sensors
Article

On-Demand Sensor Node Wake-Up Using Solar
Panels and Visible Light Communication
Carolina Carrascal 1 , Ilker Demirkol 1,2, * and Josep Paradells 1,2
1
2

*

Department of Telematics Engineering, Universitat Politecnica de Catalunya, Barcelona 08034, Spain;
carolina.andrea.carrascal@alu-etsetb.upc.edu (C.C.); josep.paradells@entel.upc.edu (J.P.)
Fundacio i2CAT, Barcelona 08034, Spain
Correspondence: ilker.demirkol@entel.upc.edu; Tel.: +34-93-4011-055

Academic Editor: Leonhard M. Reindl
Received: 26 December 2015; Accepted: 14 March 2016; Published: 22 March 2016

Abstract: To significantly reduce, or eliminate completely, the energy waste caused by the standby
(idle) mode of wireless sensor nodes, we propose a novel on-demand wake-up system, which allows
the nodes to be put into sleep mode unless their activation is truly necessary. Although there
have been many studies proposing RF-based wake-up radio systems, in this work, we develop the
first visible light communication (VLC)-based wake-up system. The developed system can extend
the existing VLC systems and can be exploited to derive new application areas such as VLC tags.
The system uses an off-the-shell indoor solar panel as receptor device of the wake-up signal as well as
for energy harvesting purposes, through which it is able to harvest enough energy for its autonomous
work. The design, implementation details and the experimental evaluation results are presented,
which include flickering characterization and wake-up range evaluations. The results show that the
developed system achieve reasonable wake-up distances for indoor environments, mainly where the
use of VLC systems are considered.
Keywords: wake-up radio; wake-up receiver; Visible Light Communication (VLC); solar panel;
energy harvesting

1. Introduction
To achieve longer battery life of wireless sensor nodes, a common method employed is
duty-cycling. In duty-cycling, the wireless nodes listen to the channel for potential incoming
communications turning their radios on periodically, and remaining in sleep mode the rest of the time.
It has been shown that this approach can better preserve energy, however, it causes energy waste when
the device wakes up and there is no data to transmit or receive. A more efficient solution consists of
on-demand communication, i.e., totally asynchronous and rendezvous-less communication, achieved
by wake-up radios. Wake-up radios are small and low- or no-power devices that activate the wireless
sensor nodes when an external wake-up call occurs. The use of a wake-up receiver permits the wireless
nodes to remain in sleep mode as long as possible. It has been shown that wake-up radio enables a
more energy-efficient approach compared to duty-cycling mechanisms, due to the elimination of the
useless idle periods of the node [1].
The harvesting of energy from the environment, however, becomes an interesting option
in wake-up receiver devices in the path towards an energy-autonomous communication system.
A relatively less explored energy source for harvesting energy is indoor lighting, although many
wireless communication applications happen in indoor environments. With current indoor solar
panel technology, the amount of power harvested is limited (e.g., less than 90 µW by the solar panel
evaluated in this study for a light intensity of 200 lux). However this amount of power can be used for
very low power communication devices such as wake-up receivers, e.g., less than 30 µW is required
Sensors 2016, 16, 418; doi:10.3390/s16030418

www.mdpi.com/journal/sensors

Sensors 2016, 16, 418

2 of 19

by the system proposed in this study. In this work, we propose the use of solar panel to develop a
wake-up communication system using Visible Light Communication (VLC) as communication channel
of the wake-up signal. Our goal is to create a wake-up system where an indoor solar panel is used
both as the receptor of the VLC signal and also as energy harvester. To the best of our knowledge,
this is the first wake-up communication system that exclusively uses light harvesting, paving the way
to a new research direction.
In a preliminary work, we presented the first stage in the development of this system [2], where we
described the general wake-up system and the components used. In this work we describe the complete
system and use the indoor solar panel also for harvesting the energy for the wake-up device operation.
This novel wake-up communication system will then allow: (i) the use of light as a wireless channel
for the wake-up signal; (ii) to put the wireless sensor nodes to deep sleep mode to conserve energy,
until it is really necessary to enter in active mode; (iii) to enable the wake-up device to work through
an addressable wake-up, i.e., waking up only the wireless nodes that are destined; (iv) to contribute
the research direction of battery-less wake-up receiver by using the solar panel both for charging and
communication purposes; and (v) to enable novel applications such as light-driven localization, asset
control, etc.
To devise a feasible and efficient solution, we evaluated different system component alternatives
through physical experiments. Specifically, we investigated the performance trade-offs brought by
the solar panel size, the use of a photodiode as a communication receptor alternative, and the choice
of correlator for addressable wake-ups. The characterization of the off-the-shelf indoor solar panel
chosen is performed for its frequency response and current, voltage and power generated for different
load values, based on which operational frequency of the system is chosen along with the wake-up
receiver circuit design.
A low-cost and flexible VLC wake-up transmitter solution is developed through a small
form-factor, wirelessly controllable VLC transmitter module devised in this paper. Through physical
experiments, the complete system performance is evaluated in terms of wake-up probabilities achieved
per transmitter-receiver distances, resiliency of the system to interference from fluorescent light,
which is one of the most common lighting technology used in the office environments. A crucial
challenge of VLC is flickering of the light, especially for low data-rate communication such as the
one targeted in this work. The flickering characterization of the system is presented, along with the
evaluation of possible mitigation solutions. Finally, different system configurations such as bit duration
settings, use of capacitors for energy storage are assessed for detailed system performance evaluations.
With the final configuration proposed, a 14 m wake-up distance is achieved in scenarios without
interferences. Moreover, with the proposed flickering mitigation techniques, a 7 m wake-up distance is
achieved with no noticeable flickering. According to the experiment results, the developed system
provides feasible operational distance for many indoor applications, such as asset control, tracking
and monitoring of devices.
The rest of the document is structured as follows: Section 2 gives an overview of the state of art of
related work. Section 3 presents the system design and the implementation details of the wake-up
system. In Section 4, we present the experimental results. Finally, Section 5 concludes the paper and
presents potential future work.
2. Related Work
The use of visible light as medium of communication is known as Visible Light Communication
(VLC). Current efforts in the IEEE 802.15.7 study group [3] includes low-rate communication, which is
one focus of this current paper, specifically using VLC as a channel to wake up a sleeping wireless
node. In [4], the use of a solar cell as a simultaneous Visible Light Communication and optical energy
receiver is demonstrated. In that work, a characterization of a solar panel is made and a distance of
40 cm is achieved with a VLC data rate of 3 Kbps. The experiment is set with the use of a frequency
generator at transmitter side and a solar cell and oscilloscope at receiver side. In this work, we propose

Sensors 2016, 16, 418

3 of 19

a complete system including transmitter and receiver that can be easily implemented as part of a
wirelessly controlled lighting infrastructure, with a wake-up distance of up to 14 m.
Wireless wake-up system studies in the literature mainly focus on radio frequency (RF) triggered
approaches. There are few studies in the literature where a different communication mode other than
RF is used for wake-up communication. A studied mode is Free-Space Optical (FSO) communication,
which covers both the VLC and the infrared (IR) spectrum. In [5], a device that uses a FSO
communication as a wake-up channel is described. The device is designed to be used within indoor
networks, and requires line of sight with a highly directional communication between transmitter
and receiver. The wake-up receiver is equipped with photodiodes, which are used as signal receivers.
The transceiver in a receive mode has a power consumption of 317 µW. This proposal reaches a
communication range of 15 m with a transmitter with a power consumption of 16.5 mW. Compared
to [5], the VLC wake-up receiver presented in this paper includes the following novel aspects: (i) the use
of an indoor solar panel as receptor of the signal, which we show to perform better than photodiodes
for wake-up purposes; and (ii) the harvesting of light energy for feeding purposes.
In [6], an active wake-up receiver that works with infrared signals is proposed. Depending on the
transmitter signal intensity, the operational range of wake-up signal reaches up to 30 m. This device
has a consumption of 12 µW while waiting for an incoming signal. Ambient light is found to saturate
the receiver and increase the power consumption. In the case of our implementation, we exploit the
lighting infrastructure for lighting, communication and energy harvesting purposes.
In [7], a 695 pW standby power optical wake-up receiver for wireless sensor nodes is introduced.
The receiver consumes 695 pW in standby mode and in active mode 140 pJ/bit at 91 bps. A pulse
width modulated communication encoding scheme is used, and chip-ID masking enables selective
batch-programming and synchronization of multiple sensor nodes. Using a laser at transmitter side,
a distance of 50 m is reached. Using a 3 W LED focused it reach a distance of 6 m. In [8], a 4.4 µW
wake-up receiver using ultrasonic communication is presented. The device achieves a distance of up
to 8.6 m for a data rate of 0.25 kb/s with ´18 dBm signal power going into the transmit transducer.
In our work, we use solar panel which paves way to a wake-up receiver of ultra-low power
consumption. A comparison of the different wake-up communication modes is given in Table 1.
For comparison purposes, the details of a recent study on RF-based wake-up receiver [9] is
also included.
Table 1. An Illustrative Comparison of Different Wake-up Communication Mode Studies.
[9]

[6]

[7]

[8]

This Work

Comm. Mode

RF

Infrared

VLC

Ultrasound

VLC

Power Source at Receiver

Sensor’s
battery

Coin cell
battery

Sensor’s
battery

Sensor´s
battery

Power harvested
from light

Receiver Power

22.9 µW

12 µW

695 pW

4.8 µW

~95 µW

16 µW

87.9 mW (Only
LED controller)

Transmitter Power

N/A

~60 mW (IR
remote control)

3.5 mW (Only
laser, controller
power N/A)

Max Data Rate

200 Kbps

N/A

91 bps

250 bps

1.12 Kbps

Range

´75 dBm
(Sensitivity)

30 m

50 m

8.6 m

14 m

Requirements

Battery
required for
the receiver

Limited
interference

Highly
directional link
for laser

Other energy
harvesting
source

Targeted for
indoor systems

3. Solar Panel and VLC Based Wake-Up Communication System: Design and Implementation
3.1. System Design
Wake-up communication systems mainly consist of four components: Wake-up transmitter
(WuTx), wake-up receiver (WuRx), the data communication modules attached to WuTx and to the

