A Batteryless 19µW MICS/ISM-

Band Energy Harvesting Body

Area Sensor Node SoC

Fan Zhang, Yanqing Zhang, Jason Silver, Yousef

Shakhsheer, Manohar Nagaraju, Alicia Klinefelter,

Jagdish Pandey, James Boley, Eric Carlson,

Aatmesh Shrivastava, Brian Otis, Benton Calhoun

University of Washington, Seattle, WA

University of Virginia, Charlottesville, VA

1

• Wireless body area sensor nodes (BASN) enable

inexpensive continuous monitoring of patients

• Battery replacement for body-worn devices may not be

feasible or desirable

Information

Assessment,

Treatment

Introduction

Commercial-off-the-shelf (COTS) wireless BASN

Other Components: Amplifiers, ADC, Power, Interface

• RF Transceiver

• Matching network

• RF crystal

• Antenna connector

(bottom) Radio

• PAVG > 20 mW (continuous transmission)

< 2 hours of battery life (33mAh)

• Memory

• Microcontroller

• Watch Crystal

(top)

Digital Logic Memory

3

Custom low-power BASN IC

Rai, ISSCC 2009. Morrison, DAC 2010

• Digitize & transmit ECG, EMG, neural data

• No signal processing

• Battery life: 3 days (100% duty cycle (DC)) 4

Custom IC Battery

(33 mAh)

Opportunity: energy harvesting

COLD

HOTn

p

LOAD

HEAT FLOW

0

0.01

0.02

0.03

0.04

0.05

5 10 50 100 1 k

Room Temp

Ou

tput

Vo

ltage (

V)

Load Resistance (Ω) 5

On-body measurement of

COTS thermoelectric generator (TEG) (4 x 4 cm2)

0

0.02

0.04

0.06

0.08

0.1

5 Ω 10 Ω 50 Ω 100 Ω 1 kΩ

6°C

ForearmBack of NeckChestShoulderAbdomenLower ThighLower BackUpper Calf

Opportunity: energy harvesting

COLD

HOTn

p

LOAD

HEAT FLOW

0

0.01

0.02

0.03

0.04

0.05

5 10 50 100 1 k

Room Temp

Ou

tput

Vo

ltage (

V)

Load Resistance (Ω)

~60µW ~200µW

6

On-body measurement of

COTS thermoelectric generator (TEG) (4 x 4 cm2)

0

0.02

0.04

0.06

0.08

0.1

5 Ω 10 Ω 50 Ω 100 Ω 1 kΩ

6°C

ForearmBack of NeckChestShoulderAbdomenLower ThighLower BackUpper Calf

PROPOSED:

• NO battery harvest power

• Extract information

• Selective transmission

Our proposed solution

7

CONVENTIONAL:

• Battery

• No signal processing

• Transmission at 100% DC

ADC RFAFE

Boost

Converter

Voltage

Regulation

µController

Memory

DSP

Power Mgmt.

