Content uploaded by Tommaso Polonelli
Author content
All content in this area was uploaded by Tommaso Polonelli on Dec 09, 2015
Content may be subject to copyright.
IEEE SENSORS JOURNAL, VOL. 15, NO. 5, MAY 2015 2963
A Low Cost, Highly Scalable Wireless Sensor
Network Solution to Achieve Smart LED
Light Control for Green Buildings
Michele Magno, Member, IEEE, Tommaso Polonelli, Luca Benini, Fellow, IEEE,
and Emanuel Popovici, Senior Member, IEEE
Abstract— Reducing energy demand in the residential and
industrial sectors is an important challenge worldwide.
In particular, lights account for a great portion of total energy
consumption, and unfortunately a huge amount of this energy
is wasted. Light-emitting diode (LED) lights are being used to
light offices, houses, industrial, or agricultural facilities more effi-
ciently than traditional lights. Moreover, the light control systems
are introduced to current markets, because the installed lighting
systems are outdated and energy inefficient. However, due to
high costs, installation issues, and difficulty of maintenance;
existing light control systems are not successfully applied to
home, office, and industrial buildings. This paper proposes a low
cost, wireless, easy to install, adaptable, and smart LED lighting
system to automatically adjust the light intensity to save energy
and maintaining user satisfaction. The system combines motion
sensors and light sensors in a low-power wireless solution using
Zigbee communication. This paper presents the design and imple-
mentation of the proposed system in a real-world deployment.
Characterization of a commercial LED panel was performed
to evaluate the benefit of dimming for this light technology.
Measurements of total power consumption over a continuous six
months period (winter to summer) of a busy office were acquired
to verify the performance and the power savings across several
weather conditions scenarios. The proposed smart lighting system
reduces total power consumption in the application scenario by
55% during a six month period and up to 69% in spring months.
These figures take also into account individual user preferences.
Index Terms—Overlay networks, wireless sensor networks,
power electronics, LED lighting control, power management,
energy efficiency.
I. INTRODUCTION
ENERGY saving and environmental friendliness/
awareness is a hot topic in current research. In fact,
Carbon dioxide (CO2)emissions are strongly associated with
energy consumption, these originated from the combustion
of hydrocarbons (oil, natural gas and coal) either directly
Manuscript received August 5, 2014; revised November 26, 2014; accepted
December 2, 2014. Date of publication December 22, 2014; date of current
version April 1, 2015. The associate editor coordinating the review of this
paper and approving it for publication was Dr. Viktor Gruev.
M. Magno and L. Benini are with the Department of Electrical, Electronic
and Information Engineering, University of Bologna, Bologna 40126, Italy,
and also with the Swiss Federal Institute of Technology Zurich, Zurich 8092,
Switzerland (e-mail: michele.magno@iis.ee.ethz.ch; luca.benini@unibo.it).
T. Polonelli is with the Department of Electrical, Electronic and Infor-
mation Engineering, University of Bologna, Bologna 40126, Italy (e-mail:
tommaso.polonelli@studio.unibo.it).
E. Popovici is with Department of Electrical and Electronics Engineering,
University College Cork, Cork, Ireland (e-mail: e.popovici@ucc.ie).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JSEN.2014.2383996
burned (transport and heating) or for generation of electricity
in power plants [1]. Lighting systems are a major source
of electricity consumption in the world. In Europe, the
amount of electrical energy used in illuminating buildings is
considerable, about 40% and leads to approximately 35% of
carbon dioxide emissions [2]. In recent years the European
Union EU has actively promoted political campaigns toward
energy efficiency. While previous research and industrial
works have shown that simple lighting controls using motion
sensors, such as PyorelectricInfraRed (PIR) sensors, are
effective at reducing the amount of electrical energy used for
lighting buildings, advanced lighting control strategies have
the potential to achieve even greater energy savings, better
quality of service and offer many advantages over simple
on/off controls. However, until present, advanced control
strategies, such as dimming light according to the day lighting
or load shedding, which require a more systems-oriented
approach, have been less successful. This is especially due
to the high cost of installation and maintenance and the
impossibility of retrofitting [3].
On the technological side, Light Emitting Diode (LED) is
rapidly becoming a commonly used solid-state light source
technology in general lighting applications. This is due to
its longer lifetime, reduced power consumption, and having
no poison mercury content compared with the conventional
fluorescent lamps [4], [5]. In addition, dimming control is
often needed to regulate lighting levels for individual human
needs or preferences as well as to achieve energy savings.
Novel driver systems are improving the dimmable features to
achieve this goal and are increasingly commercially available.
This new technology is boosting interest in controlling the
light to reduce power consumption. The market for lighting
controls in residential and commercial buildings has entered
a period of dramatic transformation. The demand for both
wireless and local controls, such as occupancy sensors; photo-
sensors; and networked controls rises, and the adoption rate of
the LED lighting systems begins to climb as well. According
to a new report from Navigant Research, worldwide revenue
from networked lighting controls will grow from $1.7 billion
annually in 2013 to more than $5.3 billion by 2020 [6].
With the advance of wireless sensor network (WSN) tech-
nology, it is now easier than ever to monitor and control
houses, offices and industrial buildings. WSN is the back-
bone of a large variety of cyber-physical systems (CPS)
applications in environmental monitoring, healthcare, security,
1530-437X © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2964 IEEE SENSORS JOURNAL, VOL. 15, NO. 5, MAY 2015
and industrial domains, among others, due to the flexible
distribution of WSN devices [7]–[9]. Each device embod-
ies a networked node that integrates computing, wireless
communication, power management and sensing capability in
order to collect and process data from sensors, generally col-
laborating to coordinate activities [10]. WSN in combination
with LED lights and novel drivers reduces the power con-
sumption of the illumination in several application scenarios
by several orders of magnitude [11]. WSN has the potential
to achieve a low cost and ultra high power saving system.
