You are on page 1of 8

2019 International Conference on Mechatronics, Remote Sensing, Information Systems and Industrial Information Technologies (ICMRSISIIT) | 978-1-7281-0486-7/20/$31.

00 ©2019 IEEE | DOI: 10.1109/ICMRSISIIT46373.2020.9405879

Nature-inspired search method for IoT-based water leakage


location detection system

Emmanuel Freeman1, Daniel Ayitey Quaye2, Israel Edem Agbehadji3 and Richard C.
Millham4
1,2
Faculty of Computing and Information Systems, Ghana Communication Technology University
2
Coventry University, UK
3
Honorary Research Associate, Faculty of Accounting and Informatics, Durban University of Technology,
South Africa. / Postdoctoral Research Fellow, Office of the Deputy Vice Chancellor: Research, Innovation and
Engagement, Central University of Technology, Bloemfontein, Free state, South Africa
E-mail: israeldel2006@gmail.com
4 ICT and Society Research Group, Department of Information Technology, Durban University of Technology, South
Africa.
E-mail: richardm1@dut.ac.za

Corresponding e-mail: efreeman@gtuc.edu.gh

Abstract residents (occupants) bear. The water leakage is


Detecting water leakage and reporting the leakage susceptible to contamination if it’s not resolved on
timely is one of the challenges faced by water time. The prime cause of water leakage in most cities
distributing companies in developing nations. This and urban areas is the bottleneck associated with the
paper employs a nature-inspired search method for heavy demands and pressure of water supply for
optimal location search to water leakage sources for both domestic and commercial consumers. Thereby,
Ghana Water and Sewage Services Company. The causing excessive pressure on the maximum
nature-inspired method aid in mapping IoT-edge threshold on the pipelines as prescribed by the
computing devices used by the water and sewage manufacturer [12]. Again, issues associated with
workers, especially in the cities and urban areas, to construction activities such as road constructions,
find and locate leaked pipelines. Once, a water and buildings, underground telecommunication cabling
sewage worker receive a trigger from any location and underground electrical wiring also cause water
that there are leakages, the algorithm generates the leakages in most urban areas. This eventually cause
location data which automatically processes the damage to the underground pipelines causing
distance, geographic location, and direction. In this outbursts of the water pipelines hence, the water
paper, a nature-inspired algorithm based on the leakages [9] [13].
behaviour of a Kestrel bird was used as the mapping One of the interventions to detect water leakage
function. The developed algorithm was tested and is the use of a water leakage detector electronic
evaluated using pre-defined and random locations gadget, which is intended to identify the leakage of
based on their latitudes and longitudinal data. The water and send alert to either the water consumers or
proposed algorithm was evaluated against BAT water company for immediate resolution to avoid
algorithm. Findings from the data indicated that the water loss and contamination. Despite this
proposed KSA used as mapping function, was intervention, most developing nations are unable to
successful in providing the optimal distance after deployed this in their water operation systems partly
receiving water leakage geographic location data due to the cost to deploy those devices. This
from IoT edge computing devices. challenge is the motivation for this paper, and we
proposed a water leakage detector using IoT devices
Keywords: nature-inspired search method, water because of the less cost involved in deploying such
leakage detection, Internet of Things (IoT), Kestrel- devices. Because water leakage is a typical issue that
Based Search Algorithm could lead to the high cost of utility bills, operational
cost and water contamination. Again, finding the
location to the source of the water leakage can
1. Introduction
sometimes be challenging without the use of IoT
Water leakage is one of the challenges facing water
devices and the public. Deploying IoT-edge devices
distribution companies. The more water spilled (or
and location optimization systems are imperative in
sprinkled), the more the cost of its repairs the
water distribution systems. In this paper, we again

