RFID FOR HUMAN ACTIVITY SENSING - New York Chapter...Leverage received signal strength and phase information collected by the Low-Level Reader Protocol (LLRP) Near-field detection

Electrical and Computer Engineering

RFID FOR HUMAN ACTIVITY SENSING CHALLENGES, SOLUTIONS AND APPLICATIONS

Shiwen MaoAuburn University, Auburn, AL, USAJoint work with Chao Yang and Xuyu Wang

IEEE ComSoc DL Talk @ The IEEE ComSoc New York Chapter


Outline

Radio Frequency Identification (RFID) preliminaries

RFID based sensing: how it works

Challenges and solutions

Applications• Respiration monitoring indoors• Respiration monitoring in driving environments• Drowsy driving detection• Human skeleton monitoring

Image Source: https://www.indiamart.com/prithingatraders/rfid-tag-card.htmlImage Source: https://www.sparkfun.com/products/14147

2


Overview of an RFID System

3

RFID ReaderRFID Tags Data Processing System

Image Source: http://www.passive-rfid-tags.com/52/i-995.htmlImage Source: https://www.britannica.com/technology/computer


A Brief History

4

1920s

1930s-1950s

1960s

1970s

1980s-1990s

2000s Foundation stablished. Radar was developed.

RFID prototype is invented for distinguishing enemy

aircraft.

Applying RFID device to markets beyond the

military.

More institutions are involved in

developing RFID.

RFID is widely used by consumers and companies

globally.

System miniaturization. Cost of RFID continues to fall.

RadarRFID was invented, UK: distinguish enemy aircraft in WWII

Electronic Article Surveillance, “1-bit tags”

eToll, animal and vehicle tracking, and factory automation

Transportation, retail, supply chain

Miniaturization, low-cost, authentication


Stored Data based Applications

5

• Person identification

• Vehicle parking monitoring

• Fast-lane and E-Zpass road toll system

• Secure entry cards

• Supply chain management

• Food production control

Image Source: https://medicalfuturist.com/rfid-implant-chip/Image Source: https://www.wikiwand.com/en/Electronic_Product_CodeImage Source: https://pilotonline.com/news/local/transportation/article_62a3b00e-64fb-11e8-88d9-5fbb5a27dbe8.html

An EPC RFID tag used by Wal-MartElectronic Product Code (EPC): a universal identifier providing a unique identity for every physical object anywhere in the world

Communications è deliver stored data


RF Sensing Applications

6

• Indoor localization

• Temperature estimation

• Gesture recognition

• Vital sign monitoring

• Driving fatigue detection

Leverage received signal strength and phase information collected by the Low-Level Reader Protocol (LLRP)

Near-field detection system

Low-cost passive RFID tags

Low sampling rate

Frequency hopping

Mutual coupling

Characteristics:

Wireless Channel è RF phase angle, Doppler frequency, and Peak RSSI


Outline






7


Vital Sign Monitoring with RFID

d=D＋A

d=D➖

A

Phase:

8

' = mod

✓2⇡d

�+ ↵T + ↵R + ↵Tag, 2⇡

◆

<latexit sha1_base64="/yB+CNg9D4k4/Evel+R/CCu66xU=">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</latexit>

Translation from phase to distance d𝜑


Outline






9


'(fi, d) = mod

✓2⇡fid

c+ ↵i, 2⇡

◆

<latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">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</latexit><latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">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</latexit><latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">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</latexit><latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">AAACOnicbVBNSxxBFOzRJOqqcdWjl0eWwEpEZiQQQQSJF48KWRW2l+VNT89OY89M0/1GWIb9XV78Fd5yyCUHRbz6A9L7cfAjBQ1FVT1ev4qNVo7C8HcwN//h46eFxaXG8srq57Xm+sa5KysrZEeUurSXMTqpVSE7pEjLS2Ml5rGWF/HV8di/uJbWqbL4RUMjezkOCpUqgeSlfvOMX6M1mYJ22lc7kGzDIfC8TIBrmVIbeGpR1HvAjQKfgGRUixF8A47aZDge4QcwtblVg4y2+81WuBtOAO9JNCMtNsNpv3nHk1JUuSxIaHSuG4WGejVaUkLLUYNXThoUVziQXU8LzKXr1ZPTR/DVKwmkpfWvIJioLydqzJ0b5rFP5kiZe+uNxf953YrS/V6tClORLMR0UVppoBLGPUKirBSkh56gsMr/FUSGvizybTd8CdHbk9+T873dyPOz762jn7M6FtkW+8LaLGI/2BE7YaeswwS7YX/YPXsIboO/wWPwNI3OBbOZTfYKwfM/PqqqGA==</latexit>

Challenge 1: Channel Hopping

Channel i on frequency fi :

Raw phase data

Channel index in a period of 30 seconds

Initial Phase Offset

10


Channel Hopping Effect Mitigation

Raw phase data

'(fi, d) = mod

✓2⇡fid

c+ ↵i, 2⇡

◆

<latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">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</latexit><latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">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</latexit><latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">AAACOnicbVBNSxxBFOzRJOqqcdWjl0eWwEpEZiQQQQSJF48KWRW2l+VNT89OY89M0/1GWIb9XV78Fd5yyCUHRbz6A9L7cfAjBQ1FVT1ev4qNVo7C8HcwN//h46eFxaXG8srq57Xm+sa5KysrZEeUurSXMTqpVSE7pEjLS2Ml5rGWF/HV8di/uJbWqbL4RUMjezkOCpUqgeSlfvOMX6M1mYJ22lc7kGzDIfC8TIBrmVIbeGpR1HvAjQKfgGRUixF8A47aZDge4QcwtblVg4y2+81WuBtOAO9JNCMtNsNpv3nHk1JUuSxIaHSuG4WGejVaUkLLUYNXThoUVziQXU8LzKXr1ZPTR/DVKwmkpfWvIJioLydqzJ0b5rFP5kiZe+uNxf953YrS/V6tClORLMR0UVppoBLGPUKirBSkh56gsMr/FUSGvizybTd8CdHbk9+T873dyPOz762jn7M6FtkW+8LaLGI/2BE7YaeswwS7YX/YPXsIboO/wWPwNI3OBbOZTfYKwfM/PqqqGA==</latexit><latexit sha1_base64="EBvUOVNi1YBvQhI99J7Oe7w/z1I=">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</latexit>