However, particular attention must be paid during the design
process of hardware and software. For example, it is important
to develop low power wireless sensor nodes, with small form
factor and cost which can be easily fitted inside the driver
casing. These features allow the future embedded devices to
be attached to the driver, adding the wireless communication
and “smart” capabilities to the driver, to achieve an automat-
ically or remotely controlled system. The novel driver can
be controlled using distributed sensors in the environmental
area to increase the quality of the control reducing the power
consumption and increasing the quality of the service. The
control of lights becomes fast and rich-featured, going beyond
on/off, dimming, to color (or color temperature) change and
scene setting, with intelligence to react to human mood and
activity, and adapt to environments and scenarios.
The contributions of this paper are as follows:
•A methodology for deploying low power sensor networks
to enhance the power consumption of LED lights using
novel, ultra-low power hardware architecture and smart
distributed algorithm. The concept of using light sensors
and WSN in LED control is not new, however using it to
directly control a LED driver with distributed intelligence
and allowing retrofit is a novel contribution.
•Experimental validation of the proposed approach. The
power consumption characterization of panels according
to the dimming and the average energy reduction in a
real-life, long-term deployment is presented.
The remainder of this paper is organized as follows: Section II
describes recent related work in the area. Section III describes
the models and devices of the proposed approach, describ-
ing the nodes and the network architectures, respectively.
Section IV presents the algorithms implemented in the whole
system. Section V describes the implemented approach, along
with measurements, comparative evaluation, and validation.
Section VI concludes the paper.
II. RELATED WORK
Research on monitoring, control and energy efficiency in
the lighting domain has been prolific in recent years; with a
variety of solutions and techniques proposed. The two main
approaches are given by wired and wireless systems. Wired
controllable lighting systems can measure the artificial and
daylight illumination through the use of sensors in a controller
area network [12] or a set of data logger devices [13] to modify
the light intensity and hence its energy consumption. However,
due to the presence of bundles of cable to perform data
communication the wired devices are much more costly, espe-
cially due to the installation and maintenance. Moreover the
wired system is limited to retrofitting the existing light system
in buildings. To overcome this installation cost and issues,
wireless technology has become a more popular alternative on
the demand-side energy management, monitoring and control
in buildings. WSN is the enabling technology for building
energy control as it is much easier and flexible to install
and implement than wired networks. By using the combina-
tion of advanced WSN-based controls and DC grid powered
LED lighting systems, the advantageous features generated
from this combined technology should lead to greater energy
savings at the demand-side of the green smart building [14].
Recently, wireless sensor networks have been applied
to energy conservation applications such as light
control [15], [16]. In [16], a trade-off between energy
consumption and users’ satisfaction using light controls
was studied. The authors applied utility functions which
considered users’ location and lighting preferences so that
illumination could be adjusted as to maximize the total
utilities. However, it did not consider the fact that people may
require different illumination levels for different activities.
The logic of lighting control systems may include factors
such as daylight intensity, which can be measured by light-
sensitive sensors [17]. In [18] the authors defined several
user requirements and cost functions. Their goal was to
adjust lights to minimize the total cost of energy supplied.
However, the result was appliedtoentertainmentandmedia
production systems rather than to buildings. In [19], light
control using wireless sensors to reduce energy consumption
in commercial buildings is introduced. In these previous
works, lighting devices are adjusted depending on ambient
daylight intensity and/or motion sensors. This approach is
conceptually similar to the proposed system. However our
work presents a comprehensive, long term (over 6 months)
in-field evaluation of power savings, during several seasons
and weather conditions. Moreover the control algorithm is not
explained in [19] and it is not possible know if the algorithm
uses distributed or local decision making. Finally in [19] there
is no data about the power consumption of the wireless system
and its associated cost. In [21] and [22], a lighting control
system is proposed that considers both users’ preferences and
energy conservation. This system assumes that the location
of each user is known via a wireless sensor that is carried by
each user that also detects local light intensity. An additional
assumption is that there is no obstacle between whole lighting
devices and fixed sensors. In [22] their model is designed
for “point-link” light sources, such as LEDs. In [23] a smart
lighting system where the ambient light at the user’s location
is controlled in real-time to give users the best indoor light
experience but in energy efficient manner is proposed. This
approach is very similar to the one proposed in this paper
and the benefits are demonstrated. However, the network
cannot control the LEDs’ driver directly as it needs a digital
addressable lighting interface (DALI) controller and this
makes the system not ultra low power and very expensive.
Also, the driver cannot be made wireless. Another standard
for the lighting control is KNX [27], which is used to add
intelligence to buildings. However, there is no native wireless
communication and it has a very high cost with tens of
MAGNO et al.: LOW COST, HIGHLY SCALABLE WSN SOLUTION TO ACHIEVE SMART LED LIGHT CONTROL 2965
Fig. 1. Typical application scenario of Smart Lighting with the topologies
of devices used: i) Coordinator of network connected to a host device;
ii) Router to monitor the enviroment with light and motion sensors; and End
Device connected to the pannel to adapt the light intensity to save energy
achiving the optimal level of brightness in the area.
thousands of dollars for the basic installation. Our solution
has been designed with low power, low cost, flexibility
and scalability in mind. It is based on a low cost wireless
sensors node, which uses the Zigbee standard to increase
accessibility and scalability, the intelligence is distributed
and it can directly control the LED driver to increase the
accuracy and reactivity of the system. Finally the motion
sensor and the light sensor monitor the surroundings to give
the best user comfort using the lowest power consumption.
An interesting approach for sensor-less lighting control
using neural network algorithm has been proposed in [24].
However the network still needs a central intelligence unit
and the DALI connection, moreover the power saving results
are much lower than in the proposed approach. Several
solutions for intelligent lighting applications using wireless
modules have been proposed also in commercial products
where NXP/Jennic [25] and Eshelon/LonWorks [26] are the
most important examples. Even for commercial products
the technology is still in its infancy since there are many
options to be validated in practice by real applications.
Surprisingly, there are no commercial low cost products
on the market that offer the functionality features listed
in this proposed smart lighting system such as flexibility,
adaptability/ease of use (suitable for several commercial
drivers), robustness, distributed intelligent, directly plug-in
LED’s drivers. A commercial wireless solution is Eyenut [30].
