Available via license: CC BY 4.0
Content may be subject to copyright.
Research Article
International Journal of Distributed
Sensor Networks
2017, Vol. 13(4)
ÓThe Author(s) 2017
DOI: 10.1177/1550147717705785
journals.sagepub.com/home/ijdsn
Energy-aware data compression
and transmission range control for
energy-harvesting wireless sensor
networks
Jun Min Yi
1
, Eom Ji Oh
1
, Dong Kun Noh
1
and Ikjune Yoon
2
Abstract
Energy-harvesting nodes are now being employed in wireless sensor networks to extend the lifetime of the network by
harvesting energy from the surrounding environments. However, unpremeditated energy consumption can incur energy
problems, such as the blackout of nodes (due to their exceeding energy consumption over the amount of harvested
energy) or inevitable disposal of harvested energy (in excess of the battery capacity). In this article, we propose an adap-
tive data compression and transmission range extension scheme that minimizes the blackout of sensor nodes and
increases the amount of data collected at the sink node using the harvested energy efficiently. In this scheme, each node
estimates the amount of harvested and consumed energy. When it determines that its remaining energy will exceed its
storage capacity, it exploits the energy to compress the data or increase the transmission range. At this point, of the
two methods, the method that can more effectively increase the network performance can be selected. The results of
experiments conducted indicate that the proposed scheme significantly reduces the extent of node blackouts and
increases the data collection rate of the sink node.
Keywords
Wireless sensor networks, energy-harvesting, data compression, transmission range, blackout time
Date received: 12 December 2016; accepted: 29 March 2017
Academic Editor: Jian Peng
Introduction
Wireless sensor networks (WSNs) are used to obtain vari-
ous types of data for monitoring environments and condi-
tions in (1) hard-to-reach areas such as disaster areas,
military zones, and underwater areas, (2) wide areas such
as forests, farmland, and seas, and (3) structures such as
buildings and bridges. Sensor nodes that are typically
applied in these WSNs use limited-capacity batteries as
energy sources. Consequently, these sensor nodes have a
finite lifetime and encounter maintenance-related issues
such as the battery needing to be replaced or blackout
nodes needing to be recharged. To overcome such issues,
studies on minimizing the energy consumption of sensor
nodes in battery-based WSNs have been carried out.
1,2
Recently, energy-harvesting nodes that recharge
their batteries by harvesting energy from the surround-
ing environment have been developed to solve the
problem of limited lifetime of nodes.
3
Because these
energy-harvesting nodes can theoretically operate inde-
finitely, energy-related research direction in WSNs has
1
Department of Software Convergence, Soongsil University, Seoul, South
Kore a
2
Department of Smart Systems Software, Soongsil University, Seoul,
South Korea
Corresponding author:
Ikjune Yoon, Department of Smart Systems Software, Soongsil University,
Seoul 06978, South Korea.
Email: ikjune.yoon@gmail.com
Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (http://www.uk.sagepub.com/aboutus/
openaccess.htm).
been changed. Whereas traditional battery-based
WSNs focus on minimizing the energy consumption of
nodes to maintain the network for a long time, energy-
harvesting WSNs focus on the efficient use of harvested
energy to meet the quality-of-service (QoS) require-
ments of each WSN applications.
4–7
For example, if a
WSN application requires real-time data, the harvested
energy should be mainly used to reduce data latency.
8
If such an application needs to manage important data,
the harvested energy should be used to increase data
reliability.
9
General WSNs have the hot-spot problem,
10,11
which is that the nearer nodes to the sink node con-
sume more energy. Several schemes have been studied
to overcome the problem such as data compression
12,13
or detouring around the hot spot.
14,15
Such solutions
consume additional energy to compress data or extend
transmission range. In energy-harvesting WSNs, how-
ever, they can be applied effectively using the surplus
harvested energy.
In this study, the surplus harvested energy is used to
maximize the amount of data obtained by the sink node
by reducing the energy consumption of the nodes near
the sink node. To this end, the following two methods
are utilized.
1. Data compression method. In this method, the
surplus harvested energy is used to compress
data. Typically, sensor nodes consume the great-
est amount of energy for data transmission.
Therefore, data compression which reduces the
volume of data can decrease the amount of
energy consumption significantly. It also leads
to a decrease in the energy consumption of the
overall network because the data compression
of a node affects the reduction in transmission
energy of all relay nodes to the sink due to the
multi-hop way of data transmission in WSNs.
However, it should be noted that compression
itself requires some amount of energy.
2. Transmission range control method. In this
method, the surplus harvested energy is used to
increase the transmission range and induce a
decrease in the number of data transmission
hops. As a result, it can reduce the total amount
of energy consumed by intermediate nodes.
This method is particularly useful in preventing
load concentration on the nodes near the sink
node. Generally, the nodes near the sink node
have to transmit a greater amount of data than
other nodes, thus increasing the probability of
blackout. As a state of blackout leads to the
consecutive blackout of neighbor nodes, as well
as a rapid decrease in the data collection rate,
load concentration on these nodes should be
prevented.
16,17
By increasing the transmission
range, in this method, data tend to be directly
transmitted to the sink node instead of passing
through intermediate nodes. However, it also
should be considered that it consumes some
extent of energy to expand the transmission
range.