'(fi, d) =2⇡fid

c+ ↵i

<latexit sha1_base64="0B3nzAERZMGOdVYo93EmoGTyFdw=">AAACHHicbZBNS8NAEIY3ftb6VfXoZbAIiiJJFfQiFL14rGBtoSlhstnYxU2y7G4KJfSHePGvePGgiBcPgv/Gbe3BrxcWHt6ZYXbeUAqujet+OFPTM7Nz86WF8uLS8spqZW39Wme5oqxJM5GpdoiaCZ6ypuFGsLZUDJNQsFZ4ez6qt/pMaZ6lV2YgWTfBm5THnKKxVlA59PuoZI/DThzwfYh24RT8WCEtauBLDtaFaFjQIeyBj0L2MOBBpeoeuGPBX/AmUCUTNYLKmx9lNE9YaqhArTueK023QGU4FWxY9nPNJNJbvGEdiykmTHeL8XFD2LZOBHGm7EsNjN3vEwUmWg+S0HYmaHr6d21k/lfr5CY+6RY8lblhKf1aFOcCTAajpCDiilEjBhaQKm7/CrSHNhpj8yzbELzfJ/+F69qBZ/nyqFo/m8RRIptki+wQjxyTOrkgDdIklNyRB/JEnp1759F5cV6/WqecycwG+SHn/RMwS5+U</latexit><latexit sha1_base64="0B3nzAERZMGOdVYo93EmoGTyFdw=">AAACHHicbZBNS8NAEIY3ftb6VfXoZbAIiiJJFfQiFL14rGBtoSlhstnYxU2y7G4KJfSHePGvePGgiBcPgv/Gbe3BrxcWHt6ZYXbeUAqujet+OFPTM7Nz86WF8uLS8spqZW39Wme5oqxJM5GpdoiaCZ6ypuFGsLZUDJNQsFZ4ez6qt/pMaZ6lV2YgWTfBm5THnKKxVlA59PuoZI/DThzwfYh24RT8WCEtauBLDtaFaFjQIeyBj0L2MOBBpeoeuGPBX/AmUCUTNYLKmx9lNE9YaqhArTueK023QGU4FWxY9nPNJNJbvGEdiykmTHeL8XFD2LZOBHGm7EsNjN3vEwUmWg+S0HYmaHr6d21k/lfr5CY+6RY8lblhKf1aFOcCTAajpCDiilEjBhaQKm7/CrSHNhpj8yzbELzfJ/+F69qBZ/nyqFo/m8RRIptki+wQjxyTOrkgDdIklNyRB/JEnp1759F5cV6/WqecycwG+SHn/RMwS5+U</latexit><latexit sha1_base64="0B3nzAERZMGOdVYo93EmoGTyFdw=">AAACHHicbZBNS8NAEIY3ftb6VfXoZbAIiiJJFfQiFL14rGBtoSlhstnYxU2y7G4KJfSHePGvePGgiBcPgv/Gbe3BrxcWHt6ZYXbeUAqujet+OFPTM7Nz86WF8uLS8spqZW39Wme5oqxJM5GpdoiaCZ6ypuFGsLZUDJNQsFZ4ez6qt/pMaZ6lV2YgWTfBm5THnKKxVlA59PuoZI/DThzwfYh24RT8WCEtauBLDtaFaFjQIeyBj0L2MOBBpeoeuGPBX/AmUCUTNYLKmx9lNE9YaqhArTueK023QGU4FWxY9nPNJNJbvGEdiykmTHeL8XFD2LZOBHGm7EsNjN3vEwUmWg+S0HYmaHr6d21k/lfr5CY+6RY8lblhKf1aFOcCTAajpCDiilEjBhaQKm7/CrSHNhpj8yzbELzfJ/+F69qBZ/nyqFo/m8RRIptki+wQjxyTOrkgDdIklNyRB/JEnp1759F5cV6/WqecycwG+SHn/RMwS5+U</latexit><latexit sha1_base64="0B3nzAERZMGOdVYo93EmoGTyFdw=">AAACHHicbZBNS8NAEIY3ftb6VfXoZbAIiiJJFfQiFL14rGBtoSlhstnYxU2y7G4KJfSHePGvePGgiBcPgv/Gbe3BrxcWHt6ZYXbeUAqujet+OFPTM7Nz86WF8uLS8spqZW39Wme5oqxJM5GpdoiaCZ6ypuFGsLZUDJNQsFZ4ez6qt/pMaZ6lV2YgWTfBm5THnKKxVlA59PuoZI/DThzwfYh24RT8WCEtauBLDtaFaFjQIeyBj0L2MOBBpeoeuGPBX/AmUCUTNYLKmx9lNE9YaqhArTueK023QGU4FWxY9nPNJNJbvGEdiykmTHeL8XFD2LZOBHGm7EsNjN3vEwUmWg+S0HYmaHr6d21k/lfr5CY+6RY8lblhKf1aFOcCTAajpCDiilEjBhaQKm7/CrSHNhpj8yzbELzfJ/+F69qBZ/nyqFo/m8RRIptki+wQjxyTOrkgDdIklNyRB/JEnp1759F5cV6/WqecycwG+SHn/RMwS5+U</latexit>

'(fi+1, d) =2⇡fi+1d

c+ ↵i+1

<latexit sha1_base64="x9uyHcporzDTyp0P3yUNAK0SYh0=">AAACKHicbZBNS8NAEIY3flu/qh69DBahUpGkCHoRRS8eFWwVmlImm41d3CTL7kYoIT/Hi3/Fi4giXv0lbtsc1Dqw8PC+M8zOG0jBtXHdT2dqemZ2bn5hsbK0vLK6Vl3faOs0U5S1aCpSdRugZoInrGW4EexWKoZxINhNcH8+9G8emNI8Ta7NQLJujHcJjzhFY6Ve9cR/QCX7HOpRL+cNr9iDcBeOwY8U0rwJvuRQOhAWOS2gAT4K2cex2KvW3H13VDAJXgk1UtZlr/rqhynNYpYYKlDrjudK081RGU4FKyp+pplEeo93rGMxwZjpbj46tIAdq4QQpcq+xMBI/TmRY6z1IA5sZ4ymr/96Q/E/r5OZ6Kib80RmhiV0vCjKBJgUhqlByBWjRgwsIFXc/hVoH21ExmZbsSF4f0+ehHZz37N8dVA7PSvjWCBbZJvUiUcOySm5IJekRSh5JM/kjbw7T86L8+F8jlunnHJmk/wq5+sb3X2kCA==</latexit><latexit sha1_base64="x9uyHcporzDTyp0P3yUNAK0SYh0=">AAACKHicbZBNS8NAEIY3flu/qh69DBahUpGkCHoRRS8eFWwVmlImm41d3CTL7kYoIT/Hi3/Fi4giXv0lbtsc1Dqw8PC+M8zOG0jBtXHdT2dqemZ2bn5hsbK0vLK6Vl3faOs0U5S1aCpSdRugZoInrGW4EexWKoZxINhNcH8+9G8emNI8Ta7NQLJujHcJjzhFY6Ve9cR/QCX7HOpRL+cNr9iDcBeOwY8U0rwJvuRQOhAWOS2gAT4K2cex2KvW3H13VDAJXgk1UtZlr/rqhynNYpYYKlDrjudK081RGU4FKyp+pplEeo93rGMxwZjpbj46tIAdq4QQpcq+xMBI/TmRY6z1IA5sZ4ymr/96Q/E/r5OZ6Kib80RmhiV0vCjKBJgUhqlByBWjRgwsIFXc/hVoH21ExmZbsSF4f0+ehHZz37N8dVA7PSvjWCBbZJvUiUcOySm5IJekRSh5JM/kjbw7T86L8+F8jlunnHJmk/wq5+sb3X2kCA==</latexit><latexit sha1_base64="x9uyHcporzDTyp0P3yUNAK0SYh0=">AAACKHicbZBNS8NAEIY3flu/qh69DBahUpGkCHoRRS8eFWwVmlImm41d3CTL7kYoIT/Hi3/Fi4giXv0lbtsc1Dqw8PC+M8zOG0jBtXHdT2dqemZ2bn5hsbK0vLK6Vl3faOs0U5S1aCpSdRugZoInrGW4EexWKoZxINhNcH8+9G8emNI8Ta7NQLJujHcJjzhFY6Ve9cR/QCX7HOpRL+cNr9iDcBeOwY8U0rwJvuRQOhAWOS2gAT4K2cex2KvW3H13VDAJXgk1UtZlr/rqhynNYpYYKlDrjudK081RGU4FKyp+pplEeo93rGMxwZjpbj46tIAdq4QQpcq+xMBI/TmRY6z1IA5sZ4ymr/96Q/E/r5OZ6Kib80RmhiV0vCjKBJgUhqlByBWjRgwsIFXc/hVoH21ExmZbsSF4f0+ehHZz37N8dVA7PSvjWCBbZJvUiUcOySm5IJekRSh5JM/kjbw7T86L8+F8jlunnHJmk/wq5+sb3X2kCA==</latexit><latexit sha1_base64="x9uyHcporzDTyp0P3yUNAK0SYh0=">AAACKHicbZBNS8NAEIY3flu/qh69DBahUpGkCHoRRS8eFWwVmlImm41d3CTL7kYoIT/Hi3/Fi4giXv0lbtsc1Dqw8PC+M8zOG0jBtXHdT2dqemZ2bn5hsbK0vLK6Vl3faOs0U5S1aCpSdRugZoInrGW4EexWKoZxINhNcH8+9G8emNI8Ta7NQLJujHcJjzhFY6Ve9cR/QCX7HOpRL+cNr9iDcBeOwY8U0rwJvuRQOhAWOS2gAT4K2cex2KvW3H13VDAJXgk1UtZlr/rqhynNYpYYKlDrjudK081RGU4FKyp+pplEeo93rGMxwZjpbj46tIAdq4QQpcq+xMBI/TmRY6z1IA5sZ4ymr/96Q/E/r5OZ6Kib80RmhiV0vCjKBJgUhqlByBWjRgwsIFXc/hVoH21ExmZbsSF4f0+ehHZz37N8dVA7PSvjWCBbZJvUiUcOySm5IJekRSh5JM/kjbw7T86L8+F8jlunnHJmk/wq5+sb3X2kCA==</latexit>

