You are on page 1of 11

Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx

Contents lists available at ScienceDirect

Journal of King Saud University –


Computer and Information Sciences
journal homepage: www.sciencedirect.com

Variable sampling interval for energy-efficient heterogeneous precision


agriculture using Wireless Sensor Networks
Yousef Hamouda ⇑, Mohammed Msallam
Department of Computer Science, Al-Aqsa University, P.O. Box 4051, Gaza, Palestine

a r t i c l e i n f o a b s t r a c t

Article history: This study aims to examine the problem of sampling interval selection for the precision agriculture using
Received 8 February 2018 Wireless Sensor Networks (WSNs). The Wireless Sensor Networks is introduced for limited energy sensor
Revised 18 April 2018 nodes and deployed in an agricultural farm, which is divided into a number of heterogeneous agricultural
Accepted 19 April 2018
areas. A Variable Sampling Interval Precision Agriculture (VSI-PA) system is proposed to measure and
Available online xxxx
monitor the agricultural parameters for appropriate agricultural activities e.g. water irrigation. At every
time step, the sensor nodes are selected for each area to sense the soil moisture and temperature. After
Keywords:
that, a variable sampling interval is calculated independently for each individual area, according to the
Variable sampling interval
Energy-efficient
soil temperature of the area. Analysis of simulation results show that, compared with other fixed sam-
Irrigation system pling interval schemes, the proposed VSI-PA system provides a significant improvement in energy con-
Precision agriculture sumption while maintaining a small variation of the soil moisture regardless of the values of soil
Wireless sensor networks temperature. Therefore, the crop yields have been enhanced.
Ó 2018 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction of applying PA inputs (i.e. farming parameters and resources) at


the right location and time to optimize the farming production,
Wireless Sensor Networks (WSNs) have become an emerging improve the resource utilization, provide online monitoring,
phenomenon in industry, both for civil and military purposes. They reduce the agriculture cost, and reduce the human power which
consist of hundreds or thousands of tiny sensor nodes, connected is subjected to minimizing the environmental impact (Crary
to each other via wireless communication protocols (Wieselthier et al., 1990; Leea et al., 2010; Paek et al., 2014). Farming parame-
et al., 2002; Potdar et al., 2009). The sensor nodes cooperate ters and resources include site-specific application of water, fertil-
(Zhao et al., 2003), together to sense, compute, and transmit the izer doses, pesticide, soil moisture, herbicides, and air temperature
information from harsh physical environments to external base (Adam et al., 2000). Thus, WSNs are employed in PA to monitor,
station or sink. Nowadays, commercial and industrial fields employ optimize, and measure different farming and sowing parameters
WSNs for a wide range of applications, such as healthcare, target and resources to take the appropriate action (Lee, 1998; Yazdi
tracking, environmental monitoring, weather monitoring, decision et al., 2000). One of the PA applications is the irrigation, which is
support systems, inventory control, wildlife habitat monitoring, the process of delivering water to the soil. Water requirements
smart homes, bio-chemical material detection, disaster manage- for plants vary based on crop species, crop growth stage, soil mois-
ment, surveillance, and agriculture (Arampatzis et al., 2005). ture, soil temperature, and soil type (Wan, 2001). In traditional irri-
WSNs have been recently employed in modern agriculture and gation, farmers irrigate plants without any accurate calculations.
farming. Precision Agriculture (PA) refers to automation in agricul- This usually results in wasting the water and irrigating crops
ture using information technology. PA is defined as the techniques unevenly (Ali, 2011). Due to the fact that traditional irrigation is
normally performed without precise knowledge of the soil, it leads
⇑ Corresponding author. to erratic water supply for the plants, which results in fewer yields,
E-mail address: ye.hamouda@alaqsa.edu.ps (Y. Hamouda). time consuming and more human efforts. There is clearly a need to
Peer review under responsibility of King Saud University. do something about soil moisture and soil temperature. They have
to be adjusted and regulated precisely to avoid water overflow and
the risk of evaporation.
To achieve some of the mentioned PA features and benefits, this
Production and hosting by Elsevier paper proposes Variable Sampling Interval Precision Agriculture

https://doi.org/10.1016/j.jksuci.2018.04.010
1319-1578/Ó 2018 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
2 Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx

(VSI-PA) system using WSNs. The main contributions for this paper are scheduled to sleep for a while to save energy. Energy-efficient
are as follows: (1) To automatize irrigation decision making routing protocol is developed in (Maurya and Jain, 2017) for
through sensing the agricultural parameters so that crop yields will automatized irrigation system, where the entire area is divided
be enhanced, water usage will be save, and labor workers will be into region-based fixed clusters. In (Nayak et al., 2014), harnessing
significantly reduced; (2) To improve energy consumption, the wind energy is used to charge the batteries of sensor nodes. In
sampling interval, which is the time between two sensing or mea- addition, adaptive power-aware routing protocol is adopted to
suring agricultural parameters, is adaptively calculated based on improve the energy efficiency and network lifetime. In (Sahota
the current sensed temperature; (3) To distinguish and treat differ- et al., 2010), Media Access Control (MAC), a layer is designed to
ent areas separately based on the type and soil conditions. increase the energy efficiency through wake-up synchronization.
The paper is divided into five sections. After this introduction, Moreover, network layer is designed through periodic data collec-
Section 2 explores and critically analyses work related to PA. Sec- tion to minimize the energy consumption.
tion 3 introduces detailed description of VSI-PA system. Section 4 Having discussed that, the above-mentioned literature adopted
describes a simulation-based evaluation. Finally, Section 5 summa- fixed sampling interval. Thus, damage is caused to crops when
rizes the paper. value of fixed sampling interval is large and high energy is con-
sumed when its value is small.
In this paper, a VSI-PA system is proposed to overcome the lim-
2. Related work itations of the current research. The VSI-PA system considers the
whole area of different types of plants. In addition, a specific sam-
Precision agriculture using WSNs is designed and developed pling interval is mapped for each plant area separately. Further-
(Diedrichs et al., 2014; Le and Tan, 2015). In (Diedrichs et al., more, the VSI-PA adopts an adaptive mechanism to calculate
2014), WSNs based on IEEE 802.15.4 are developed to measure sampling intervals for each plant’s area. The sampling interval is
the temperature. In (Le and Tan, 2015), the environmental param- adaptively calculated and mapped to each plant’s area according
eters including temperature, air humidity, light, soil moisture, and to the current agricultural parameters and requirements for each
pH level are monitored and controlled. In (Hamouda and Elhabil, plant to reduce the network consumed energy and improve the
2017), Greenhouse Smart Management System (GSMS) is devel- crop yields.
oped to control, manage and monitor the agriculture inside green-
houses. The ambient relative humidity and temperature are
measured using WSN. Agricultural monitoring systems using 3. VSI-PA detailed description
WSNs are developed by Xu et al. (2015); Khedo et al. (2014); Das
et al. (2009); Shinghal et al. (2010). In (Xu et al., 2015), soil temper- 3.1. Problem formalization
ature, soil moisture, atmospheric temperature, atmospheric
humidity, CO2 concentration, PH values and illumination intensity The agricultural farm (A) is divided into ‘‘m” areas (ai ), where
are monitored. Alarms are sent when the parameters exceed i ¼ 1; 2; . . . ; m so that heterogeneous plants can be planted in the
threshold values. In (Khedo et al., 2014), the temperature, humidity farm. The agricultural farm is a set of areas (ai ) as fol-
and carbon dioxide level are measured and sent to farmers. Disease lows:A ¼ fa1 ; a2 ; . . . ; am g. Each area (ai ) has a set (Sai ðkÞ) of
monitoring of grape vine crops is monitored in (Das et al., 2009). In ‘‘nai ðkÞ” sensor nodes as follows: Sai ¼ fsi1 ; si2 ; . . . ; sinai g, where sij
(Shinghal et al., 2010), the potato plants are monitored individually is the ‘‘jth” sensor node in area (ai ) and j ¼ 1; 2; . . . nai . Therefore,
to improve crop production. The automatic irrigation systems the set and the total number of sensor node for the agricultural
using WSNs have been receiving great attention. The soil moisture farm are defined as Snet and ‘‘N” respectively, where
and soil temperature are measured and monitored in (Tarange P
Snet ¼ Sa1 [ Sa2 [ . . . [ Sam and N ¼ m i¼1 nai . It is assumed that there
et al., 2015; Haule and Michael, 2014; Patel et al., 2011). In (Patel
are no conflict nodes which are defined as the sensor nodes that
et al., 2011), the amount of water required for irrigation is deter-
can sense the agricultural parameters for more than one area.
mined to save water and to increase productivity. The soil water
Therefore, Sap \ Saq ¼ U for all p  m and q  m, where U is the
content for irrigation is measured in (Milind and Bhaskar, 2014).
empty set. At each time step (k), VSI-PA system selects a sensing
Threshold values for irrigation are predefined according to the crop
set for each area (Ss ðai ; kÞ) of ‘‘ns ðai ; kÞ” sensor nodes to sense the
type and plant growth stage. The system proposed in (Ramya et al.,
agricultural parameters, where Ss ðai ; kÞ ¼ fsi1 ; si2 ; . . . ; sins ðai ;kÞ g,
2012) measures temperature, soil moisture, humidity and the level
of water tank used for cardamom plants irrigation. An irrigation ns ðai ; kÞ 6 nai , and Ss ðai ; kÞ  Sai . The area and sensor node locations
system is developed in (Nandurkar et al., 2014) to monitor the are defined as Lai ¼ ½ xai yai T and Lsij ¼ ½ xsij ysij T , respectively. At
crops and allow irrigation as needed. An intelligent cultivation each time step (k), the variable sampling interval of area (ai ) is also
management system is presented in (ZENG et al., 2016) to improve calculated according to the agricultural parameters and denoted as
crop yield. Temperature, humidity, light, and concentration of CO2 Dtv ðai ; kÞ.
are sensed to monitor and control the garden. Pumps and valves
are automatically opened for irrigation when the measurements
exceed threshold values. In (Bennis et al., 2015), a model for drip 3.2. Sensor node structure
irrigation system is proposed to overcome the blockage of water
emitter and the broken pipes through monitoring the soil mois- The sensor nodes are classified into three categories which are
ture, temperature and pressure. In (Hamouda, 2017), Smart Irriga- Normal Sensor Node (NSN), Master Sensor Node (MSN) and Sink
tion Decision Support (SIDS) system using fuzzy logic is developed Sensor Node (SSN). The NSN has limited-resources in terms of pro-
for water irrigation. The soil temperature and moisture are sensed cessing and energy. As shown in Fig. 1, the NSN is equipped with
and employed to compute the amount of irrigation time using sensors, to measure soil moisture and temperature, wireless com-
fuzzy logic. munication unit, to wirelessly communicate with other sensor
Energy-efficient precision agriculture has been discussed in nodes, and limited processing capabilities, to cooperate with in-
many studies. In (Math and Dharwadkar, 2017), energy-efficient network processing (Wieselthier et al., 2002; Potdar et al., 2009).
framework is developed for precision agriculture. The sensor nodes The MSN and SSN are full-resources nodes with higher processing
are classified into coordinator and remote nodes. The remote nodes and energy capabilities. However, the MSN and SSN do not support

