Autonomous Sensor Node Powered over WiFi: A Use Case Study

Florian Grante, Ghalid Abib, Muriel Muller and Nel Samama

epartement Electronique et Physique (EPh), T

ecom SudParis, Institut Polytechnique de Paris,

19 Rue Marguerite Perey, 91120, Palaiseau, France

Keywords:

RF Electromagnetic Energy, Energy Harvesting, WiFi, Autonomous Sensor, Schottky Diode,

RF/DC Rectiﬁer, Boost, Energy Budget Analysis, Green IoT.

Abstract:

This paper presents a new approach to the sizing of a radio frequency energy harvesting system. Starting

from a sensor node to deﬁne our system constraints such as its energy requirement or its supply voltage, an

energy budget analysis protocol is developed to validate whether it is possible to power such a system with a

Radio Frequency to Direct Current (RF/DC) converter. We study a converter capable of harvesting RF signals

in the Industrial, Scientiﬁc and Medical (ISM) band at 2.45 GHz whose ambient power has been previously

characterized. Finally, the duty cycle is determined, i.e., how long would it take the converter to recover the

required energy in order to power the sensor node.

1 INTRODUCTION

The Internet of Things (IoT) is gradually revolution-

izing our daily lives as well as in the industrial, med-

ical, connected city projects and electricity distribu-

tion with smart grids. But the explosion in the num-

ber of connected objects also implies the explosion in

the production of batteries, the main source of electric

energy for these objects. Batteries being a depletable

source of energy and requiring regular maintenance,

there is a need to ﬁnd alternative energy sources that

we can group together as the concept of Energy Har-

vesting.

Among the different harvestable energy sources,

the most common one is the solar energy with the de-

velopment of small organic solar panels. They have

the advantage of making ﬂexible panel and do not

require rare-earth elements in its design. Also, me-

chanical energy is well suited for products such as

connected switches. We can also ﬁnd research teams

working on thermoelectric energy sources based on

the Seebeck effect or on the source that we will focus

on in this paper, which is the Radio Frequency (RF)

signal. RF sources like WiFi, GSM, TV transmitters,

... emit electromagnetic waves and surround our envi-

ronment. A lot of them are wasted and could be con-

verted into Direct Current (DC) energy to power spe-

ciﬁc low consumption electronic devices like a sensor

node, leading to a green IoT.

During the last 10 years, RF energy harvesting

state of the art has evolved from a converter sys-

tem using TV signals (Parks et al., 2013), which was

commonly used and powerful in the early 2010’s, to

more research in the Industrial, Scientiﬁc and Medi-

cal (ISM) 2.45 GHz and GSM 900 / 1800 bands (Ho

et al., 2016) because of the emergence of WiFi, 3G,

4G systems...

To our knowledge, few works take into account

the sizing of the harvester system for a dedicated ap-

plication or a sensor node according to its energy re-

quirements. A well-optimized ”wake-up” principle

allows the sensor to be powered for a short period

to measure and send data, then turned off to allow

the converter to charge a capacitor for energy stor-

age. But we would like to go further and propose a

new approach in the study of RF/DC converters by

concretely characterizing the energy requirement of

an application such as powering an IoT sensor node

and thus discuss the feasibility of such a system.

We base this work on WiFi energy harvesting

system in the 2.45 GHz band and we are focus-

ing only on the necessary circuits after the antenna.

This latter will not be discussed here because we can

rely on other works such as (Kurvey and Kunte, ),

(Krakauskas et al., ) or (Shaker et al., ) if needed. We

will contextualize our needs in Section 2 where we

will present the energy requirement of a sensor node

and the available ambient WiFi electromagnetic en-

ergy. Then, we will study the main issue of a RF/DC

converter which is the low output voltage level in Sec-

tion 3. Finally, a calculation model based on energy

budget analysis will be deﬁned in Section 4. It is ca-

Grante, F., Abib, G., Muller, M. and Samama, N.