However, the sensor node is limited to movement sensors,
meaning light sensors are excluded, and there is no real
distributed intelligence. Moreover, the solution has a high
cost of thousands of dollars for the base station and hundreds
of dollars for the wireless devices.
This paper shows the design, development and accurate
measurements of a whole low power and low cost wireless
sensor network to achieve power saving through automatic
control and demonstrates its benefits in terms of power saving
and scalability using in-field experimental results.
III. DEVICES AND METHODS
Figure 1 illustrates a conceptual scheme of the pro-
posed system. It consists of groups of LED panels managed
by multiple sensors (motion and light) and distributed
intelligence. The nodes communicate wirelessly through a
Zigbee [31] (using IEEE 802.15.04 MAC protocol) mesh
network with a coordinator, several routers and several
End Devices (EDs). Each panel has a wireless controller
(Zigbee ED) directly connected to its driver to be set the light
intensity through a pulse-width modulation (PWM) signal.
The PWM signal is used to encode the level of the LED
brightness with the width of the pulse (duration) of micro-
controller signal as explained better in next subsection. The
value of the PWM is decided by a control unit, given by
one of the distributed routers provided with sensors. Each
router uses the sensors’ data to adapt the intensity according
with the user’s preferences with the goal of maximizing the
energy saving and users’ preferences. The Zigbee network in
a mesh configuration permits building a scalable and modular
system easily extendable, and allows each sub group of lights
to be completely independent and flexible in terms of area
monitored/controlled. In fact, each router has a flexible and
controllable number of associated ED’s and LED panels,
which it can control under the same conditions. This allows
having different areas with different controls in order to
increase the power saving driven by users’ preference.
The whole network is managed by one supervision unit, the
Zigbee coordinator that both manages the network and ensures
that all network devices are working properly. Furthermore
this unit works as a gateway with a remote host (laptop,
wall embedded devices, Wireless Lan/Bluetooth devices, and
so on), to enable human interaction. Thus, it is possible
both to acquire users’ preferences to adapt the dimming of
the lights in desirable values and to enable a graphical user
interface for the management and for visualizing the energy
saving data for each group of LEDs or single device. This
is an important feature as the percentage of energy savings
depends on several factors but the most important is the users’
preference, and the user can evaluate this graphically. Other
import factors affecting the power saving are the position
of each group of panels, i.e. a room with a big windows
south facing saves more energy than a basement, the weather
conditions, season, geographical location, etc.
The primary objective of the proposed approach is to reduce
the power consumption of a generic (and also existing) LED
light system using a flexible network deployed in the same
target field reducing cost of installation and guaranteeing smart
and green buildings with and high return of the investment
saving energy.
In this work, all the devices needed for the network were
designed, developed and deployed in the field around the
CC2530 chip from Instruments (TI). This chip supports the
ZigbeePRO stack solution, with a small form factor and
sufficient computational resources to execute the proposed
algorithms. The developed devices include two chips from
Texas Instruments: an MSP430 microcontroller where the
firmware can be developed and implemented, and a CC2530
which is in charge of the whole communication and the
Zigbee stack. The device also includes an optional external
board to be connected through an USB port for programming
and testing. However only the coordinator uses the USB port
during the deployment, to be interfaced with the remote host
2966 IEEE SENSORS JOURNAL, VOL. 15, NO. 5, MAY 2015
Fig. 2. End Device architecture and node developed to be plugged directly
in a commercial driver to control and be supplied.
and no more external hardware is required for networking. The
router is equipped with sensors to monitor the controlled area
while the end device is interfaced with the commercial light
driver.
In the proposed smart lighting system the most important
elements are:
•The LED panels, highly efficient white LED for illumi-
nation;
•The CC2530 that provides the management of ZigBee
and is present in each node of the network;
•The MSP430 for the control of the LED panels’ smooth-
ing and where the distributed intelligence is implemented.
MSP430 is present in all the nodes;
•A dimmable commercial driver for the LED, which
provides a highly dimmable range (up to 89%) and an
accrate control (constant current) for the smoothing.
•A light and PIR sensors, used by the router to monitor
and control the brightness value.
In the following subsection the wireless network and the three
architectures of the nodes are presented.
A. Wireless Driver Device.
In each LED panel a new device is needed to enable
the wireless control. The sole purpose of this device is to
control through PWM the driver LED providing an accurate
smoothing of the light and to communicate with the wireless
network. As mentioned earlier, the node is built around the
CC2530 and MSP430 from TI, where the CC2530 chip
is used for the network and the MSP430 on board is the
core intelligence which manages the radio chip and where
the firmware is running. Figure 2 shows the architecture of
the device developed which includes the electronic circuits
to control the industrial driver and to be supplied from an
external power supply from 3V to 24V which can also come
from the driver itself depending of the model. To allow the
power stage to convert and give a stable 3V supply to the
node, a step down low dropout (LDO) regulator with an
ultralow quiescent current TLV70433 from Texas Instruments
was used. This chip has a very low quiescent current with
high conversion efficiency and it is optimized specifically for
the MPS430.
Fig. 3. Router Architecture.
PWM to Driver Block in Figure 2 is the most important part
of the end device and it is needed to convert the PWM signal
generated from the microcontroller in a 0-10V signal needed
to control the commercial LED driver. The 0-10V control is
one of the earliest and simplest electronic lighting control
signaling systems and it is included in the most commercial
drivers. Due to this interface the node can be adaptable to a
wide range of commercial drivers with the 0-10V port, and
can be incorporated directly into the driver as figure shows.
To achieve this goal it is sufficient to insert a P-MOS transistor
in a Common Collector configurationbetween the PWM signal
of microcontroller and the 0-10V driver’s input. For the end
devices, we do not have any sensors on board as the PWM
value is decided from the router which controls more than
one device in the same group and it will be presented in next
subsection. This has the benefit to bringing flexibility in the
deployment and more reliable feedback on the light in the
monitored area.
