Content uploaded by Sabrine Khriji
Author content
All content in this area was uploaded by Sabrine Khriji on Feb 13, 2019
Content may be subject to copyright.
Measuring Energy Consumption of a Wireless
Sensor Node during Transmission: panStamp
Sabrine Khriji∗†, Rym Ch´
eour‡, Martin Goetz∗, Dhouha El Houssaini∗, Ines Kammoun†and Olfa Kanoun∗
∗Chemnitz University of Technology, Measurement and Sensor Technology, Chemnitz, Germany
†National School of Engineers of Sfax, University of Sfax, LETI Laboratory, Sfax, Tunisia
‡National School of Engineers of Sfax, University of Sfax, Computer and Embedded System Laboratory, Sfax, Tunisia
Email: ∗sabrine.kheriji@etit.tu-chemnitz.de, ‡rym.cheour@enis.tn, ∗martin.goetz@etit.tu-chemnitz.de,
∗dhouha.el-houssaini@etit.tu-chemnitz.de, †ines.kammoun@ieee.org, ∗kanoun@ieee.org
Abstract—In most of the real-life application scenarios, where
Wireless Sensor Networks (WSNs) are deployed, the energy is
a limited and valuable resource. Many energy-based strategies
were proposed in the literature to reduce the power consumption
of sensor nodes and thus enhance the network lifetime. When
those algorithms are developed and tested with some well-known
network simulators like NS3, OPNET, the energy estimation
aspect is not a key criterion, because these simulators have their
own specific energy models. By implementing these algorithms in
a real sensor node, the energy consumption can be measured
and compared to the results provided by other previously
implemented approaches. This paper proposes a method for
estimating the energy consumption of a low power wireless node
panStamp NRG 2.0. The proposed model calculates the consumed
energy of the system, at run time for different operation cycles.
Besides, since the biggest amount of energy is consumed during
the communication, this paper aims to characterize the power
consumption in the transmission state. A proper metric of the
energy consumed per byte transmission pattern of the packets is
provided. The results can be used to derive not only the energy
consumption for other protocols but also the energy needed to
transmit one byte as well as the payload size are given.
Keywords—Wireless sensor networks, energy estimation, com-
munication, energy model, packet transmission, run time,
panStamp.
I. INTRODUCTION
Wireless sensor networks consist of tiny sensor nodes
with limited energy and resources such as power processing,
memory, radio bandwidth and communication capabilities [1].
A key issue in large scale WSN is the amount of power
required to maintain an efficient working of the network for
a long duration since it is impracticable to change the battery
for the network with thousands of physically embedded sensor
nodes [2]. For this purpose, energy consumption has to be
minimized for optimal operation and network coverage by
investigating the energy awareness in each stage of wireless
sensor network design and working process [3]. With energy
efficiency being a major issue in the design of WSNs, re-
searchers have thoroughly investigated new algorithms such
as communication protocols to save energy consumption [4].
These algorithms are developed and tested with some well-
known network simulators like NS3 [5], OPNET [6]. However,
the energy estimation aspect is not considered as a main
criterion since these simulators have their specific energy
models. Furthermore, for real implementation the measurement
of residual battery for a sensor node is problematic. Current
wireless sensor node such as the IRIS [7], the ESB [8] and the
panStamp [9] do not enable hardware mechanisms to measure
the energy consumption of the sensor mote at run time. In
general, it is possible to calculate the battery voltage of every
sensor node but, this method cannot provide a good estimation
for the current consumption. In this paper, we perform a
wireless communication between a System on Chip (SoC)
named panStamp which is a low power module that is widely
used in many automation and WSN projects [10]. Optimizing
all the aspects of WSN, in particular, the communication and
the networking is the overarching goal.
This paper presents a study of panStamp packet transmis-
sion in data mode using CC11XX radio channel for the trans-
mission of CCPACKET from the transmitter to the receiver.