Multiply the data in new channel fi by fi/fi+1<latexit sha1_base64="HtUyy81vuI4yuFdo9luwBNnbi6s=">AAAB8nicbZBNS8NAEIYn9avWr6pHL8EiCEJNRNBj0YvHCtYW0lA22027dLMbdidCCf0ZXjwo4tVf481/47bNQVtfWHh4Z4adeaNUcIOe9+2UVlbX1jfKm5Wt7Z3dver+waNRmaasRZVQuhMRwwSXrIUcBeukmpEkEqwdjW6n9fYT04Yr+YDjlIUJGUgec0rQWkHc4+dxL+dn/qRXrXl1byZ3GfwCalCo2at+dfuKZgmTSAUxJvC9FMOcaORUsEmlmxmWEjoiAxZYlCRhJsxnK0/cE+v03Vhp+yS6M/f3RE4SY8ZJZDsTgkOzWJua/9WCDOPrMOcyzZBJOv8ozoSLyp3e7/a5ZhTF2AKhmttdXTokmlC0KVVsCP7iycvweFH3Ld9f1ho3RRxlOIJjOAUfrqABd9CEFlBQ8Ayv8Oag8+K8Ox/z1pJTzBzCHzmfP5PTkMc=</latexit><latexit sha1_base64="HtUyy81vuI4yuFdo9luwBNnbi6s=">AAAB8nicbZBNS8NAEIYn9avWr6pHL8EiCEJNRNBj0YvHCtYW0lA22027dLMbdidCCf0ZXjwo4tVf481/47bNQVtfWHh4Z4adeaNUcIOe9+2UVlbX1jfKm5Wt7Z3dver+waNRmaasRZVQuhMRwwSXrIUcBeukmpEkEqwdjW6n9fYT04Yr+YDjlIUJGUgec0rQWkHc4+dxL+dn/qRXrXl1byZ3GfwCalCo2at+dfuKZgmTSAUxJvC9FMOcaORUsEmlmxmWEjoiAxZYlCRhJsxnK0/cE+v03Vhp+yS6M/f3RE4SY8ZJZDsTgkOzWJua/9WCDOPrMOcyzZBJOv8ozoSLyp3e7/a5ZhTF2AKhmttdXTokmlC0KVVsCP7iycvweFH3Ld9f1ho3RRxlOIJjOAUfrqABd9CEFlBQ8Ayv8Oag8+K8Ox/z1pJTzBzCHzmfP5PTkMc=</latexit><latexit sha1_base64="HtUyy81vuI4yuFdo9luwBNnbi6s=">AAAB8nicbZBNS8NAEIYn9avWr6pHL8EiCEJNRNBj0YvHCtYW0lA22027dLMbdidCCf0ZXjwo4tVf481/47bNQVtfWHh4Z4adeaNUcIOe9+2UVlbX1jfKm5Wt7Z3dver+waNRmaasRZVQuhMRwwSXrIUcBeukmpEkEqwdjW6n9fYT04Yr+YDjlIUJGUgec0rQWkHc4+dxL+dn/qRXrXl1byZ3GfwCalCo2at+dfuKZgmTSAUxJvC9FMOcaORUsEmlmxmWEjoiAxZYlCRhJsxnK0/cE+v03Vhp+yS6M/f3RE4SY8ZJZDsTgkOzWJua/9WCDOPrMOcyzZBJOv8ozoSLyp3e7/a5ZhTF2AKhmttdXTokmlC0KVVsCP7iycvweFH3Ld9f1ho3RRxlOIJjOAUfrqABd9CEFlBQ8Ayv8Oag8+K8Ox/z1pJTzBzCHzmfP5PTkMc=</latexit><latexit sha1_base64="HtUyy81vuI4yuFdo9luwBNnbi6s=">AAAB8nicbZBNS8NAEIYn9avWr6pHL8EiCEJNRNBj0YvHCtYW0lA22027dLMbdidCCf0ZXjwo4tVf481/47bNQVtfWHh4Z4adeaNUcIOe9+2UVlbX1jfKm5Wt7Z3dver+waNRmaasRZVQuhMRwwSXrIUcBeukmpEkEqwdjW6n9fYT04Yr+YDjlIUJGUgec0rQWkHc4+dxL+dn/qRXrXl1byZ3GfwCalCo2at+dfuKZgmTSAUxJvC9FMOcaORUsEmlmxmWEjoiAxZYlCRhJsxnK0/cE+v03Vhp+yS6M/f3RE4SY8ZJZDsTgkOzWJua/9WCDOPrMOcyzZBJOv8ozoSLyp3e7/a5ZhTF2AKhmttdXTokmlC0KVVsCP7iycvweFH3Ld9f1ho3RRxlOIJjOAUfrqABd9CEFlBQ8Ayv8Oag8+K8Ox/z1pJTzBzCHzmfP5PTkMc=</latexit>

'(fi+1, d)fi

fi+1=

2⇡fid

c+ ↵i+1 ·

fifi+1

<latexit sha1_base64="j3BzrAUewNVwBjLFPQbo/kDSvTw=">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</latexit><latexit sha1_base64="j3BzrAUewNVwBjLFPQbo/kDSvTw=">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</latexit><latexit sha1_base64="j3BzrAUewNVwBjLFPQbo/kDSvTw=">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</latexit><latexit sha1_base64="j3BzrAUewNVwBjLFPQbo/kDSvTw=">AAACVHicbVFNS8NAEN2kftT6VfXoZbAIiiJJEfQiiF48KlgVmhImm41dukmW3Y1QQn6kHgR/iRcPbtuAnwMLj/fezOy+jaTg2njem+M25uYXFptLreWV1bX19sbmnc4LRVmP5iJXDxFqJnjGeoYbwR6kYphGgt1Ho8uJfv/ElOZ5dmvGkg1SfMx4wikaS4XtUfCESg457CVhyQ/86hDifQgShbRMQl6VNV3BWc12IZAcrAZxVdIKDiBAIYc480FA49x8DYBvE8J2xzvypgV/gV+DDqnrOmy/BHFOi5RlhgrUuu970gxKVIZTwapWUGgmkY7wkfUtzDBlelBOQ6lg1zIxJLmyJzMwZb93lJhqPU4j60zRDPVvbUL+p/ULk5wOSp7JwrCMzhYlhQCTwyRhiLli1IixBUgVt3cFOkSbh7H/0LIh+L+f/BfcdY98i2+OO+cXdRxNsk12yB7xyQk5J1fkmvQIJc/k3SGO47w6H27DnZ9ZXafu2SI/yl37BPJAsUc=</latexit>

'(fi+1, d)fi

fi+1=

2⇡fid

c+ ↵i +�↵i

<latexit sha1_base64="S5J1A6WqeBMssKW5JM9VqNj1ptg=">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</latexit><latexit sha1_base64="S5J1A6WqeBMssKW5JM9VqNj1ptg=">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</latexit><latexit sha1_base64="S5J1A6WqeBMssKW5JM9VqNj1ptg=">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</latexit><latexit sha1_base64="S5J1A6WqeBMssKW5JM9VqNj1ptg=">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</latexit>

11


Calibrated phase

Channel Hopping Effect Mitigation (cont’d)