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx 3

Fig. 1. Normal Sensor Node (NSN) Component and Architecture.

the sensing capability. They are used for data exchange and the sleeping mode. NSN is triggered to enter the communicating
communication. mode. According to the messages received by sensor node, the
Since the energy efficiency is a crucial feature in WSNs, the sen- operation modes are changed as shown in Fig. 2.
sor node operates in one of predefined modes which are sensing
mode (Sen.) to sense soil moisture and temperature; communicat- 3.3. VSI-PA framework
ing mode (Comm.) to transmit, or receive data from other node;
processing mode (Proc.) to perform the processing tasks; and As shown in Fig. 3, the agriculture farm, with width, W and
sleeping mode to listen for possible communication. The sensing, length, L is divided into smaller areas in which the heterogeneous
communicating and processing modes are active modes because plants are planted. NSNs are the sensor nodes that sense the agri-
the sensor node consumes energy in sensing, transmitting, receiv- culture parameters. MSNs receive the information from NSNs and
ing, and processing activities. On the other hand, the sleeping relay it to SSN. Finally, the SSNs forward the measurements to
mode is a low-energy communication channel in which sensor the server and farmers via Internet, cellular network, or satellite
node listens to channel status for transmission initialization, or communication. The server performs processing and storage
waits for traffic from other sensor nodes (Miller and Vaidya, requirements for VSI-PA. Therefore, the server sends the suitable
2005). Therefore, sensor nodes in sleeping mode have to be trig- triggers for the actuators, such as triggering the valves and pumps
gered to ‘‘wakeup” for communicating, processing or sensing. The for irrigation. Furthermore, the server performs algorithm that
state diagram of sensor node shown in Fig. 2 explains the changing adaptively calculates the next sampling interval, and sends the
among the sensor node modes. The default mode of sensor node is appreciate commands to the farm.

3.4. Sensor nodes selection Algorithm

The irrigation system is used for precision agriculture purposes.


At each time step (k), the soil moisture and soil temperature for
each area (ai ) is measured and denoted as M ai ðkÞ and T ai ðkÞ respec-
tively. Like most research into precision agriculture, the irrigation
using VSI-PA system triggers the valves, pumps, and other actua-
tors when the measured agricultural parameters exceed a prede-
fined threshold value. Therefore, the VSI-PA system activates the
irrigation process when M ai ðkÞ of ai reaches to a predefined thresh-
old value, which is set based on crop requirements (e.g. size, age
and type). Thus, the soil moisture should be monitored periodi-
cally. However, soil moisture is measured in discrete times for each
area separately. The sensor nodes used to measure the agricultural
parameters are firstly selection as shown in Algorithm 1. At each
time step (k), the nearest sensor nodes set (Ss ðai ; kÞ) to the area ai
center with energy level bigger than predefined threshold value
Eth is selected to sense the soil moisture and temperature. The
Euclidean Distance (ED) is used to measure the distance between
the sensor node and the area center according to the relation:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dij ¼ ½Lsij  Lai T ½Lsij  Lai  ð1Þ

The average readings of the selected ‘‘ns ðai ; kÞ” sensors of ai are
recorded as the current Mai ðkÞ and T ai ðkÞ according to the following
Fig. 2. State Diagram for Sensor Node. relations:-

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
4 Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx

Fig. 3. VSI-PA framework.

nsX
ðai ;kÞ Algorithm 2, the variable sampling interval at each time step is
M sij ðkÞ selected according to the soil temperature. The sampling interval
j¼1
M ai ðkÞ ¼ ð2Þ is reduced when the temperature is high. This means that the rate
ns ðai ; kÞ of soil moisture reduction is also high, and in turn, the soil moisture
is measured in time before it drops significantly below the threshold
nsX
ðai ;kÞ
value, and causes damage to the plants and reduction of the crop
T sij ðkÞ
yields. On the other hand, the sampling interval is increased when
j¼1
T ai ðkÞ ¼ ð3Þ the temperature is low. This means that the rate of soil moisture
ns ðai ; kÞ
reduction is also low, and consecutively. The overall energy con-
where Msij ðkÞ and T sij ðkÞ are the soil moisture and temperature read- sumption by the network is reduced, and the network lifetime is
ings of sensor (sij ) in area (ai ). improved. Therefore, the variable sampling interval (Dtv ðai ; kÞ) and
the soil temperature (T ai ðkÞ) of area (ai ) at time k are assumed to
Algorithm 1: Sensor Node Selection be limited into a range as Dtmin 6 Dt v ðai ; kÞ  Dtmax and T min 
T ai ðkÞ 6 T max respectively. The variable sampling interval (Dt v ðai ; kÞ)
1: for each area ai do:
at time step (k) is assumed to be selected according to an inverse
2: set ‘ ¼ 0;
liner function of T ai ðkÞ according to the following equation:
3: while ‘  ns ðai ; kÞ do:
4: for each sensor node sij 2 Sai ðkÞ do:  
Dt min  Dt max
5: calculate ED (dij ) to the area centre based on Eq. (1); Dtv ðai ; kÞ ¼ ðT ai ðkÞ  T max Þ þ Dtmin ð4Þ
T max  T min
6: end for;
7: find the sensor node sij with minimum ED; According to Eq. (4), when the soil temperature is a minimum
8: if (Er ðai ; sij ; kÞ > Eth ) do: value (T min ), the sampling interval is a maximum value (Dtmax ),
and vice versa. The parameters T min , T max , Dtmin and Dtmax are cho-
9: add sij to Ss ðai ; kÞ;
sen according to crop and soil types.
10: measure M sij ðkÞ and T sij ðkÞ;
11: end if; Algorithm 2: Variable Sampling Interval Selection
12: remove sij from next search;
1: for each area ai do:
13: increment ‘: ‘ ¼ ‘ þ 1;
2: initialization:
14: end while;
3: initialize the time step k ¼ 0;
15: compute the average readings M ai ðkÞ andT ai ðkÞbased
4: set independently,Dt v ðai ; 0Þ;
on Eq. (2) and (3);
5: advance the time,t ai ðk þ 1Þ ¼ Dt v ðai ; 0Þ;
16: end for;
6: define values for T min ; T max ; Dt min ; Dt max ;
7: end initialization;
8: while (stop condition is not valid) do:
3.5. Sampling interval selection algorithm
9: compute the average readingT ai ðkÞaccording to
Algorithm 1;
Unlike PA schemes using a fixed sampling interval, the proposed
10: calculate the sampling intervalDt v ðai ; kÞbased on
VSI-PA scheme adaptively calculates the sampling interval to
Eq. (4);
improve the energy efficiency and maintain the soil moisture of
11: advances the timetai ðk þ 1Þ ¼ t ai ðkÞ þ Dt v ðai ; kÞ;
the plants with improved crop yields. Each area is assumed to have
12: end while;
different soil temperature variation. However, all plants are
13: end for;
assumed to have the same water consumption. As shown in