The packet is transferred using the modem and a serial protocol
and later energy measurement is carried-out on these packets
in order to find-out the energy required for transmitting one
byte. Also, the target of this work is to develop an energy
model to measure the residual battery of panStamp NRG
2.0. This method estimates the energy consumption in run
time. As the power consumption values and the transition cost
related to each state of each component cannot be found in the
datasheet. Besides, as traditionally the power consumption for
the different states are extracted from datasheet to determine
the operating time, the exact measurements on its energy con-
sumption are needed [11]. The main advantage of our method
over the existing algorithms is that the energy consumption
measurement during the transmission and the reception phase
considers real experiments, not only the characteristics of
the node described by the specifications of the manufacturer.
This leads to a realistic power consumption model for WSN
devices. Moreover, this approach enables a measurement of the
transition overhead during the state transistions. This enables
more accurate send and receive models as well as a better
energy and node lifetime calculation.
As a second contribution to the proposed method, the
’Time break even’ Tbe between the different states is estimated.
The duration and the current consumption of each transition
between two states are recorded. Finally, estimating the life-
time of a device is based on both the battery model and the
consumption model, that’s why we focus our work not only
on a non-linear model but also on different power modes. So,
during the active mode, we estimate the energy consumption
274
2018 IEEE 32nd International Conference on Advanced Information Networking and Applications
1550-445X/18/$31.00 ©2018 IEEE
DOI 10.1109/AINA.2018.00050
at both transmit, receive and sleep modes.
The remainder of the paper is organized as follows; Section
II briefly presents relevant previous works. Section III de-
scribes the considered energy consumption estimation method.
Section V analyzes the experimental results obtained with the
panStamp. Finally, we conclude the paper and give directions
for future work.
II. RELATED WORKS
The target behind the development of an energy based
protocols for the sensor networks is the measurement of energy
with exact values. PanStamp and other hardware platforms
do not provide hardware energy measurement capabilities [9].
Usually the energy consumption is estimated using a software
method [12]. By reading ADC, it is possible to measure the
battery voltage and to predict also the residual energy from
the battery voltage.
Several available studies focus on estimating the lifetime
of sensors through simulation like [13] that presents a method
operating on a compiled firmware images and modeling the
complex behavior of batteries through the MSPSim/Cooja node
emulator and network simulator. The temperature is a variable
parameter which affects the battery lifetime. Assuming that it
is constant can lead to some inaccuracies.
Therefore, to estimate the residual energy in the battery,
the energy consumption model of [14] is based on a current
consumption specifications of the IRIS node. The proposed
model determines the different hardware states as well as the
time spent in each state by each hardware component such as
the radio (transmission, reception and sleep mode) whose total
energy is calculated at the base of the IRIS datasheet by using
a linear energy model. However, this linear energy model, the
basis for estimating the residual energy of the IRIS node at
run time compared to an ideal battery mode results in a large
gap between the measured values.
Authors in [15] estimate the lifetime of WSN nodes
through the use of embedded analytical battery models.
Mainly, they present a software-level approach to estimate not
only the load state but also the voltage of the batteries depend-
ing on the use of a temperature-dependent voltage model in
low-power MCUs. Nevertheless, relying on a prediction model
results on energy profiles different from those of the sensor
network.
Most of works that estimate the amount of residual energy
in the battery are based on ideal conditions that do not take
into account the non-linearity of the battery or the variation of
the voltage. This very simple model used by default is however
not realistic. Indeed, these values were not constant so thinking
about voltage measurements will be certainly incorrect due
to the non-linearity involved in the discharge curve of the
battery that depends on how much current it draws as shown
in the discharge curve of our selected battery datasheet [16].
In addition, if the rechargeable batteries are used, it remains
constant at a particular level and the level changes only if the
battery voltage is reduced by the considerable amount. These
factors will lead to inaccurate energy estimation based on the
proposed energy model discussed lately.
Many scientific works [17], [18] and [19] assume a constant
value of supply voltage due to its hard approximation or
measurement. In [20], a battery model is used in which the
discharge curve of the battery is approximated to a linear
curve. The battery capacity depends also on temperature and
the initial capacity of the battery. The capacity equation of
the battery is derived from the linear approximated model
of the discharge curve. That method has not been tested or
verified via experimental measurement. Therefore, until now,
supply voltage presents a huge problem in its approximation
or in its measure due to different current values as well as
temperature and initial energy value. Thus, assuming it as
a constant is not perfectly accurate, nevertheless, estimating
energy consumption will be more or less acceptable.