Sensors 2016, 16, 418

4 of 19

WuRx. In certain implementations, WuTx and/or WuRx is part of the data communication module
as in [1]. WuTx and WuRx hardware might also be implemented jointly into a wake-up transceiver.
In this work, our implementation will include the four components, implementing separate WuTx and
WuRx components.
Sensors 2016, 16, 418
4 of 18
Figure 1 illustrates the wake-up communication system operation. First, the transmitter sends a
signal intended illustrates the wake-up communication system operation. First, the transmitter sends a
Figure 1 for the wireless node (Step 1). The wake-up receiver is constantly sensing the channel
(Step 2) and uponfor the wireless node (Step 1). The wake-up receiver is constantly sensingcompares the
signal intended reception of a signal, it demodulates the data within the signal and the channel
address included in the data to its address. If the addresses match, the WuRx triggers a pulse through
(Step 2) and upon reception of a signal, it demodulates the data within the signal and compares the
an address included in the data to itsto wake up the wirelessmatch,attached, notifying pulse through
interrupt pin (Step 3) in order address. If the addresses node the WuRx triggers a the node about
an an interrupt pin (Step 3) in order to wake up the wireless wakes up (Step 4) and is ready to perform
incoming communication. Finally, the wireless node node attached, notifying the node about an
theincoming communication. Finally,of a wireless node wakes up (Step 4) and is ready to perform the
data communication. The use the wake-up device permits the nodes to remain in an energy
data communication. The use of a wake-up node’s MCU/CPU put to a remain in mode to
conserving stage as long as possible, where thedevice permits the nodes to deep-sleepan energybe
conserving stage as long as possible, and the main data transceiver to a deep-sleep mode to be
only triggered by an external interruptwhere the node’s MCU/CPU putis switched off. To achieve an
only triggered by an external interrupt and the main data transceiver is switched off. To achieve an
energy-efficient operation with a low form factor overhead, the wake-up receivers are required to have
energy-efficient operation with a low form factor overhead,
a small footprint with an ultra-low or no power consumption.the wake-up receivers are required to
have a small footprint with an ultra-low or no power consumption.
3
1

",

4

frr

Figure 1. Wake-up communication system operation.
Figure 1. Wake-up communication system operation.

In this study, we we targetuse use visible light ascommunication medium (Step 1), while an indoor
In this study, target to to visible light as the the communication medium (Step 1), while an
solar-panel is used as theused as the wake-up signal (Step 2). To realize 2). To realize communication
indoor solar-panel is receptor of receptor of wake-up signal (Step the wake-up the wake-up
communication system targeted, the required system components Figure 2. At in transmitter side,
system targeted, the required system components are presented inare presented theFigure 2. At the
thetransmitter side, theare a Control Module a Control Module with a logic for the modulation of Drive
main components main components are with a logic for the modulation of the light, a LED the
light, for varying the amount of current passing through passing through LED used for LED
Module a LED Drive Module for varying the amount of current the LED, and a the LED, and a indoor
used purposes.
lightingfor indoor lighting purposes.

Figure 2. 2. Global systemview of the solar-panel-based VLC wake-up system.
Figure Global system view of the solar-panel-based VLC wake-up system.

The receiver side is comprised an indoor solar panel, which works as both a receptor and as
The receiver side is comprised ofof an indoor solar panel, which works as both a receptorand as an
an energy harvester. The other components of the receiver modules in charge charge of the signal
energy harvester. The other components of the receiver are theare the modules in of the signal detection,
detection, its demodulation and correlation, i.e., comparison with the local ID code (address). Once
the wake-up interrupt signal is triggered and consequently the wireless node is awoken, the
communication between transmitter and receiver can be performed using the main data
communication interface of the nodes. The design presented in this work can be easily adapted to
work with different application requirements. For example, by adding a secondary reception path,

Sensors 2016, 16, 418

5 of 19

its demodulation and correlation, i.e., comparison with the local ID code (address). Once the wake-up
interrupt signal is triggered and consequently the wireless node is awoken, the communication
between transmitter and receiver can be performed using the main data communication interface of
the nodes. The design presented in this work can be easily adapted to work with different application
Sensors 2016, 16, 418
5 of 18
requirements. For example, by adding a secondary reception path, once the receiver node is awoken,
the system receiver node is awoken, the communication through the use of LED and indoor solar panel.
once the can support downlink data system can support downlink data communication through the
use of LED and indoor solar panel.
3.2. System Implementation Details
3.2. System Implementation Details
3.2.1. WuRx Receptor: Indoor Solar Panel
3.2.1. WuRx Receptor: Indoor Solar Panel
For the design of our system, the choice, and therefore the characterization, of the indoor solar
panel is crucial since of our system, the choice, and therefore To harvest the energy to feed thesolar
For the design this device has two important tasks: the characterization, of the indoor WuRx
device andcrucial since this device has two important tasks: To of power that can be harvested by the
panel is to receive the VLC signal. Note that, the amount harvest the energy to feed the WuRx
indoor solar panel is a determining factor forthat, selection of thepower receiver be harvested by the
device and to receive the VLC signal. Note the the amount of other that can components.
After preliminary determining selected the selection of the other receiver components.
indoor solar panel is a studies, we factor for an off-the-shelf indoor solar panel for wide-range of
availability and for its good operational performance as indicated in solar panel for The chosen solar
After preliminary studies, we selected an off-the-shelf indoor the following. wide-range of
availability and for its good operational performance as indicated in the following. The chosen solar
panel is SC3726I-8-1 (Blue Solar Company, Dongguan, China), which is an amorphous silicon solar
panel is rectangular shape of size 3.6 cm ˆ 2.6 cm and of thickness amorphous silicon circuit
cell with aSC3726I-8-1 (Blue Solar Company, Dongguan, China), which is an 0.1 cm. The open solar
cell with a rectangular shape of size 3.6 to × around 5.6 thickness short The open circuit voltage
voltage of the solar panel is measured cm be2.6 cm and of V and its 0.1 cm. circuit current is 34.6 µA.
In of the solar panel is measured to be around 5.6 V and itsindoor solar currentand compare orderother
order to characterize the performance of the chosen short circuit panel is 34.6 μA. In it to to
characterize the tests were conducted: (i) current, voltage and power characteristics for different
alternatives, four performance of the chosen indoor solar panel and compare it to other alternatives,
four values; (ii) frequency current, characterization characteristics for different resistor values;
resistortests were conducted: (i)responsevoltage and powerof the indoor solar panel for different VLC
(ii) frequency response characterization of the indoor with a photodiode used as a modulation
modulation frequencies; (iii) a performance comparisonsolar panel for different VLC VLC receiver;
frequencies; (iii) a performance comparison with a photodiode used as a VLC receiver; and (iv) its
and (iv) its performance comparison with an off-the-shelf indoor solar panel with a larger surface area.
performance comparison with an off-the-shelf indoor solar panel with a larger surface area.
The first test lets us find out the maximum power of the solar panel in an indoor environment
The first test lets us find out the maximum power of the solar panel in an indoor environment
with a measured light intensity of 200 lux. This light intensity value is chosen as it corresponds to a
with a measured light intensity of 200 lux. This light intensity value is chosen as it corresponds to a
typical indoor environment light intensity level [10]. As shown in Figure 3, different resistor values
typical indoor environment light intensity level [10]. As shown in Figure 3, different resistor values
were tested and the maximum power is found to to be 96.9 μW withload of of 150 KΩ. For the second
be 96.9 µW with a a load 150 KΩ. For the second test,
were tested and the maximum power is found
the goal was to identify the responseresponse of the indoor solardifferent frequencies to characterize
of the indoor solar panel to panel to different frequencies to
test, the goal was to identify the
and choose operational frequency for VLC communication. In the test performed, light modulated
characterize and choose operational frequency for VLC communication. In the test performed, light
with different frequencies were projected to the panel.the panel. The frequency responsesolar panel is
modulated with different frequencies were projected to The frequency response of the of the solar
characterized by the amplitude of the AC component, i.e., Vpp (peak-to-peak voltage) of the signal
panel is characterized by the amplitude of the AC component, i.e., Vpp (peak-to-peak voltage) of the
relayed by the solarthe solar panel.
signal relayed by panel.
100
90

Voltage | Current | Power

80

Voltage (V)

70

Current (µA)

60

Power (µW)
50
40
30
20
10
0
10

100

1000

10000

Load RL (KΩ, log scale)

Figure Current, voltage and power characteristics of indoor solar panel evaluated for different
Figure 3. 3. Current, voltageand power characteristics of indoor solar panel evaluated for different
resistor values.
resistor values.

As seen in Figure 4, lower frequencies provide higher Vpp values, as also was observed in [4].
In the test, a 10 W LED was used, with a distance of 1 m between LED and the indoor solar panel,
and no interference from other sources of light scenario exists. Based on the measured frequency
response and the operational frequencies supported by the capabilities of the other receiver
components discussed later, we chose to work at 21 kHz. As shown in Section 4, this choice enables
the system to work with decent wake-up ranges.

Sensors 2016, 16, 418

6 of 19

As seen in Figure 4, lower frequencies provide higher Vpp values, as also was observed in [4].
In the test, a 10 W LED was used, with a distance of 1 m between LED and the indoor solar panel, and no
interference from other sources of light scenario exists. Based on the measured frequency response and
the operational frequencies supported by the capabilities of the other receiver components discussed
later, we chose to work at 21 kHz. As shown in Section 4, this choice enables the system to work with
Sensors 2016, 16, 418
6 of 18
decent wake-up ranges.
Sensors 2016, 16, 418

6 of 18

Amplitude of AC omponent (mV)
Amplitude of AC omponent (mV)

1000
1000
800
800
600
600
400
400
200
200
0
0 0
0

5

10

15

20

25

30

35

40

5

10

Frequency (KHz)
15
20
25

30

35

40

Frequency (KHz)
Figure 4. 4. Indoor solarpanel responseto different frequencies.
Figure Indoor solar panel response to different frequencies.
Figure 4. Indoor solar panel response to different frequencies.