B. Router for Monitoring and Decisions Making.
This device is in charge of the most important workload
in the network with the following main duties: i) manage
the routing protocol of the Zigbee stack, monitoring the
evniromental parameters throughout the sensors, ii) take the
decision on the light intensity, and iii) send the control
configuration to the panels that are assigned under its control
during the network configuration. Figure 3 shows the hardware
architecture of the router node, which is very similar to the
end device, where instead of the PWM driver control there
is the infrared sensor (PIR) block. This block includes the
sensor and its coupling circuit which generates an interrupt
when an object moves in the filed of view. The PIR used
is the Panosonic EW - AMN34111J which garantees a fast
and accurate interrupt for any moving object in the range
of 10m. The interrupt generated by the PIR block is connected
directly to a General Purpose Input Output (GPIO) pin of
the MSP430 to recieve the interrupt. The light sensor on the
board is enabled to monitor the luminosity in the area of
interest and is an input to the algorithm. The light sensor is
the SSFH 5711 a high accuracy ambient light sensor from
Osram. The smart control of the light is managed by the low
power microprocessor (MSP430) acquiring by ADC the light
sensor data and computing the light intensity according to the
implemented power policy and the user preferences. As for the
MAGNO et al.: LOW COST, HIGHLY SCALABLE WSN SOLUTION TO ACHIEVE SMART LED LIGHT CONTROL 2967
Fig. 4. Coordinator board connected to the remote host user interface and
monitoring application.
end device, the routing protocol is managed using the CC2530
with the ZigbeePRO stack.
C. Base Control Station
The base control station is the hub of the proposed system
as it allows the visualization of the lighting system and the
setting of important parameters such as the users’ preferences.
The role of the coordinator is only to manage the network and
allow the user interface through a remote host. The device is
provided with interface to be connected with UART to USB
ready to use as showed in Figure 4. Thanks to the interface and
the remote host it is possible to set the users’ preferences, and
monitor the whole network and store all the data to evaluate
the power saving.
D. Wireless Sensor Network
One of primary goals in designing the proposed system
was the scalability, the low power and a standardized network
for commercial application. ZigBee is a wireless communi-
cation technology based on the IEEE802.15.4 standard for
communication among multiple devices in a wireless personal
area network (WPAN). The ZigBee alliance has developed
low-cost, low power consumption, wireless communication
standard, and the CC2530 chipset was chosen. Therefore, this
standard is designed to be more affordable than other WPANs
(Wi-Fi or Bluetooth) for developing low power embedded sys-
tems for consumer electronics, home and building automation,
industrial controls, PC peripherals, medical sensor application,
toys and games. The ZigBee architecture is made up of a
set of blocks called layers; each layer performs a specific set
of services for the layer above. The IEEE802.15.4 standard
defines the two lower layers: the physical (PHY) layer and the
medium access control (MAC) sub-layer. The ZigBee Alliance
builds on this foundation by providing the network (NWK)
layer and the framework for the application layer, such as the
ZigBee device objects (ZDO) and the manufacturer-defined
application object.
TI provided the Z-Stack to use easily the Zigbee stack
implemented on the CC2530. The network is built to transfer
information from the router to the panels and from the user
interface to the distributed routers who will perform the
algorithm to select the dimming value of the lamps. The
LEDs’ light is associated to only one router that controls them
as described by Figure 6. With the distributed approach, the
routers can decide the brightness level without continuously
sending and receiving messages to the central host. In this
way, the system saves energy for the transmission increasing
together the reactivity as the router is the closer parents of the
controlled panels. This is especially true when the network is
expanded and the number of nodes and messages exchanged
increase [31].
A mesh network was chosen to maximize the scalability of
the network to a dimension of more than 64000 wireless nodes.
TheZigBeewirelesscommunication network has been imple-
mented using the CC2530 chip and the home automation PRO
Stack already implemented on the chip. The developed devices
have a PCB antenna and provide an operation range of tens of
meters indoor and outdoor with selectable output power from
−22 dBm to 4.5 dBm according with application scenario. The
peak power during transmission, while covering more than
20 meters with an indoor application and +4.5dBm output
power, is around 100mW at 3.3V. However, the firmware can
adapt the transmission power runtime according to the received
signal strength indicator. In this work we are not using an
adaptive power transmission and we fixed the power output at
+4.5dBm.
IV. LIGHTING CONTROL ALGORITHM
As it was presented in the previous section there are three
different devices which need three different algorithms to
work properly. The network software is a critical part of the
system (Figure 7 and Figure 8). The Z-STACK from TI was
used to work with the ZigbeePRO protocol with the CC2530.
In this section are presented only the algorithms needed for
the smart light control residing on the three node topologies:
End Device, Router, Coordinator (Figure 5).
A. End Device Algorithm
Figure 5 shows the main flowchart of the algorithm. The
main task of the network management is to receive and set the
right brightness for the LED panel (Figure 7). Thus, after the
device joins the network, a router is associated to it. From this
instant it waits for the PWM value decided from the router’s
own algorithm and sets the LED light intensity of the panel.
After the value is set, the radio goes into standby mode for
energy saving. The wake up time to get a new luminosity
value can be selected by the user as this affects the response
time, in the proposed approach 500ms was selected, since it
is a good trade-off between power saving and reactivity. This
simple procedure with the above mentioned hardware allows
every commercial driver to be controlled through a standard
Zigbee network.
B. Router and Control Algorithm
The router algorithm is somewhat more complex than for
the end device. The core of smart lighting intelligence is
distributed to each router which then controls one or more
2968 IEEE SENSORS JOURNAL, VOL. 15, NO. 5, MAY 2015
Fig. 5. Flow charts of the three devices. The Network management is not included in the flow chart and the Z-STACK from TI library were used to work
with the ZigbeePRO protocol.
Fig. 6. Overview of the Zigbee network for the proposed system with one
coordinator, one router who controls the panel according to the light and
motion senosrs and the controlled end devices wich adapt the LED light
intensity trough the PWM driver.