Autonomous Sensor Node Powered over WiFi: A Use Case Study.

DOI: 10.5220/0009804101270132

In Proceedings of the 17th International Joint Conference on e-Business and Telecommunications (ICETE 2020) - DCNET, OPTICS, SIGMAP and WINSYS, pages 127-132

ISBN: 978-989-758-445-9

127

pable of determining the data transmission frequency

based on ambient WiFi energy measurements.

2 CONSTRAINTS

Before taking an in-depth look at used RF/DC con-

verter, we need to contextualize the study by deﬁning

the different requirements for the proper functioning

of an IoT sensor node and a characterization of the

electromagnetic environment around the 2.45 GHz

band.

2.1 Sensor Node

Let us consider here a sensor node as a platform

capable of measuring physical quantities and trans-

mitting its data by radio waves. Great progresses

have been made in recent years on the architectures

and power consumption of microcontrollers, Micro-

Electro-Mechanical-System (MEMS) sensors and RF

transceivers with the emergence of the term ”ultra low

power”.

We will base our study on a platform developed by

ON Semiconductor: RSL10-SOLARSENS-GEVK

(Figure 1). This platform includes the BME280 en-

vironmental sensor from Bosch Sensortec and the

RSL10SIP, a system in a package developed by ON

Semiconductor and consisting of an ARM Cortex-

M3 microcontroller and a Bluetooth Low Energy

transceiver. A white paper (Bruno Damien, 2019) is

dedicated to the feasibility of solar energy harvest-

ing to power this sensor node. Therefore, we want

to know if it is possible to operate this platform by re-

placing the solar panel with a RF/DC converter to ob-

tain a sensor node powered from the WiFi harvested

electromagnetic energy.

The ﬁrst thing to note is the supply voltage. In-

deed, the solar panel already provides a voltage in

the operating range of the platform, which therefore

charges a capacitor storing the energy necessary for

one cycle of data sending. Once a threshold volt-

age (V = 2.6 V ) is reached, a voltage regulator is

activated so that the capacitor can power the sensor

node. Therefore, a voltage higher than 2.6 V must be

reached in order to activate the energy management

system.

Next, we need to quantify the energy required. As

the platform is not permanently powered, ON Semi-

conductor explains there are three phases in the soft-

ware process, each of which has its own energy con-

sumption:

• Boot, consuming 120 µJ

Figure 1: ON Semiconducor RSL10-SOLARSENS-GEVK

sensor platform.

• Measurements, consuming 20 µJ

• Transmission, consuming 40 µJ

As described in the white paper, the required energy

(E) for the three phases above is about 180 µJ. To

store this energy, a very common 100 µF capacitor (C)

can be used. Indeed, the accumulated energy (E

cap

)

in a capacitor is given by Equation (1) and is equal

to 338 µJ in our case, which is enough given to the

required energy.

cap

.C.V

(1)

We will consider an energy of 200 µJ instead of 180 µJ

for the rest of the paper to simplify the calculations.

We can now look at the required conditions for a

proper functioning of the node:

• Have a DC voltage (V ) of 2.6 V.

• Recover an energy (E) of 200 µJ.

Nevertheless, these conditions are obtained by ON

Semiconductor using a solar panel. We now want

to satisfy these conditions when harvesting ambient

WiFi signal.

2.2 Ambient WiFi Power

In France, the legislation limits the transmission

power on all channels in the 2.45 GHz ISM band to

100 mW (20 dBm) (leg, 2018). We want to estimate

the ambient energy that could be recovered, i.e., the

amount of energy that passes through the antenna.

Hence, we did a measurement campaign on a WiFi

router in a normal use. The measurements were per-

formed using the spectrum analyzer Aaronia Spectran

HF-2025E with the OmniLOG30800 antenna at a dis-

tance of 3 meters from the transmitting router. The de-

vice scans all the WiFi channels and Figure 2 shows

the maximum power measured at each scan over a

7 hours period.