Note that the use of harvested energy in each node
should be prioritized for their basic operations (such as
data sensing, processing, and transmission), and only
surplus energy should be used to fulfill the QoS require-
ments of WSN applications. Therefore, the proposed
scheme first calculates the residual energy by estimating
the amount of energy harvested and consumed in a
node during a certain period. Then, it uses the expected
surplus energy to maximize the network performance,
for example, amount of data obtained at the sink node.
To this end, the data compression method or transmis-
sion range extension method are utilized. To facilitate
more efficient energy usage, the method that is more
useful in the network performance can be selected, or
both methods can be applied when residual energy is
quite a lot.
The remainder of this article is organized as follows.
Section ‘‘Related work’’ discusses the existing studies
conducted with the objective of solving the energy-
related problems of WSNs. Section ‘‘Energy-adaptive
data compression and transmission range adjustment
scheme’’ introduces the energy-adaptive data compres-
sion and transmission range extension scheme proposed
in this study. Section ‘‘Performance evaluation’’ com-
pares the performance of the proposed scheme with
that of other existing schemes. Section ‘‘Conclusion’’
presents the conclusions of this study.
Related work
In a WSN, several small sensor nodes are placed over a
wide area. As the sensor nodes are cheap and small,
they have many limitations on hardware such as the
processor, memory, and battery. Consequently, studies
geared toward developing methods of effectively using
the limited energy (such as energy-harvesting, data
compression, and transmission range adjustment) are
being actively carried out.
3
Data compression in a WSN
Sensor nodes typically consume the greatest amount of
energy for data transmission. The amount of energy
consumed for data transmission varies according to the
volume of data and the transmission range. It can be
reduced by decreasing the volume of data. Particular,
in a WSN, data are transmitted in a manner of multi-
hop, the reduction in data volume decreases the
2International Journal of Distributed Sensor Networks
workload of the relay nodes and increases the lifetime
of the overall network.
18
Sadler and Martonosi
19
proposed the Sensor
Lampel–Ziv–Welch (S-LZW) algorithm, derived by
lightening the LZW algorithm, a lossless compression
algorithm designed by Welch,
20
for sensor nodes. They
also proposed another compression algorithm called
S-LZW with Burrows–Wheeler Transform (S-LZW-
BWT) that conducts invertible BWT
21
before compres-
sion by S-LZW. Yoon et al.
22
proposed an energy-
aware data compression scheme that adjusts the data
collection frequency and selects either the S-LZW or
S-LZW-BWT compression algorithm according to the
energy level of the energy-harvesting node in order to
increase the accuracy of the sensed data.
Petrovic et al.
23
proposed a data-funneling scheme
that compresses the setup and control packets used for
routing to reduce the transmission energy consumption.
In this scheme, data are compressed using the coding
by ordering scheme, which is a lossless compression
method. Arici et al.
24
proposed PINCO, an in-network
compression scheme that reduces redundancy in the
data collected from sensors, thereby decreasing the
wireless communication among the sensor nodes and
saving energy. In this scheme, compressed data can be
recompressed without decompressing, thereby eliminat-
ing data decompression and recompression overhead.
The scheme proposed in this article reduces the
workload of the relay nodes by transferring the data
compressed by S-LZW.
19
Figure 1 shows the flow chart
of S-LZW compression scheme. However, in contrast
to the methods applied in previous studies, data com-
pression is conducted only using surplus energy har-
vested from their environments. Thus, nodes are not
adversely affected.
Transmission range adjustment in a WSN
Transmission range is a dominant property in determin-
ing the amount of energy consumed for data transmis-
sion in a WSN. When the transmission range of sensor
nodes is adjusted, the energy consumption for transmis-
sion and the number of hops vary according to the dis-
tance, thus affecting the network lifetime.
Noh
25
proposed a transmission range determination
scheme for the solar-powered WSN to increase the
routing efficiency. Jeong et al.
26
proposed a scheme that
establishes an appropriate wireless transmission range
for communication environments using the received sig-
nal strength indication (RSSI) values selected through a
Figure 1. Flow chart of S-LZW compression scheme.
20
Yi et al. 3
reliability determination process without additional
hardware.
Hou and Li
27
confirmed that network performance
is enhanced when transmission range is appropriately
adjusted, by comparing the most forward progress
(MFP) and the nearest forward progress (NFP)
schemes in which the transmission range is fixed, with
the MFP with variable radius (MVR) scheme in which
the transmission range can be adjusted. Lin et al.
28
pro-
posed adaptive transmission power control (ATPC),
which controls transmission power according to the
changes in external environments, to solve the existing
problem of inefficient wireless communication using
fixed transmission range.
Our proposed scheme extends the transmission
range of operating nodes when surplus energy is gener-
ated to decrease the number of relay hops. As a result,
it reduces the amount of energy consumed in relay
nodes. Although nodes can collide with each other dur-
ing transmission due to the extended transmission
range, this collision problem can be solved by applying
time division multiple access (TDMA), which facilitates
mutual communication without signal overlapping by
dividing the same frequency band according to time.
29
For this reason, collision in the link stage is not
reflected in the proposed scheme.
Energy-harvesting WSN
Energy-harvesting nodes solve the problem of limited
energy by harvesting various types of eco-friendly
energy near them, storing it in a rechargeable battery,
and supplying the energy to a system. The amount of
available energy varies according to the characteristics
of the energy source and the energy storage employed.