12

Channel i

Channel i+1

Convert to Channel i

Time

Data Num=

Data Num=

Ni<latexit sha1_base64="AEFtVY9kPOHarXUEtmXq3BbM6Ls=">AAAB6nicbZBNS8NAEIYn9avWr6pHL4tF8FQSEeqx6MWTVLQf0Iay2U7apZtN2N0IJfQnePGgiFd/kTf/jds2B219YeHhnRl25g0SwbVx3W+nsLa+sblV3C7t7O7tH5QPj1o6ThXDJotFrDoB1Si4xKbhRmAnUUijQGA7GN/M6u0nVJrH8tFMEvQjOpQ85Iwaaz3c9Xm/XHGr7lxkFbwcKpCr0S9/9QYxSyOUhgmqdddzE+NnVBnOBE5LvVRjQtmYDrFrUdIItZ/NV52SM+sMSBgr+6Qhc/f3REYjrSdRYDsjakZ6uTYz/6t1UxNe+RmXSWpQssVHYSqIicnsbjLgCpkREwuUKW53JWxEFWXGplOyIXjLJ69C66LqWb6/rNSv8ziKcAKncA4e1KAOt9CAJjAYwjO8wpsjnBfn3flYtBacfOYY/sj5/AEglo2u</latexit><latexit sha1_base64="AEFtVY9kPOHarXUEtmXq3BbM6Ls=">AAAB6nicbZBNS8NAEIYn9avWr6pHL4tF8FQSEeqx6MWTVLQf0Iay2U7apZtN2N0IJfQnePGgiFd/kTf/jds2B219YeHhnRl25g0SwbVx3W+nsLa+sblV3C7t7O7tH5QPj1o6ThXDJotFrDoB1Si4xKbhRmAnUUijQGA7GN/M6u0nVJrH8tFMEvQjOpQ85Iwaaz3c9Xm/XHGr7lxkFbwcKpCr0S9/9QYxSyOUhgmqdddzE+NnVBnOBE5LvVRjQtmYDrFrUdIItZ/NV52SM+sMSBgr+6Qhc/f3REYjrSdRYDsjakZ6uTYz/6t1UxNe+RmXSWpQssVHYSqIicnsbjLgCpkREwuUKW53JWxEFWXGplOyIXjLJ69C66LqWb6/rNSv8ziKcAKncA4e1KAOt9CAJjAYwjO8wpsjnBfn3flYtBacfOYY/sj5/AEglo2u</latexit><latexit sha1_base64="AEFtVY9kPOHarXUEtmXq3BbM6Ls=">AAAB6nicbZBNS8NAEIYn9avWr6pHL4tF8FQSEeqx6MWTVLQf0Iay2U7apZtN2N0IJfQnePGgiFd/kTf/jds2B219YeHhnRl25g0SwbVx3W+nsLa+sblV3C7t7O7tH5QPj1o6ThXDJotFrDoB1Si4xKbhRmAnUUijQGA7GN/M6u0nVJrH8tFMEvQjOpQ85Iwaaz3c9Xm/XHGr7lxkFbwcKpCr0S9/9QYxSyOUhgmqdddzE+NnVBnOBE5LvVRjQtmYDrFrUdIItZ/NV52SM+sMSBgr+6Qhc/f3REYjrSdRYDsjakZ6uTYz/6t1UxNe+RmXSWpQssVHYSqIicnsbjLgCpkREwuUKW53JWxEFWXGplOyIXjLJ69C66LqWb6/rNSv8ziKcAKncA4e1KAOt9CAJjAYwjO8wpsjnBfn3flYtBacfOYY/sj5/AEglo2u</latexit><latexit sha1_base64="AEFtVY9kPOHarXUEtmXq3BbM6Ls=">AAAB6nicbZBNS8NAEIYn9avWr6pHL4tF8FQSEeqx6MWTVLQf0Iay2U7apZtN2N0IJfQnePGgiFd/kTf/jds2B219YeHhnRl25g0SwbVx3W+nsLa+sblV3C7t7O7tH5QPj1o6ThXDJotFrDoB1Si4xKbhRmAnUUijQGA7GN/M6u0nVJrH8tFMEvQjOpQ85Iwaaz3c9Xm/XHGr7lxkFbwcKpCr0S9/9QYxSyOUhgmqdddzE+NnVBnOBE5LvVRjQtmYDrFrUdIItZ/NV52SM+sMSBgr+6Qhc/f3REYjrSdRYDsjakZ6uTYz/6t1UxNe+RmXSWpQssVHYSqIicnsbjLgCpkREwuUKW53JWxEFWXGplOyIXjLJ69C66LqWb6/rNSv8ziKcAKncA4e1KAOt9CAJjAYwjO8wpsjnBfn3flYtBacfOYY/sj5/AEglo2u</latexit>

fifi+1

<latexit sha1_base64="QTX/bo4J3uswzjNsgTqj7KUgih4=">AAAB/HicbZDLSsNAFIZP6q3WW7RLN4NFEISSiKDLohuXFewF2hAm00k7dDIJMxMhhPgqblwo4tYHcefbOG2z0NYDM3z8/znMmT9IOFPacb6tytr6xuZWdbu2s7u3f2AfHnVVnEpCOyTmsewHWFHOBO1opjntJ5LiKOC0F0xvZ37vkUrFYvGgs4R6ER4LFjKCtZF8uz4MJSZ56LPCXDk7d4vCtxtO05kXWgW3hAaU1fbtr+EoJmlEhSYcKzVwnUR7OZaaEU6L2jBVNMFkisd0YFDgiCovny9foFOjjFAYS3OERnP190SOI6WyKDCdEdYTtezNxP+8QarDay9nIkk1FWTxUJhypGM0SwKNmKRE88wAJpKZXRGZYJOGNnnVTAju8pdXoXvRdA3fXzZaN2UcVTiGEzgDF66gBXfQhg4QyOAZXuHNerJerHfrY9FascqZOvwp6/MHCmCVAQ==</latexit><latexit sha1_base64="QTX/bo4J3uswzjNsgTqj7KUgih4=">AAAB/HicbZDLSsNAFIZP6q3WW7RLN4NFEISSiKDLohuXFewF2hAm00k7dDIJMxMhhPgqblwo4tYHcefbOG2z0NYDM3z8/znMmT9IOFPacb6tytr6xuZWdbu2s7u3f2AfHnVVnEpCOyTmsewHWFHOBO1opjntJ5LiKOC0F0xvZ37vkUrFYvGgs4R6ER4LFjKCtZF8uz4MJSZ56LPCXDk7d4vCtxtO05kXWgW3hAaU1fbtr+EoJmlEhSYcKzVwnUR7OZaaEU6L2jBVNMFkisd0YFDgiCovny9foFOjjFAYS3OERnP190SOI6WyKDCdEdYTtezNxP+8QarDay9nIkk1FWTxUJhypGM0SwKNmKRE88wAJpKZXRGZYJOGNnnVTAju8pdXoXvRdA3fXzZaN2UcVTiGEzgDF66gBXfQhg4QyOAZXuHNerJerHfrY9FascqZOvwp6/MHCmCVAQ==</latexit><latexit sha1_base64="QTX/bo4J3uswzjNsgTqj7KUgih4=">AAAB/HicbZDLSsNAFIZP6q3WW7RLN4NFEISSiKDLohuXFewF2hAm00k7dDIJMxMhhPgqblwo4tYHcefbOG2z0NYDM3z8/znMmT9IOFPacb6tytr6xuZWdbu2s7u3f2AfHnVVnEpCOyTmsewHWFHOBO1opjntJ5LiKOC0F0xvZ37vkUrFYvGgs4R6ER4LFjKCtZF8uz4MJSZ56LPCXDk7d4vCtxtO05kXWgW3hAaU1fbtr+EoJmlEhSYcKzVwnUR7OZaaEU6L2jBVNMFkisd0YFDgiCovny9foFOjjFAYS3OERnP190SOI6WyKDCdEdYTtezNxP+8QarDay9nIkk1FWTxUJhypGM0SwKNmKRE88wAJpKZXRGZYJOGNnnVTAju8pdXoXvRdA3fXzZaN2UcVTiGEzgDF66gBXfQhg4QyOAZXuHNerJerHfrY9FascqZOvwp6/MHCmCVAQ==</latexit><latexit sha1_base64="QTX/bo4J3uswzjNsgTqj7KUgih4=">AAAB/HicbZDLSsNAFIZP6q3WW7RLN4NFEISSiKDLohuXFewF2hAm00k7dDIJMxMhhPgqblwo4tYHcefbOG2z0NYDM3z8/znMmT9IOFPacb6tytr6xuZWdbu2s7u3f2AfHnVVnEpCOyTmsewHWFHOBO1opjntJ5LiKOC0F0xvZ37vkUrFYvGgs4R6ER4LFjKCtZF8uz4MJSZ56LPCXDk7d4vCtxtO05kXWgW3hAaU1fbtr+EoJmlEhSYcKzVwnUR7OZaaEU6L2jBVNMFkisd0YFDgiCovny9foFOjjFAYS3OERnP190SOI6WyKDCdEdYTtezNxP+8QarDay9nIkk1FWTxUJhypGM0SwKNmKRE88wAJpKZXRGZYJOGNnnVTAju8pdXoXvRdA3fXzZaN2UcVTiGEzgDF66gBXfQhg4QyOAZXuHNerJerHfrY9FascqZOvwp6/MHCmCVAQ==</latexit>