Fig. 7. Pseudocode of the End Device Algorithm to set the PWM value of
the LED driver.
end devices. To achieve this important goal, the router has as
main blocks, the communication and control algorithm on it.
The communication block is in charge of receiving data
from the network about the user’s preferences and send data
about the status of the controlled panels to the remote host.
Fig. 8. Pseudocode of the Control Algorithm distributed in each wireless
sensors’ node.
As the network is a mesh, the information can hop to other
routers before reaching the coordinator which monitors the
status of the panels and manages the errors.
Figure 5 shows the control algorithm running on the sensor
node device. The control algorithm is the core of the smart
lighting application setting the dimming value of the panel.
To achieve this goal the microcontroller acquires the light
sensor data through the ADC port containing the brightness
of the room. The purpose of this measurement is to ensure an
optimal level of illumination in the room according to the user
preferences and existing standards for lighting. The algorithm
controls the brightness value of the panels setting the PWM
to achieve the optimal value of luminosity in the area and
save energy. As Figure 8 shows in case the brightness of the
room is higher than the desirable user’s value the PWM value
MAGNO et al.: LOW COST, HIGHLY SCALABLE WSN SOLUTION TO ACHIEVE SMART LED LIGHT CONTROL 2969
is decreased. On the other hand, PWM value is increased
when the monitored brightness is lower than the desirable
user’s value. Moreover with energy savings in mind, the
algorithm includes a PIR management routine which identifies
the presence of people from the motion sensor. The control
algorithm turns off the group of LED panels to prevent waste
if no movement is detected for a certain amount of time (which
is set by the user). In the same way the PIR management
interrupt routine turns the panels on quickly and wakes the
control routine up if any movement is detected. This feature
enables the capability to switch on the LED panels only when
necessary, avoiding the waste of energy.
The main challenge with the sensors’ node is its correct
placement in the monitored area, because this can affect
the performance of the whole system. For instance, if the
light sensor is under direct sunlight or panel light it can
give wrong feedback to the control algorithm. Moreover it
is also important to avoid shadow places or places where the
temporary shadows are generated from people’s movements.
For these reasons the best positions were found on the
ceiling in the middle of the light groups. Here the light sensor
is not directly affected by the external environmental light
or LED panels and from random shadows. In this position
the Panasonic EW - AMN34111J PIR sensor also provided
optimal performance covering 10 meters with a wide angle
(around 120°) detecting all the movements in the field of view
with a fast response of few milliseconds. Concerning the PIR
sensors it is also important to avoid undesirable switching off
if no movement is detected from a long of time. To avoid this
condition the TIMEOUT_TIME of switching on has to be
chosen carefully. For our deployment a conservative value of
45minutes was chosen during the office time 8am-6pm, and
5minutes outside this time interval. As it will be presented
in the experimental results section, the deployment in a real
office was active for a continuous 6 months period with full
user satisfaction who did not notice any difference with the
traditional system without the smart control.
C. Coordinator Algorithm
The main role of the coordinator, over setup and control of
the Zigbee WSN, is to connect the wireless devices deployed
in the building with a remote host which provides the user
interface. The coordinator also sends the user preferences
to the routers and collects the status information from the
routers to store the monitored status in a remote database.
The communication is done through the UART port of the
microcontroller and the UART to USB converter that allows
connecting the dongle to every host with an USB interface.
Thus, the coordinator works as a gateway and it is required for
a graphical display of the results and user input. Furthermore,
data on wireless device operations are associated with the
LEDs light address; consequently, all faults and the state are
easily identified.
The graphical interface enables monitoring the state of the
system with the state of the lights and the power consumption
of each controlled LED light (individual energy consumption
meter) Figure 4. As the host interface also stores the dimming
Fig. 9. Deployment of the system in the VerdeLED company offices.
value of all the panels the user or network manager can have
an overlook of the power consumption and working time of
every panel in a graphical vision. The program is also equipped
with a management system that acts in case of no acknowl-
edgements are sent from the panel to highlight the errors.
V. EXPERIMENTAL RESULTS
All prototypes have been developed, tested and deployed in
variable real-life conditions to verify the overall functionality,
the scalability and the robustness of the network and seek
better performance. This section describes an experimental
evaluation of the system as applied to indoor offices. Firstly,
the measurements of power consumption of the devices in
different states are presented. Evaluation of the power saving
energy of a commercial LED panel VER-P6060-43-840 from
VerdeLED with the dimmable driver LPF-40D-42 from Mean
Well is presented. Secondly, the section presents power con-
sumption measurements done during 6 months of continuous
work in the company office where the smart lighting system
was deployed as the primary and only light system.
Figure 9 shows the development system implemented in
an office with the goal of testing it in real conditions while
Figure 1 shows the floor plan and where the sensor nodes
were placed. In this implementation, 25 wireless devices were
directly connected to the power supply of LED panels, so is
possible to cover the whole office presented in Figure 1. The
Zigbee network has been deployed in an office together with
2 Wi-Fi internet access points and several users’ phones and
PC connected to the access points. Under these conditions the
system was working for 6 months without any interruption
showing a high robustness to Wi-Fi interferences.
The positioning of panels and sensors was done with a
preliminary analysis on solar irradiation within the office
as explained in section IV.B, and taking into consideration
the work time difference of the employees in this office.
This has permitted an accurate positioning of the sensors
that, in combination with the configuration of the network
(connections between LED panels and sensor) by user
interface, have provided a stable and robust solution working at
writing time for 6 months without any interruption of services.
2970 IEEE SENSORS JOURNAL, VOL. 15, NO. 5, MAY 2015
Fig. 10. LED panel +LED driver power consumption according to the the
PWM signal of the microcontroller.