Accordingly, schemes that maximize utilization of the
harvested energy with an efficient energy-schedule have
been proposed and investigated in energy-harvesting
WSNs.
Sudevalayam and Kulkarni
3
and Encarnacion and
Yang
30
studied the characteristics of several eco-
friendly energy sources, methods of harvesting them,
and their energy conversion efficiency. Among the
energy sources, solar energy was found to have a particu-
larly higher power density (approximately 15 mW/cm
2
)
than other energy sources,
31
thus making it the most
desirable energy source. Generally, the solar-powered
nodes can satisfy the power requirements of typical sen-
sor nodes (1–10 mW) except some applications requiring
extremely high performance.
Kansal et al.
32
proposed an energy model for energy-
harvesting nodes and a model that determines the
energy boundary, which facilitates energy-harvesting
nodes with infinite lifetimes. Shaikh and Zeadally
33
investigated a system that can effectively use the har-
vested energy based on the energy estimation models.
In our proposed scheme, the system design assumes
that the solar-powered node satisfies the energy require-
ments of typical sensor nodes, which was verified in the
studies described above.
Energy-adaptive data compression and
transmission range adjustment scheme
In our proposed scheme, a node estimates the amount
of energy harvested and consumed, and these estimated
values are used to expect its residual energy, which will
be utilized to increase the amount of data obtained by
the sink node when it exists. To this end, a scheme that
applies data compression, transmission range extension,
or both methods is proposed. The proposed scheme is
not applicable for real-time applications, but for delay
tolerant networks (DTNs),
34
in which each node peri-
odically senses data, conducts data processing such as
compression, and transfers the data. And data compres-
sion is performed using the S-LZW
19
compression
scheme.
Each node operates in basic mode, selective mode,
or hybrid mode according to the energy estimated for
the subsequent period. In basic mode, nodes perform
only basic operations (sensing, processing, and trans-
mission). In selective mode, either data compression or
transmission range extension is selectively performed.
Additional energy is consumed to compress the data
when the data compression method is applied.
However, the energy consumed by relay nodes for data
transmission can be reduced by decreasing the volume
of data transmitted. As the length of the data transmis-
sion path increases, the amount of energy consumed
decreases due to the higher number of nodes through
which the data have to pass. In the transmission range
adjustment method, more energy is consumed to
extend the transmission range. However, as the number
of hops in the transmission path decreases, load con-
centration of nodes near the sink node is effectively
prevented, and the problem of energy imbalance
between nodes is solved. In other words, when the
transmission range is extended, data tend to be directly
transferred to the sink node instead of being trans-
mitted through the nodes near the sink node. When
surplus energy is sufficient, the hybrid mode, in which
both methods are utilized simultaneously, can be
applied to maximize the effects. Figure 2 gives an over-
view of the proposed scheme.
Energy model for energy-harvesting nodes
Energy-harvesting nodes estimate the residual energy of
their rechargeable batteries by simultaneously considering
the amount of energy harvested and consumed, in contrast
to the energy models for battery-based nodes. In this
4International Journal of Distributed Sensor Networks
article, the energy model proposed by Yang et al.
35
is used
to estimate the amount of residual energy.
When the current time is t, the estimated residual
energy E0
residualiðt;ptxÞof node iafter p
tx
is as shown in
equation (1)
E0residualiðt;ptxÞ=EresidualiðtÞEconsumeiðt;ptxÞ
+Eharvestiðt;ptxÞð1Þ
where EresidualiðtÞis the amount of residual energy of
node iat time t, and Econsumeiðt;ptxÞand Eharvestiðt;ptxÞ
are the amounts of energy consumed and harvested by
node ifrom time tduring period p
tx
, respectively.
EresidualiðtÞis a measurable value in the node, and
Eharvestiðt;ptxÞcan be estimated using existing
schemes
36,38
to estimate the amount of harvested
energy efficiently. We adopted weather-conditioned
moving average (WCMA)
38
scheme to estimate har-
vested energy. WCMA is able to effectively take into
account both the current and past days’ weather condi-
tions, obtaining a relative mean error of only 10%.
When coupled with energy management algorithm, it
can achieve gains of more than 90% in energy utiliza-
tion with respect to exponentially weighted moving
average (EWMA).
37
The amount of energy consumed
Econsumeiðt;ptxÞcan be estimated through equation (2)
Econsumei=Etransmissioni+Esystemið2Þ
where Etransmissioniis the amount of energy consumed for
data transmission, and Esystemiis all the energy con-
sumed with the exception of that consumed for data
transmission. Esystemiincludes the energy consumed in a
state of reception or waiting; thus, it applies to all nodes
similarly. The value of Etransmissionican significantly vary
according to the amount of data transferred, as shown
in equation (3)
39
Etransmissioni=Sibda
ið3Þ
where S
i
is the amount of data transferred by node i,b
is the amount of energy consumed to transfer a 1-m bit,
ais path loss, and d
i
is the transmission range of node
i.S
i
can be derived by adding the volume Sdataiof data
that node isensed and Srelayiof data transferred from
other nodes. In this regard, the amount of energy con-
sumed by node ican be represented, as shown in equa-
tion (4)
Econsumei=ðSdatai+SrelayiÞbda
i+Esystemið4Þ
Srelayiis estimated based on the weighted moving aver-
age of the historical data relayed by node i. The esti-
mated residual energy E0
residual of a node at a certain
point in time can be calculated using equations (1)–(4).
