score or a mismatch score.

11 . The system of claim 10 , wherein the processor is further configured for:
computing a plurality of candidate mappings, wherein each candidate mapping is between a candidate target segment of the training TSSTI and an additional candidate target segment of the additional training TSSTI; computing a plurality of mapping scores associated with the candidate mappings, wherein each mapping score associated with a candidate mapping is normalized based on a length associated with the candidate target segment and the additional candidate target segment; and computing the mapping based on the plurality of candidate mappings and the associated mapping scores.

12 . The system of claim 11 , wherein the processor is further configured for:
determining a particular candidate mapping associated with an optimal mapping score among all the candidate mappings, wherein the optimal mapping score comprises at least one of: a maximum similarity score or a minimum mismatch score; choosing the particular candidate mapping as the mapping between the two TSSTI; and choosing the associated optimal mapping score as the associated mapping score between the two TSSTI.

13 . The system of claim 11 , wherein the processor is further configured for:
computing iteratively the plurality of mapping scores; determining a series of partial mappings between subsets of the training TSSTI and subsets of the additional training TSSTI; and computing iteratively a quantity associated with each partial mapping, wherein each partial mapping is a mapping between a partial segment of the training TSSTI and a partial segment of the additional training TSSTI, wherein the series of partial mappings comprises the plurality of candidate mappings.

14 . The system of claim 13 , wherein the processor is further configured for:
in an initial iteration, initializing iteration by determining an initial partial mapping between a single STI of the training TSSTI and a single STI of the additional training TSSTI and computing an initial quantity associated with the initial partial mapping; in subsequent iterations, computing iteratively the quantity associated with each partial mapping based on a value associated with a partial mapping computed in a previous iteration, wherein the partial mapping is between a first segment of the training TSSTI and a second segment of the additional training TSSTI, and the additional partial mapping is between a third segment of the training TSSTI and a fourth segment of the additional TSSTI, wherein the third segment is a subset of the first segment, wherein the fourth segment is a subset of the second segment; and computing the plurality of mapping scores based on the quantities associated with the series of partial mappings.

15 . The system of claim 13 , wherein the processor is further configured for:
in an initial iteration, initializing iteration by determining a group of initial partial mappings comprising a mapping between a single STI of the training TSSTI and a single STI of the additional training TSSTI, and computing initial quantities associated with the group of initial partial mappings; in subsequent iterations, co

Assignees

• Hu Yuqian
• Zhang Feng
• Wang Beibei
• Wu Chenshu
• Liu K J Ray
• Au Oscar Chi Lim

Inventors

• Hu Yuqian
• Zhang Feng
• Wang Beibei
• Wu Chenshu
• Liu K J Ray
• Au Oscar Chi-Lim

Classifications

• G06N20/00

  Machine learning · CPC title

• G01S13/886

  for alarm systems (alarms with electrical actuation G08B13/22) · CPC title

• G01S7/415 Primary

  Identification of targets based on measurements of movement associated with the target · CPC title

• G01S13/765

  with exchange of information between interrogator and responder · CPC title

• G08B21/043 Primary

  detecting an emergency event, e.g. a fall · CPC title