A. Power Measurements
The first step to evaluate the energy saving achievable
with our approach was to understand the power consumption
reduction dimming a panel, Figure 10 shows the characteristic
of power consumption/dimming of each LED panel which
include the consumption of driver as well. As explained
earlier, the microcontroller sets different PWM signals for
the brightness of the panel. The characteristic of the panel
was measured directly with an AC power meter changing
smoothly the PWM signal at step of 1%. This data shows
the importance of the dimming not to waste energy as it is
possible to save up to 99% of power by just dimming the
light. The characteristic shows also the limit of the driver to
smooth the light down to more than 87% of PWM. Below this
value the panel is switched off so no light is provided. During
the tests, several commercial drivers from different producers
were evaluated and the selected driver LPF-40D-42 was the
one with the best performance in terms of range (0-87%) and
accuracy of constant current in output (useful to guarantee long
LED panel lifetime). In order to evaluate the ultra-low power
consumption due to the extra hardware needed to add the smart
light wireless control of our approach, the end device and
routers power consumption were measured in several states.
Table I shows the wireless nodes’ current consumption (End
Device and router) with 5V power supply. Measurement of the
wireless sensor’s power consumption was performed, setting
the clock of the MCU at 1MHz, and assuming the node can
be in one of the three configurations shown in the table.
The maximum power consumption due to the new hardware
is 100mW. This is a negligible consumption compared to the
power saving. In fact, as Figure 10 shows with a dimming of
only 20% the system is saving already around 20W. Moreover
the ED has two power saving modes to save energy, one is
switching the radio ON periodically according to the Zigbee
stack to reach only 0.500mW; the second one is manually
switched off by the user. The power supply of the ED is
associated with the power supply of the LED panel so when
the user switches off the Panel the node will consume zero in
this case.
TAB L E I
NODES’CURRENT AND POW ER CHARACTERISTICS
Fig. 11. Average power consumption measured in three different weather
condition day on 20 panels of 40W each.
B. Power Saving Evaluation
To evaluate the proposed system in terms of power saving
a real office was used as testbed. Four separated groups of
5 LED panels each were controlled by four routers. The
user preference was set to 600Lux, a common value to have
good quality of light. The network was run continuously for
6 months and the coordinator saved all the states of the PWM
signal of each ED. Thanks to the characteristic of the panel
in Figure 10 it is possible to know the instantaneous power
of each panel during the day and night. Figure 11 shows the
average power saving and consumption of three days from
8am to 18pm (during the open time of the office) for all the
20 panels to evaluate the benefits in terms of power saving
and the influence of the weather conditions. The data were
compared with an office scenario without the smart control
and the Zigbee network. Without the smart control the average
power consumption of each panel is 40W, as all the panels
are fully on to the max power during the time office. The
MAGNO et al.: LOW COST, HIGHLY SCALABLE WSN SOLUTION TO ACHIEVE SMART LED LIGHT CONTROL 2971
TAB L E II
POWER SAV I N G MEASURMENT COLLECTED DURING SIX DAY S
power consumption of 40W is the power consumption without
the wireless control system and takes into count only the
power of LED driver and the LED panel. As we presented in
previous sections when the smart control is active, the panels
are dimmed according to the user preference, the brightness in
the room and the motion detection sensor. The figure shows
the average power consumption of each panel during three
different days with 3 different weather conditions. This power
was evaluated as total power measured to supply the entire
panels divided by the number of the panels. The three plots
shows the influence of the weather conditions and the power
saving. In fact, the average power consumption is only 9W
against the 40W in a sunny condition, while it is 12W in
variable conditions and 17.5W in cloudy conditions.
TABLE II shows the power saving during six day for all the
areas of the office. This table shows how the power saving is
affected by the external light and the presence of the people in
the room. For example, the day 5 was cloudy so the external
light brightness was not very high then the power saving is
much lower. Moreover the showroom is the least used room
in the office as there is not a stable/regular presence of people
inside so the power saving is always higher there. These results
show how important it is to have different group of lights
controlled by separate routers to have a more efficient control
and power saving. Finally taking into account only the 6 days
for space reason, the overall energy saved was around 43% due
to the proposed approach compared with the same deployment
without it. This value does not take into account the night
period, supposing the user is always reminded to switch off
the office lights or lights being switched off due to inactivity.
Another important parameter is the achieved power savings
according to different periods of the year.
TABLE III presents the average power consumption of the
each of the 20 panels. The data was acquired during 2 weeks
in December 2013, when the deployment started and during
TABLE III
POWER SAV I N G I N TWO WEEKS IN MAY 2014 AND DECEMBER 2014
2 weeks of May 2014. The table shows how each day the
power saving changes according to the weather condition and
also that the average power consumption in December was
much higher than in May. This is due mainly to the worse
weather conditions and the lower sun light hours in winter
with respect to the spring.
Finally, in order to evaluate better the benefit of the pro-
posed approach, data of power consumption in-field for long
period has been acquired. At the time the paper has been
written the system has been running without any interruption
in that office for 6 months from December 2013 to May 2014.
The average energy saving in this period was around 55%. It is
estimated that the power saving will be higher during a full
year as in spring and summer the day light will decrease the
average usage of the panel and the smart light system will be
optimize the dimming level to minimize the waste of energy.
Concering the low cost, in the proposed solution all the parts
were evaluated in terms of production cost for a volume of
10000 units. For this not so high volume the system costs 200$
for the cordination and base station, 50$ for the sensors’ node,
and only 15$ for the end devices, which has to be connected to
the LED Driver. TABLE IV shows a comparsion with the most
popular light control system and highlights the significantly
lower cost of the proposed solution.
VI. CONCLUSION AND FUTURE WORK
A novel system to control LED lighting with a low cost and
low power wireless sensor network has been proposed. The
2972 IEEE SENSORS JOURNAL, VOL. 15, NO. 5, MAY 2015
TAB L E IV
COST COMPARISON OF DIFFERENT CONTROLLING SYSTEM
method requires the deployment of complementary sensors
with Zigbee radio that generate a PWM signal to control
existing commercial LED drivers, which can significantly
reduce the power consumption of the LED lighting. The use
of a light sensor and a PIR sensor in combination with the
user preferences allows the distributed intelligence to save
energy reducing the light intensity. Because many fixtures of
LED lights are already placed, this solution is also suitable
for retrofitting. Moreover the network is flexible and scalable
due to the Zigbee radio. Experimental results indicate that
the proposed system outperforms the state-of-the-art with
a significant reduction of power consumption and cost for
the single and groups of LED lights using the low power,
scalable WSN. It has been shown that this approach decreases
the power consumption in a real life office application by
more than 55% throughout 6 months (in an unpredictable Irish
weather scenario). The prototypes are ready to be inserted in a
commercial driver to enable wireless capability and distributed
control.