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx 5

3.6. VSI-PA communication protocol


8: GTrig message is received from SSN;
Protocol 1 shows that the communication protocol runs in 9: schedule the event: communication;
NSNs. Firstly, each NSN identifies its MSN and enters sleeping 10: Event 2: Communication
mode waiting for activation. NSNs send periodically (Beacon) mes- 11: turn on communication channel (i.e.
sages to its MSN containing its current states such as the remaining communicating mode);
battery level and timestamp for time synchronization. When the 12: relay GTrig message to sensing NSNs (multicasting);
MSN sends a group-triggering message (GTrig) to NSNs, the NSNs 13: schedule event: Readings;
wake up to sense and send the agricultural parameters such as soil 14: Event 3: Readings
moisture and temperature. NSNs send the measurement readings 15: waiting for measurements from the sensing group;
to MSN through the sensing messages (SMeg) message. 16: SMeg message is received from the sensing group;
17: Determine the best route to SSN via MSNs (multi-
Protocol 1: Communication Protocol for NSN hop communication);
1: Initialization 18: relay SMeg message to SSN;
2: identify the current MSN; 19: schedule event: Sleeping;
3: set Timer_Beacon; 20: Event 4: Sleeping
4: schedule event: sleeping; 21: shutdown the communication channel;
5: end initialization; 22: turn on the low-energy communication channel (i.e.
6: while (true) do: sleeping mode);
7: Event 1: Sensing 23: Event 5: Beacon
8: GTrig message is received from MSN;9: turn on 24: Timer_Beacon is expired;
the sensing circuits (i.e. sensing mode); 25: turn on the energy communication channel (i.e.
9: sense the agricultural parameters; communicating mode);
10: schedule event: communication; 26: Beacon message is received from the NSNs;
11: Event 2: Communication 27: record the NSNs states and synchronize the time
12: turn on the communication channel (i.e. with NSNs;
communicating mode); 28: schedule event: sleeping;
13: send SMeg message to the MSN; 29: Event 6: Information
14: schedule event: sleeping; 30: Req_GInfo message is received from SSN;
15: Event 3: Sleeping 31: turn on the energy communication channel (i.e.
16: shutdown the sensing circuit and communication communicating mode);
channel; 32: send Res_GInfo to the SSN containing the NSNs
17: turn on the low-energy communication channel (i.e. states;
sleeping mode); 33: schedule event: sleeping;
18: Event 4: Beacon 34: end while;
19: Timer_Beacon is expired;
20: turn on the communication channel (i.e. 3.7. Energy consumption model
communicating mode);
21: send Beacon message to MSN; Energy is consumed during sensing and communication activi-
22: schedule event: sleeping; ties. As in (Wang and Chandrakasan, 2002), the energy consump-
23: end while; tion of the sensor (si ) to transmit l -bit message to the sensor (sj )
over a distance dij is:

wTX ðl; dij Þ ¼ a1 :l þ a2 :l:dij


2
Protocol 2 shows the communication protocol runs in MSNs. Firstly, ð5Þ
MSN identifies its current NSNs and enters the sleeping mode wait-
where a1 is the electronic energy required to transmit one bit and
ing for activation. The MSN wakes up periodically to receive (Bea-
con) messages from NSNs, records NSNs states, and synchronizes
a2 is the related to radio energy. The energy consumption to receive
l -bit message is:
the time with NSNs. The MSN forwards the (GTrig) messages
received from SSN to its NSNs. On the other hand, MSN forwards wRX ¼ a3 :l ð6Þ
the (SMeg) messages received from its NSNs to SSN. When MSN
receives a request from SSN containing the sensing group informa- 4. Results and discussion
tion (Req_GInfo) message, it replies a response to SSN containing the
sensing group information (Req_GInfo) message that give informa- 4.1. Simulation assumption
tion about the NSNs states to be used for VSI-PA algorithms.
In this section, the performance of the proposed VSI-PA system
Protocol 2: Communication Protocol for MSN is evaluated using event-driven simulation. C++ is used to build the
1: Initialization simulation environment using core i5 of 2.5 GHz processor and 4
2: identify the current NSNs; MB memory. The simulation results are averaged over 20 runs with
3: set Timer_Beacon; different random sensor placement using fixed sensor density, and
4: schedule event: Sleeping; the simulator is stopped after 130 days. The agricultural farm of
5: end initialization; 200 m  200 m is divided into equal four areas (m ¼ 4) with 100
6: while (true) do: m  100 m for each area. Four MSNs are placed in the center of
7: Event 1: Activation each area, and one SSN is placed in the most left-top corner of
the agricultural farm. For each time step, three NSNs
(ns ðai ; kÞ ¼ 3) are selected for each area to sense the solid moisture,