Ni+1<latexit sha1_base64="TQyP5Io1C9Ky6SApEDjGXBCTyZ8=">AAAB7nicbZBNS8NAEIYn9avWr6pHL4tFEISSSEGPRS+epIL9gDaUzXbSLt1swu5GKKE/wosHRbz6e7z5b9y2OWjrCwsP78ywM2+QCK6N6347hbX1jc2t4nZpZ3dv/6B8eNTScaoYNlksYtUJqEbBJTYNNwI7iUIaBQLbwfh2Vm8/odI8lo9mkqAf0aHkIWfUWKt938/4hTftlytu1Z2LrIKXQwVyNfrlr94gZmmE0jBBte56bmL8jCrDmcBpqZdqTCgb0yF2LUoaofaz+bpTcmadAQljZZ80ZO7+nshopPUkCmxnRM1IL9dm5n+1bmrCaz/jMkkNSrb4KEwFMTGZ3U4GXCEzYmKBMsXtroSNqKLM2IRKNgRv+eRVaF1WPcsPtUr9Jo+jCCdwCufgwRXU4Q4a0AQGY3iGV3hzEufFeXc+Fq0FJ585hj9yPn8AvhKPKg==</latexit><latexit sha1_base64="TQyP5Io1C9Ky6SApEDjGXBCTyZ8=">AAAB7nicbZBNS8NAEIYn9avWr6pHL4tFEISSSEGPRS+epIL9gDaUzXbSLt1swu5GKKE/wosHRbz6e7z5b9y2OWjrCwsP78ywM2+QCK6N6347hbX1jc2t4nZpZ3dv/6B8eNTScaoYNlksYtUJqEbBJTYNNwI7iUIaBQLbwfh2Vm8/odI8lo9mkqAf0aHkIWfUWKt938/4hTftlytu1Z2LrIKXQwVyNfrlr94gZmmE0jBBte56bmL8jCrDmcBpqZdqTCgb0yF2LUoaofaz+bpTcmadAQljZZ80ZO7+nshopPUkCmxnRM1IL9dm5n+1bmrCaz/jMkkNSrb4KEwFMTGZ3U4GXCEzYmKBMsXtroSNqKLM2IRKNgRv+eRVaF1WPcsPtUr9Jo+jCCdwCufgwRXU4Q4a0AQGY3iGV3hzEufFeXc+Fq0FJ585hj9yPn8AvhKPKg==</latexit><latexit sha1_base64="TQyP5Io1C9Ky6SApEDjGXBCTyZ8=">AAAB7nicbZBNS8NAEIYn9avWr6pHL4tFEISSSEGPRS+epIL9gDaUzXbSLt1swu5GKKE/wosHRbz6e7z5b9y2OWjrCwsP78ywM2+QCK6N6347hbX1jc2t4nZpZ3dv/6B8eNTScaoYNlksYtUJqEbBJTYNNwI7iUIaBQLbwfh2Vm8/odI8lo9mkqAf0aHkIWfUWKt938/4hTftlytu1Z2LrIKXQwVyNfrlr94gZmmE0jBBte56bmL8jCrDmcBpqZdqTCgb0yF2LUoaofaz+bpTcmadAQljZZ80ZO7+nshopPUkCmxnRM1IL9dm5n+1bmrCaz/jMkkNSrb4KEwFMTGZ3U4GXCEzYmKBMsXtroSNqKLM2IRKNgRv+eRVaF1WPcsPtUr9Jo+jCCdwCufgwRXU4Q4a0AQGY3iGV3hzEufFeXc+Fq0FJ585hj9yPn8AvhKPKg==</latexit><latexit sha1_base64="TQyP5Io1C9Ky6SApEDjGXBCTyZ8=">AAAB7nicbZBNS8NAEIYn9avWr6pHL4tFEISSSEGPRS+epIL9gDaUzXbSLt1swu5GKKE/wosHRbz6e7z5b9y2OWjrCwsP78ywM2+QCK6N6347hbX1jc2t4nZpZ3dv/6B8eNTScaoYNlksYtUJqEbBJTYNNwI7iUIaBQLbwfh2Vm8/odI8lo9mkqAf0aHkIWfUWKt938/4hTftlytu1Z2LrIKXQwVyNfrlr94gZmmE0jBBte56bmL8jCrDmcBpqZdqTCgb0yF2LUoaofaz+bpTcmadAQljZZ80ZO7+nshopPUkCmxnRM1IL9dm5n+1bmrCaz/jMkkNSrb4KEwFMTGZ3U4GXCEzYmKBMsXtroSNqKLM2IRKNgRv+eRVaF1WPcsPtUr9Jo+jCCdwCufgwRXU4Q4a0AQGY3iGV3hzEufFeXc+Fq0FJ585hj9yPn8AvhKPKg==</latexit>

�↵i<latexit sha1_base64="4JPh39dwkaKh9jKa2jKEQ5mhGAI=">AAAB+XicbZDLSsNAFIYn9VbrLerSzWARXJVEBF0WdeGygr1AE8LJdNIOnVyYOSmU0Ddx40IRt76JO9/GaZuFVn8Y+PjPOZwzf5hJodFxvqzK2vrG5lZ1u7azu7d/YB8edXSaK8bbLJWp6oWguRQJb6NAyXuZ4hCHknfD8e283p1wpUWaPOI0434Mw0REggEaK7Bt745LBOqBzEYQCBrYdafhLET/gltCnZRqBfanN0hZHvMEmQSt+66ToV+AQsEkn9W8XPMM2BiGvG8wgZhrv1hcPqNnxhnQKFXmJUgX7s+JAmKtp3FoOmPAkV6tzc3/av0co2u/EEmWI0/YclGUS4opncdAB0JxhnJqAJgS5lbKRqCAoQmrZkJwV7/8FzoXDdfww2W9eVPGUSUn5JScE5dckSa5Jy3SJoxMyBN5Ia9WYT1bb9b7srVilTPH5Jesj2+dHpL7</latexit><latexit sha1_base64="4JPh39dwkaKh9jKa2jKEQ5mhGAI=">AAAB+XicbZDLSsNAFIYn9VbrLerSzWARXJVEBF0WdeGygr1AE8LJdNIOnVyYOSmU0Ddx40IRt76JO9/GaZuFVn8Y+PjPOZwzf5hJodFxvqzK2vrG5lZ1u7azu7d/YB8edXSaK8bbLJWp6oWguRQJb6NAyXuZ4hCHknfD8e283p1wpUWaPOI0434Mw0REggEaK7Bt745LBOqBzEYQCBrYdafhLET/gltCnZRqBfanN0hZHvMEmQSt+66ToV+AQsEkn9W8XPMM2BiGvG8wgZhrv1hcPqNnxhnQKFXmJUgX7s+JAmKtp3FoOmPAkV6tzc3/av0co2u/EEmWI0/YclGUS4opncdAB0JxhnJqAJgS5lbKRqCAoQmrZkJwV7/8FzoXDdfww2W9eVPGUSUn5JScE5dckSa5Jy3SJoxMyBN5Ia9WYT1bb9b7srVilTPH5Jesj2+dHpL7</latexit><latexit sha1_base64="4JPh39dwkaKh9jKa2jKEQ5mhGAI=">AAAB+XicbZDLSsNAFIYn9VbrLerSzWARXJVEBF0WdeGygr1AE8LJdNIOnVyYOSmU0Ddx40IRt76JO9/GaZuFVn8Y+PjPOZwzf5hJodFxvqzK2vrG5lZ1u7azu7d/YB8edXSaK8bbLJWp6oWguRQJb6NAyXuZ4hCHknfD8e283p1wpUWaPOI0434Mw0REggEaK7Bt745LBOqBzEYQCBrYdafhLET/gltCnZRqBfanN0hZHvMEmQSt+66ToV+AQsEkn9W8XPMM2BiGvG8wgZhrv1hcPqNnxhnQKFXmJUgX7s+JAmKtp3FoOmPAkV6tzc3/av0co2u/EEmWI0/YclGUS4opncdAB0JxhnJqAJgS5lbKRqCAoQmrZkJwV7/8FzoXDdfww2W9eVPGUSUn5JScE5dckSa5Jy3SJoxMyBN5Ia9WYT1bb9b7srVilTPH5Jesj2+dHpL7</latexit><latexit sha1_base64="4JPh39dwkaKh9jKa2jKEQ5mhGAI=">AAAB+XicbZDLSsNAFIYn9VbrLerSzWARXJVEBF0WdeGygr1AE8LJdNIOnVyYOSmU0Ddx40IRt76JO9/GaZuFVn8Y+PjPOZwzf5hJodFxvqzK2vrG5lZ1u7azu7d/YB8edXSaK8bbLJWp6oWguRQJb6NAyXuZ4hCHknfD8e283p1wpUWaPOI0434Mw0REggEaK7Bt745LBOqBzEYQCBrYdafhLET/gltCnZRqBfanN0hZHvMEmQSt+66ToV+AQsEkn9W8XPMM2BiGvG8wgZhrv1hcPqNnxhnQKFXmJUgX7s+JAmKtp3FoOmPAkV6tzc3/av0co2u/EEmWI0/YclGUS4opncdAB0JxhnJqAJgS5lbKRqCAoQmrZkJwV7/8FzoXDdfww2W9eVPGUSUn5JScE5dckSa5Jy3SJoxMyBN5Ia9WYT1bb9b7srVilTPH5Jesj2+dHpL7</latexit>

�↵i ⇡1

6

NX

k=N�5

'(fi, k)�5X

k=0

'(fi+1, k)

!

<latexit sha1_base64="VjHfodQhYDAp3NnbM6yBqHUTCu0=">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</latexit>


Challenge 2: Cumulative Error in Phase Difference

Calibrated phase after frequency hopping offset mitigation

Measured phase:

Channel k

Channel k+1

Convert to Channel k

Time

Cumulative error for each conversion

Channel hopping offset process

13

The tag movements will be overwhelmed by the cumulated error

Could be removed by detrending, but the useful signal will also be removed Cumulative error in phase difference


Derivative of Raw Phase Difference

Derivative of raw phase difference.