In the third test, we compare the indoor solar panel to an off-the-shelf high-speed, highly
In the third test, we compare the indoor solar panel to an off-the-shelf high-speed, highly sensitive
sensitive PIN photodiode. The photodiode selected is the photocell off-the-shelf high-speed, highly
In the third photodiode selected is the solar panel to an BPW34 (Vishay Semiconductors,
PIN photodiode. Thetest, we compare the indoor photocell BPW34 (Vishay Semiconductors, Malvern,
Malvern, PIN USA) [11], with a photodiode selected is the photocell BPW34 (Vishayof 30 cm between
sensitive PA, photodiode. The size of 3 mm × 3 mm. In the test we use a separation Semiconductors,
PA, USA) [11], with athe photodiode/indoor solar panel.test we use a separation of 30 cm between the
the 10 W PA, and size of 3 mm ˆ 3 mm. In the Different frequencies are tested to analyze the
Malvern, LEDUSA) [11], with a size of 3 mm × 3 mm. In the test we use a separation of 30 cm between
10 Wfrequency response of both options. Figure panel. Different frequenciesphotocell responses to the
LED and the photodiode/indoor solar 5 shows the solar panel and are tested to analyze
the 10 W LED and the photodiode/indoor solar panel. Different frequencies are tested to analyze the
frequency response of both both graph we Figure 5 showssolarVpp of andAC component at the output
different frequencies. In options. Figure 5 observethe the solar panel photocell responses to different
frequency response of the options. can shows that the panel the and photocell responses to
frequencies. In the graph wethan that ofcan that thesolarthe Vpp of the ACacomponentthearound 8.6 the
of the photodiode is In the graph we the indoor Vpp of the AC component at at output of
different frequencies.lower can observe observe that panel. Starting at frequency of the output
kHz, photodiode is lower the that at the solar of the Starting at frequency a around to
photodiode is lower than that ofsignalof the indoor panel.photodiode becomes flat at ofvalue close8.6 kHz,
of thethe AC component of thanthe indoor output solar panel. Startingaat a frequency of around8.6
0 mV. Even component of the at the faster of the photodiode becomes flat a a value to
kHz, the AC though photodiodes have outputresponse, their achieved Vpp is not high enough close
the AC component of the signalsignal atathe output of the photodiode becomes flat at atvalue closefor to
the demodulating signal and have a a faster response, their be generated
proven enough for
0 EvenEven though photodiodes havefaster response, their achieved Vpp as not highwith further the
0 mV. mV. though photodiodes the wake-up interruption cannotachieved Vpp is not highenough for
tests. The use of signal and the panel also eliminates the need generated
the demodulatingan indoor solarwake-up interruption cannot generated asas proven withfurther tests.
be for any amplifier stage after the
demodulating signal and the wake-up interruption cannot be
proven with further
reception use of an indoor solar panel also eliminates the need for any amplifier stage after the
tests. The of the signal.

The use of an indoor solar panel also eliminates the need for any amplifier stage after the reception
reception of the signal.
of the signal.
500
500

Vpp of ACof AC component (mV)
Vpp component (mV)

400
400

Solar panel

300

Photodiode
Solar panel

300

Photodiode

200
200
100
100
0
0

5

0

5

0

10
Frequency (KHz)
10
Frequency (KHz)

15
15

Figure 5. Response of indoor solar panel vs. photodiode to a VLC signal modulated with different

Figure 5. Response of indoor solar panel vs. photodiode to a VLC signal modulated with
frequencies.
Figure 5. Response of indoor solar panel vs. photodiode to a VLC signal modulated with different
different frequencies.
frequencies.

The fourth test compares the indoor solar panel chosen with a larger panel from the same
manufacturer, whichcompares the indoor solar panel chosen with a larger panel from the same
The fourth test is also an amorphous silicon solar cell. Figure 6 shows the responses for the
small and the large indoor solar panels at a distance of 1.8 m from the LED, and the interferences from
manufacturer, which is also an amorphous silicon solar cell. Figure 6 shows no responses for the
external light sources. In solar panels at a distance of 1.8 m from the signal sent and the upper line
small and the large indoor the figures, the lower line represents the LED, and no interferences from
external light sources. In the figures, the lower line represents the signal sent and the upper line

Sensors 2016, 16, 418

7 of 19

The fourth test compares the indoor solar panel chosen with a larger panel from the same
manufacturer, which is also an amorphous silicon solar cell. Figure 6 shows the responses for the
small 2016, 16, 418
Sensors and the large indoor solar panels at a distance of 1.8 m from the LED, and no interferences
7 of 18
from external light sources. In the figures, the lower line represents the signal sent and the upper
represents the the signal detected by the solar As seen in Figure Figure 6, in Vpp in the case of
line representssignal detected by the solar panel. panel. As seen in 6, the Vppthethe case of the large
panel is lower is lower than small the small reason for reason for this that the is that the large
the large panelthan that of thethat of panel. Thepanel. The this behavior is behaviorlarge solar panel
maintains more stable signal level in the output, output, not the fast variations of light signal
solar panelamaintains a more stable signal level in thenot relaying relaying the fast variations of light
amplitude values to the to the Due to Due to this behavior, the maximum distance reached with
signal amplitude valuesoutput. output. this behavior, the maximum distance reached with the large
panel is expected expected to than the one reached with the small the panel, which is verified is
the large panel is to be shorterbe shorter than the one reached withsolarsmall solar panel, whichby
the experiments presented presented in Section 4. that, also that, the DC voltage level of large solar
verified by the experiments in Section 4. Note also Note the DC voltage level of large solar panel is
slightly higher as expected, although the Vpp values are important in VLC systems for a more
panel is slightly higher as expected, although the Vpp values are important in VLC systems fora more
successful demodulation.
successful demodulation.

(a)

(b)
Figure 6. The indoor solar panel responses for (a) a small solar panel with dimensions 3.6 × 2.6 × 0.1 cm
Figure 6. The indoor solar panel responses for (a) a small solar panel with dimensions 3.6 ˆ 2.6 ˆ 0.1 cm
and (b) a large indoor solar panel with dimensions 7.5 × 5.3 ˆ 0.1 cm.
and (b) a large indoor solar panel with dimensions 7.5 ˆ 5.3 × 0.1 cm.

3.2.2. WuRx Envelope Detector, Demodulator and Correlator
3.2.2. WuRx Envelope Detector, Demodulator and Correlator
We investigated the off-the-shelf ultra-low power component alternatives wake-up receiver
We investigated the off-the-shelf ultra-low power component alternatives for the for the wake-up
receiver parts, that is, fordetector, demodulator and correlator. Although many options many the total
parts, that is, for envelope envelope detector, demodulator and correlator. Although exist, options
exist, the total amount of energy consumed by all the individual components exceeds the generated
amount of energy consumed by all the individual components exceeds the amount of energyamount of
energy generated by the indoor that, the power that, the power harvested from indoor light is much
by the indoor solar panel. Notesolar panel. Noteharvested from indoor light is much lower than the
lower than creating light, creating indoor light harvesting systems. As systems. the indoor solar
solar light, the solar a challenge ona challenge on indoor light harvestingreference, As reference, the
indoor solar our implementation generates the generates the maximum power voltage of 3.76 a
panel used inpanel used in our implementationmaximum power of 96 µW with aof 96 μW with V
voltage of 3.76 V26.6 µA, as shown in Section shown in Section 3.2.1.
and a current of and a current of 26.6 μA, as 3.2.1.
Although ultra-low power components such as the 74VHC595 8-bit shift register [12] which only
Although ultra-low power components such as the 74VHC595 8-bit shift register [12] which only
requires 4 μA as input current and a source voltage of 3 V, are available in the market, to build a
requires 4 µA as input current and a source voltage of 3 V, are available in the market, to build a
wake-up receiver, several other components such as two 8-bit comparators, a rectifier, a Manchester
wake-up receiver, several other components such as two 8-bit comparators, a rectifier, a Manchester
demodulator, a local register for the local ID are required, resulting a a lack power budget to to feed
demodulator, a local register for the local ID are required, resulting in inlack of of power budget feed all
all needed components. A more power-efficient approach is to utilize an integrated chip that include
the the needed components. A more power-efficient approach is to utilize an integrated chip that
include all these components. After a detailed search, the decision was to use an off-the-shelf ultralow power, high sensitivity radio wake-up receiver, the AS3933 chip [13] manufactured by AMS, as
part of the receiver system. Feeding the electrical signal generated by the solar panel to the antenna
input of AS3933 was verified to generate a functional receiver circuit.
AS3933 is internally composed of, aside from other modules, an envelope detector, an OOK

Sensors 2016, 16, 418

8 of 19

all these components. After a detailed search, the decision was to use an off-the-shelf ultra-low power,
high sensitivity radio wake-up receiver, the AS3933 chip [13] manufactured by AMS, as part of the
receiver system. Feeding the electrical signal generated by the solar panel to the antenna input of
AS3933 was verified to generate a functional receiver circuit.
Sensors 2016, 16, 418
8 of 18
AS3933 is internally composed of, aside from other modules, an envelope detector, an OOK
demodulator, and a 16-bit correlator as required by the receiver system. The power consumption of
this device fits well with the power requirements desired in the project as it requires around 2 μA
this device fits well with the power requirements desired in the project as it requires around 2 µA
current consumption in listening mode and around 8 μA in correlation mode working at 2.4–5 V.
current consumption in listening mode and around 8 µA in correlation mode working at 2.4–5 V.
AS3933can be configured to work inin one of the bands within the possible carrier frequency ranges
AS3933 can be configured to work one of the bands within the possible carrier frequency ranges of
of between 15 kHz and kHz. We choose to to work in frequency band of of 15 kHz to 23 kHz due
between 15 kHz and 150150 kHz. We choosework in the the frequency band15 kHz to 23 kHz due to
to better response of of indoor solar panel to to low frequencies as as described in Section 3.2.1.
the the better responsethe the indoor solar panelthe the low frequenciesdescribed in Section 3.2.1.
AS3933 requires specific format for wake-up signal, which is depicted in Figure 7. The signal
AS3933 requires aaspecific format for wake-up signal, which is depicted in Figure 7. The signal
is composed of four parts: Carrier burst, a separation bit with the length of half Manchester symbol,
is composed of four parts: Carrier burst, a separation bit with the length of half Manchester symbol,
a preamble of six bits (with the 101010 sequence), and the ID code or address of the node intended
a preamble of six bits (with the 101010 sequence), and the ID code or address of the node intended to
be woken up (pattern). In In addition to the triggering of a wake-up interrupt on address match
to be woken up (pattern). addition to the triggering of a wake-up interrupt basedbased on address
(correlator ON), ON), the chip can configured to generate a wake-up interrupt without an address
match (correlatorthe chip can also bealso be configured to generate a wake-up interrupt without an
matching, using using only the detection of carrier frequency the trigger of the wake-up interrupt
address matching, only the detection of carrier frequency as as the trigger ofthe wake-up interrupt
(correlator OFF). This latter feature be be used in applications that require a wake-up of nodes
(correlator OFF). This latter feature cancan used in applications that require a wake-up of severalseveral
nodes at the same time, providing a broadcast wake-up
at the same time, providing a broadcast wake-up feature.feature.