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
6 Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx

and the average reading is recorded. The sensor reading is assumed coverage area if the distance between them is equal to or less than
to be within a sensing range of 50 m. Therefore, to increase the the radio range, which is set to 100 m. Therefore, MSN for each area
likelihood of sensing area by three sensor nodes with sensing range is within the coverage area of all NSNs in the same area. The MSN is
of 50 m at all-time steps, the number of NSNs deployed in the agri- also within the coverage area of its neighbor MSNs. The Destina-
cultural farm should be calculated properly. The sensor density of tion Sequenced Distance Vector routing (DSDV) routing protocol,
q ¼ 5:6  103 sensors=m2 can guarantee 100% that three NSNs or which is a proactive ad hoc routing protocol, has been imple-
more within sensing range inside the areas (Wang et al., 2008). mented in the simulation model for the multi-hop routing among
Thus, for a given agricultural farm of A ¼ 200  200m2 , the number MSNs and SSN (Perkins and Bhagwat, 1994). The energy model
of NSNs required to be uniformly deployed is N ¼ qA ¼ 224sensor parameters are set as follows: a1 = 100 nJ/b, a2 = 1 pJ/bm2, a3 =
nodes. 100 nJ/b. The sensing energy cost for all sensor nodes is assumed
We assume line of sight (LOS) communication between the to be 8 nJ. The energy level of each sensor node is set to 0.5 J.
nodes within the same coverage area. Two nodes are in the same The communication loads for all measurements and management
messages are assumed to use l ¼ 200KB. The VSI-PA system is used
with T min ¼ 25C ; T max ¼ 40C , Dt min ¼ 7 days and Dtmax ¼ 14 days.
For each irrigation activity, the soil moisture is assumed to rise
to 60%.
As discussed in Section 2, the related research in precision agri-
culture adopt fixed sampling interval. In the following simulation
results, these researches are referred as fixed schemes. The pro-
posed VSI-PA system is compared with the fixed schemes devel-
oped in the literature.

4.2. VSI-PA evaluation

The agricultural farm with four areas of VSI-PA system is shown


in Fig. 4. The dots represent the locations of NSNs inside the farm.
The SSN and MSNs locations with routing paths based on DSDV are
also shown. Each MSN has a path through other MSNs to send the
messages to SSN.
At each time step (k), the soil moisture (Mai ðkÞ) and soil temper-
ature (T ai ðkÞ) for each area (ai ) are measured and plotted in Figs. 5
and 6 respectiv M ai ðkÞ ely. The readings of sensor nodes are set ran-
Fig. 4. Network topology and structure for the simulated VSI-PA.
domly in the interval ½T min T max , where T min ¼ 25C ; T max ¼ 40C .

Fig. 5. Soil temperature of VSI-PA system for each area.

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx 7

The measured soil temperature (T ai ðkÞ) for each area (ai ) at time
step (k) is the average readings of selected sensor nodes (ns ðai ; kÞ)
at time step (k). Therefore, the average readings of selected sensor
nodes are plotted versus the simulated time in Fig. 5. Conse-
quently, soil moisture (Mai ðkÞ), at each time step (k) for each area,
is measured and plotted in Fig. 6. In addition, an irrigation activity
is performed at each time step (k) after measuring the current level
of soil moisture (Mai ðkÞ). It is assumed that the level of soil mois-
ture after each irrigation activity rises to 60%. The measured soil
moisture (), before irrigation activity, along with its level after irri-
gation activity (i.e. 60%) at each time step (k) for each area, are
plotted in Fig. 6.
As shown in Fig. 5, soil temperature varies from area to another.
In turn, the variable sampling interval is calculated independently
for each area according to its soil temperature readings.
Fig. 7 shows the total energy consumption versus the simulation
time of the proposed VSI-PA for each area. For all areas, the energy Fig. 7. Energy consumption of VSI-PA system for each area.
consumption increases with the simulation time because each irri-
gation activity consumes energy in sensing and communication
where irrigation activities are performed at the end of each sam- 4.4. The impact of selection algorithm
pling interval. The increase of energy consumption with simulation
time differs from area to another because the VSI-PA system calcu- As proposed in Section 3.4, the sensor nodes (ns ðai ; kÞ) is
lates the variable of sampling interval for each area separately. selected to measure the soil parameters. The plants roots are
assumed to be focusing on the center of the area particularly in
4.3. Variable sampling interval selection early stages of plants growth (Judd et al., 2015). Thus, the sensor
nodes around the area center can highly represent the area and
Fig. 8 shows sampling intervals of different schemes for each the plant requirements of water irrigation. Therefore, the closest
area. The sampling intervals of the fixed schemes are static and sensor nodes with energy level bigger than defined threshold value
do not change during the simulation time. The sampling interval are selected to detect the soil parameters.
of the proposed VSI-PA system is selected for each area indepen- Fig. 9 shows the total energy consumption of the network with
dently according to Eq. (4). simulation time for different selected nodes size. For all selected

Fig. 6. Soil moisture of VSI-PA system for each area.