Among the major disadvantages of techniques mentioned
above, is considering ideal conditions to validate the results.
Although the proposed battery model is based on technical
specifications close to the actual battery behavior, comparing
the results with an ideal battery model leads to an overes-
timation of the battery lifetime up to 40%-50% on average
as demonstrated in [21]. Besides, by working on a real plat-
forms we take into account a certain low-level implementation
aspects that are neglected with simulation such as the state
transition overhead that has an additional cost to the overall
energy consumption. In the next section,an energy model based
on a real hardware characteristics is formulated in order to
measure the remaining energy in batteries for each sensor node
with different operational mode.
III. PROPOSED ENERGETIC MODEL
The wireless platform used in this paper is the panStamp
NRG 2.0 which is recently introduced in the market. It is
based on the popular CC430F5137 SoC including MSP430
microcontroller and CC11XX radio chip (Texas Instruments,
2007) which operates at 433 and 868-915 MHz. The panStamp
has 4 KB RAM capability and 32 KB flash memory and it
operates with 2 AA batteries. Interested in energy consump-
tion of sensor nodes, a classic energy model has been used
along with many previous scientific works [22]. Although the
distance and other parameters are considered in calculating the
energy consumed, energy model does not seem so accurate. To
transmit lbits message to the receiver node over a distance d,
the energy consumed by considering the transmitter ETx and
by the receiver ERX could be calculated as follows:
ETx =Eelect ×l+εfs ×l×d2if d<d
0(1)
ETx =Eelect ×l+εamp ×l×d4if dd0(2)
ERX =Eelect ×l, (3)
while d0is the threshold distance. In addition, the energy
model contains many needed parameters that must be available
in the panStamp NRG2 datasheet. Unfortunately, due to its
limited information, that energy model cannot be used. The
shortage of Eelect which represents the electrical energy of the
circuitry needed to transmit or to receive, εfs and εamp present
respectively, as the energy consumption factor for free space
and multipath radio models are not even mentioned in our radio
sensor datasheet. An accurate energy model would ensure an
approximate realistic prediction of network lifetime and that
would be based on the real state of sensor nodes in each phase.
275
(a) (b)
Fig. 1: Test scenarios: (a) Wire guarded source-meter-measurement via Keysight E5270B device (b) Outdoor test field: for the
sender and the receiver node
For example, the energy consumed for a transmitter node can
be described in eq. (4):
ETx =V×ITx ×ΔtTx (4)
Afterwards, calculating the energy consumption of a sensor
node will be resumed in its scheduling chronogram. The total
energy consumed will be the sum of all energies in different
states (eq. (5)).
Etotal =
i
Ei,(5)
where Eiand Etotal are respectively, the energy consumption
for each state and the total energy consumed in Joule, Vrepre-
sents the supply voltage in Volt, Iis the current consumption in
Ampere, Δtis the state duration in seconds. Accurate power
estimation at early design stage is a key ingredient for any
successful design methodology at system level.
Therefore, a scheme to predict the energy of a wireless
device using Keysight E5270B precision IV analyser [23] as
seen in figure 1a. The device is connected to a Device Under
Test (DUT) using 4 wire system. This approach is used for
low resistance measurements that reduces the effect of test lead
resistance. The measurement accuracy can be improved with
this method because it avoids errors caused by wire resistance.
By doing so, a proper metric of energy consumed per byte for
each state can be given.
IV. ENERGY MEASUREMENT SETUP
Concretely, nothing more than real outdoor measurements
could inform about exact maximum distances in each config-
uration. Later, those experiments would be exploited to pick
the current consumption corresponding to size of transmitted
payload. In real environmental circumstances, deployed motes
will certainly face multiple obstacles that could limit their
sending-receiving range the likes of trees and plants. Fig. 1
displays the test field of experiments done in a park near
Stadlerplatz in Chemnitz, Germany, which contains different
types of hurdles.