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
propose a nature-inspired search method to find the dissemination of data for further processing and
optimal location of water leakage. visualizations via a network wireless sensor.
The contribution of this paper is as follows:
• employ IoT-Edge computing devices for 2.2 Nature-inspired search methods
water leakage detection In recent years, nature-inspired search methods have
• apply nature-inspired search method for the been extensively applied as an approximation
optimal location where water leakage is function because of their ability to avoid the search
detected by IoT-edge device for non-promising results and still gives a global
The remainder of this paper is categorized into the optimal search result. Conceptually, the nature-
following sections: Related work is presented in inspired search/meta-heuristic search method is the
section 2. The proposed water leakage detection application of the social hunting behavior of animals
system is presented in section 3. Section 4 presents in their natural environments [18]. Social hunting is
the nature-inspired search methods. Section 5 seen by how animals share their location
presents the process and stages of the water leakage information and converge to capture their prey. The
system. Section 6 is the push location notification meta-heuristic search method consists of a group of
hub. Section 7 presents the evaluation and search agents that share information about a feasible
discussion of results. The conclusion is presented in region by both randomization and some specified
section 8. rules [21]. The search methods rely on the iterative
evaluation of an objective function to estimate the
2. Related Work next search direction of search agents. Algorithms
2.1 Water leakage systems inspired by the behaviour of social colonies are
Generally, urban water supply infrastructure applied in studying the phenomenology of living
consists of well, pumps, transmission pipes and systems such as planning, learning, making
canals, surface water diversions, treatment and decisions, perception, etc. The advantage of a
storage facilities, and distribution network elements. Nature-inspired technique is the ability to jump out
Urban water management considers the total water of any local search that might not lead to an ideal
cluster formation. Examples of bio-inspired
cycle, facilitates the integration of water factors
early in the land planning process, and encourages techniques are namely Particle Swarm Optimization
all levels of government and industry to adopt water (PSO), Ant Colony Optimization (ACO), Wolf
management and urban planning practices that Search Algorithm (WSA), Social Spider Algorithm
benefit the community, the economy, and the (SSA), Dung beetles, etc.
environment. Globally, managers of water
companies grapple with how to use technologies to A. Swarm Intelligence
provide sustainable water systems [14]. Swarm Intelligence is a nature-inspired method
Previous studies have shed light in the areas of based on swarm behaviour such as fish and bird
water pipeline leakage monitoring applications that schools in nature. The swarm behaviour is expressed
use wireless sensors, Radio Frequency Identification in terms of how particles (e.g., Particle Swarm
(RFID) and WSN technologies [14]. The studies of Optimization) adapt and decide depending on their
[11] and [12] presented a design of a scalable and position within a search space and neighbouring
simulated water leakage monitoring system that particles [1]. The advantage of swarm behaviour is
used Radio Frequency Identification (RFID) model that, as an individual particle decides, it leads to
and Wireless Sensor Networks (WSN) innovations. emergent behaviour, which depends on the local
The studies of [13] also posit that the use of wireless interaction among particles to find a potentially
sensors can be used in the optimization process of optimal solution. Thus, swarm behaviour ensures
water leakage detection. some form of collective intelligence. The author of
Diverse technologies that have been applied in water [2] was known to have proposed a model that
leakage detection included the studies by Karray et mimics the flocking behaviour of birds based on
al. [12] that proposed a “testbed prototype” for the three simplified rules namely collision avoidance
monitoring of water pipeline topologies. The testbed with neighbouring flockmates, velocity matching
prototype used an algorithm called Leak Detection with neighbouring flockmates, and flock centring to
Predictive Kalman Filter (LPKF) for the detection of stay close to the flock [3].
pipeline leakages for continuous and long-distance
pipelines. B. Ant Colony Optimization
Again, Hammoudeh et al. [14] posit a “Linear Ant Colony Optimization (ACO) is a nature-
Wireless Sensor Network (LWSN)” that can be used inspired approach which uses the foraging
for water leakage detection that can be deployed for behaviour of real ants in their search for the shortest
monitoring and surveillance of water supply around paths to food sources. When a source of food is
borders. In [14] the LWSN network is deployed found, real ants deposit a chemical pheromone trail
through mapping services for efficient
to mark their path for other ants to traverse [1].