We notice a very heterogeneous result. We con-

sider a received power threshold equal to -30 dBm

(1 µW), value below which the received energy is con-

sidered to be negligible. Nearly 25% of the measure-

ments show a power greater or equal to -30 dBm.

WINSYS 2020 - 17th International Conference on Wireless Networks and Mobile Systems

128

Figure 2: Received WiFi power.

By time integration, we can plot the energy accumu-

lation as a function of time, as displayed by Figure

Figure 3: Accumulation of received WiFi energy.

Our spectrum analyzer received almost 25 mJ of WiFi

energy during 7 hours of measurements. Knowing

that we need 200 µJ to send data from our sensor node

(see Section 2.1), we can be optimistic about the pos-

sibility to harvest WiFi signal to power our platform

using a RF/DC converter that we will study in Section

3 RF/DC CONVERTER VOLTAGE

ISSUE

In order to provide the necessary DC power to our

sensor node, we will consider a RF/DC converter

(known also as rectiﬁer) in the 2.45 GHz ISM band.

We focus the study on the conversion circuit and not

on the antenna. Let us ﬁrst of all establish an inven-

tory of the evolution of the DC output voltage (V) and

of the conversion efﬁciency (η), in order to be able

to deﬁne the development axes on which it is neces-

sary to work. We will start with the simplest and well

known RF/DC converter which is a basic crest detec-

tor, based of a single Skyworks SMS7630 Schottky

diode as presented by Figure 4.

Figure 4: RF/DC converter.

For maximum power transfer from the antenna to the

converter, an impedance matching network must be

inserted in order to minimize signal reﬂexion and also

harmonics created by the Schottky diode due to its

non linearities, as presented on Figure 4. This match-

ing network includes transmission lines and induc-

tors as presented on Figure 5 and they are determined

thanks to simulations performed with Keysight Ad-

vanced Design System software.

We will therefore optimize the circuit parame-

ters to maximize the DC output voltage for our load.

Based on ON Semiconductor’s white papers, we can

estimate the equivalent platform load resistance (R) to

be about 1000 Ohm. Since the WiFi input power (P

)

will be converted into DC power (P

), we can be

interested in the evolution of the efﬁciency (η) (Equa-

tion (2)) and of the DC output voltage (V ) of our con-

verter. Simulation results for different arbitrary WiFi

input powers are presented in Table 1.

η =

R.P

(2)

Table 1: Output voltage and efﬁciency of the RF/DC con-

verter.

(dBm) η (%) V (mV)

-30 5.7 8

-25 12 19

-20 20 45

-15 28 94

-10 34.6 186

0 42 648

The main issue with a RF/DC converter is the DC out-

put voltage, which can be very low given the RF input

power levels. Indeed, with the reception power levels

of WiFi signals, the diode sees its efﬁciency drops be-

cause of its lack of sensitivity. We can notice that we

never obtain the needed 2.6 V output voltage since the

Autonomous Sensor Node Powered over WiFi: A Use Case Study

129

Figure 5: RF/DC converter schematic (Keysight ADS).

maximum obtained is 648 mV at 0 dBm. It is there-

fore necessary to work on a more elaborate conver-

sion system to reach acceptable voltage levels. The

state of the art on the subject mainly uses Cockcroft-

Walton voltage doubler assemblies (Kee et al., 2018)

which can be mounted in cascade to raise the voltage

to our needs.

Figure 6: One stage Cockcroft Walton voltage doubler

boost.

To limit the number of Cockcroft-Walton stages re-

quired, we can consider adding a boost like the

BQ25504 from Texas Instruments or the LTC3105

from Analog Devices. We thus lower our target volt-

age at the converter output down to about 300 mV in-

stead of 2.6 V which relaxes our constraints. Indeed,

the boost will provide an output voltage of 3.3 V with

the 300 mV input voltage provided by the converter.

It will then allow us to charge a capacitor up to our

threshold voltage of 2.6 V.