On Fig. 1(b), a sender node is placed directly in the grass.
The first test, the radio sensor nodes are configured in Tx low
power via Arduino programming. In the second test, the radio
motes will be configured in Tx high power.
Table I shows experimental results based on maximum
distance picked approximately when the first data packet is
lost.
TABLE I: Measured distance according to power transmission
Power transmission
configuration
Measured distance
(m)
Tx low power: 0 dbm 70
Tx high power: +12 dbm 110
All the observations are done keeping following settings:
•A 3.6 V battery supply was used.
•8, 16 and 32 bytes of payload are transmitted individ-
ually to compare the energy consumed by packet for
one byte of data.
•The power was set to low power mode (0 dBm).
•The distance between transmitter and receiver was
kept to 50 m.
•The averaging sample was set to one with 4000
measurements.
•The encryption was disabled.
Fig. 2(a) describes the operation cycle for the transmitter node
which includes four states which are: Wake On Radio (WOR),
transmitting (Tx), switch from Tx to sleep mode and sleep
mode. This node is supplied by 3.6 V and it transmits 32
bytes in its payload.
276
(a) (b)
Fig. 2: (a) Dynamic current consumed drawn from the panStamp during 32 bytes packet transmission in low power transmission
mode (b) Current consumed over time for T2 and T3 state
TABLE II: Transmitter operational cycle description
State Description
T1
WOR: enable CC1101 radio to periodically
wake up from sleep and listen for incoming
packet without MCU interaction
T2 Transmitting operation, packet transmision
T3 Switch state from transmission to sleep mode
T4 Sleep mode
The first interval, marked as WOR in Fig. 2(a), is the wake
on radio time which is about 1.2 s. It requires approximately
2.5 mA as current. According to the CC1101 datasheet, the
concept behind the WOR is to keep the micrcontroller (MCU)
in deep sleep, and the CC1101 radio sleeping for some time
as well but keeping a watchdog timer, including the oscillator,
running in the CC1101 chip to wake up the CC1101 once in
a while to check for data.
Afterwards, to transmit the data packet displayed in Fig. 3,
the current consumption reaches around 19.5mA. Investigating
the radio datasheet, only 16.8 mA is needed to send data packet
for 0 dBm as power transmission. That difference is explained
by the consumption of the rest of the sensor components,
specifically the microprocessor as Keysight device measures
the whole sensor node consumption not only its radio.
In the zone marked by Tx →Sleep, the radio switches
from transmitting mode to sleep mode which takes about 7 ms
and needs around 2.8 mA. Afterwards, the SoC is maintained
in sleep mode about 2 s imposed by the user. During this
interval, the required current is approximately 0.4 mA. As seen
in Fig. 2(b), with the measurement device, the ’Time break
even’ Tbe between the transmission state and sleep state is
estimated which is approximately equal to 7 ms.
V. E XPERIMENTAL RESULTS
The CC1101 [24] contains a specific packet structure
provided in Fig. 3 with a total size ntotal. The preamble pattern
is an alternating sequence of ones and zeros (10101010). The
minimum length of the preamble is programmable through the
value of MDMCFG1·NUM PREAMBLE. When enabling Tx,
the modulator will start transmitting the preamble. When the
programmed number of preamble bytes npream are transmit-
ted, the modulator sends the synchronization word and then
the data from the Tx First In First Out (FIFO) if the data is
available.
Fig. 3: CC1101 packet structure
If the Tx FIFO is empty, the modulator will continue to
send preamble bytes until the first byte is written to the Tx
FIFO. The modulator will then send the synchronization word
and then the data bytes. The synchronization word is used to
identify networks and make them “unique”. It is presented by
277
four-byte values at maximum. It is an initial 2 bytes with 2
bytes value transmitted by most radios to synchronize timings
between nodes.
In this section, a proof of the concept for the presented
energy estimation approach is carried-out for experimental
evaluation. Different test beds are necessary to fill in a com-
plete survey of measurements. Table III provides the mean
consumed current for different payload size ( 8 bytes, 16 bytes
and 32 bytes) for each operation cycle.