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
8 Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx

Fig. 8. Sampling interval of different scheme for each area.

nai ðkÞ
nodes size, the total energy consumption increases with simulation X
time because sensing and communicating activities are performed M sij ðkÞ
at each time step. However, a higher selected nodes size gives j¼1
Mr ðai ; kÞ ¼ ð8Þ
higher energy consumption because more sensor nodes are nai ðkÞ
involved for communicating and sensing activities.
Therefore, the soil moisture (M e ðai ; kÞ) and temperature
The impact of selected sensor nodes size is evaluated where soil
(T e ðai ; kÞ) errors at time step (k) for area (ai ) are defined according
moisture error, soil temperature error, and total energy consump-
to the following equations:
tion of the whole farm are calculated for difference selected nodes
size. The real measurements of soil moisture (M r ðai ; kÞ) and tem- jMr ðai ; kÞ  M ai ðkÞj
Me ðai ; kÞ ¼  100% ð9Þ
perature (T r ðai ; kÞ) at time step (k) for area (ai ) is defined as the Mr ðai ; kÞ
average of all sensor nodes readings of the area, where:
nai ðkÞ
X jT r ðai ; kÞ  T ai ðkÞj
T e ðai ; kÞ ¼  100% ð10Þ
M sij ðkÞ T r ðai ; kÞ
j¼1
M r ðai ; kÞ ¼ ð7Þ where M ai ðkÞ and T ai ðkÞ are the soil moisture and temperature
nai ðkÞ
which are calculated here using the selected sensor nodes
(ns ðai ; kÞ) according to Eqs. (2) and (3). The average soil moisture
and soil temperature errors for all areas over all simulation time
are calculated and recorded in Table 2, for selected nodes size of
1, 2, 3, 4 and 5. With increasing the selected sensor nodes size,
the average soil moisture and soil temperature errors are reduced
because more readings of soil moisture and temperature are per-
formed. On the other hand, total energy consumption of network
is increased with using more selected sensor nodes because com-
municating and sensing activities are increased with increasing
the selected sensor nodes.
As show in Table 1, the soil moisture and temperature errors at
selected nodes size of 5 are reduced by 8% compared with its value
at selected nodes size of 4. Nevertheless, total energy consumption
at selected nodes size of 5 is increased by 17% compared with its
value at selected nodes size of 4. Consequently, the cost of total
energy consumption of network due to increasing of selected
nodes size is high compared with the improvement of soil mois-
ture and soil temperature errors. Therefore, the selected sensor
Fig. 9. Total energy consumption of network for different scheme. nodes size is a designed parameter which can be selected accord-

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx 9

Table 1
Soil moisture error, soil temperature error, and total energy consumption of the
whole farm for different selected nodes size.

Selected Nodes Average Average Total Energy


(ns ðai ; kÞ) Temperature Error Moisture Error Consumption (J)
(%) (%)
1 12.39 20.42 42.804
2 8.32 13.41 58.684
3 6.61 10.22 74.606
4 5.76 9.07 89.413
5 5.31 8.44 104.86

ing to required sensing accuracy crop and current network lifetime.


For instance, if the crop is valuable and requires precise soil detec-
tion, high number of sensor nodes can be selected. In contrast, low
number of sensor nodes is selected for less valuable crop to
increase the network lifetime.

4.5. Comparison with fixed scheme Fig. 11. Average soil moisture of the whole farm for different scheme.

Fig. 10 shows the soil moisture of different scheme for each


area. In addition, Fig. 11 shows the average soil moisture of the on soil temperature variations, it maintains soil temperature with
whole agricultural farm for each different scheme. The soil mois- small values, compared with fixed scheme with sampling interval
ture reduces severely in case of using a fixed scheme with sam- of 11 days and 7 days.
pling intervals of 14 days because its sampling interval for Energy consumption of different scheme for each area is plotted
sensing the soil moisture is the highest compared with other in Fig. 12. Furthermore, Fig. 13 also shows the total energy con-
schemes. On the other hand, the soil moisture in case of using a sumption of the network for different scheme. The energy con-
fixed scheme with sampling intervals of 7 days has the lowest val- sumption increases with simulation time. A lower sampling
ues because its sampling interval for sensing the soil moisture is interval gives higher energy consumption because more sensing
the lowest compared with other schemes. However, the highest and communication activities are required to do monitoring and
values of energy consumption occur. A fixed scheme with sampling water irrigation. Therefore, Figs. 12 and 13 show that the lowest
intervals of 11 days has soil moisture behavior between the two and highest values of energy consumption occur in case of using
fixed schemes with sampling intervals of 7 and 14 days. Since the fixed schemes with sampling interval of 14 days and 7 days
the proposed VSI-PA system selects the sampling interval based respectively. The fixed scheme with sampling interval of 11 days

Fig. 10. Soil moisture of different scheme for each area.