But the useful signal (in low-frequency) is still overwhelmed by noise

Cumulative error is removed

è Affected by estimation error

è Error is removed

14


Feature Extraction (cont’d)

Filter the signal before differentiation • Low pass filter with a 5Hz cutoff frequency

The differentiation effect in the frequency domain

The differentiation operation can be modeled as a convolution with vector [F, -F]:

Derivative of the filtered phase difference

Phase difference in the frequency domainNodding features

15


Challenge 3: Environmental Movements

16

When deployed in a moving environment, e.g., driving environment

• Vehicle vibration

• Driver’s and passengers’ movements

Image Source: https://www.shutterstock.com/search/sleepy+driver


Tag Deployment and Phase Difference

Two tags horizontally attached to the back of head (e.g., on a hat) Measured phase from a single RFID tag. Calibrated phase difference between two

horizontally attached tags

17

Vehicle vibration can be mitigated by using phase difference

Nodding can be differentiated from head rotation by a proper tag deployment


19

CPD based Respiration Signal Separation

Tensor data construction with Hankelization:

It indicates that 2 sinusoidal components should be decomposed for 1 breathing signal from the generalized Hankel matrix

Generalized Hankel

Matrices+ … +

Phase Difference Data

Breathing Signal Extraction

M Components

M RowsIn our system, M = 4 for CPD.


20

CPD based Respiration Signal Separation (Cont’d)

Decomposed signals by CPD

Fused signal of all breathing related components.Calculate the proportion of the power spectrumbetween 0.2 Hz and 0.5 Hz in the frequency domain.

Find the signals with the first two largest proportions and fuse them


21

Challenge 4: Sparse RFID Data

When multiple tags are employed Slotted ALOHA or binary tree search based sampling:

• At most one phase data is sampled at a time

• Need to estimate the missing phase data

• Traditional compressed sensing method may not work well for 4 or more tags

• D. K. Klair, K. Chin, and R. Raad. “A Survey and Tutorial of RFID Anti-Collision Protocols,” IEEE Communications Surveys Tutorials, vol.12, no.3, pp.400-421, Third-Quarter 2010.• Daniel M Dobkin. The RF in RFID: uhf RFID in practice. Newnes, 2012.


22

HaLRTC for Phase Recovering

The ideal matrix we aim to recover:

m tags

n time slots

The sparse tensor built by the sparse sampled matrix

r rows in Hankelization

Recovered signals using HaLRTC and Matrix Completion (based on SVD)


23

Challenge 5: Learning Features from RFID Data

http://digital.anthro-seminars.net/2015/theory-2015/brain-inspired-deep-learning/https://www.technotification.com/2017/11/alphago-zero.html

High accuracy, operate in realtime, low complexity, labeled training data, …


Outline






24


I. Respiration Monitoring/Apnea Detection

Normal Respiration Abnormal Respiration

Apnea

Apnea

Image Source: https://www.healthination.com/health/surprising-sleep-apnea-symptoms

Image Source: https://en.wikipedia.org/wiki/Biot%27s_respiration

25


Radar and WiFi based Vital Sign Monitoring Systems

Requires special hardware, which could be expensive and complicated

Radar based Systems:

1. Doppler radar

2. Ultrawideband (UWB) radar

3. Frequency Modulated Carrier Wave (FMCW) radar

WiFi based Systems:

1. Received Signal Strength (RSS) based system: UbiBreathe

2. 60 GHz WiFi based system: Mmvital

3. Channel State Information (CSI) based system: PhaseBeat, ResBeat, TensorBeat

Low-cost hardware

Easily affected by the unstable environment

RFID based Systems:

1. The Tagbreathe system monitors breathing signals by grouping the signals with same channel index

Low cost and robust to movements in the environment.

Doesn’t work well with channel hopping (as required by FCC)

Does not consider apnea detection

26


AutoTag: Respiration Monitoring for Home Environments

AutoTag System Architecture:

1. Signal Extraction

2. CalibrationFrequency Hopping OffsetMovement DetectionDC RemovalTag SelectionDownsampling and Filtering

3. Respiration MonitoringApna DetectionRespiration Rate Estimation

27


Apnea Detection

• Consider a breathing signal in a window as a time sequence

• An LSTM network encodes the time sequence to get the sample z, , and

• After obtaining the reconstructed signal, we propose a KL divergence based method for apnea detection.

• Then the sample z can be fed to another LSTM network for decoding to obtain the reconstructed signal

�2�(x)

<latexit sha1_base64="6g1kn21yW17wYq9ESoPD0IjlqVg=">AAAB+nicbZDLSsNAFIZPvNZ6S3XpZrAIdVOSIuiy6MZlBXuBpobJdNIOnUnCzEQtsY/ixoUibn0Sd76N0zYLbf1h4OM/53DO/EHCmdKO822trK6tb2wWtorbO7t7+3bpoKXiVBLaJDGPZSfAinIW0aZmmtNOIikWAaftYHQ1rbfvqVQsjm71OKE9gQcRCxnB2li+XfIUGwjse40hu6uhyuOpb5edqjMTWgY3hzLkavj2l9ePSSpopAnHSnVdJ9G9DEvNCKeTopcqmmAywgPaNRhhQVUvm50+QSfG6aMwluZFGs3c3xMZFkqNRWA6BdZDtVibmv/VuqkOL3oZi5JU04jMF4UpRzpG0xxQn0lKNB8bwEQycysiQywx0SatognBXfzyMrRqVdfwzVm5fpnHUYAjOIYKuHAOdbiGBjSBwAM8wyu8WU/Wi/VufcxbV6x85hD+yPr8AeVukxc=</latexit><latexit sha1_base64="6g1kn21yW17wYq9ESoPD0IjlqVg=">AAAB+nicbZDLSsNAFIZPvNZ6S3XpZrAIdVOSIuiy6MZlBXuBpobJdNIOnUnCzEQtsY/ixoUibn0Sd76N0zYLbf1h4OM/53DO/EHCmdKO822trK6tb2wWtorbO7t7+3bpoKXiVBLaJDGPZSfAinIW0aZmmtNOIikWAaftYHQ1rbfvqVQsjm71OKE9gQcRCxnB2li+XfIUGwjse40hu6uhyuOpb5edqjMTWgY3hzLkavj2l9ePSSpopAnHSnVdJ9G9DEvNCKeTopcqmmAywgPaNRhhQVUvm50+QSfG6aMwluZFGs3c3xMZFkqNRWA6BdZDtVibmv/VuqkOL3oZi5JU04jMF4UpRzpG0xxQn0lKNB8bwEQycysiQywx0SatognBXfzyMrRqVdfwzVm5fpnHUYAjOIYKuHAOdbiGBjSBwAM8wyu8WU/Wi/VufcxbV6x85hD+yPr8AeVukxc=</latexit><latexit sha1_base64="6g1kn21yW17wYq9ESoPD0IjlqVg=">AAAB+nicbZDLSsNAFIZPvNZ6S3XpZrAIdVOSIuiy6MZlBXuBpobJdNIOnUnCzEQtsY/ixoUibn0Sd76N0zYLbf1h4OM/53DO/EHCmdKO822trK6tb2wWtorbO7t7+3bpoKXiVBLaJDGPZSfAinIW0aZmmtNOIikWAaftYHQ1rbfvqVQsjm71OKE9gQcRCxnB2li+XfIUGwjse40hu6uhyuOpb5edqjMTWgY3hzLkavj2l9ePSSpopAnHSnVdJ9G9DEvNCKeTopcqmmAywgPaNRhhQVUvm50+QSfG6aMwluZFGs3c3xMZFkqNRWA6BdZDtVibmv/VuqkOL3oZi5JU04jMF4UpRzpG0xxQn0lKNB8bwEQycysiQywx0SatognBXfzyMrRqVdfwzVm5fpnHUYAjOIYKuHAOdbiGBjSBwAM8wyu8WU/Wi/VufcxbV6x85hD+yPr8AeVukxc=</latexit><latexit sha1_base64="6g1kn21yW17wYq9ESoPD0IjlqVg=">AAAB+nicbZDLSsNAFIZPvNZ6S3XpZrAIdVOSIuiy6MZlBXuBpobJdNIOnUnCzEQtsY/ixoUibn0Sd76N0zYLbf1h4OM/53DO/EHCmdKO822trK6tb2wWtorbO7t7+3bpoKXiVBLaJDGPZSfAinIW0aZmmtNOIikWAaftYHQ1rbfvqVQsjm71OKE9gQcRCxnB2li+XfIUGwjse40hu6uhyuOpb5edqjMTWgY3hzLkavj2l9ePSSpopAnHSnVdJ9G9DEvNCKeTopcqmmAywgPaNRhhQVUvm50+QSfG6aMwluZFGs3c3xMZFkqNRWA6BdZDtVibmv/VuqkOL3oZi5JU04jMF4UpRzpG0xxQn0lKNB8bwEQycysiQywx0SatognBXfzyMrRqVdfwzVm5fpnHUYAjOIYKuHAOdbiGBjSBwAM8wyu8WU/Wi/VufcxbV6x85hD+yPr8AeVukxc=</latexit>