TABLE III: Drained current for different node operations and
different payload size
Mean Current Datasheet 8 bytes 16 bytes 32 bytes
WOR NA2.69 mA 2.57 mA 2.51 mA
Tx 36 mA max 20.12 mA 19.69 mA 18.2 mA
Rx 18 mA max 12.21 mA 12.19 mA 12.09 mA
Sleep mode 1-2 μA0.46 mA 0.36 mA 0.41 mA
Tx to Sleep NA2.81 mA 2.85 mA 2.72 mA
transition
The data bytes are transmitted as shown in Fig. 4. The num-
ber of samples recorded during the transmission are around 3,
5 and 8 for 8, 16 and 32 bytes respectively. From this figure, we
can calculate the number of points y required for transmitting
x bytes as illustrated in eq. (6),
y=0.1027 ×x+0.8839.(6)
20 30 40 50 60 70 80
2
4
6
8
10
Total number of transmitteed bytes
Sample points recorded during the transmission
Measured values
Linear fit
Fig. 4: Measurement of the number of points per bytes
ntotal =nover +npayl (7)
f=npayl
ntotal
(8)
We considers ntotal (eq. (7)) as the total size of a packet data
and nover as the number of bytes overhead added to each
packet by the header and the footer. nover is the summation
of preamble (npream = 1 byte), sync word (2 bytes), length (1
byte), address (1 byte) and the CRC16 (2 bytes). Out of this
a fill ratio fcan be defined like shown in eq. (8) which is an
indicator of how effective the transmitted packet is used. The
best value is 100%when no header and footer are used. The
header and footer do not contribute to the data transmission,
but they are needed for a reliable communication and routing
of data. The fill ratio fcan be increased by using more payload
bytes per packet as it can be shown in fig. 5.
Total packet length ntotal = 8 bytes
Payload length npayl = 1 byte
Payload length npayl = 5 bytes
Total packet length ntotal = 12 bytes
Fill ratio f
12.5%
41.7%
Fig. 5: Fill ratio versus the payload
0 20406080100
0
20
40
60
80
100
Payload size npayl in bytes
Fill ratio fin percent
Preamble length = 1 byte
Preamble length = 5% of payload
Optimal fill size of 100%
Fig. 6: Fill ratio versus the payload
Fig. (6) illustrates the effect for different payload byte-size
npayl. It can be seen, that higher amounts of payload implies
to a better filling ratio. To demonstrate this hypothesis, the fill
ration fis measured for a static preamble size such in the blue
curve where only 1 byte of preamble is used. The preamble is
important to detect the start of the packet. With increasing pay-
load size the probability of a wrong detected packet becomes
278
high. That’s why it is recommended to increase the preamble
size accordingly. E.g. it can be chosen to be 5%of the payload,
as shown in the red curve. It can be seen, that a high amount of
payload per packet is still desirable. For wireless sensor nodes
it is essential to use the communication effectively, as it is the
biggest contributor to the energy consumption.
npacket req =npayl
nqueue
(9)
ntrans msg =(npayl +nover )·npacket req (10)
nqueue describes the maximum amount of bytes that one
node ca send per data packet. It depends of the hardware
specifications of a sensor node. In some cases, this number is
smaller than the total bytes available to be send. So, the data is
send as fast as possible, resulting in a small payload size and
a big number of send packets. For each packet, the overhead
needs to be transmitted resulting in a lot of overall transmitted
overhead. Thus, it is important to manage the transmission
process. npacket req presents the number of packets to send
the total bytes npayl in consideration of the permitted nqueue .
It can be calculated in eq. 9. ntrans msg , as calculated in eq.
10, defines the total number of transmitted message to send
one complete data information.
Fig. 7 shows an example scenario for a queue of
nqueue =16-bytes needed to be transmitted. In this case, the
nover is assumed to be 8 bytes. Sending the 16 bytes in one
packet, needs 24 bytes in total. Sending it in 2 packets, requires
32 bytes in total. In the worst case a payload of 1-byte data is
transmitted with 16 packets and results in 144 bytes in total.