Now that we have some answers about our voltage

constraint, let us have a look on the energy. Having

seen in Section 2.2 that it was possible to power our

system given the amount of the received RF energy at

the antenna, is it still energetically viable after con-

version?

4 ENERGY HARVESTING DUTY

CYCLE

Let’s assume for this part that we have solved our volt-

age concerns, so we can store energy using a capaci-

tor. Then, what would be the duty cycle (T), i.e., how

long would it take the converter to store the required

energy (E) of 200 µJ in the capacitor in order to power

the sensor node to measure and send the data?

4.1 Theoretical Model

To answer to this question, we will deﬁne the duty

cycle (T ) according to Equation (3).

T =

η.P

(3)

Firstly, we will start by considering an ideal converter

with 100% efﬁciency (η = 1). Table 2 presents the

duty cycle T for different P

. So, for P

equal to

-30 dBm, T will be equal to 200 seconds. Conse-

quently, we won’t manage to get better than 200 sec-

onds (3 minutes and 20 seconds) between two trans-

missions with a permanent harvested RF power of

-30 dBm.

Table 2: Duty cycle (T ) for different P

with an ideal con-

verter.

(dBm) T (s)

-30 200

-20 20

-10 2

0 0.2

When considering our real converter, whose efﬁ-

ciency is given in Table 1, the estimated duty cycle

(T ) is presented in Table 3.

Table 3: Duty cycle (T ) for different P

with a real con-

verter.

(dBm) η (%) T (s)

-30 5.7 3509

-20 20 100

-10 34.6 6

0 42 0.48

We therefore notice a drastic drop in performance

with a duty cycle increasing from 200 seconds to

3509 seconds (around 58 minutes) for P

equal to

-30 dBm.

To get to the end of our approach, let’s assume

that we use the LTC3105 boost from Analog Devices

as mentioned in Section 3. The component datasheet

speciﬁes a minimum efﬁciency (η

) of 65% for an in-

put voltage lower than 1 V which should be the case in

WINSYS 2020 - 17th International Conference on Wireless Networks and Mobile Systems

130

the light of the study carried out in Section 3. Then,

let us compute the full efﬁciency (η.η

) to estimate

the duty cycle (T) in this worst case, as presented in

Table 4.

Table 4: Duty cycle (T ) when taking into account the boost

efﬁciency.

(dBm) η.η

(%) T (s)

-30 3.7 5406

-20 13 154

-10 22.5 9

0 27.3 0.73

For a constant RF input power of -30 dBm, the re-

sult indicates a duty cycle of 5406 seconds (around

1 hours and 30 minutes) between two transmissions

when using our RF converter associated to the DC

boost.

The next step is to consider a non-constant har-

vested RF power to have a closer approach for a real

conditions use case.

4.2 Application of the Model

In order to determine the obtained amount of DC en-

ergy harvested from a real WiFi signal, like one dis-

played on Figure 2, we have ﬁrst to model the efﬁ-

ciency η = F(P

) of our converter. We can apply a

polynomial ﬁtting on the data given in table 1 and

then, use the obtained model to determine the con-

verter efﬁciency associated to the measured WiFi in-

put powers. The available DC energy could be ob-

tained after time integration of P

, determined using

Equation (2) and displayed on Figure 7. Vertical lines

indicate each time the capacitor stores enough energy

(200 µJ) to power the sensor node.

Figure 7: Amount of DC energy available to power the sen-

sor node.

We can consider 11 transmissions on a 7-hour mea-

surement of ambient WiFi signal. It is equivalent to

one data transmission every 38 minutes and 11 sec-

onds on average.

5 CONCLUSION

The objective of this work is to propose and study an

energy budget analysis protocol to power an IoT sen-

sor node by harvesting ambient electromagnetic WiFi

waves. For that, a classical crest detector based on

a Schottky diode is choosed as a RF/DC converter.

After determining the sensor power requirements and

characterizing the available ambient WiFi energy, the

RF/DC converter is investigated in order to optimize

its conversion efﬁciency through simulations.