Figure 7. AMS AS3933 wake-up signal format.
Figure 7. AMS AS3933 wake-up signal format.

The next step to complete the receiver system is to devise necessary circuitry to connect its
The next step to complete the receiver system is to devise necessary circuitry to connect its
components. We deduced two feasible ways to connect the components and complete the WuRx
components. We deduced two feasible ways to connect the components and complete the WuRx
design (Figure 8). In the first circuit design (Figure 8a), a capacitor, C, is used as the storage of the
design (Figure 8). In the first circuit design (Figure 8a), a capacitor, C, is used as the storage of the
power and is connected directly to the VCC and GND pins of AS3933 chip. A diode, D, is used to
power and is connected directly to the VCC and GND pins of AS3933 chip. A diode, D, is used to
prevent the capacitor from discharging through the solar cell and aaresistor R is used where aaportion
prevent the capacitor from discharging through the solar cell and resistor R is used where portion
of the voltage is used and the variation of the signal is detected. The signal that falls on R is used as
of the voltage is used and the variation of the signal is detected. The signal that falls on R is used as
incoming signal and is fed to the AS3933 through LF1P and LFN pins, where LF1P is the input of
incoming signal and is fed to the AS3933 through LF1P and LFN pins, where LF1P is the input of
antenna channel 11and LFN is the ground for the antenna in AS3933 chip [13].
antenna channel and LFN is the ground for the antenna in AS3933 chip [13].
The second WuRx circuit design alternative (Figure 8b)8b)ais a simplified version offirst one and
The second WuRx circuit design alternative (Figure is simplified version of the the first one
would work well if well if enough illumination is available in the deploymentthis case no additional
and would work enough illumination is available in the deployment place. In place. In this case no
additional are required, even the capacitor can be removed. be removed. This circuit has controlled
componentscomponents are required, even the capacitor can This circuit has been tested inbeen tested
in controlled indoor scenarios successfully; however, capacitor no capacitor to regulate the to the
indoor scenarios successfully; however, as there is no as there is to regulate the voltage inputvoltage
input to the chip, sunlight it to sunlight the AS3933 chip. AS3933 chip. As an illustrative exposed
chip, exposing it toexposing might damagemight damage theAs an illustrative example, when example,
when exposed indoor solar panel is measured to have a short circuit a short of 0.737 mA and an open
to sunlight, the to sunlight, the indoor solar panel is measured to have current circuit current of 0.737 mA
and an open voltage of 6.07 V, the latter being higher than the AS3933 maximum operational Hence,
voltage of 6.07 V, the latter being higher than the AS3933 maximum operational voltage of 5 V. voltage
of the Hence, for the developed system, circuit WuRx circuit design option is chosen, considering
for 5 V. developed system, the first WuRx the firstdesign option is chosen, considering the potential
the potential indirect sunlight reception through
indirect sunlight reception through windows, etc. windows, etc.
The choice of the small indoor solar panel and the AS3933 chip components enable small
footprint and very low cost (around US $ 4.50) receiver system. The prototype of the receiver system
for the WuRx design 1 is pictured in Figure 9.

Sensors 2016, 16, 418

9 of 19

Sensors 2016, 16, 418

9 of 18

Sensors 2016, 16, 418

9 of 18

(a)

(a)

(b)
Figure 8. Two possible wake-up receiver circuit design options: (a) WuRx design 1 and (b) simplified
Figure 8. Two possible wake-up receiver circuit design options: (a) WuRx design 1 and (b) simplified
WuRx design 2.
WuRx design 2.

(b)
The choice of the small indoor solar panel and the AS3933 chip components enable small footprint
Figure 8. cost (around US $ 4.50) receiver system. The prototype design 1 and (b) simplified
and very lowTwo possible wake-up receiver circuit design options: (a) WuRxof the receiver system for the
WuRx design pictured in Figure 9.
WuRx design 1 is 2.

Figure 9. Prototype developed for WuRx design 1.

3.2.3. Wake-Up Transmitter
Unlike the case of the receiver, for the design of the transmitter there are no power restrictions;
yet, we opt for a low-cost, small footprint wireless node as the Control Module of the LED, enabling
Figure
Figure 9. Prototype developed for WuRx design 1.
fast and inexpensive integration 9. Prototype developed for into existing environments. We choose a
of the developed system WuRx design 1.
Z1 device (Zolertia, Barcelona, Spain) for this purpose, due to its low cost and small form factor, while
3.2.3. Wake-Up Transmitter
Transmitter
providing support to low power wireless communication protocols such as IEEE 802.15.4 and ZigBee [14].
For the case of the receiver, for the design of the we developed a circuit interfacing the Z1
Unlike theimplementation of the LED drive module, transmitter there are no power restrictions;
digital ports, a low-cost, small footprint wireless node as the Control Module of the LED, enabling
yet, we opt for which consists of a summation stage, a voltage divider and a LED drive controller as
detailed in Figure integration of the developed system into to generate digital signals choose
fast and inexpensive10. We used three digital ports of Zolertia Z1existingenvironments. We and an a
inexpensive integration of the developed system into existing environments. We
choose
adder circuit to sum these signals for an Amplitude Shift
(ASK) cost and
values of
a Z1 device (Zolertia, Barcelona, Spain) thisthis purpose, Keying itscost modulation. The form factor,
Z1 device (Zolertia, Barcelona, Spain) for for purpose, due due to
to its low lowand small smallfactor, while
form
the resistors in the adder have been chosen to be 15 KΩ to deliver a current value of 0.2 mA from
while providing supportpower wireless wireless communication such as IEEE 802.15.4 and ZigBee [14].
providing support to low to low power communication protocols protocols such as IEEE 802.15.4 and
each port in order to protect the Zolertia module (the amount of current exiting from all the output
ZigBee [14]. implementation of the LED drive module, we developed a circuit interfacing the Z1
For the
pins of Z1 should not be greater than 1.5 mA [14]). Z1 is able to produce only digital signals with the
For the implementation of a LED drive stage, a voltage divider and a LED drive controller as
digital ports, which consistsof thesummationmodule, we developed a circuit interfacing the Z1 digital
levels of 0 V and 3 V for low and high logical levels, respectively. Hence, the election of the adder
ports, which consists ofWe used threestage, a voltage divider and a to generatecontroller as detailed an
a
detailed in Figureus withsummation digital ports of Zolertia Z1 LED drive digital signals and in
circuit will help 10.
the generation of ASK signals and different levels of light intensity, which is
Figure circuitdeveloping flickering mitigation approaches as explained in signals and an adder values to
We used three digital ports an Amplitude generate digital Section 4.
adder 10.for to sum these signals forof Zolertia Z1 toShift Keying (ASK) modulation. The circuit of
crucial
sumresistors in the adder have been chosen to be 15 KΩ to deliver The values of the resistors in the
the these signals for an Amplitude Shift Keying (ASK) modulation. a current value of 0.2 mA from
adder have been chosen to be 15 Zolertia module current valueof current exiting from all in order to
each port in order to protect the KΩ to deliver a (the amount of 0.2 mA from each port the output
protect Z1 should not be greater than 1.5 mA [14]). Z1 is able to produce only digitalof Z1 should not
pins of the Zolertia module (the amount of current exiting from all the output pins signals with the
be greater than 1.5 mA [14]). Z1 is able to produce only digital signals with the election of V and 3 V
levels of 0 V and 3 V for low and high logical levels, respectively. Hence, the levels of 0 the adder
for lowwill help us with the generation of ASK signals and different levels of circuit will helpwhich is
circuit and high logical levels, respectively. Hence, the election of the adder light intensity, us with

crucial for developing flickering mitigation approaches as explained in Section 4.

Sensors 2016, 16, 418

10 of 19

the generation of ASK signals and different levels of light intensity, which is crucial for developing
Sensors 2016, 16, 418
10 of 18
flickering mitigation approaches as explained in Section 4.
Sensors 2016, 16, 418
10 of 18

Figure 10.
drive circuit developed.
Figure 10. LED drive circuit developed.
Figure 10. LED drive circuit developed.