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
10 Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx

Fig. 12. Total energy consumption of different scheme for each area.

sumes less energy compared with fixed scheme with sampling


interval of 11 days and 14 days. The fixed scheme with sampling
interval of 14 days has the lowest energy consumption. However,
it does not provide a good constancy of soil moisture as shown
in Fig. 10 and 11.
Table 2 shows the network lifetime, number of dead nodes,
total energy consumption, average soil moisture, and the standard
deviation of soil moisture variations for fixed schemes and pro-
posed VSI-PA system. Since the standard deviation of the proposed
VSI-PA system is the lowest, it provides the lowest soil moisture
variations where the soil moisture is measured in time before it
drops significantly and causes damage to the plants so that the
crop yields are improved. In case of using fixed schemes with sam-
pling interval of 7 days and 11 days, the sensing and communica-
tion activities are performed frequently in small and fixed
intervals, which lead to the increase of energy consumption and
number of dead nodes, and in turn, reduce the network lifetime.
However, the proposed VSI-PA system performs sensing and com-
Fig. 13. Total energy consumption of network for different scheme.
munication activities when they are needed according to the calcu-
lated sampling interval that depends on agricultural parameters.
consumes energy between the fixed schemes with sampling inter- Nevertheless, the fixed schemes with sampling interval of 14 days
val of 7 and 14 days. Since the proposed VSI-PA system selects can provide better performance in term of the energy consump-
sampling interval based on the soil temperature variations, it con- tion, number of dead nodes, and network lifetime, compared with

Table 2
Network lifetime, dead nodes count, total energy consumption, average soil moisture and standard deviation of the whole farm for different scheme.

Approach Lifetime (days) Dead Nodes Total Energy Consumption (J) Soil Moisture (Average) Soil Moisture (Standard Deviation)
Dt f ¼ 7 days 70 10.84 114.09 38.97% 2.44%
Dt f ¼ 11 days 80 6.29 75.947 28.34% 4.18%
Dt f ¼ 14 days 74.67 4.64 56.875 18.75% 6.33%
VSI-PA System 80.67 5.46 74.606 29.74% 1.67%

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010
Y. Hamouda, M. Msallam / Journal of King Saud University – Computer and Information Sciences xxx (2018) xxx–xxx 11