Types of operating modes
Nodes determine their operating modes at the end of
each round according to E0
residuali. In each mode, nodes
perform the operations shown in Figure 3. Each opera-
tion can be described as follows:
1. Basic mode. This operating mode is selected
when the energy of a node is insufficient. In this
mode, uncompressed data are transferred in the
normal transmission range to reduce the energy
consumption of the node.
2. Selective mode. This mode is selected when sur-
plus energy is estimated to be harvested while
the node is operating in basic mode during the
Figure 2. Overview of the proposed scheme.
Yi et al. 5
present round. In this mode, the surplus energy
is used to perform data compression or trans-
mission range extension.
3. Hybrid mode. This mode is selected when sur-
plus energy is estimated to be harvested while
the node is operating in selective mode during
the present round. In this mode, the node
applies the data compression and transmission
range extension methods at the same time.
Change of operating mode
Each node estimates the amount of E0
residualiat every
data transmission period p
tx
and selects an appropriate
operating mode for the next round.
Node currently operating in basic mode. First, it is assumed
that the node operated in basic mode during the previ-
ous round and that it is to operate in basic mode during
the subsequent round. If E0
residualiis estimated to exceed
the battery capacity Efulliduring the next round, the
storage capacity is deemed insufficient to store the har-
vested energy. As a result, some energy is inevitably dis-
carded without being used
E0
residualiðt;ptxÞ.Efullið5Þ
Thus, when equation (5) is satisfied, a transition is
made from basic mode to selective mode to prevent the
energy harvested being discarded due to the insufficient
energy storage capacity.
Node currently operating in selective mode. Meanwhile, it is
assumed that the node operated in selective mode dur-
ing the previous round and that it is to operate in the
same mode during the subsequent round. If E0
residualiat
the end of the round is estimated to be lower than the
minimum amount of energy Ebottomifor node operation,
the node can be in a state of blackout during period p
tx
.
Thus, the node transitions to basic mode to save energy
E0
residualiðt;ptxÞ\Ebottomið6Þ
That is, when equation (6) is satisfied, it means
E0
residualiis estimated to be lower than the minimum
amount of energy Ebottomiassuming the node keeps its
current (selective) mode during the subsequent round.
Therefore, the node transitions from the current selec-
tive mode to basic mode. Conversely, if E0
residualiis esti-
mated to exceed Efulliwhen the node is assumed to
operate in current (selective) mode during the subse-
quent round, the node selects hybrid mode since this
node can be assumed to have enough energy.
Otherwise, it stays in its current mode.
Node currently operating in hybrid mode. It is assumed that
the node operated in hybrid mode during the previous
round and that it is to operate in the same mode during
the subsequent round. If E0
residualiat the end of the
round is estimated to be lower than the minimum
amount Ebottomiof energy for node operation, the node
can be in a state of blackout during period p
tx
. To pre-
vent this, the node transitions to selective mode.
Algorithm 1 and Figure 4 show the pseudo-codes for
node operations adjusted according to the modes, and
finite state machine (FSM), respectively.
Method selection in selective mode
A node operating in selective mode, as shown in
Figure 4, performs either data compression or trans-
mission range extension. It selects the more efficient
method by calculating the efficiency of each method. In
the proposed scheme, the estimated value of energy
consumed by all relay nodes from a target node to a
sink node is used as an index for efficiency. In other
words, between two methods, the method that derives
the lower estimated value is selected and used.
In this section, a model that estimates the amount of
energy consumed by relay nodes when using each
method respectively is proposed. If data are transferred
to the sink node through the Hhop when a node i
transfers data, the sum Ehopiof energy consumed by
relay nodes can be calculated by applying equation (7)
Ehopi=ðSdatai+SrelayiÞbda
iHð7Þ
Figure 3. Node operations according to modes in the proposed scheme.
6International Journal of Distributed Sensor Networks
If the data compression method is used, data are
compressed with a compression ratio of R
compression
,
and Sdataiis reduced to Sdataið1RcompressionÞ, and
EcompressionðSdataiÞis the amount of consumed energy to
compress Sdataibits of data. Accordingly, the energy
consumption Ecompression
hopiof relay nodes using this
method can be represented, as shown in equation (8)
Ecompression
hopi=ðSdataið1RcompressionÞ+SrelayiÞ
bda
iH+EcompressionðSdataiÞð8Þ
However, if the number of hops is reduced to
H
reduced
due to the extended transmission range dexi
resulting from application of the transmission range
adjustment method, the energy consumption Erange
hopiof
the relay nodes in this method is derived through equa-
tion (9)
Erange
hopi=ðSdatai+SrelayiÞbda
exiðHHreducedÞð9Þ
By comparing the energy consumption estimated for
relay nodes using the results of equations (8) and (9), a
more efficient method for reducing the energy con-
sumption of relay nodes can be found
Ecompression
hopi
.Erange
hopið10Þ
If equation (10) is satisfied, the transmission range
adjustment method is more efficient, and thus, the node
selects this method. If not, the node selects the data
compression method.