RF Kick-StartTEG

Signal Path

Power Path

VBOOST

ECG

EMG

EEG

Functional diagram

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

SIGNAL PATH

AFE Processing

RF

DPM

TEG

Boost

ROM

Boot

EN

ER

GY

HA

RV

ES

TIN

G

8

DPM: Digital Power Management

AFE: Analog Front-End

Startup and energy harvesting

RF-kick

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

SIGNAL PATH

AFE Processing

RF

DPM

TEG

Boost

ROM

Boot

EN

ER

GY

HA

RV

ES

TIN

G

Startup and energy harvesting

RF-kick

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

SIGNAL PATH

AFE Processing

RF

DPM

TEG

Boost

ROM

Boot

EN

ER

GY

HA

RV

ES

TIN

G

POR

Signal processing path

11

TEG

Boost

ROM

Boot

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

EN

ER

GY

HA

RV

ES

TIN

G

SIGNAL PATH

AFE Processing

RF

DPM

Signal processing path

12

TEG

Boost

ROM

Boot

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

EN

ER

GY

HA

RV

ES

TIN

G

SIGNAL PATH

AFE Processing

RF

DPM

Flexible signal path

Power management

13

TEG

Boost

ROM

Boot

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

EN

ER

GY

HA

RV

ES

TIN

G

SIGNAL PATH

AFE Processing

RF

DPM

Power management

14

TEG

Boost

ROM

Boot

POWER

MANAGEMENT

Regulate VBOOST

Monitor Stoplight

EN

ER

GY

HA

RV

ES

TIN

G

SIGNAL PATH

AFE Processing

RF

DPM

DVS

System block diagram

15

Energy harvesting

16

ROM

Hybrid energy harvesting

VBOOST: 1.35V

6 stages

Rectifier

Over-Voltage Clamp

N1

Boost Converter

RF:

-10 dBm

VTEG:

30 mV

Gate

Control

TEG

Power Mgmt.

Storage CapN1

Power-On Reset (POR)

Off-chip

Off-chip

17

Capable of thermal, photovoltaic, and/or RF energy harvesting

1cm

• Measured efficiency of 38% for Vin=30mV, Vout=1.35V

• Requires minimum voltage (VKILL) to sustain conversion

18

Boost converter

Carlson, Strunz, Otis, JSSC 2010

Measured RF kick-start

0 2 4 6 8 100

0.5

1

1.5

time (s)

Vo

ltag

e (

V)

RF Pulse, -10dBm

Boost Converter

turns on

VBOOST

VTEG

• Wireless RF pulse provides one-time kick-start

• The node runs indefinitely thereafter 19

Boot-up sequence

• POR issues reset at 1.0V

• Upon boot-up, the chip fetches instructions from the ROM

(default) or the RAM

• If VBOOST < VKILL , RF burst can ‘revive’ chip to default

algorithm in ROM

POR

Programmed?

132B ROM

(AFib Detection)

1404B RAM

(Reprogrammable)

Yes No

1.2 V

1 V

VBOOST POR out

20

Instruction memory (IMEM)

Signal path

21

Analog front-end (AFE)

22

AmpOut+

AmpOut-

LNA

Cs

Cf

In+

In-

AmpIn+

AmpIn- GND

Cfilter

Out+

Out-

Gm Gm

Cfilter

Gm Gm

Chopper

modulatorPseudo-resistor

Cs

Cf

Cfilter

Cfilter

VgaOutVGA

CsVgaIn+

VgaIn-Cs

Cf

GmVCM

AmpIn+

Vbias1

Vbias2

VDD

GND

AmpIn-

Vbias3

Am

pO

ut-

Am

pO

ut+

Vcs

CMFB

VgaIn+

Vbias4

Vbias5

VDD

GND

VgaIn-

Vbias6

Vg

aO

ut

Vcs

Cf

Gm

• Chopper-stabilized

low-noise amp

• Variable-gain amp

• 6 programmable

gain 40 – 78 dB

• 3 µW / channel

1 of 4 channels

Flexible signal path

23

Example of Mixed Path

ENV

DetectFIR

Example Custom Path

RR+

AFibFIR

Generic Path

MCU

Flexible Architecture for Data Processing

MCU

Event-Based Burst

Store and Burst

Stream

If event

Processed

Data

Flexible Architecture for Data Transmission

Data for TX

4kB

DMem

4kB

DMem

ECG

EMG

EEG

AFE

Data processing Data transmission

• Data processing: max flexibility (generic path) or max efficiency (biosignal accelerators)

• Data transmission: supports modes from streaming (100% DC) to rare event detection (~0% DC)

MCU: microcontroller

TX

MCU

RR+AFib Accel.

30-Tap FIR Accel.

0 50 100 150 200 Delay (µs)

Measure

d E

nerg

y/O

p (

pJ)

Energy efficient accelerators

24

Accelerators:

• Programmable FIR

• Heart rate (R-R) extraction

• Atrial Fibrillation (AFib) detection

• Band energy extraction

• Direct memory access (DMA)

• Packetizer

Energy Efficiency / Sample

110x

6800x

4000x

0

1

2

3

4

30 Tap FIR MCU 6.3 nJ

Accel 57.6 pJ

Env. Detect MCU 3.6 nJ

Accel 530 fJ

R-R Extract MCU 12 pJ

Accel 3 fJ

25

DPM: signal path control

ADC

DMA/SRAM

Bio-signal

Accelerators

Packetizer

Power and

Channel control

Sa

mp

ling

ra

te c

on

tro

l

Power/clock gate,

clock rate, and

bus control

Du

ty c

ycle

, d

ata

ra

te c

on

tro

l

Dig

itiz

ed V

BO

OS

T

DPM Chip

program

LNA

VBOOST

VGA

IMEM

MCU vs. DPM

26

DPM

MCU

IMEM

• Execution of instructions toggles automatically

between MCU and DPM.