The maximum level of the added signal is 9 V, hence a simple voltage divider is used after the
hence a simple voltage
The maximum level of the added signal is 9 V, hence a simple voltage divider is used after the
adder circuit, to limit the maximum voltage to 4.5 V. Finally, the LED drive controller uses the 2N3904
circuit, to limit the maximum voltage to 4.5 V. Finally, the LED drive controller uses the 2N3904
Finally,
adder
V.
NPN transistor for converting the variation of the voltage in its base connection into a proportional
NPN transistor for converting the variation of the voltage in its base connection into a proportional
variation of the current to flow through the Collector-Emitter connections. The last component of the
flow
Collector-Emitter connections.
variation of the current to flow through the Collector-Emitter connections. The last component of the
transmitter is a LED light, for which we opted for a LED light supporting a maximum power of 10 W.
transmitter is a LED light, for which we opted for a LED light supporting a maximum power of 10 W.
we opted for a LED light supporting a maximum power of 10 W.
The voltage employed (20 V) results in a LED light power consumption of 3.2 W and the LED drive
The voltage employed (20 V) results in a LED light power consumption of 3.2 W and the LED drive
circuit including the Z1 node is measured to consume 87.9 mW.
is measured to consume 87.9 mW.
circuit including the Z1 node is measured to consume 87.9 mW.
The resulting solar panel based VLC wake-up communication system is shown in Figure 11.
The resulting solar panel based VLC wake-up communication system is shown in Figure 11.
communication system is shown in Figure 11.
Note that although specific components are chosen for the wireless node and indoor solar panel for
specific
are
Note that although specific components
chosen for the wireless node and indoor solar panel for
our implementation and for the physical experimentations presented in this paper, any off-the-shelf
our implementation and for the physical experimentations presented in this paper, any off-the-shelf
wireless node and indoor solar panel can be used within this system.
wireless node and indoor solar panel can be used within this system.
wireless node and indoor solar panel can be used within this system.

Figure 11. Resulting system view of the solar-panel-based wake-up system.
Figure 11. Resulting system view of the solar-panel-based wake-up system.
Figure 11. Resulting system view of the solar-panel-based wake-up system.

4. System Evaluations
4. System Evaluations
One of the challenges in Visible Light Communications (VLC) is the flickering, which may
One of the challenges in Visible Light Communications (VLC) is the flickering, which may
appear when the source of light is modulated. Especially, for low data-rate VLC communications,
appear when the source of light is modulated. Especially, for low data-rate VLC communications,
such as for wake-up communication or camera communication, where low carrier frequency is used,
such as for wake-up communication or camera communication, where low carrier frequency is used,

Sensors 2016, 16, 418

11 of 19

4. System Evaluations
One of the challenges in Visible Light Communications (VLC) is the flickering, which may appear
when the source of light is modulated. Especially, for low data-rate VLC communications, such as for
wake-up communication or camera communication, where low carrier frequency is used, this challenge
requires careful analysis. Hence, we first investigate the flickering characteristics of the developed
system, and propose solutions for its mitigation. Then, we perform operational range evaluations
considering different configurations of the system.
4.1. Flickering Characterization
When a light source is modulated, flicker may appear. Flickering is the fluctuation in the brightness
of light and can become an annoying effect, and develop negative physiological changes in humans [3].
In [15], the Maximum Flickering Time Period (MFTP) is defined to be the maximum time period over
which the light intensity can change without being perceived by the human eye. An optimal flickering
frequency does not exist, however a frequency greater than 200 Hz (i.e., MFTP < 5 ms) is considered
to be safe.
In IEEE 802.15.7 VLC standard, which specifically focuses on high data rate VLC communication,
flickering is classified as intra-frame flickering and inter-frame flickering. Intra-frame flickering is the
fluctuation in the brightness within a data frame. Mitigation of intra-frame flickering is accomplished
by the use of length limiting coding or modulation scheme, e.g., Manchester encoding or Variable
Pulse Position Modulation (VPPM) [16–18]. On the other hand, inter-frame flickering is the brightness
fluctuation between the data frame transmissions. Inter-frame flickering mitigation is accomplished by
transmitting an idle pattern between data frames whose average brightness is equal to that of the data
frames [16]. In order to mitigate flickering, light dimming can be used [3]. Light dimming consists in
the control of the brightness of the light source in order to satisfy a constant brightness.
In order to characterize the flickering the following tests were performed: (i) the first test evaluates
the kind of signal generated by the Zolertia Z1 and how it induces flickering; (ii) a second test analyzes
how dimming affects the range of the system; (iii) a third test helps us to measure the light brightness
generated by each part of the frame; and (iv) finally a fourth test was performed in order to control
inter-frame flickering.
The first test was carried out using the Zolertia Z1 for the generation of a constant signal of
21 kHz of frequency, which is the carrier frequency chosen. The signal was fed to a LM3409 LED
driver module (Texas Instruments, city, state abbrev if USA, country). A flickering appears in the LED,
even accomplishing the rule of the use of an MFTP lower than 5 ms (for a signal of 21 kHz of frequency,
the period is 0.0047 ms). The test then was repeated generating the signal with a Function Generator
with a frequency of 21 kHz. In this second case, no flickering appears in the LED.
In the case of Z1, the flickering is found to be due to the “gaps” introduced by the Z1 into
the signal. We conclude that the gaps introduced are due to the fact that Contiki kernel employed
in the Z1 does not provide a real-time multi-processing, instead, an event-driven programming is
implemented and the processes run in cooperative context, whereas the interrupts and real-time
timers run in the preemptive context. In cooperative context the kernel wait for the finalization of the
task but in preemptive context the kernel temporarily interrupt the task being carried out in order
to execute another task of higher priority level. The interrupted task is resumed at a later time [19].
As the conclusion of this first test, we infer that in order to avoid flickering, a node with an MCU and
operating system supporting real-time multi-processing is necessary. One of the future works, hence,
will be to investigate such a solution.
One of the methods for flickering mitigation is dimming the LED light. In the second test we
generated the same signal with three different levels of dimming with the format of the signal required
by the AS3933 (Figure 7). For dimming the signal, we use a simple voltage divider after the output
of the Zolertia Z1 pins. The levels of dimming chosen were limited by the characteristics of the Z1.
This test was performed using the configuration option 1 of the receiver Figure 8. In the no dimming

Sensors 2016, 16, 418

12 of 19

case, transmitted signal Vpp is 2 V. In the case that the signal was dimmed to have a Vpp of around 1 V,
we observe that the flickering decreases slightly, yet, this results in a wake-up range shorter than the
Sensors 2016, 16, 418
12 of 18
one without dimming (detailed operational wake-up range tests are presented in Section 4.2). Finally,
with the signal that has a Vpp of approximately 0.6 V, few flickering is perceived in the LED; however
that even if are flickering is considerably reduced, in order to we conclude that even the flickering
no wakeups the triggered at any distance. From this second test, reach longer distances,if theamount of
dimming should be limited.
is considerably reduced, in order to reach longer distances, the amount of dimming should be limited.
In the third test, we quantify the light intensity levels generated by different frame components
In the third test, we quantify the light intensity levels generated by different frame components
of the format shown in Figure 7, using a light meter. The signals tested are one with a frequency of
of the format shown in Figure 7, using a light meter. The signals tested are one with a frequency of
21 kHz (Figure 12a), one with OOK modulation with bit duration defined in this study (Figure 12b)
21 kHz (Figure 12a), one with OOK modulation with bit duration defined in this study (Figure 12b)
and a frequency of 2.625 kHz which corresponds to the bit rate (Figure 12c). As point of reference the
and a frequency of 2.625 kHz which corresponds to the bit rate (Figure 12c). As point of reference the
intensity of light was measured when the signal is not modulated, which is achieved by feeding the
intensity of light was measured when the signal is not modulated, which is achieved by feeding the
LED with constant DC signal (Figure 12d). The measurements done at a distance of 1 of 1 m and
LED with aaconstant DC signal (Figure 12d). The measurements are are done at a distance m and with
with no interference from sources of light.
no interference from other other sources of light.

Figure 12. Light intensity measurements for different frame parts of (a) carrier burst; (b) OOK signal
Figure 12.
frame parts of (a) carrier burst; (b) OOK signal
used for preamble and pattern; (c) OOK with the data bit rate chosen, and (d) for a reference signal of
used for preamble and pattern; (c) OOK with the data bit
chosen, and
reference signal of
always on.
always on.

As seen in Figure 12, different modulations yield different light intensities, which might result
As seen in Figure 12, different modulations yield different light intensities, which might result in
in flickering. The OOK modulated signal accordingthe the duration of the the system is one one with
flickering. The OOK modulated signal according to to bit bit duration of system is the the with the
the highest illuminance level among the modulated signals. This be be explained by the that in the
highest illuminance level among the modulated signals. This can canexplained by the fact fact that in
the OOK modulated signal, half of the time the light remains higher level of intensity, as we choose
OOK modulated signal, half of the time the light remains in the in the higher level of intensity, as we
choose the LED on for the for corresponding to the logical logical zero. The reason is use the use
to keep to keep the LED ondatathe data corresponding to thezero. The reason is that the that of high
of high level for logical zero allows the lights lights to during most of time, reducing flickering and
voltage voltage level for logical zero allows the to be on be on during most of time, reducing flickering
and increasing the intensity level light, light, a result result an increased operational range. Note
increasing the intensity level of the of the and asand as aan increased operational range. Note that the
that the carrier burst of the frame corresponding to Figure 12a provides 12a provides a different
carrier burst of the frame corresponding to the signal inthe signal in Figure a different intensity level
intensity level used for the preamble and preamble and in Figure 12b.
than the signal than the signal used for thepattern shown pattern shown in Figure 12b.
As stated before, the flickering can be categorized as intra-frame flickering and inter-frame
As stated before, the flickering can be categorized as intra-frame flickering and inter-frame
flickering. ForFor intra-frame flickering, splitting the sub-frames and add some “compensation
intra-frame flickering, splitting the frame in frame in sub-frames and add some
flickering.
bits” between them is recommended [3]. The compensation bits will help to maintain maintain
“compensation bits” between them is recommended [3]. The compensation bits will help to the same
brightness during the transmission of the frame. In frame. In the case of our implementation, it is
the same brightness during the transmission of thethe case of our implementation, it is not possible
to possible to add compensation bits inside since the since the requirements of the frame format of
notadd compensation bits inside the frame, the frame, requirements of the frame format of AS3933
cannot cannot be changed. in our in our implementation, we propose to signal signal in points,
AS3933 be changed. Instead,Instead,implementation, we propose to dim thedim thein certain certain
helping to maintain the level of brightness in the LED. In Figure Figure 13, the signal using formats
points, helping to maintain the level of brightness in the LED. In13, the signal using the two the two
is shown: Figure 13a shows the signal without dimming and Figure 13b presents our dimming
formats is shown: Figure 13a shows the signal without dimming and Figure 13b presentsour dimming
solution applying aacompensation by adjusting the signal level of thethe logic zero data bits. Also use
solution applying compensation by adjusting the signal level of logic zero data bits. Also the the
use of Manchester encoding is recommended for intra-frame flickering mitigation and implemented in
of Manchester encoding is recommended for intra-frame flickering mitigation [3],[3], and implemented
in proposed frame shown in in Figure
thethe proposed frame shown Figure 13. 13.
For inter-frame flickering compensation, it is necessary to introduce an idle pattern between
frames. It is important to choose an idle pattern with a similar level of brightness with the frame. As
a trivial solution, the idle pattern is defined to use the same format with the data frame. That is, the
idle pattern is also composed of a carrier burst, separation bit, preamble bits and pattern bits (or
wake-up code), using a special wake-up code in the pattern. The full stream is then built following
the format shown in Figure 14, where the wake-up frames are included between the streams of 50
idle patterns.