In addition, if a certain node is currently isolated
from other nodes and can have a neighboring node in
the transmission range extended, the node preferen-
tially uses the transmission range extension method for
a connectivity when surplus energy exists in the pro-
posed scheme.
Performance evaluation
The SolarCastalia
40
simulator was used to analyze the
performance of the scheme proposed in this study. The
results obtained were compared with those obtained by
other four schemes: Normal, Comp, TR, Adaptive
Compression, and Transmission (ACT) schemes.
41
Normal scheme is a naive WSN scheme, which trans-
mits uncompressed packets using normal transmission
range. It is the same as basic mode in the proposed
scheme. Comp scheme uses S-LZW
19
data compression
to reduce the size of all sensed data. Even though it
consumes additional energy for compression, it can
save transmission energy because it transmits smaller
data. TR scheme transmits data with a comparatively
extended transmission range than Normal scheme. It
can shorten the length of routing paths by consuming
more transmission energy. ACT
41
scheme performs
data compression or transmission range extension to
obtain more data when surplus energy is expected. In
this scheme, nodes estimate its remaining energy, and
choose more efficient method between data compres-
sion and transmission range extension by the estima-
tion result. Specifically, the amount of data obtained
by the sink node (which is the most critical perfor-
mance metric) and other performance metrics closely
related to the metric dictated above, including the num-
ber of blackout nodes and the degree of residual energy
balancing of the nodes, were compared according to
the schemes.
Simulations were conducted by randomly placing
100 to 500 nodes in a 120 m 3120 m field. The trans-
mission range was established as 8–13 m for basic data
Algorithm 1 Algorithm of the proposed method.
1: function ChangeMode()
2: if E0
residualiðt;ptxÞEfullithen
3: UpperMode()
4: else if E0
residualiðt;ptxÞEbottomithen
5: LowerMode()
6: else
7: Do not change current mode
8: end if
9: end function
10: function UpperMode()
11: if current_mode g = Basic_mode then
12: current_mode Selective_mode
13: if Ecompression
hopi
\Erange
hopithen
14: transmission_scheme compression
15: if neighbor_node = [then
16: transmission_scheme distance
17: end if
18: else if Ecompression
hopi
.Erange
hopithen
19: transmission_scheme distance
20: end if
21: else if current_mode = Selective_modethen
22: current_mode Hybrid_mode
23: transmission_scheme compression&distance
24: end if
25: end function
26: function LowerMode()
27: if current_mode = Selective_mode then
28: current_mode Basic_mode
29: transmission_scheme normal
30: else if current_mode = Hybrid_mode then
31: current_mode Selective_mode
32: if Ecompression
hopi
\Erange
hopithen
33: transmission_scheme compression
34: if neighbor_node = Ø then
35: transmission_scheme distance
36: end if
37: else if Ecompression
hopi
.Erange
hopithen
38: transmission_scheme distance
39: end if
40: end if
41: end function
Yi et al. 7
transmission and as 14–22 m for extended transmis-
sion. Data compression was performed by applying the
S-LZW scheme.
19
Each experiment lasted 10 days
(1000 rounds), and the mean value derived through 10
or more repeated experiments was used for compari-
son. The main parameters used in the experiments are
given in Table 1.
Figure 5 shows the network topology of 300 nodes
and examples of operating modes of each node when
the proposed scheme is applied.
Number of blackout nodes
The throughput of the network can be estimated based
on the blackout state of a node. When many nodes are
in a state of blackout, they can neither sense data nor
relay the data receiving from their neighboring nodes.
For this reason, the number of blackout nodes should
be minimized.
The number of blackout nodes generated per round
was measured to identify the number of blackout nodes
in the proposed scheme, when the number of nodes is
300. The number of blackout nodes generated was com-
pared by increasing the number of nodes by 100 to ver-
ify scalability.
Figure 6 shows the average number of blackout
nodes generated per round. The proposed scheme gen-
erated the lowest number of blackout nodes compared
to the other schemes. Specifically, it shows 62.2% lower
Figure 4. Plan of finite state machine (FSM) regarding the mode variation in the proposed scheme.
Table 1. Simulation environment.
Parameters Values
Field size 120 m 3120 m
Node 200/300/400/500
Deploy Random
Radio range 8–22 m
TX power 27 dBm at 3 V
0 dBm at 3 V
RX power 0 dBm at 3 V
Baud rate 250 kbps
Super capacitor 20 F
TX period 20 min
Energy-harvesting period 1–3 mW
Figure 5. Plan of node placement (300 nodes).
8International Journal of Distributed Sensor Networks
number of blackout nodes than the TR scheme, 31.4%
than the Normal scheme, 17.7% than the Comp
scheme, and 7.9% than the ACT scheme.
This result is derived because the selective or simul-
taneous application of the data compression and trans-
mission range adjustment methods reduced the volume
of data transferred to the sink node and the length of
data transmission path, thus decreasing the workload
of the relay nodes the most. The number of blackout
nodes in the TR scheme was higher by 81.6% than that
in the Normal scheme because the greatest amount of
energy was consumed for data transmission and recep-
tion owing to the excessive transmission range in the
former scheme.