µ�(x)<latexit sha1_base64="558LursGZyNz8wA+n80sH4iY9Qw=">AAAB9HicbZDLSgMxFIbP1Futt6pLN8Ei1E2ZEUGXRTcuK9gLtEPJpJk2NMmMSaZYhj6HGxeKuPVh3Pk2ZtpZaOsPgY//nMM5+YOYM21c99sprK1vbG4Vt0s7u3v7B+XDo5aOEkVok0Q8Up0Aa8qZpE3DDKedWFEsAk7bwfg2q7cnVGkWyQczjakv8FCykBFsrOX3RNLvNUYMVZ/OS/1yxa25c6FV8HKoQK5Gv/zVG0QkEVQawrHWXc+NjZ9iZRjhdFbqJZrGmIzxkHYtSiyo9tP50TN0Zp0BCiNlnzRo7v6eSLHQeioC2ymwGenlWmb+V+smJrz2UybjxFBJFovChCMToSwBNGCKEsOnFjBRzN6KyAgrTIzNKQvBW/7yKrQuap7l+8tK/SaPowgncApV8OAK6nAHDWgCgUd4hld4cybOi/PufCxaC04+cwx/5Hz+ACxBkQk=</latexit><latexit sha1_base64="558LursGZyNz8wA+n80sH4iY9Qw=">AAAB9HicbZDLSgMxFIbP1Futt6pLN8Ei1E2ZEUGXRTcuK9gLtEPJpJk2NMmMSaZYhj6HGxeKuPVh3Pk2ZtpZaOsPgY//nMM5+YOYM21c99sprK1vbG4Vt0s7u3v7B+XDo5aOEkVok0Q8Up0Aa8qZpE3DDKedWFEsAk7bwfg2q7cnVGkWyQczjakv8FCykBFsrOX3RNLvNUYMVZ/OS/1yxa25c6FV8HKoQK5Gv/zVG0QkEVQawrHWXc+NjZ9iZRjhdFbqJZrGmIzxkHYtSiyo9tP50TN0Zp0BCiNlnzRo7v6eSLHQeioC2ymwGenlWmb+V+smJrz2UybjxFBJFovChCMToSwBNGCKEsOnFjBRzN6KyAgrTIzNKQvBW/7yKrQuap7l+8tK/SaPowgncApV8OAK6nAHDWgCgUd4hld4cybOi/PufCxaC04+cwx/5Hz+ACxBkQk=</latexit><latexit sha1_base64="558LursGZyNz8wA+n80sH4iY9Qw=">AAAB9HicbZDLSgMxFIbP1Futt6pLN8Ei1E2ZEUGXRTcuK9gLtEPJpJk2NMmMSaZYhj6HGxeKuPVh3Pk2ZtpZaOsPgY//nMM5+YOYM21c99sprK1vbG4Vt0s7u3v7B+XDo5aOEkVok0Q8Up0Aa8qZpE3DDKedWFEsAk7bwfg2q7cnVGkWyQczjakv8FCykBFsrOX3RNLvNUYMVZ/OS/1yxa25c6FV8HKoQK5Gv/zVG0QkEVQawrHWXc+NjZ9iZRjhdFbqJZrGmIzxkHYtSiyo9tP50TN0Zp0BCiNlnzRo7v6eSLHQeioC2ymwGenlWmb+V+smJrz2UybjxFBJFovChCMToSwBNGCKEsOnFjBRzN6KyAgrTIzNKQvBW/7yKrQuap7l+8tK/SaPowgncApV8OAK6nAHDWgCgUd4hld4cybOi/PufCxaC04+cwx/5Hz+ACxBkQk=</latexit><latexit sha1_base64="558LursGZyNz8wA+n80sH4iY9Qw=">AAAB9HicbZDLSgMxFIbP1Futt6pLN8Ei1E2ZEUGXRTcuK9gLtEPJpJk2NMmMSaZYhj6HGxeKuPVh3Pk2ZtpZaOsPgY//nMM5+YOYM21c99sprK1vbG4Vt0s7u3v7B+XDo5aOEkVok0Q8Up0Aa8qZpE3DDKedWFEsAk7bwfg2q7cnVGkWyQczjakv8FCykBFsrOX3RNLvNUYMVZ/OS/1yxa25c6FV8HKoQK5Gv/zVG0QkEVQawrHWXc+NjZ9iZRjhdFbqJZrGmIzxkHYtSiyo9tP50TN0Zp0BCiNlnzRo7v6eSLHQeioC2ymwGenlWmb+V+smJrz2UybjxFBJFovChCMToSwBNGCKEsOnFjBRzN6KyAgrTIzNKQvBW/7yKrQuap7l+8tK/SaPowgncApV8OAK6nAHDWgCgUd4hld4cybOi/PufCxaC04+cwx/5Hz+ACxBkQk=</latexit>

28


Implementation

• Commodity RFID reader ImpinjR420 equipped with a directional antenna

• ALN-9740 tags

• Frequency hopping is from 902.5 to 927.5 MHz with 50 channels, which is FCC-compliant

Hardware:

29

L. Almaszydeh, K. Elleithy, M. Faezipour, and A. Abushakra, "Apnea detection based on respiratory signal classification," Procedia Comput. Sci., vol.21, pp.310-316, 2013.

Baseline scheme:


Experiment Results

w/ interference

w/o interference

TN rate: success rate that breathingsignal is correctly detected TP rate: success rate that apnea is correctly detected

CDFs of breathing rate estimation error

< 0.1 bpm 0.32 bpm 0.46 bpm

30

88% 92%

56%

92%


II. Respiration Monitoring in Driving Environments

[1] B. Warwick, N. Symons, X. Chen, and K. Xiong, “Detecting driver drowsiness using wireless wearables,” in Proc. IEEE MASS 2015, Dallas, TX, Oct. 2015, pp. 585–588.

Image Source: https://www.gps.com.au/fleet-management-benefits/driver-safety-and-driver-fatigue-managementImage Source: https://www.transpoco.com/blog/2015/01/10/tackle-driver-fatigue/

More challenging in driving environments:

• Signal is corrupted by strong noises

• Channel hopping effect

• Vehicle vibration

• Movements of drivers and passengers

Indicator of driving fatigue [1]

Image Source: https://www.shutterstock.com/search/sleepy+driver

31


Related Work

Requires sufficient lightingMay raise security and privacy concerns

Video based:

1. Video is captured by a video camera [1]

2. Detect respiration by analyzing the chest movements in the video

Ultrasound based:

1. Leverage audio devices on smartphones [2]

2. Not requiring lighting inside the vehicle

3. Low cost (smartphone embedded system)

Sensitive to noises in the driving environment

RF based:

1. Leverage WiFi CSI [3]

2. Monitor chest movement with UWB Radar [4]

3. No requirements for lighting

Sensitive to the movements of drivers and passengers

[1] J. Solaz, J. Laparra-Hern´andez, D. Bande, N. Rodr´ıguez, S. Veleff, J. Gerpe, and E. Medina, “Drowsiness detection based on the analysis of breathing rate obtained from real-time image recognition,” Transportation Research Procedia, vol. 14, pp. 3867–3876, Apr. 2016.

[3] W. Jia and H. Peng, “Wifind: Driver fatigue detection with fine-grained Wi-Fi signal features,” in Proc. IEEE GLOBECOM 2017, Singapore, Dec. 2017, pp. 1–6.

[4] Z. Yang, M. Bocca, V. Jain, and P. Mohapatra, “Contactless breathing rate monitoring in vehicle using UWB Radar.” in Proc. RealWSN Workshop at SenSys 2018, Shenzhen, China, Nov. 2018, pp. 13–18.

[2] X. Xu, J. Yu, Y. Chen, Y. Zhu, L. Kong, and M. Li, “BreathListener: Fine-grained breathing monitoring in driving environments utilizing acoustic signals,” in Proc. ACM MobiSys 2019, Seoul, Republic of Korea, June 2019, pp. 54–66.