Sensors 2016, 16, 418
Sensors 2016, 16, 418

13 of 19
13 of 18

(a)
Sensors 2016, 16, 418

13 of 18

(b)
Figure 13. Intra-frame flickering compensation. (a) Signal without dimming; (b) Signal with dimming
Figure 13. Intra-frame flickering compensation. (a) Signal without dimming; (b) Signal with dimming
(a)
applied to the logic zero bits.
applied to the logic zero bits.

For inter-frame flickering compensation, it is necessary to introduce an idle pattern between
frames. It is important to choose an idle pattern with a similar level of brightness with the frame. As a
trivial solution, the idle pattern is Complete framethe same format with the data frame. That is, the idle
Figure 14. defined to use format of VLC communication.
(b)
pattern is also composed of a carrier burst, separation bit, preamble bits and pattern bits (or wake-up
Figure 13. Intra-frame flickering compensation. (a) Signalfull intensity generatedfollowing the format
code), usingfourth test, measurements of the amount of lightstream is then(b) Signal with dimming
As the a special wake-up code in the pattern. The without dimming; built by this frame stream
applied to the logic zero the wake-upthe frame we used: (i) a datathe streams of 50 idle patterns.
shown in FigureFor where bits.
are performed. 14, the generation of frames are included between stream using frames without

dimming (Figure 14a) and (ii) a data stream using frames with dimming in the logic zero bits (Figure
14b). The results are shown in Table 2. Again a distance of 1 m between the LED and the light intensity
meter was established and no interference by external sources of light exist.
Figure 14. Complete frame format of VLC communication.
Table 2. Illumination Level of the Frame Stream Using Different Levels of Dimming.
No. the fourth test, measurements of thethe amount of light intensity generated by this stream
Description
As
As the fourth test, measurements of amount of light intensity generated by this frameLux
frame
Case (i)
Data stream (Figure 14) using frames without(i) a data (Figure 14a)
47.5–65.5
are performed. For the generation of the frame we used: dimming(i) a datausing frames without
stream stream using frames
stream are performed. For the generation of the frame we used:
Case (ii)
Data stream (Figure data stream using frames with dimming bits (Figure 14b)
45.5–53.5
dimming (Figure 14a) and (ii) a14) using frames with dimming in the logic zero in the logic zero bits (Figure

without dimming (Figure 14a) and (ii) a data stream using frames with dimming in the logic zero bits
14b). The results results are shown 2. Table a Again a of 1 m between the LED the the light the light
(Figure 14b). Theare shown in Table in Again2. distance distance of 1 m between and LED and intensity
As can be observed in no results presented in Table 2, in the case (ii) the variation
meter was established and the interference by external external sources of light exist. in illuminance
intensity meter was established and no interference by sources of light exist.
is 7 lux less than the one in case (i). Also, in the case (i), the illuminance range varies between the
values reached by a signal of Level of the Frame13a) and a signal modulated Dimming.
21 kHz (Figure Stream Using Different Levels of using OOK with bit
Table 2. Illumination Level of the Frame Stream Using Different Levels of Dimming.
Table 2. Illumination
duration of the system (Figure 13b). In case (ii) the intensity levels are closer to the ones reached by
No.
Description
Lux
the carrier burst signal of 21 kHz, which provides a more stable signal in terms of light intensity.
No.
Description
Lux
Case (i)
Data stream (Figure 14) using frames without dimming (Figure 14a)
47.5–65.5
It is important Data stream (Figure different frames without i.e., idle patterns were also tested (i.e.,
to mention that 14) using inter-frame, dimming (Figure 14a)
Case (i) Data stream (Figure 14) using frames with dimming in the logic zero bits (Figure 14b) 47.5–65.5
Case (ii)
45.5–53.5
Manchester symbols, frequency of 21 kHz, frequency of 2.65 kHz among others, all with different
Case (ii) Data stream (Figure 14) using frames with dimming in the logic zero bits (Figure 14b) 45.5–53.5
levels of dimming). However, none of them mitigates flickering as much as the approach presented
As can be observed in the results presented in Table 2, in the case (ii) the variation in illuminance
in Figure 13b. Based on the results of this test we decided to dim the signal in zero logic bits, use
is 7 lux can be observed inin case (i). Also, in the case (i), thethe case (ii) the variation inbetween the
As less than the one the results presented in Table 2, in illuminance range varies illuminance
Manchester coding for intra-frame flickering mitigation, and use the idle pattern described for intervalues less than the signal of 21 kHz (Figure 13a) (i), a signal modulated using OOK with bit
is 7 lux reached by a one in case (i). Also, in the caseand the illuminance range varies between the
frame flickering mitigation in Figure 13b. The dimming in the signal is applied to the wake-up frame
duration of the system (Figure kHz In case (ii) and a signal levels are using OOK ones reached by
values reached by a signal of 21 13b).(Figure 13a) the intensity modulatedcloser to the with bit duration
and also to the idle pattern.
the carrier burst signal of 21 kHz, (ii) the intensity levels stable signal in ones of light intensity.
of the system (Figure 13b). In casewhich provides a moreare closer to the termsreached by the carrier
After this set of tests, the format of the wake-up signal derived is shown in Figure 15. For the
burstIt is important to which provides a more stable signal in terms of light intensity.
signal of 21 kHz, mention that different inter-frame, i.e., idle patterns were also tested (i.e.,
generation of the signal, three ports (P4.0, P4.2 and P4.7) of the Zolertia Z1 are used. The objective of
Manchester symbols, to mention thatkHz, frequency of 2.65 kHz among others, were also tested
It is important frequency of 21 different inter-frame, i.e., idle patterns all with different
using three ports is the possibility of creating a signal of three levels of intensity of brightness. Note
levels of dimming). However, none them mitigates flickering as much as the approach presented
(i.e., Manchester symbols, frequency of 21 kHz, frequency of 2.65 kHz among others, all with different
that with the Zolertia Z1 ports it is possible to generate only rectangular signals putting the pins in
in Figure 13b. Based on the results
this test we decided to dim the signal in zero logic bits, use
levels of dimming). However, none of them mitigates flickering as much as the approach presented
high or low level.
Manchester coding for intra-frame flickering mitigation, and use the idle pattern describedlogic bits,
in Figure 13b. Based on the results of this test we decided to dim the signal in zero for interIn Figure 15, both the AMS proposal for the signal and our proposal are depicted. We propose a
frame flickering coding for intra-frame flickering mitigation, and use applied pattern described for
use Manchester mitigation in Figure 13b. The dimming in the signal is the idle to the wake-up frame
change in the way the signal is sent: as we are working with light, the default state (or zero logic bits)
and also to flickering mitigation in Figure 13b. The dimming in the signal is applied to the wake-up
inter-frame the idle pattern.
of the signal should have a “higher level” than the one proposed by AMS. With this small change the
After also to of idle pattern.
frame and this setthe tests, the format of the wake-up signal derived is shown in Figure 15. For the
lights are in “on” state for longer, which brings several advantages such as the reduction of flickering,
generation of the signal, three ports (P4.0, P4.2 and P4.7) of the Zolertia Z1 are used. The objective of
having more light to be harvested by the indoor solar panel without any change in the AS3933
using three ports is the possibility of creating a signal of three levels of intensity of brightness. Note
behavior.
that with the Zolertia Z1 ports it is possible to generate only rectangular signals putting the pins in
high or low level.
In Figure 15, both the AMS proposal for the signal and our proposal are depicted. We propose a
change in the way the signal is sent: as we are working with light, the default state (or zero logic bits)

Sensors 2016, 16, 418

14 of 19

After this set of tests, the format of the wake-up signal derived is shown in Figure 15. For the
generation of the signal, three ports (P4.0, P4.2 and P4.7) of the Zolertia Z1 are used. The objective of
using three ports is the possibility of creating a signal of three levels of intensity of brightness. Note
that with the Zolertia Z1 ports it is possible to generate only rectangular signals putting the pins in
Sensors 2016, 16, 418
14 of 18
high or low level.

Figure 15. Transmitted signal format: AMS proposal vs. our proposal.
Figure 15. Transmitted signal format: AMS proposal vs. our proposal.