The number of blackout nodes in the Comp scheme
is lower by 16.7% than that in the Normal scheme and
by 54.1% than that in the TR scheme. This result is
obtained because the transmission method based on
data compression reduces the volume of data to be
transferred in the Comp scheme compared to that in
the Normal and TR schemes, thus leading to a decrease
in the amount of energy consumed for data
transmission.
The number of blackout nodes in the ACT scheme
is relatively lower than other schemes. The reason is
that the ACT scheme can apply one between compres-
sion and transmission range adjustment method adap-
tively. However, the number of blackout nodes in the
proposed scheme is slightly lower than the ACT scheme
because the proposed scheme manages the harvested
energy more sophisticatedly.
Figure 7 shows the cumulative number of blackout
nodes generated when 200, 300, 400, and 500 nodes are
used in a field of the same size, respectively. When 200
nodes are used, the number of blackout nodes in the
proposed scheme is lower by 78.7% than that in the
TR scheme. When 300, 400, and 500 nodes are used,
the cumulative number of blackout nodes in the pro-
posed scheme is lower by 69.2 to 77.1% than that in
the TR scheme.
For 200, 300, 400, and 500 nodes used, the proposed
scheme generates a lower number of blackout nodes
ranging from 18% to 34% than the Normal scheme,
8% to 26% than the Comp scheme, and 2% to 23%
than the ACT scheme. This result is obtained because
the proposed scheme selects the more efficient of the
two methods or applies both when it determines a
method for specific operation.
Amount of data obtained by a sink node
The amount of data obtained per round and the overall
amount of data obtained according to the change in
the number of nodes were compared to examine the
amount of data obtained by the sink node.
Figure 8 shows the amount of data obtained from
the sink node per round when 300 nodes were used in
the experiment. The proposed scheme obtained a
greater amount of data by 34.1% to 127% per round
on average. This result was obtained because this
scheme minimizes the blackout of nodes and increases
the rate of successful data transmission, as shown in
Figure 6.
Figure 9 compares the amount of data obtained
according to the schemes and changes in the density of
nodes. For 500 nodes, the proposed scheme obtained a
greater amount of data by 32.6% than the Normal
scheme, by 18.9% than the Comp scheme, by 61% than
the TR scheme, and by 6.8% than the ACT scheme.
For 200, 300, and 400 nodes, the proposed scheme
obtains greater amounts of data by 19.4%, 46.5%, and
27.7% than the Normal scheme, by 7.8%, 28.3%,
and 17.8% than the Comp scheme, by 30.5%, 73%,
and 71.9% than the TR scheme, and by 6.2%, 8.9%,
Figure 6. Comparison of number of blackout nodes according
to the transmission schemes (300 nodes).
Figure 7. Comparison of blackout nodes according to the
number of nodes and transmission schemes.
Yi et al. 9
and 7.6% than the ACT scheme. The proposed scheme
exhibits the highest data collection ratio regardless of
the number of nodes because it ensures scalability
through local decision.
Amount of residual energy
The state of residual energy of the nodes close to the
sink node and outer nodes far from the sink node was
also examined to verify the effects of the proposed
scheme on the residual energy balancing of nodes.
Figures 10 and 11 show the state of the residual
energy according to the schemes and rounds based on
the nodes located one to three hops from the sink node
and those located one to three hops from the outside
end of the sink node for 300 nodes used, respectively.
In Figure 10, for the outer nodes located far from the
sink node, in all the schemes, they operated in a state
of virtually full recharge; however, the energy level was
lower for the proposed scheme than for other schemes.
In Figure 11, for the nodes located close to the sink
node, the proposed scheme and ACT scheme main-
tained the highest residual energy, followed by the
Comp scheme, Normal scheme, and TR scheme. This
result was obtained because the two schemes reduce the
workload of relay nodes by utilizing the surplus energy
of outer nodes.
Figure 12 compares the state of average residual
energy of the nodes close to the sink node to that of
outer nodes located far from the sink node according to
the number (200, 300, 400, and 500) of nodes. As the
maximum number of hops varies according to the num-
ber of nodes, the standard of close and outer nodes of
sink nodes is determined by the front and back of 30%
based on the overall number of hops.
As the number of nodes increases, the transmission
data amount of the relay nodes also increases. As a
Figure 9. Comparison of amount of data obtained according
to the transmission schemes and number of nodes.
Figure 8. Comparison of amount of data obtained according
to the transmission schemes based on the round (20 min) (300
nodes).
Figure 10. Residual energy comparison of outer nodes (300
nodes).
Figure 11. Residual energy comparison of inner nodes (300
nodes).
10 International Journal of Distributed Sensor Networks
result, the nodes close to the sink node generally show
a lower amount of energy. Furthermore, as the number
of nodes increases, the difference in energy among the
nodes close to the sink nodes decreases in accordance
with the schemes. This result is obtained because as the
number of nodes increases, the number of blackout
nodes also increases and the amount of data transferred
to relay nodes decreases in other schemes. In contrast,
as the number of blackout nodes generated is signifi-
cantly low in the proposed scheme, relay nodes can still
transfer a significant amount of data.
These experimental results verify that the proposed
scheme effectively controls the blackout of relay nodes
using the surplus energy of nodes. Specifically, the pro-
posed scheme utilizes the energy of outer nodes (which
consume relatively less energy) to reduce the load of
nodes close to the sink nodes (which require high-
energy consumption), thereby decreasing the number
of blackout nodes and facilitating residual energy bal-
ancing. Consequently, this scheme leads to a significant
increase in the data collection ratio of networks.