Operation DPM Energy MCU Energy

NOP 0.7 pJ 1.46 pJ

Control Signals 2.8 pJ 2.92 pJ

Branch Commands 2.9 pJ 4.38 pJ

Generic processing

(e.g. add, multiple)

Control instructions

(e.g. power/clock-gate)

Frequency multiplying TX

Pandey, Otis, JSSC’11

27

• Frequency multiplication:

synthesis at low frequency, transmission at high frequency

• Edge-combiner based frequency multiplier

• ILRO-based edge generator

• PA integrates the edge combiner

BFSK data modulation

FSK Data

CL CL

ΔC

• Quartz reference clock is pulled using DC (~200ppm)

• Δf is multiplied by 9x (~100kHz)

“1” “0”

28

Measured Data Modulation

Power management

29

0.5 V

Bandgap

0.5

V0

.5V

1.0

V

1.2

V

F1

F1

F2

F2

0.25-1.0V

VBOOST: 1.35V Linear Regulators Switched-Cap Regulator

Bias Gen

VREF, IREF

30

Voltage regulation

1.2V PADS, AFE

1.0V TX LO

0.5V TX PA

0.5V DPM, MEM, ACCEL

Variable 0.25-1.0V in 50mV, enabling DVS.

Power management scheme

VREF

MU

XChannel 1

VBOOST

Channel 2

Channel 3Channel 4

ADC

SEL

DPM

• VBOOST is scaled and digitized

• DPM compares VBOOST vs. VTHR (programmable)

• DPM chooses reconfigurable modes (“stoplight”)

Closed-loop power management

VTEG (V)

VBoost VDD, AFE

VGA Out (V)

Supply (V)

TX Duty-

Cycle

Green Yellow Red Yellow Green

Time (s)

32

MODES AFE Process Data Mem. Inst. Mem. Transmit VTHR

1.35 V

1.3 V

1.1 V

0.9 V

Green

Yellow /

Red Sto

pli

gh

t

Continuous transmission of ECG

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8time (s)

Real ECG

Reconstructed ECG

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8time (s)

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8time (s)

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

0.5

1

TX

D

ata

2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

0.5

1

RX

D

ata

2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

0.5

1

time (ms)

RX

Clo

ck

1.798 1.7981 1.7982 1.7983 1.7984 1.7985 1.7986

0

0.5

1

TX

EN

1.7979 1.798 1.7981 1.7982 1.7983 1.7984 1.7985 1.7986 1.7987

0

0.5

1

TX

data

time (s)

Header(5b) Data(16b) CRC(3b)

655msVBOOST

sample

2

1

1

0

1

01

0

1

01

0

1

0.5

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8Time (s)

Time (s)

Vo

lta

ge

(mV

)T

X

Da

ta

RX

Da

ta

RX

Clo

ck

AD

C i

n(V

)T

X

Da

ta

TX

EN

Raw-Data Mode:

R-R Mode:

1.2ms

650µs

TX DATA

RX DATA

RX CLK

Voltage

(mV)

Time (s)

• ECG signal measured from a healthy human subject

• Wireless link demonstrated between the custom IC and a

commercial receiver (TI CC1101)

• 397 µW from VBOOST

33

R-R interval extraction of ECG

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

0.4

0.6

0.8

1

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

0

0.5

1

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

0

0.5

1

1.798 1.7981 1.7982 1.7983 1.7984 1.7985 1.7986

0

0.5

1

1.7979 1.798 1.7981 1.7982 1.7983 1.7984 1.7985 1.7986 1.7987

0

0.5

1

655 ms

650 µs

Header Data CRC

VBoost

sample

ADC IN

(V)