4.2. Wake-Up Range Evaluations proposal for the signal and our proposal are depicted. We propose a
In Figure 15, both the AMS
change in the way the signal is sent: as we are working with light, the default state (or zero logic bits)
For a wake-up communication system, the wake-up range determines the applications that can
of the signal should have a “higher level”In thisthe one proposed by AMS. With this small change the
be supported by the developed system. than section, we depict the wake-up range evaluations of
lights are system state for longer,then propose several advantages such as the reduction of flickering,
the base in “on” configuration, which brings and evaluate alternative settings to the base system
having more lightimprove the system performance. panel without any change in the AS3933 behavior.
configuration to to be harvested by the indoor solar
In the first set of tests, we use the signal format proposed in Figure 15 to evaluate the wake-up
4.2. Wake-Up Range Evaluations
probabilities per distance in presence of interference from other light sources (office fluorescent lights
usedForthis test as interference). The main evaluation environment used is an office/laboratory with
in a wake-up communication system, the wake-up range determines the applications that can
a supported by 2.4 developed system. In this desks, we depict the wake-up range evaluations
besize of 10 × 4 × the m (L × W ×H) that includessection,bookshelves and electronic equipment. The
WuTx is fixed on a desk and the WuRx is displaced evaluate alternative settings to the base system
of the base system configuration, then propose and within the evaluation environment for the range
evaluations. For the calculation of the wake-up probability, for each evaluated distance, we send the
configuration to improve the system performance.
signal 50 timesset of we count use the signalof timesproposed in Figure 15 to evaluate the wake-up
In the first and tests, we the number format the wake-up signal was triggered. The range
evaluations are done with bit duration of 381.264 μs other is the value finally chosen for this
probabilities per distance in presence of interference fromwhich light sources (office fluorescent lights
implementation; however the range main evaluation environment used is an office/laboratory with
used in this test as interference). Theevaluations with bit duration of 755.786 μs are also presented for
comparison. Different bit durations are assessed for desks, bookshelves and electronic equipment.
a size of 10 ˆ 4 ˆ 2.4 m (L ˆ W ˆH) that includes two important issues: With shorter bit duration
the WuTx is fixed light desk and the WuRx is displaced within the evaluation to reach a higher
The flickering of the on a source is lower, and with shorter bit duration it is possibleenvironment for
transmission bit rate. For the calculation of for the receiver is the one shown in Figure distance,
the range evaluations.The configuration used the wake-up probability, for each evaluated 8 and the
frame structure and the signal formats are the number of times and 15, respectively. In triggered.
we send the signal 50 times and we count the ones in Figures 14the wake-up signal was this case a
full signal (carrier burst done with bit duration of 381.264 µs which sent, in other words, we set the
The range evaluations are+ separation bit + preamble + pattern) wasis the value finally chosen for this
internal correlator setting of AS3933 to ON to wake up only one 755.786 µs are also presented are
implementation; however the range evaluations with bit duration of targeted device. The results for
depicted in Different
comparison.Figure 16. bit durations are assessed for two important issues: With shorter bit duration
the flickering of the light source is lower, and with shorter bit duration it is possible to reach a higher
transmission bit rate. The configuration used for the receiver is the one shown in Figure 8 and the
frame structure and the signal formats are the ones in Figures 14 and 15 respectively. In this case a full
signal (carrier burst + separation bit + preamble + pattern) was sent, in other words, we set the internal
correlator setting of AS3933 to ON to wake up only one targeted device. The results are depicted
in Figure 16.

the flickering of the light source is lower, and with shorter bit duration it is possible to reach a higher
transmission bit rate. The configuration used for the receiver is the one shown in Figure 8 and the
frame structure and the signal formats are the ones in Figures 14 and 15, respectively. In this case a
full signal (carrier burst + separation bit + preamble + pattern) was sent, in other words, we set the
internal correlator setting of AS3933 to ON to wake up only one targeted device. The results are
Sensors 2016, 16, 418
15 of 19
depicted in Figure 16.

Figure 16. Probability of wake-up vs. distance for short and long bit duration, under interference from
Figure 16. Probability of wake-up vs. distance for short and long bit duration, under interference from
fluorescent light with correlator setting ON.
fluorescent light with correlator setting ON.
Sensors 2016, 16, 418
15 of 18

With the short bit duration a maximum distance of 0.6 m was reached and with the longer bit
With the short bit duration a maximum distance of 0.6 m was reached and with the longer bit
duration the maximum distance is found to be 3.1 m. The further distance reached with thethe longer
duration the maximum distance is found to be 3.1 m. The further distance reached with longer bit
bit duration is due to the fact that each bit has longer exposure and consequently a higher brightness
duration is due to the fact that each bit has a a longer exposure andconsequently a higher brightness
in the source of light, increasing the probability of successful demodulation. Note that the wake-up
in the source of light, increasing the probability of successful demodulation. Note that the wake-up
probabilities fall down sharply once the operational distance limits are reached. The negative impact
probabilities fall down sharply once the operational distance limits are reached. The negative impact
of interference on the performance can be explained by the solar panel’s capacitive behavior resulting
of interference on the performance can be explained by the solar panel’s capacitive behavior resulting
in slow responses to fast signal level changes caused by interference.
in slow responses to fast signal level changes caused by interference.
The number of signal level changes are also increased further with the short bit duration setting,
The number of signal level changes are also increased further with the short bit duration setting,
resulting in a limited wake-up distance performance. Given the increased flickering caused by longer
resulting in a limited wake-up distance performance. Given the increased flickering caused by longer
bit durations due to the higher intra-frame brightness differences, we choose the shorter bit duration
bit durations due to the higher intra-frame brightness differences, we choose the shorter bit duration
and focus to improve the system range for this setting.
and focus to improve the system range for this setting.
The next set of tests are performed to observe the effect of the solar panel size on the wake-up
The next set of tests are performed to observe the effect of the solar panel size on the wake-up
probabilities, evaluating the distances reached by large and small solar panels. The results are shown
probabilities, evaluating the distances reached by large and small solar panels. The results are shown
in Figure 17, where we observe that the maximum reached reached with light interference
in Figure 17, where we observe that the maximum distance distance with fluorescent fluorescent light
interference is larger panel larger panel and 3 m for the small panel. this behavior is that large solar
is 2.5 m for the 2.5 m for theand 3 m for the small panel. The reason for The reason for this behavior is
that maintains more maintains a more stable signal level in the output, not relaying the fast
panellarge solar apanel stable signal level in the output, not relaying the fast variations of light signal
variations values signal amplitude values to the in Figure also
amplitude of light to the output, as also illustratedoutput, as 6. illustrated in Figure 6.

Figure 17. Probability of wake-up vs. distance for small and large indoor solar panels, under
Figure 17. Probability of wake-up vs. distance for small and large indoor solar panels, under
interference from fluorescent light with correlator setting ON.
interference from fluorescent light with correlator setting ON.

The third set of tests are done to evaluate the wake-up distances for the case of no interference
and using the two possible configurations for the transmitted signal. Note that this setup corresponds
to an operational indoor environment, where the interference can be controlled, such as when the
lighting system consists of coordinated LEDs. In the first configuration a signal with carrier burst,
separation bit, preamble and pattern is sent (correlator ON), using the two different bit durations
considered. We identify the probability of the successful wake-up of the signal versus the distance
from transmitter to receiver. In a similar way, in the second configuration only a carrier burst with
frequency of 21 kHz was sent, i.e., with the correlator OFF. A very large seminar room is used for this
set of evaluations to allow the experimentation of a wide range of distances. The results are shown in

Sensors 2016, 16, 418

16 of 19

The third set of tests are done to evaluate the wake-up distances for the case of no interference and
using the two possible configurations for the transmitted signal. Note that this setup corresponds to an
operational indoor environment, where the interference can be controlled, such as when the lighting
system consists of coordinated LEDs. In the first configuration a signal with carrier burst, separation
bit, preamble and pattern is sent (correlator ON), using the two different bit durations considered.
We identify the probability of the successful wake-up of the signal versus the distance from transmitter
to receiver. In a similar way, in the second configuration only a carrier burst with frequency of 21 kHz
was sent, i.e., with the correlator OFF. A very large seminar room is used for this set of evaluations to
allow the experimentation of a wide range of distances. The results are shown in Figure 18. With the
correlator ON, we reach a distance of 14 m for the case of long bit duration, and 7 m for the case of
short bit duration; and with the correlator OFF and using a burst carrier with duration of 300 carrier
frequency periods, the maximum distance achieved was 15.5 m. As it was expected, the distance
reached with the correlator OFF is longer because the AS3933 only has to identify the presence of a
carrier frequency of 21 kHz and is not affected by the bit duration or the format of the frame. Again
Sensors 2016, 16, 418
16
we observe that the wake-up probability falls down quickly once the limited distance is reached. of 18

Figure 18. Probability of wake-up vs. distance for the correlator ON and OFF settings, for short and
18.
under no interference scenario.
long bit duration and under no interference scenario.

A further test in the LED is lower for the case of with bit duration than for the case of long bit
The flickeringwith the correlator setting OFF and shortfluorescent light interference in the office
environment the case of short bit duration, a high number of false wake-ups is observed due to the
duration. In is also performed. In this case, the perception of flickering is mainly noticed near the
optical both of the test environment. This power AS3933 versus it is to use the correlator in order to
LED. Innoise cases, using the solar panel toshows how importantusing the USB for the same purpose,
avoid any undesirable behavior of the
the distances achieved were the same. system.
A final test was developed using the WuRx and with fluorescent 2, presented in Figure office
A further test with the correlator setting OFF design options 1 and light interference in the8. The
test was done also performed. In this case, a high number of false wake-ups is observed due to the
environment is in the office environment without light interference. In the case of receiver design
option noise of the test environment. This maximum distance reached is also 7 correlator in receiver
optical 1, once the capacitor is charged the shows how important it is to use the m. Using the order to
design option 2, a maximum distance of 60 cm
avoid any undesirable behavior of the system. was reached; however using the same configuration
with A final test was developed using as power source, a distance and m presented in Figure 8. The test
the receiver being fed with USB the WuRx design options 1 of 7 2, was reached. Figure 19 shows
the result in the office environment without light interference. In the case of receiver design option 1,
was done of this test.
once the capacitor is charged the maximum distance reached is also 7 m. Using the receiver design
option 2, a maximum distance of 60 cm was reached; however using the same configuration with the
receiver being fed with USB as power source, a distance of 7 m was reached. Figure 19 shows the result
of this test.