ACKNOWLEDGMENT
The authors would like to thank NANOTERA ICYSOC,
EU project PHIDIAS (G.A. 318013), Irish Research
Council (iLUX project) and VerdeLED.
REFERENCES
[1] R. Smale, M. Hartley, C. Hepburn, J. Ward, and M. Grubb, “The impact
of CO2emissions trading on firm profits and market prices,” Climate
Policy, vol. 6, no. 1, pp. 31–48, 2006.
[2] E. G. Dascalaki, K. Droutsa, A. G. Gaglia, S. Kontoyiannidis, and
C. A. Balaras, “Data collection and analysis of the building stock and
its energy performance—An example for Hellenic buildings,” Energy
Buildings, vol. 42, no. 8, pp. 1231–1237, Aug. 2010.
[3] D. Caicedo and A. Pandharipande, “Distributed illumination control
with local sensing and actuation in networked lighting systems,” IEEE
Sensors J., vol. 13, no. 3, pp. 1092–1104, Mar. 2013.
[4] S.-R. Lim, D. Kang, O. A. Ogunseitan, and J. M. Schoenung, “Poten-
tial environmental impacts of light-emitting diodes (LEDs): Metallic
resources, toxicity, and hazardous waste classification,” Environ. Sci.
Technol., vol. 45, no. 1, pp. 320–327, Jan. 2011.
[5] B. Von Neida, D. Maniccia, and A. Tweed, “An analysis of the energy
and cost savings potential of occupancy sensors for commercial lighting
systems,” J. Illuminating Eng. Soc., vol. 3, no. 2, pp. 111–125, 2001.
[6] Networked Lighting Controls Will Surpass $5.3 Billion in Annual
Revenue by 2020. [Online]. Available: http://www.navigantresearch.com/
newsroom/networked-lighting-controls-will-surpass-5-3-billion-in-ann-
ual-revenue-by-2020, accessed Sep. 2013.
[7] T. Torfs et al., “Low power wireless sensor network for building
monitoring,” IEEE Sensors J., vol. 13, no. 3, pp. 909–915, Mar. 2013.
[8] M. Magno, D. Boyle, D. Brunelli, E. Popovici, and L. Benini, “Ensuring
survivability of resource-intensive sensor networks through ultra-low
power overlays,” IEEE Trans. Ind. Informat., vol. 10, no. 2, pp. 946–956,
May 2014.
[9] M. Magno, N. Jackson, A. Mathewson, L. Benini, and E. Popovici,
“Combination of hybrid energy harvesters with MEMS piezoelectric and
nano-Watt radio wake up to extend lifetime of system for wireless sensor
nodes,” in Proc. 26th Int. Conf. Archit. Comput. Syst. (ARCS), Feb. 2013,
pp. 1–6.
[10] S. D. T. Kelly, N. K. Suryadevara, and S. C. Mukhopadhyay, “Towards
the implementation of IoT for environmental condition monitoring in
homes,” IEEE Sensors J., vol. 13, no. 10, pp. 3846–3853, Oct. 2013.
[11] J. Byun, I. Hong, B. Lee, and S. Park, “Intelligent household LED
lighting system considering energy efficiency and user satisfaction,”
IEEE Trans. Consum. Electron., vol. 59, no. 1, pp. 70–76, Feb. 2013.
[12] S. Matta and S. M. Mahmud, “An intelligent light control system
for power saving,” in Proc. 36th Annu. Conf. IEEE Ind. Electron.
Soc. (IECON), Nov. 2010, pp. 3316–3321.
[13] J. Lu, D. Birru, and K. Whitehouse, “Using simple light sensors
to achieve smart daylight harvesting,” in Proc. 2nd ACM Workshop
Embedded Sens. Syst. Energy-Efficiency Building, Zürich, Switzerland,
2010, pp. 73–78.
[14] M. Erol-Kantarci and H. T. Mouftah, “Wireless sensor networks for
cost-efficient residential energy management in the smart grid,” IEEE
Trans. Smart Grid, vol. 2, no. 2, pp. 314–325, Jun. 2011.
[15] F. O’Reilly and J. Buckley, “Use of wireless sensor networks for fluo-
rescent lighting control with daylight substitution,” in Proc. Workshop
Real-World Wireless Sensor Netw. (REANWSN), Stockholm, Sweden,
Jun. 2005.
[16] V. Singhvi, A. Krause, C. Guestrin, J. H. Garrett, Jr., and H. S. Matthews,
“Intelligent light control using sensor networks,” in Proc. ACM Int.
Conf. Embedded Netw. Sensor Syst. (SenSys), San Diego, CA, USA,
Nov. 2005, pp. 218–229.
[17] Y.-J. Wen, J. Granderson, and A. M. Agogino, “Towards embed-
ded wireless-networked intelligent daylighting systems for commer-
cial buildings,” in Proc. IEEE Int. Conf. Sensor Netw., Ubiquitous,
Trustworthy Comput. (SUTC), Taichung, Taiwan, Jun. 2006, pp. 1–6.
[18] H. Park, M. B. Srivastava, and J. Burke, “Design and implementation
of a wireless sensor network for intelligent light control,” in Proc. 6th
Int. Symp. Inf. Process. Sensor Netw. (IPSN), Cambridge, MA, USA,
Apr. 2007, pp. 370–379.
[19] S. H. Hong, S. H. Kim, J. H. Kim, Y. G. Kim, G. M. Kim, and
W. S. Song, “Integrated BACnet–ZigBee communication for building
energy management system,” in Proc. 39th Annu. Conf. IEEE Ind.
Electron. Soc. (IECON), Nov. 2013, pp. 5723–5728.