32


RFID based Respiration Monitoring

• Leverage received signal strength and phase information collected by the Low-Level Reader Protocol (LLRP)

• Robust to surrounding noises (wearable tags)

• Low-cost passive RFID tags

Characteristics:Respiration Monitoring in Driving Environments

Illustration of the respiration monitoring mechanism

Signal corrupted by strong driving environmental noises and channel hopping effect

33


System Overview

Data collection

Breathing data preprocessingPhase variation extractionSparse tensor constructionHaLRTCVehicle vibration mitigationDownsampling and Hankelization

Canonical Polyadic Decomposition

Respiration signal reconstruction34


35

Experiment Configuration

System setup in a car in our experiments

Commodity RFID reader Impinj R420 equipped with a polarized antenna S9028PCR

ALN-9740 tags

Frequency hopping among 50 channels from 902.5 to 927.5 MHz, which is FCC-compliant

MSI laptop computer with a Navidia GTX 1080 GPU and Intel Core i7-6820HK CPU

Five volunteers (1 female and 4 males) are tested


36

Experiment Results

Different driving

scenarios

Different numbers of passengers

Different deployment locations of

antenna

Different numbers of tags


III. Drowsy Driving Detection

EEG WiFi and UltrasoundCamera

Collect eyelid movements using a camera [2]

Require sufficient lighting in the car

Wearing sunglasses could affect the performance

Easy to deploy, contact-free, and high accuracy

High fatigue detection accuracy [1]

Special devices are required and attached to the head

Image Source: https://www.verywellhealth.com/what-is-an-eeg-test-and-what-is-it-used-for-3014879

Exploit the channel state information (CSI) (WiFind) [3], and backscattered ultrasound signal (D3-Guard) [4]

Sensitive to interference from surroundings, such as passengers and vehicle vibration

Does not require lighting

37

[1] B. T. Jap, S. Lal, P. Fischer, and E. Bekiaris, “Using EEG spectral components to assess algorithms for detecting fatigue,” Elsevier Expert Systems with Applications, vol. 36, no. 2, pp. 2352–2359, Mar. 2009.

[2] W.-B. Horng, C.-Y. Chen, Y. Chang, and C.-H. Fan, “Driver fatigue detection based on eye tracking and dynamic template matching,” in Proc. IEEE ICNSC’04, Taipei, Taiwan, Mar. 2004, pp. 7–12.

[3] W. Jia and H. Peng, “Wifind: Driver fatigue detection with fine-grained Wi-Fi signal features,” in Proc. IEEE GLOBECOM 2017, Singapore, Dec. 2017, pp. 1–6.[4] Y. Xie, F. Li, Y. Wu, S. Yang, and Y. Wang, “D 3-Guard: Acousticbased drowsy driving detection using smartphones,” in Proc. IEEE INFOCOM’19, Paris, France, Apr./May 2019, pp. 1–9.


NodTrack: Nodding Detection in Driving Environments

RFID phase data measurement

Movement feature extractionPhase difference estimationChannel Hopping OffsetLow-pass FilteringDerivative Calculation

Offline training

Online drowsiness detection

Architecture of the NodTrack system

38


Driving Fatigue Detection

The recurrent variational autoencoder model for driving fatigue detection.

LSTM model:

Suitable for learning both long-range and short-range dependency

Generate mean vector and the variance vector with two linear modules

39

Autoencoder:

Calculating latent vector z


Driving Fatigue Detection (cont’d)

Sampling with a sliding window of 2 s

Input signal is reconstructed with the autoencoder

Calculating the divergence between the input and reconstructed signal

40

Nodding signal Normal driving signal

Original and Reconstructed Signals


Experimental Setting

Commodity RFID reader Impinj R420 equipped with a polarized antenna S9028PCR

ALN-9740 tags

Frequency hopping among 50 channels from 902.5 to 927.5 MHz , which is FCC-compliant

MSI laptop computer with a Navidia GTX 1080 GPU and Intel Core i7-6820HK CPU

Hardware:

41


CDFs of the mean absolute errors forreconstructed normal driving and nodding signals

0.21

Experimental Study: Results

42

Detection accuracy in two scenarios: TPrates (nodding) and TN rates (normal driving)

97.23% 96.23%

91.48%

95.38%91.26%


Experimental Study: Results (cont’d)

Impact of the window size used for training

Accuracy in different driving scenarios

Accuracy for different drivers

Impact of the number of passengers in the vehicle 43


https://www.komando.com/security-privacy/9-million-webcams-could-be-hacked/497628/https://www.vedosoft.com/featured/surveillance-cameras-and-privacy/https://medium.com/@victoriamazo/3d-human-pose-estimation-ce1259979306

Human pose tracking becomes an important topic in computer vision

• Millions of security cameras at risk of being hacked

Security and privacy concerns:

• Being watched by a video camera all day

IV. Human Pose Detection and Tracking

44


Related Work

Require expensive and complicated hardware

Radar based:1. Frequency-Modulated

Continuous Wave (FMCW) radar-based system [1]

2. High accuracy and more robust to environment noise than Wi-Fi based systems

Wi-Fi based:1. 2D pose estimation for

multiple people [1], and 3D pose generation [2]

2. Contact-free pose estimation, and wide range detection.

3. Low-cost hardware

Sensitive to the interference from environment

RFID based:1. RF-Kinect based on the RF

hologram technique [4], and 3D limbs movement tracking with RFID array [5]

2. Good performance for single limb tracking

Not suitable for the full body skeleton reconstruction

[3] M. Zhao, T. Li, M. Abu Alsheikh, Y. Tian, H. Zhao, A. Torralba, and D. Katabi, “Through-wall human pose estimation using radio signals,” in Proc. IEEE CVPR 2018, Salt Lake City, UT, June 2018, pp. 7356–7365.

[1] F. Wang, S. Zhou, S. Panev, J. Han, and D. Huang, “Person-in-WiFi: Fine-grained person perception using WiFi,” in Proc. IEEE ICCV 2019, Seoul, Republic of Korea, Oct. 2019, pp. 5452–5461.[2] W. Jiang, H. Xue, C. Miao, S. Wang, S. Lin, C. Tian, S. Murali, H. Hu, Z. Sun, and L. Su, “Towards 3D human pose construction using WiFi,” in Proc. ACM MobiCom’20, London, UK, Sept. 2020, pp. 1–14.

[4] C. Wang, J. Liu, Y. Chen, L. Xie, H. B. Liu, and S. Lu, “RF-Kinect: A wearable RFID-based approach towards 3D body movement tracking,” Proc. ACM Int., Mobile, Wearable Ubiquitous Technol., vol. 2, no. 1, Mar. 2018.[5] H. Jin, Z. Yang, S. Kumar, and J. I. Hong, “Towards wearable everyday body-frame tracking using passive RFIDs,” Proc. ACM Interactive, Mobile, Wearable Ubiquitous Technol., vol. 1, no. 4, pp. 1–23, Dec. 2018.

45


Vision-aided 3D Human Pose Estimation with RFID


RFID Phase Data Collection

Kinect Skeleton Data Collection

RFID Data PreprocessingPhase variation calibrationDownsampling and synchronizationInput Tensor ConstructionLow Rank Tensor Completion

Skeleton Reconstruction Deep kinematic neural networkUsing Kinect data in training

RFID-Pose System Overview


RFID Phase Distortion Mitigation and Data Imputation

Raw phase sampled from one of the RFID tags by a single reader antenna

Flow chart of RFID data preprocessing


The deep kinematic neural network

Human Pose Reconstruction with RFID DataKinect Data

• Recurrent Autoencoder

• Forward kinematic layer

• Kinect data: labeled, for training


Pose tracking experiments Pose estimation when the subject is walking

Implementation and Evaluation


Pose estimation when the subject is squatting Pose estimation when the subject is twisting

Experiment Results


Overall pose estimation accuracy Estimation errors for different types of motions

Accuracy Results

2.83 3.75

5.44


53

Conclusions

RFID based human activity sensing and identification

Prototyping with commodity RFID devices and evaluation in real world scenarios, where a highly accurate and robust performance are demonstrated

A multidisciplinary area with the joint play of wireless communications, hardware devices, signal processing, machine learning, and healthcare

Machine learning: faster training, generalization, explainable, open-source dataset

As understandings of wireless communications are gained and advanced infrastructure deployed, wireless application is becoming an important playground


This work is supported in part by the National Science Foundation (NSF) under Grants ECCS-1923163 and CNS-1822055, and through the Wireless Engineering Research and Education Center (WEREC) at Auburn University.

For more information: http://www.eng.auburn.edu/~szm0001/

http://www.eng.auburn.edu/~szm0001/

RFID FOR HUMAN ACTIVITY SENSING - New York Chapter...Leverage received signal strength and phase information collected by the Low-Level Reader Protocol (LLRP) Near-field detection

Documents