the proposed VSI-PA system. However, it has the worst standard T. D. Le, D. H. Tan, Design and deploy a wireless sensor network for precision
agriculture, Conference on Information and Computer Science (NICS), 2015 2nd
deviation of soil moisture variations among all other schemes.
National Foundation for Science and Technology Development, 16-18 Sept.
2015; Ho Chi Minh City, Vietnam, pp. 294-299.
5. Conclusion K. Lee, A Smart Transducer Interface for Sensors and Actuators, Smart Transducer
Interface Standard-IEEE 1451.2 Standard 1998.
Leea, W.S., Alchanatis, V., Yang, C., Hirafuji, M., Moshou, D., Li, C., 2010. Sensing
This paper explores and critically analyses the much debated technologies for precision specialty crop production. Comput. Electron. AGR 74,
research in precision agriculture. VSI-PA system with variable sam- 2–33.
Math, R.K., Dharwadkar, N.V., 2017. A Wireless Sensor Network based Low Cost and
pling interval is proposed to overcome the shortcomings of other Energy Efficient Frame Work for Precision Agriculture. IEEE International
research that use fixed sampling intervals. The main objectives of Conference on Nascent Technologies in Engineering (ICNTE), pp. 1–6.
the proposed VSI-PA system are as follows: improving of the total Maurya, S., Jain, V.K., 2017. Energy-efficient network protocol for precision
agriculture: using threshold sensitive sensors for optimal performance. IEEE
energy consumption, increasing the network lifetime, while main-
Consumer Electr. Mag. 6, 42–51.
taining the soil moisture, and enhancing the crop yields. At each Milind, P.K., Bhaskar, P.C., 2014. Microcontroller based adaptive irrigation system
time step, the agriculture parameters are measured and monitored using WSN for variety crops and development of insect avoidance system for
better yield. Int. J. Res. Eng. Technol. (IJRET) 03, 308–312.
for each area to irrigate and select the next sampling interval. The
Miller, M.J., Vaidya, N.H., 2005. AMAC protocol to reduce sensor network energy
sampling interval for each individual area is calculated according consumption using a wakeup radio. IEEE Trans. Mobile Comp. 4, 228–242.
to the current soil temperature measurement of that area. The sim- Nandurkar, S.R., Thool, V.R., Thool, R.C., 2014. Design and Development of Precision
ulation results show that the proposed VSI-PA improves the energy Agriculture System Using Wireless Sensor Network. First International
Conference on Automation, Control, Energy and Systems (ACES), Hooghy,
consumption and maintains the soil moisture regardless of the soil India, pp. 1–6.
temperature variation, compared with other fixed sampling inter- Nayak, A., Prakash, G., Rao, A., 2014. Harnessing Wind Energy to Power Sensor
val schemes. Networks for Agriculture. IEEE International Conference on Advances in Energy
Conversion Technologies (ICAECT), pp. 221–226.
J. Paek, J. Hicks, S. Coe, R. Govindan, Image-Based Environmental Monitoring Sensor
Declarations of interest Application Using an Embedded Wireless Sensor Network, SENSORS-
BASEL2014, 14, pp. 15981-16002.
Patel, J.B., Bhatt, C.B., Patel, B., Parwani, K., C., 2011. Sohaliya, Field Irrigation
None. Management System Using Wireless Sensor Network. International Conference
on Engineering, Nirma University, Ahmedabad, Gujarat, India, pp. 1–4.
References Perkins, C.E., Bhagwat, P., 1994. Highly Dynamic Destination Sequence-Vector
Routing (DSDV) for Mobile Computers, SIGCOMM ’94. Proceedings of the
Conference on Communications Architectures, Protocols and Applications,
Adam, M.L., Cook, S., Corner, R., 2000. Managing uncertainty in site-specific London, United Kingdom, pp. 234–244.
management: what is the best model? Precis Agric 2, 39–54. Potdar, V., Sharif, A., Chang, E., 2009. Wireless Sensor Networks. A Survey. IEEE
Ali, M.H., 2011. Practices of Irrigation & On-farm Water Management, Practices of International Conference on Advanced Information Networking and
Irrigation & On-farm Water Management, 2. Springer. Applications Workshops, Bradford, UK, pp. 636–641.
Arampatzis, T., Lygeros, J., Manesis, S., 2005. A Survey of Applications of Wireless Ramya, V., Palaniappan, B., George, B., 2012. INT J COMPUT APPL. Int. J. Comp. Appl.
Sensors and Wireless Sensor Networks. Proceedings of the IEEE 13th 53, 36–43.
Mediterranean Conference on Control and Automation, Limassol, Cyprus, pp. Sahota, H., Kumar, R., Kamal, A., Huang, J., 2010. An energy-efficient wireless sensor
719–724. network for precision agriculture. IEEE Symposium on Computers and
Bennis, I., Fouchal, H., Zytoune, O., Aboutajdine, D., 2015. Drip irrigation system Communications (ISCC), 347–350.
using Wireless Sensor Networks. Conf. Comp. Sci. Inf. Syst. (FedCSIS), 1297– Shinghal, K., Noor, A., Srivastava, N., Singh, R., 2010. Wireless sensor networks in
1302. agriculture: for potato farming. Int. J. Eng. Sci. Technol. 2, 3955–3963.
Crary, S.B., Baer, W.G., Cowles, J.C., Wise, K.D., 1990. Digital compensation of high- Tarange, P.H., Mevekari, R.G., Shinde, P.A., 2015. Web Based Automatic Irrigation
performance silicon pressure transducers. Sensor Actuat A-Phys. 21, 70–72. System Using Wireless Sensor Network and Embedded LINUX Board.
I. Das, C. Naveen, S. S. Yadav, A. Kodilkar, N. G. Shah, S. N. Merchant, U. B. Desai, WSN International Conference on Circuit, Power and Computing Technologies
Monitoring of Weather and Crop Parameters for Possible Disease Risk (ICCPCT), Nagercoil, India, pp. 1–5.
Evaluation for Grape Farms - Sula Vineyards, A Case Study, Presented in the Wan, S.S., 2001. Fundamentals of soil physics. Natl. Inst. Comp. Transl., 136–145
Geomatrix-2009 conferemce, 2009, IIT Bombay, Mumbai, Oral Session Five. Wang, A., Chandrakasan, A., 2002. Energy-efficient DSPs for wireless sensor
Diedrichs, A.L., Tabacchi, G., Grünwaldt, G., Pecchia, M., Mercado, G., Antivilo, F.G., networks. IEEE Signal Proc. Mag. 19, 68–78.
2014. Low-Power Wireless Sensor Network for Frost Monitoring in Agriculture Wang, W., Srinivasan, V., Wang, B., Chua, K.C., 2008. Coverage for target localization
Research. IEEE Biennial Congress of Argentina (ARGENCON), Bariloche, in wireless sensor networks. IEEE T Wirel. Commun. 7, 667–676.
Argentina, pp. 525–530. J. E. Wieselthier, G. D. Nguyen, A. Ephremides, Resource management in energy-
Hamouda, Y.E.M., 2017. Smart Irrigation Decision Support Based on Fuzzy Logic limited, bandwidth-limited, transceiver-limited wireless networks for session
Using Wireless Sensor Network. International Conference on Promising based multicasting, COMPUT NETWORKS ISDN2002, 39, 2002, pp. 113-131.
Electronic Technologies (ICPET), Deir El-Balah, Palestine, pp. 109–113. J. Xu, J. Zhang, X. Zheng, X. Wei, J. Han, Wireless Sensors in Farmland Environmental
Hamouda, Y.E.M., Elhabil, B.H., 2017. Precision Agriculture for Greenhouses Using a Monitoring, International Conference on Cyber-Enabled Distributed Computing
Wireless Sensor Network. IEEE International Conference on Information and and Knowledge Discovery (CyberC), 17-19 Sept. 2015;Xi’an, China, pp. 372-379.
Communication Technology (PICICT), Gaza, Palestinian, pp. 78–83. Yazdi, N., Mason, A., Najafi, K., Wise, K.D., 2000. A generic interface chip for
Haule, J., Michael, K., 2014. Deployment of Wireless Sensor Networks (WSN) in capacitive sensors in low-power multi-parameter microsystems. Sensor Actuat.
Automated Irrigation Management and Scheduling Systems: A Review. Pan A-Phys. 84, 351–361.
African Conference on Science, Computing and Telecommunications (PACT), Zeng, X.H., Long, M.S., Qing, L.I.U., Wang, X.A., Luo, W.L., 2016. Intelligent
Arusha, Tanzania, pp. 86–91. Management System for Small Gardens Based on Wireless Sensor Network, In
Judd, L.A., Jackson, B.E., Fonteno, W.C., 2015. Advancements in root growth International Conference on P2P, Parallel, Grid, Cloud and Internet Computing.
measurement technologies and observation capabilities for container-grown Springer International Publishing, Spain, pp. 49–56.
plants. Plants 2015 (4), 369–392. Zhao, F., Liu, J., Liu, J., Guibas, L., Reich, J., August 2003. Collaborative signal and
K. K. Khedo, M. R. Hosseny, M. Z. Toonah, PotatoSense: A wireless sensor network information processing: an information directed approach. Proc. IEEE 11, 1199–
system for precision agriculture, IST-Africa Conference Proceedings; 7-9 May 1209.
2014, Le Meridien Ile Maurice, Mauritius, pp. 1-11.

Please cite this article in press as: Hamouda, Y., Msallam, M. Variable sampling interval for energy-efficient heterogeneous precision agriculture using
Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2018), https://doi.org/10.1016/j.jksuci.2018.04.010

You might also like