[20] V. Jelicic, M. Magno, D. Brunelli, G. Paci, and L. Benini, “Context-
adaptive multimodal wireless sensor network for energy-efficient gas
monitoring,” IEEE Sensors J., vol. 13, no. 1, pp. 328–338, Jan. 2013.
[21] M.-S. Pan, L.-W. Yeh, Y.-A. Chen, Y.-H. Lin, and Y.-C. Tseng, “A WSN-
based intelligent light control system considering user activities and
profiles,” IEEE Sensors J., vol. 8, no. 10, pp. 1710–1721, Oct. 2008.
[22] L.-W. Yeh, C.-Y. Lu, C.-W. Kou, Y.-C. Tseng, and C.-W. Yi,
“Autonomous light control by wireless sensor and actuator networks,”
IEEE Sensors J., vol. 10, no. 6, pp. 1029–1041, Jun. 2010.
[23] Y. K. Tan, T. P. Huynh, and Z. Wang, “Smart personal sensor network
control for energy saving in DC grid powered LED lighting system,”
IEEE Trans. Smart Grid, vol. 4, no. 2, pp. 669–676, Jun. 2013.
[24] D. Tran and Y. K. Tan, “Sensorless illumination control of a networked
LED-lighting system using feedforward neural network,” IEEE Trans.
Ind. Electron., vol. 61, no. 4, pp. 2113–2121, Apr. 2014.
[25] M. Tomoroga, I. Jivet, and D. Nicoara, “Inteligent tele-management
of street lighting equiped with HID lamps,” in Proc. 10th Int. Symp.
Electron. Telecommun. (ISETC), Nov. 2012, pp. 7–10.
[26] Q. Y. Zhou, Z. L. Zhang, and F. Y. Cui, “Research on the integrated
control teaching system of building based on fieldbus,” Appl. Mech.
Mater., vol. 278, pp. 1952–1955, Jan. 2013.
[27] KNX Lighting Control. [Online]. Available:
http://www.knxuk.org/index.php?option=com_content&view=article&id
=40&Itemid=143, accessed Jun. 2014.
[28] Redwood Control System. [Online]. Available:
http://www.redwoodsystems.com, accessed Jun. 2014.
[29] Lutron Electronics, Inc.—Dimmers And Lighting Controls. [Online].
Available: http://www.lutron.com/en-US/Pages/default.aspx, accessed
Sep. 2014.
MAGNO et al.: LOW COST, HIGHLY SCALABLE WSN SOLUTION TO ACHIEVE SMART LED LIGHT CONTROL 2973
[30] Eyenut Ingenuous Control. [Online]. Available: http://www.eyenut.co.uk,
accessed Sep. 2014.
[31] P. Baronti, P. Pillai, V. W. C. Chook, S. Chessa, A. Gotta, and Y. F. Hu,
“Wireless sensor networks: A survey on the state of the art and the
802.15.4 and ZigBee standards,” Comput. Commun., vol. 30, no. 7,
pp. 1655–1695, May 2007.
Michele Magno (M’13) received the master’s and
Ph.D. degrees in electronics engineering from the
University of Bologna, Bologna, Italy, in 2004 and
2010, respectively. He is currently a Post-Doctoral
Researcher with the Swiss Federal Institute of Tech-
nology Zurich, Zurich, Switzerland, and a Research
Fellow with the University of Bologna. The most
important themes of his research are on wireless
sensor networks, power management techniques, and
extension of lifetime of batteries-operating devices
and embedded video surveillance. He has collabo-
rated with several universities and research centers, such as the University
College Cork and Tyndall Institute, Cork, Ireland, Imperial College London,
London, U.K., the University of Trento, Trento, Italy, the Politecnico di Turin,
Turin, Italy, and the University of British Columbia, Vancouver, BC, Canada.
He has authored over 30 papers in international journals and conferences.
Tommaso Polonelli received the bachelor’s degree
in electronic engineering from the University of
Bologna, Bologna, Italy, in 2013. He is currently
pursuing the master’s degree in electronic engineer-
ing at the University of Bologna. He has collabo-
rated with the University of Cork, Cork, Ireland,
where he has been for three months during his
bachelor’s thesis project. He is also collaborating
with the University of Bologna to transfer a smart
lighting control using wireless sensor network to the
VerdeLED company in Ireland
Luca Benini is currently a Full Professor with
the University of Bologna, Bologna, Italy, and the
Chair of Digital Circuits and Systems with the Swiss
Federal Institute of Technology Zurich, Zurich,
Switzerland. He served as the Chief Architect for the
Platform2012/STHORM project with STMicro-
electronics, Grenoble, France, from 2009 to 2013.
He has held visiting and consulting researcher
positions at the École Polytechnique Fédérale de
Lausanne, Lausanne, Switzerland, IMEC, Leuven,
Belgium, Hewlett-Packard Laboratories, Palo Alto,
CA, USA, and Stanford University, Stanford, CA, USA. His research interests
are in energy-efficient system design and multicore system-on-a-chip design.
He is also active in the area of energy-efficient smart sensors and sensor net-
works for biomedical and ambient intelligence applications. He has authored
over 700 papers in peer-reviewed international journals and conferences, four
books, and several book chapters. He is a member of the Academia Europaea.
Emanuel Popovici (SM’08) received the
Dipl.-Ing. degree in computer engineering from
Politehnica University Timisoara, Timisoara,
Romania, in 1997, and the Ph.D. degree in
microelectronic engineering from the National
University of Ireland, in 2002. He has been a Senior
Lecturer with the Department of Electrical and
Electronic Engineering, University College Cork,
Ireland, since 2002. His research interests include
embedded system design for low power, reliable,
and secure computing and communications
with applications in smart LED lighting systems, biomedical, agricultural,
structural health, energy control and optimization, learning, and entertainment
domains. He has authored widely in these areas, with over ten co-authored
papers being distinguished at national and international levels. He also
co-supervised three interdisciplinary teams who achieved top awards in the
IEEE/IBM Smarter Planet Challenge in 2011, 2013, and 2014, respectively.