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
Pheromone trail is an odorous substance that is used constructing social web, spiders provide a habitat
as a medium of indirect communication between that serves as a trap to hunt down prey. Any prey that
ants. The pheromone depends on the distance, is found/trapped on the web causes some vibration
quantity, and quality of the food source. However, which can be detected by the neighbouring spider.
the pheromone trail which decays or evaporates with As an entangled prey tries to detach itself from the
time prevents real ants from converging web it amplifies the intensity and frequency of the
prematurely; thereby real ants can explore different vibration which intends to trigger spiders’
terrains in their habitat. In a situation where an ant is awareness of a prey’s presence on the social web [6].
lost, it moves at random in search of a laid Thus, spiders become more aware of the presence of
pheromone trail and it is likely to follow the path that prey and inform neighbouring spiders to cooperate
reinforces the pheromone trail. Thus, real ants can in hunting.
make probabilistic decisions to update their
pheromone. E. Dung Beetle Algorithm
Dung Beetle Algorithm (DBA) is also a nature-
C. Wolf Search Algorithm inspired algorithm that is based on the behaviour of
Wolf Search Algorithm (WSA), is another nature- dung beetles. The dung beetle is an insect with a
inspired algorithm that mimics the preying very tiny brain (similar to a grain of rice) that feeds
behaviour of wolf, which includes the ability to hunt on the dung of herbivorous animals. The
independently by remembering their trait (indicating characteristics of dung beetle include the ability to
that wolf has memory); only merge with its peer if roll a ball of dung, perform a dance by combining
the peer is in a better position (meaning there is trust internal cue of direction and distance with external
among wolves to never prey on each other); only reference from its environment and then orient
attracts to prey within its visual range; escape themselves using the celestial polarized pattern; and
randomly upon the appearance of a hunter; and the path integration. That is adding all changes in
use of scent marks to demarcate its territory and position and distance to update its direction. The
communicate with another wolf [4]. A variant of dung beetle is known to use minimal computational
WSA is the Wolf Search Algorithm with Minus Step power for navigation and orientation using celestial
Previous (WSA-MP) which indicates a wolf with a polarization patterns [23].
memory to remember the previous best position and
avoid the old positions taken that do not guarantee 3. Proposed Water Leakage Detection
an optimum solution.
System Model
The proposed solution employs IoT device for water
D. Social Spider Algorithm
leakage detection that quickly detects leakages in
Social Spider Algorithm (SSA) is based on the social
pipelines and sends a notification to water
behaviour of the spider [5]. Social spiders are management companies for the problem to be
known to cooperate with their neighbours using their rectified. The structure of the proposed solution is
web building and preying foraging ability. By shown in Figure 1.

Figure 1: Structure of water leakage detection system (Authors, 2020)

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
The water leak detection system consists of a water
sensor, micro-controller, power supply, GSM
module, actuator, LED and SMS text described as
follows:
a. Water Sensor
The sensor yields and simple voltage relative to the
measure of fluid presented to it. The sensor operates
on 5Vand needs less than 20mA for the operating
power current, which means the Arduino can easily
provide the power needed so no external power is
needed to power it.
b. Arduino Microcontroller
It is an open-source hardware and software single-
board microcontroller device. The board design uses Whenever there is a leakage in the water flowing,
a variety of microprocessors and controllers. The location data is sent from the edge device to the
board has Arduino Uno microcontroller which is water company. The algorithm to do the location
programmed in C++. The Arduino is programmed to mapping between the source of water leakage and
receive data from the flow sensor and it has a power the water management company is based on the
supply unit. The Arduino acts as the brain of the proposed nature-inspired search method.
system. The flow meter is utilized to record the
weight of water in the tanks of the pipelines 4. Proposed Nature-inspired search
associating the siphon to the tank.
method for water leakage
c. Wi-Fi Module
The nature-inspired search method to find the
The GSM model is used for communication because
optimal location to the area of water leakage is based
of its affordability and its miniature as compared
on the behavior of a bird, called kestrel [20]. The
with the XBee Wi-Fi module system. GSM module
kestrel bird applies hovering (flight) and perch
reports water leakages. This system consists of an
techniques for hunting in an unknown location. The
actuator installed in the main water pump of a
hover technique gives it a global view of the location
residence and a device based on sensors to detect
of preys whereas the perch mode enables it to locally
water leakages. A microcontroller processes the
view a specific prey for capture. The kestrel-based
readings from the water sensor. If an unusual flow
search algorithm (KSA) is one of the novel's nature-
of water is detected, the system issues a warning
inspired search methods that use three simplified
message by sending a short text message on the
rules namely the improve rule, check rule and reduce
coordinates of the source of leakage to the mobile
rule. The KSA was chosen as a search method
phone of the user or system of the water
because the random encircling, the use of half-life of
management company. An electrical actuator at the
radioactive substances which hitherto is an idea in a
user’s end stops the flow of water so that no further
science discipline such as chemistry, and the
leakage occurs. It has the following advantages:
imitative behavior of another kestrel with a good
• Early leakage detection position have demonstrated to give optimal search
• Alerting users of leakages results when compared with other nature-inspired
• The actuator stops the flow of water to algorithms such as the WSA-MP, BAT, and PSO
prevent further leakage [21]. The novelty lies in the application of the KSA
to water leakage detection, which is our unique
d. Cloud Infrastructure contribution to knowledge such that, we simulated
The data on referencedID and targetID is then stored different location data to evaluate the effectiveness
on the cloud infrastructure for future reference. of the proposed approach. The advantage of the
Table 1 presents the functions of each module of the KSA is its ability to adapt to changes, random
water leakage system. parameterization to control the search for the global
Table 1: Module and function of the water leakage optimal. The KSA is applied to simulate the random
system location of all sources of water leakage and then
maps these locations to the location of the office of
the water company (target location).
The objective function selects the optimal search
position subject to the target location. The fitness
𝑘
objective function 𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗(𝑥𝑖+1 , ) was based
on equirectangular approximation formulations and
haversine formula as performance is imperative for
the proposed algorithm and the selection of the
most/near-optimal solution.