TX EN

TX

DATA

Time (s)

• Every 5s, VBOOST is sampled to check for sufficient energy

• DPM enables RF crystal oscillator (20ms) and TX (650µs)

• 19 µW from VBOOST

• Powered from a 30mV input 34

Time (s) 0 1 93 95 97 99 101 103 105 107

1

…

…

…

AFib begins Chip detects AFib

0.8

0.6

0.4

0.2

0

0.5

0

AFib detection of ECG

• When a rare AFib occurs, TX is enabled to transmit the

last 8 beats of ECG (in the data memory).

• 19 µW from VBOOST

• Powered from a 30mV input 35

Input E

CG

Sig

nal

(V)

AF

ib D

ete

ct

(V)

0 20 40 60 800

100

200

300

400

500

Selective transmission

AFE

(4µA)

Supply

Regulation

(3µA)

Digital

(4.6µA)

TX

(0.14µA) Clock

Gen

(2µA)

36

Effective Transmit Rate (kbps)

Estim

ate

d T

ota

l chip

Pow

er

(µW

)

Battery-free with TEGs today

• Selective TX and ULP circuits enable energy harvesting

With next-generation

TEGs

TX DC: 0.013%

SoC die photo

0.13 µm CMOS 2.5

mm

3.3 mm 37

This Work Kim

VLSI'11

Rai

ISSCC’09

Verma

JSSC’10

Yan

JSSC'11

Chen

ISSCC'10

Sensors ECG, EMG,

EEG ECG

Neural, ECG,

EMG, EEG EEG

ECG,

TIV

Temp,

Pressure

Supply

Voltage

30 mV,

-10 dBm 1.2 V 1 V 1 V 1.2 V 0.4/0.5 V

E Harvesting Thermal, RF Solar

Supply Reg.

AFE 4-ch 3-ch 1-ch 18-ch 4-ch N/A

TX datarate 200 kb/s 100 kb/s

1 Mbps

(On-body

link)

TX PDC

(100% on) 160 µW 400 µW 2.8 mW

TX POUT -18.5 dBm -16 dBm -6 dBm

TX band 402 / 433

MHz

402 / 433

MHz 20-40 MHz

Power Mgmt.

Clock +

Power gating,

DPM

Clock

gating

Power

gating

DVS

Comparison with prior work

Comparison with prior work (cont.) This Work

Kim

VLSI'11

Rai

ISSCC’09

Verma

JSSC’10

Yan

JSSC'11

Chen

ISSCC'10

Micro-

processor

1.5 pJ/Inst

200kHz

(8b RISC ISA)

28.9 pJ/Inst

73kHz

(32b COR-

M3)

Accelerator

Prog. FIR,

AFIB, DMA,

Env. Det.,

Packetizer

4x SIMD,

FIR, DMA,

Encryption ASIC DSP

FIR,

Packetizer,

Compression

Memory 5.5kB (0.3-

0.7V) 42kB (1.2V) 20kB (1.2V) 5kB (0.4V)

Dig. Power 2.1µW ~12µW N/A 2.1µW 500µW 2.1µW

(MCU)

Total Power 19µW 31.1µW 500µW 77.1µW 2.4mW 7.7µW

Note on

Total Power

(includes):

8b ADC, DSP

(R-R extract),

TX at 0.013%

DC

12b ADC,

DSP

(heart beat

detection)

8b ADC,

TX at

100%

DC

12b ADC,

DSP (EEG

feature

extraction)

10b ADC,

DSP (data

comp, FIR),

SRAM, TX at

5% DC

Data

acquisition,

DSP (DFT),

SRAM

Technology 130nm 180nm 130nm 180nm 180nm 180nm

Conclusion

1. First wireless biosignal processing chip powered solely from a TEG with RF kickstart enabling battery-free operation

2. Our chip integrates state-of-the-art MICS/ISM transmitter, AFE, powertrain, and biosignal accelerators

3. These blocks are intelligently controlled to enable programmable datapath and closed-loop power management

40

Thank you!

41