Figure 19. Probability of wake-up vs. distance for the correlator ON for the different receiver
configuration under no interference scenario.

A final test was developed using the WuRx design options 1 and 2, presented in Figure 8. The
test was done in the office environment without light interference. In the case of receiver design
option 1, once the capacitor is charged the maximum distance reached is also 7 m. Using the receiver
design option 2, a maximum distance of 60 cm was reached; however using the same configuration
Sensorsthe receiver being fed with USB as power source, a distance of 7 m was reached. Figure 1917 of 19
with 2016, 16, 418
shows
the result of this test.

Figure 19. Probability of wake-up vs. distance for the correlator ON for the different receiver
Figure 19. Probability of wake-up vs. distance for the correlator ON for the different receiver
configuration under no interference scenario.
configuration under no interference scenario.

The short distance reached with the receiver design option 2 is easily explained by the fact that
The short distance reached with the receiver design option 2 is easily explained by the fact that
with the indoor solar panel it is not possible to harvest enough power for the operation of the AS3933
with the indoor solar panel it is not possible to harvest enough power for the operation of the AS3933
from
from the amount of light emitted by this single LED at at longer distances. Hence, using a capacitor
amount of light emitted by this single LED longer distances. Hence, using a capacitor and
storing the energy (WuRx option 1) is necessary for the battery-less execution of
proposed
and storing the energy (WuRx option 1) is necessary for thebattery-less execution of the proposed
VLC-based wake-up system.
it is clear that the distances achieved by the
VLC-based wake-up system. Based on these test results, it is clear that the distances achieved by the
proposed system are enough for many indoor
proposed system are enough for many indoor applications, such as asset control, tracking of devices,
security systems, and others.
security systems, and others.
5. Conclusions and Future Work
5. Conclusions and Future Work
A wake-up receiver system that uses VLC for communication and an indoor solar panel for energy
A wake-up receiver system that uses VLC for communication and an indoor solar panel for
harvesting and reception reception of the VLCbeen presented. The use of VLC for use of VLC for
energy harvesting and of the VLC signal has signal has been presented. The communication
avoids the problem of radio interferences with other communication signals and preventing the
congestion in the radio channels. Also, the wake-up receiver enables an on-demand communication
system avoiding the unnecessary wake-up of the wireless sensor nodes attached. In the tests performed
a wake-up distance of 7 m is reached in case of no external interference, which is a targeted distance for
VLC systems. However using longer bit durations, it is possible to reach longer distances of around
14 m, which exceeds the targeted range of distance.
The system also has the novelty of the use of an indoor solar panel as the VLC signal receiver,
allowing the use of the indoor solar panel for a different purpose than the one they were built for.
Also, the use of an indoor solar panel permits harvesting the necessary power from the light for the
energy-autonomous operation of the wake-up receiver. This allows the wake-up receiver to be in
constant and independent activity without affecting the battery lifetime of the wireless sensor node,
saving in this way significant power and consequently extending the lifetime of the sensor device
connected to wake up receiver and the network.
Also the addressable wake-up feature of the developed system enables on-demand wake-up of
only the targeted node. Nevertheless, it was shown that the wake-up receiver can also generate a
wake-up signal using the carrier burst, which can be useful in environments where it is desirable to
wake up several sensor nodes at once.
Using an indoor solar panel in the system also brings a set of new issues and challenges,
for example, the restriction in the use of high frequencies for the transmission and consequently
lower bit rates; and the use of ultra-low power components for the receiver due to limited energy
harvested from indoor lighting systems. In addition, the tests performed using a small and a large size
indoor solar panel show a better frequency response for the small one, which permits the use of the
system in applications where the small size of the device is a desirable characteristic.

Sensors 2016, 16, 418

18 of 19

Flickering mitigation is still a challenge that must be investigated further. In this study, the
flickering was minimized through several actions: Intra-frame flickering mitigation, inter-frame
flickering mitigation, dimming of the signal in the zero logic bits, the use of a shorter bit duration time
and the reduction of the length of the signal sent.
It is important to mention that upon the triggering of the wake up signal, it is possible to transmit
data using the VLC channel for downlink data communication, which is a future work topic. Also,
the study of flickering mitigation in VLC communications can be furthered, along with the developing
of a transmitter with more levels of dimming, and the possibility of inserting compensation bits within
the frame. Another future work involves the detailed study of the use of capacitors to accumulate the
energy harvested by the solar panel for a longer wake-up range.
Acknowledgments: This work was supported in part by the Spanish Government’s Ministerio de Economía y
Competitividad through project TEC2012-3253, RYC-2013-13029 and FEDER.
Author Contributions: All of the authors have participated in designing, writing and revising the intellectual
content of this article. Carolina Carrascal performed the experiments.
Conflicts of Interest: The authors declare no conflict of interest.

References
1.

2.

3.

4.

5.

6.

7.

8.
9.
10.

11.
12.

Oller, J.; Demirkol, I.; Casademont, J.; Paradells, J.; Gamm, U.; Reindl, L. Has Time Come to Switch from
Duty-Cycled MAC Protocols to Wake-up Radio for Wireless Sensor Networks? IEEE/ACM Trans. Netw. 2015,
PP. [CrossRef]
Carrascal, C.; Demirkol, I.; Paradells, J. A Novel Wake-Up Communication System using Solar Panel
and Visible Light Communication. In Proceedings of the IEEE Global Communications Conference
(GLOBECOM), Austin, TX, USA, 8–24 December 2014; pp. 461–467.
IEEE Standard for Local and Metropolitan Area Networks–Part 15.7: Short-Range Wireless Optical
Communication Using Visible Light. Available online: http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnu
mber=6016195&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6016195
(accessed on 21 March 2016).
Kim, S.-M.; Won, J.-S. Simultaneous reception of visible light communication and optical energy using a
solar cell receiver. In Proceedings of the International Conference on ICT Convergence (ICTC), Jeju, Korea,
14–16 October 2013; pp. 896–897.
Mathews, J.; Barnes, M.; Young, A.; Arvind, D.K. Low power wake-up in wireless sensor networks using free
space optical communications. In Proceedings of the 4th International Conference on Sensor Technologies
and Applications, Venice, Italy, 18–25 July 2010; pp. 256–261.
Häkkinen, T.; Vanhala, J. Ultra-low power wake-up circuit for short-range wireless communication.
In Proceedings of the IET 4th International Conference on Intelligent Environments, Seattle, WA, USA,
21–22 July 2008; pp. 1–4.
Kim, G.; Lee, Y.; Bang, S.; Lee, I.; Kim, Y.; Sylvester, D.; Blaauw, D. A 695 pW standby power optical wake-up
receiver for wireless sensor nodes. In Proceedings of the IEEE Custom Integrated Circuits Conference (CICC),
San Jose, CA, USA, 9–12 September 2012; pp. 1–4.
Yadav, K.; Kymissis, I.; Kinget, P. A 4.4 µW Wake-Up Receiver using Ultrasound Data Communications.
In Proceedings of the Symposium on VLSI Circuits (VLSIC), Honolulu, HI, USA, 15–17 June 2011; pp. 212–213.
Moazzeni, S.; Sawan, M.; Cowan, G. An Ultra-Low-Power Energy-Efficient Dual-Mode Wake-Up Receiver.
IEEE Trans. Circuits Syst. I Regul. Pap. 2015, 62, 517–526. [CrossRef]
Reynolds, S.J.; Black, D.W.; Borin, S.; Breuer, G.; Burmeister, L.F.; Fuortes, L.J.; Smith, T.F.; Stein, M.A.;
Subramanian, P.; Thorne, P.S.; et al. Indoor Environmental Quality in Six Commercial Office Buildings in the
Midwest United States. Appl. Occup. Environ. Hyg. 2001, 16, 1065–1077. [CrossRef] [PubMed]
BPW34 Silicon PIN Photodiode Datasheet. Vishay Semiconductors. Available online: http://www.vishay.com/
docs/81521/bpw34.pdf (accessed on 21 March 2016).
74VHC595 8-Bit Shift Register with Output Latches Datasheet. Fairchild Semiconductor. Available online:
https://www.fairchildsemi.com/datasheets/74/74vhc595.pdf (accessed on 21 March 2016).

Sensors 2016, 16, 418

13.
14.
15.

16.
17.

18.
19.

19 of 19

AS3933—3D Low Frequency Wake-up Receiver Datasheet. Austriamicrosystems. Available online:
http://ams.com/eng/content/download/66224/570460/36238 (accessed on 21 March 2016).
Zolertia Z1 datasheet. Available online: http://zolertia.sourceforge.net/wiki/images/e/e8/Z1_RevC_Data
sheet.pdf (accessed on 21 March 2016).
Berman, S.M.; Greenhouse, D.S.; Bailey, I.L.; Clear, R.; Raasch, T.W. Human electroretinogram responses to
video displays, fluorescent lighting and other high frequency sources. Optom. Vis. Sci. 1991, 68, 645–662.
[CrossRef] [PubMed]
Rajagopal, S.; Roberts, R.D.; Lim, S.-K. IEEE 802.15.7 Visible Light Communication: Modulation Schemes
and Dimming Support. IEEE Commun. Mag. 2012, 50, 71–82. [CrossRef]
Deng, X.; Wu, Y.; Arulandu, K.; Zhou, G.; Linnartz, J. Performance comparison for illumination and visible
light communication system using buck converters. In Proceedings of the IEEE Globecom Workshops
(GC Wkshps), Austin, TX, USA, 8–12 December 2014; pp. 547–552.
Yoo, J.; Kim, B.W.; Jung, S. Modelling and analysis of M-ary variable pulse position modulation for visible
light communications. IET Optoelectron. 2015, 9, 184–190. [CrossRef]
Dunkels, A.; Schmidt, O.; Voigt, T.; Ali, M. Protothreads: Simplifying Event-Driven Programming of
Memory-Constrained Embedded Systems. In Proceedings of the 4th International Conference on Embedded
Networked Sensor Systems, Boulder, CO, USA, 1–3 November 2006; pp. 1–4.
© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons by Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).