Conclusion
An energy consumption policy which is carefully
designed by considering both the QoS requirements of
application and the characteristics of the energy source
is required in energy-harvesting WSNs. This article
proposed an efficient energy utilization scheme that
maximizes the amount of data obtained by the sink
node in energy-harvesting WSNs. When surplus energy
is available, it is utilized for data compression, trans-
mission range extension, or a hybrid method that com-
bines both methods. Experimental results show that
this process reduces the energy consumption of each
node and facilitates residual energy balancing, thereby
increasing the amount of data obtained in the network.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial sup-
port for the research, authorship, and/or publication of this
article: This research was supported by the Ministry of
Science, ICT and Future Planning (MSIP), Korea, under the
Information Technology Research Center (ITRC) support
program (IITP-2017-2012-0-00646) supervised by the Institute
for Information & communications Technology Promotion
(IITP).
References
1. Yick Y, Mukherjee B and Ghosal D. Wireless sensor
network survey. Comput Netw 2008; 52(12):
2292–2330.
2. Akyilidiz IF, Su W, Sankarasubramaniam Y, et al. Wire-
less sensor networks: a survey. Comput Netw 2002; 38(4):
393–422.
3. Sudevalayam S and Kulkarni P. Energy harvesting sen-
sor nodes: survey and implications. IEEE Commun Surv
Tut 2011; 13(3): 443–461.
4. Yoo H, Shim M and Kim D. Dynamic duty-cycle
scheduling scheme for energy-harvesting wireless sen-
sor networks. IEEE Commun Lett 2012; 16(2):
202–204.
5. Vullers RJM, Schaijk RV, Visser HJ, et al. Energy har-
vesting for autonomous wireless sensor networks. IEEE
Solid-State Circ Mag 2010; 2(2): 29–38.
6. Basagni S, Naderi MY, Petrioli C, et al. Wireless sensor
networks with energy harvesting. In: Basagni S, Conti
M, Giordano S, et al. (eds) Mobile ad hoc networking: the
cutting edge directions. Hoboken, NJ: John Wiley &
Sons, 2013, pp.701–736.
7. Stojmenovic I. Handbook of sensor networks: algorithms
and architectures. Hoboken, NJ: John Wiley & Sons,
2005.
8. Stonebraker M, Cetintemel U and Zdonik S. The 8
requirements of real-time stream procession. ACM Spec
Int Group Manage Data 2005; 34(4): 42–47.
9. Bertozzi D, Benini L and De Micheli G. Error control
schemes for on-chip communication links: the energy-
reliability tradeoff. IEEE T Comput Aid D 2005; 24(6):
818–831.
10. Rooshenas A, Rabiee HR, Movaghar A, et al. Reducing
the data transmission in wireless sensor networks using
the principal component analysis. In: Proceedings of the
6th international conference on intelligent sensors, sensor
networks and information processing(ISSNIP), Brisbane,
QLD, Australia, 7–10 December 2010, pp.133–138. New
York: IEEE.
11. Olivares T, Tirado P, Orozco-Barbosa L, et al. Simula-
tion of power-aware wireless sensor network architec-
tures. In: Proceedings of the 1st ACM international
workshop on performance monitoring, measurement, and
Figure 12. Energy comparison according to the schemes based
on the number and location of nodes.
Yi et al. 11
evaluation of heterogeneous wireless and wired networks
(PM2HW2N), Torremolinos, 2–6 October 2006,
pp.32–39. New York: ACM.
12. Kimura N and Latifi S. A survey on data compression in
wireless sensor networks. Inform Technol Coding Comput
2005; 2:8–13.
13. Srisooksai T, Keamarungsi K, Lamsrichan P, et al. Prac-
tical data compression in wireless sensor networks: a sur-
vey. J Netw Comput Appl 2012; 35(1): 37–59.
14. Huang CF and Tseng YC. The coverage problem in a
wireless sensor network. Mobile Netw Appl 2005; 10(4):
519–528.
15. Hu X, Li Y and Xu H. Multi-mode clustering model for
hierarchical wireless sensor networks. Physica A 2017;
469: 665–675.
16. Kubisch M, Karl H, Wolisz A, et al. Distributed algo-
rithms for transmission power control in wireless sensor
networks. In: Proceedings of the IEEE wireless communi-
cations and networking (WCNC), New Orleans, LA, 16–
20 March 2003, pp.558–563. New York: IEEE.
17. Demirkol I, Ersoy C and Alagoz F. MAC protocols for
wireless sensor networks: a survey. IEEE Commun Mag
2006; 44(4): 115–121.
18. Zordan D, Melodia T and Rossi M. On the design of
temporal compression strategies for energy harvesting
sensor networks. IEEE T Wirel Commun 2016; 15(2):
1336–1352.
19. Sadler CM and Martonosi M. Data compression algo-
rithms for energy-constrained devices in delay tolerant
networks. In: Proceedings of the 4th international confer-
ence on embedded networked sensor systems (SenSys),
Boulder, CO, 31 October–3 November 2006, pp.265–278.
New York: ACM.
20. Welch TA. A technique for high-performance data com-
pression. IEEE Comput 1984; 17(6): 8–19.