Even if the availability of electromagnetic energy

is non constant, our work shows that powering such a

system with ambient WiFi signals is energetically vi-

able despite a low converter efﬁciency of 3.7% with a

RF input power of -30 dBm. According to the charac-

teristics of our sensor node, it seems that it is possible

to consider 11 data transmissions on a 7-hour harvest-

ing of ambient WiFi signal.

However, this study does not allow us to operate

the sensor node since an optimization work is neces-

sary at the converter level to reach the 300 mV thresh-

old. It will also be necessary to characterize the per-

formance of the converter as a function of the distance

to the nearest WiFi transmitting terminal. The use of a

converter that extends over several frequency ranges,

such as for example combining the 2.45 GHz ISM

band with GSM 1800 (Berg

es et al., 2015) can be a

way to improve performances.

A study of the different sensors and microcon-

trollers (Mouapi and Hakem, 2016) in the industry

will also have to be carried out to compare and ensure

that we have an optimal software layer with regard to

power consumption.

REFERENCES

(2018). Les niveaux d’exposition WiFi.

http://www.radiofrequences.gouv.fr/

les-niveaux-d-exposition-a73.html.

Berg

es, R., Fadel, L., Oyhenart, L., Vigneras, V., and Taris,

T. (2015). A dual band 915MHz/2.44GHz RF energy

harvester. In 2015 European Microwave Conference

(EuMC), pages 307–310. ISSN: null.

Bruno Damien, Tobias Raimar, L. M. (2019). Continu-

ous Harvesters and ON Semiconductor’s Low-Power

rf technology close the gap in environmental and ac-

celerometer Sensors for iot.

Autonomous Sensor Node Powered over WiFi: A Use Case Study

131

Ho, D.-K., Kharrat, I., Ngo, V.-D., Vuong, T.-P., Nguyen,

Q.-C., and Le, M.-T. (2016). Dual-band rectenna for

ambient RF energy harvesting at GSM 900 MHz and

1800 MHz. In 2016 IEEE International Conference

on Sustainable Energy Technologies (ICSET), pages

306–310. ISSN: null.

Kee, C. P., Olule, L., and Gnanagurunathan, G. (2018).

Microstrip Patch Antenna and Three-stage Cockcroft-

Walton Rectenna for Wi-Fi Energy Harvesting. In

2018 IEEE International RF and Microwave Confer-

ence (RFM), pages 242–245. ISSN: null.

Krakauskas, M., Sabaawi, A. M., and Tsimenidis, C. C.

Suspended patch microstrip antenna with cut rect-

angular slots for RF energy harvesting. In 2014

Loughborough Antennas and Propagation Conference

(LAPC), pages 304–307. ISSN: null.

Kurvey, M. and Kunte, A. Design and optimization of

stepped rectangular antenna for RF energy harvest-

ing. In 2018 International Conference on Commu-

nication information and Computing Technology (IC-

CICT), pages 1–4. ISSN: null.

Mouapi, A. and Hakem, N. (2016). Performance evalua-

tion of wireless sensor node powered by RF energy

harvesting. In 2016 16th Mediterranean Microwave

Symposium (MMS), pages 1–4. ISSN: 2157-9830.

Parks, A. N., Sample, A. P., Zhao, Y., and Smith, J. R.

(2013). A wireless sensing platform utilizing ambi-

ent RF energy. In 2013 IEEE Topical Conference

on Biomedical Wireless Technologies, Networks, and

Sensing Systems, pages 154–156. ISSN: null.

Shaker, M. F., Ghali, H. A., Elsheakh, D. M. N., and El-

sadek, H. A. E. Multiband coplanar monopole antenna

for energy harvesting. In 2018 IEEE International

Symposium on Radio-Frequency Integration Technol-

ogy (RFIT), pages 1–3. ISSN: null.

WINSYS 2020 - 17th International Conference on Wireless Networks and Mobile Systems

132