16 bytes
8 bytes
8 bytes
Fig. 7: Packet size comparison for sending a 16 bytes-sized
payload in one or two packets.
It is recommendable to wait as long as possible and
accumulate data to send bigger payloads and less packets. This
uses less transmitted bytes, less bandwidth and less energy on
transmitter and receiver nodes. It also decreases the time the
transmission channel is blocked.
TABLE IV: Summary of energy consumption for different
payloads
Number of
payload
Total number of
transmitted bytes
Number of
points
Energy per
payload byte
Sampling
period (ms)
8 22 3 21.7 μJ1.2
16 38 5 21.3 μJ1.2
32 70 8 18.5 μJ1.2
As seen from the table IV, the energy required for the
transmission of one byte is more when the total data bytes
transmitted are less. The energy required per byte transmis-
sion for 8 bytes is more compared to 32 bytes. Hence, as
conclusion, the more the data bytes transmitted, the less is the
energy consumption per byte. The current required to transmit
a data packet between two nodes decreases when the size of
payload increases as seen in table III.
The objectives behind such measurements and analysis is
to estimate the node lifetime. This will give an overview of
the battery energy. This is very helpful to manage the network
lifetime by predicting the best time to change the drained
batteries.
CONCLUSION
An accurate energy estimation at an early design stage is
a key ingredient for any successful design methodology at
the system level. This paper proposes a scheme to predict
the energy of a wireless device using a Precision Analyzer.
With the measurements of the device, we can determine how
much energy is consumed for each state and the required time
for each operational cycles. Some power saving modes that
seemed useful while reading the datasheet cannot be used to
save energy. A proper metric of consumed energy per byte
transmission pattern of packets is also provided.
Using these measurement results, we can improve the
precision of the energy component of the panStamp wireless
platform. Further works aim to use the obtained measurements
in order to save the energy of such sensor node.
ACKNOWLEDGMENT
This work is under a cooperation between the National
School of Engineering of Sfax, Tunisia and the Technical
University of Chemnitz, Germany. The authors would like to
convey thanks to the TU Chemnitz for hosting them. They
would like also to thank the DAAD foundation for supporting
their work.
This work is done within the “Landesinnovation-
sstipendium” funded by the S¨
achsische Aufbaubank (SAB) and
the European Social Fund (ESF).
REFERENCES
[1] Sabrine Khriji, Dhouha El Houssaini, Mohamed Wassim Jmal, Christian
Viehweger, Mohamed Abid, and Olfa Kanoun. Precision irrigation
based on wireless sensor network. IET Science, Measurement &
Technology, 8(3):98–106, 2014.
[2] Victor Shnayder, Mark Hempstead, Bor-rong Chen, Geoff Werner Allen,
and Matt Welsh. Simulating the power consumption of large-scale
sensor network applications. In Proceedings of the 2nd international
conference on Embedded networked sensor systems, pages 188–200.
ACM, 2004.
[3] Rym Ch´
eour, Mohamed Wassim Jmal, Olfa Kanoun, and Mohamed
Abid. Evaluation of simulators tools and power-aware scheduling model
for wireless sensor networks. IET Computers & Digital Techniques,
11(5):p. 173 – 182, 2017.
279
[4] Mianxiong Dong, Kaoru Ota, and Anfeng Liu. Rmer: Reliable and
energy-efficient data collection for large-scale wireless sensor networks.
IEEE Internet of Things Journal, 3(4):511–519, 2016.
[5] Gustavo Carneiro. Ns-3: Network simulator 3. In UTM Lab Meeting
April, volume 20, 2010.
[6] Xinjie Chang. Network simulations with opnet. In Simulation Con-
ference Proceedings, 1999 Winter, volume 1, pages 307–314. IEEE,
1999.
[7] Crossbows data sheet of iris mote. http://www.memsic.com/userfiles
/files/Datasheets/WSN/IRIS/Datasheet.pdf, December 2017.