21. Burrows M and Wheeler DJ. A block-sorting lossless data
compression algorithm. Report for systems research center
no. 124, 10 May 1994. Palo Alto, CA: SRC.
22. Yoon I, Yi J, Jeong S, et al. Dynamic sensing-rate control
scheme using a selective data-compression for energy-
harvesting wireless sensor networks. IEMEK J Embed
Syst Appl 2016; 11(2): 79–86.
23. Petrovic D, Shah RC, Ramchandran K, et al. Data fun-
neling: routing with aggregation and compression for
wireless sensor networks. In: Proceedings of 1st IEEE
conference on sensor network protocols and applications
(SNPA), Anchorage, AK, 11 May 2003, pp.156–162.
New York: IEEE.
24. Arici T, Gedik B, Altunbasak Y, et al. PINCO: a pipe-
lined in-network compression scheme for data collection
in wireless sensor networks. In: Proceedings of IEEE con-
ference on computer communications and networks
(ICCCN), Dallas, TX, 20–22 October 2003, pp.539–544.
New York: IEEE.
25. Noh DK. Transmission range determination with a
timeslot-based energy distribution scheme for solar-
energy harvesting sensor systems. In: Jung HK, Kim JT,
Sahama T, et al. (eds) Future information communication
technology and applications. Amsterdam: Springer, 2013,
pp.661–669.
26. Jeong HJ, Lee WY and Joe I. Transmission power con-
trol scheme through reliability judgment of RSSI in wire-
less sensor network. In: Proceedings of symposium of the
Korean Institute of Communications and Information
Sciences, Jeju, Korea, 20–22 June 2012, pp.362–363.
Seoul, Korea: KICS.
27. Hou TC and Li V. Transmission range control in multi-
hop packet radio networks. IEEE T Commun 1986; 34(1):
38–44.
28. Lin S, Zhang J, Zhou G, et al. ATPC: adaptive transmis-
sion power control for wireless sensor networks. In: Pro-
ceedings of the 4th international conference on embedded
networked sensor systems (SenSys), Boulder, CO, 31
October–3 November 2006, pp.223–236. New York:
ACM.
29. Arisha K, Youssef M and Younis M. Energy-aware
TDMA-based MAC for sensor networks. In: Karri R
and Goodman D (eds) System-level power optimization
for wireless multimedia communications. New York:
Springer, 2002, pp.21–40.
30. Encarnacion NN and Yang H. A simple energy harvest-
ing algorithm for wireless sensor networks. J inform Com-
mun Convergence Eng 2012; 10(4): 359–364.
31. Roundy SJ. Energy scavenging for wireless sensor nodes
with a focus on vibration to electricity conversion. PhD
Thesis, University of California, Berkeley, CA, 2003.
32. Kansal A, Potter D and Srivastava MB. Performance
aware tasking for environmentally powered sensor net-
works. In: Proceedings of the joint international conference
on measurement and modeling of computer systems (SIG-
METRICS), New York, 10–14 June 2004, pp.223–234.
New York: ACM.
33. Shaikh FK and Zeadally S. Energy harvesting in wireless
sensor networks: a comprehensive review. Renew Sust
Energ Rev 2016; 55: 1041–1054.
34. Fall K. A delay-tolerant network architecture for chal-
lenged internets. In: Proceedings of the conference on
applications, technologies, architectures, and protocols for
computer communications (SIGCOMM), Karlsruhe, 25–
29 August 2003, pp.27–34. New York: ACM.
35. Yang Y, Wang L, Noh DK, et al. Solarstore: enhancing
data reliability in solar-powered storage-centric sensor
networks. In: Proceedings of the 7th international confer-
ence on mobile systems, applications, and services
(MobiSys), Krakow, 22–25 June 2009, pp.333–346. New
York: ACM.
36. Kansal A, Hsu J, Zahedi S, et al. Power management in
energy harvesting sensor networks. ACM T Embed Com-
put Syst 2007; 6(4): 1–38.
37. Cox DR. Prediction by exponentially weighted moving
averages and related methods. J R Stat Soc 1961; 23:
414–422.
38. Piorno JR, Bergonzini C, Atienza D, et al. Prediction and
management in energy harvested wireless sensor nodes.
In: Proceedings of the 1st IEEE conference on wireless
communication, vehicular technology, information theory
12 International Journal of Distributed Sensor Networks
and aerospace & electronic systems technology (VITAE),
Aalborg, 17–20 May 2009, pp.6–10. New York: IEEE.
39. Melodia T, Pompili D and Akyildiz I. Optimal local
topology knowledge for energy efficient geographical
routing in sensor networks. In: Proceedings of the 23rd
annual joint conference on computer and communications
societies (INFOCOM), Hong Kong, China, 7–11 March
2004, pp.1705–1716. New York: IEEE.
40. Yi JM, Kang MJ and Noh DK. SolarCastalia-solar
energy harvesting wireless sensor network simulator. Int
J Distrib Sens N 2015; 2015: 1–10.
41. Yi JM, Oh E, Noh DK, et al. Energy-aware data com-
pression and transmission range control scheme for
energy-harvesting wireless sensor networks. IEMEK J
Embed Syst Appl 2016; 11(4): 243–249.
Yi et al. 13