[8] Jochen Schiller, Hartmut Ritter, Achim Liers, and Thiemo Voigt.
Scatterweb-low power nodes and energy aware routing. In Hawaii
International Conference on System Sciences, 2005.
[9] panstamp nrg2. http://panstamp.com/store/index.php, December 2017.
access.
[10] L Elmahni, L Bouhouch, and A Moudden. The development of an
advanced it system dedicated to a photovoltaic pumping station. Journal
of Theoretical and Applied Information Technology, 93(1):88, 2016.
[11] Marc Kramer and Alexander Geraldy. Energy measurements for micaz
node. University of Kaiserslautern, Kaiserslautern, Germany, Technical
Report KrGe06, 2006.
[12] Ram Prasadh Narayanan, Thazath Veedu Sarath, and Vellora Veetil Vi-
neeth. Survey on motes used in wireless sensor networks: Performance
& parametric analysis. Wireless Sensor Network, 8(04):51, 2016.
[13] Wilfried Dron, Simon Duquennoy, Thiemo Voigt, Khalil Hachicha, and
Patrick Garda. An emulation-based method for lifetime estimation
of wireless sensor networks. In IEEE International Conference on
Distributed Computing in Sensor Systems (DCOSS), pages 241–248.
IEEE, 2014.
[14] A Sivagami, K Pavai, D Sridharan, and SAV Satya Murty. Estimating
the energy consumption of wireless sensor node: Iris. International J.
of Recent Trends in Engineering and Technology, 3(4), 2010.
[15] Leonardo M Rodrigues, Carlos Montez, Gerson Budke, Francisco
Vasques, and Paulo Portugal. Estimating the lifetime of wireless sensor
network nodes through the use of embedded analytical battery models.
Journal of Sensor and Actuator Networks, 6(2):8, 2017.
[16] Ravishankar Rao, Sarma Vrudhula, and Daler N Rakhmatov. Battery
modeling for energy aware system design. Computer, 36(12):77–87,
2003.
[17] Kalyan Sasidhar, R Sreeresmi, and P Rekha. A wsn lifetime improve-
ment algorithm reaping benefits of data aggregation and state transitions.
In Global Humanitarian Technology Conference-South Asia Satellite
(GHTC-SAS), 2014 IEEE, pages 201–205. IEEE, 2014.
[18] Chinh D Nguyen, Tan D Tran, Nghia D Tran, Tue Huu Huynh,
and Duc T Nguyen. Flexible and efficient wireless sensor networks
for detecting rainfall-induced landslides. International Journal of
Distributed Sensor Networks, 2015, 2015.
[19] Marcelo M Carvalho, Cintia B Margi, Katia Obraczka, and JJ Garcia-
Luna-Aceves. Modeling energy consumption in single-hop ieee 802.11
ad hoc networks. In Computer Communications and Networks, 2004.
ICCCN 2004. Proceedings. 13th International Conference on, pages
367–372. IEEE, 2004.
[20] C Behrens, O Bischoff, M Lueders, and R Laur. Energy-efficient
topology control for wireless sensor networks using online battery
monitoring. Advances in Radio Science, 5(C. 1):205–208, 2007.
[21] Konstantin Mikhaylov and Jouni Tervonen. Novel energy consumption
model for simulating wireless sensor networks. In 4th International
Congress on Ultra Modern Telecommunications and Control Systems
and Workshops (ICUMT), pages 15–21. IEEE, 2012.
[22] Sabrine Khriji, Dhouha El Houssaini, Olfa Kanoun, and In´
es Kammoun
Jemal. Energy efficient routing based on cooperative communication
for wireless sensor network. In Sensorica 2017, At Mhlheim an der
Ruhr, page 36. IEEE WORKSHOP, 2017.
[23] Keysight technologies. http://literature.cdn.keysight.com/litweb/pdf/5992-
0731EN.pdf?id=2612931, December 2017. access.
[24] Cc1101 low-power sub-1 ghz rf transceiver.
http://www.ti.com/lit/ds/symlink/cc1101.pdf, December 2017. access.
280