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
The KSA algorithm to find the optimal location
is implemented as follows:

Initially: load location coordinates of all IoT-edge devices 𝑥𝑖 and target 𝑥𝑗


Start: Set model parameters
Algorithm 5.1: Improve rule
initialize 𝑥⃗(𝑡) coordinate (0,0)
Pick location coordinate of IoT-edge device 𝑥𝑖
Pick location coordinate of target 𝑥𝑗
2
Compute 𝛽𝑜 𝑒 −𝛾𝑟 (1)
Find 𝛾 at time t from the reduce rule
𝑘
Compute 𝑓𝑡+1 (2)
𝑘 −𝛾𝑟 2 𝑘
Compute position 𝑥𝑖+1 = 𝑥⃗(𝑡) + 𝛽𝑜 𝑒 (𝑥𝑗 − 𝑥𝑖 ) + 𝑓𝑡+1 (3)
Display results of the optimal position

Algorithm 5.2: Reduce rule


Compute 𝛾𝑡 = 𝛾𝑜 𝑒 −𝜑𝑡 (4)
𝑙𝑛 0.5
Compute 𝜑 = (5)
−𝑡1
2
𝜑 > 1, 𝑡𝑟𝑎𝑖𝑙 𝑖𝑠 𝑛𝑒𝑤
𝑖𝑓 𝜑 → {
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

Algorithm 5.3: Check rule


For t=1 to Max_itr
𝑘
Compute 𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗(𝑥𝑖+1 ) using Algorithm 5.4
𝑖+1
𝒊𝒇 𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗 < 𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗 𝑖
𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗 𝑖 = 𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗 𝑖+1
𝑘
𝑁𝑒𝑤𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 (𝑥𝑖+1 ) = 𝑂𝑙𝑑𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 (𝑥𝑖𝑘 )
End if
𝑘
Update position 𝑥𝑖+1
End for
Output: Display ranked list of location coordinates between targe and IoT-edge device

𝑘
Algorithm 5.4: Fitness objective function 𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑂𝑏𝑗(𝑥𝑖+1 ) = (φ, λ)
//Comments: where φ is change in latitude that is (𝛥𝜑 =φ2-φ1), λ is change in longitude that is (𝛥𝜆 =λ2-λ1), R is
earth’s radius (mean radius R = 6,371km)
For each position: Evaluate
Fitness Function-1 using Equirectangular approximation:
Δφ
𝑥 = ∆λ ∗ cos ( ) (6)
2
𝑦 = ∆φ (7)
𝑑 = 𝑅 ∗ √𝑥 2 + 𝑦 2 (8)

Fitness Function-2 using Haversine formula:


𝑎 = 𝑠𝑖𝑛²(𝛥𝜑/2) + 𝑐𝑜𝑠 𝜑1 ⋅ 𝑐𝑜𝑠 𝜑2 ⋅ 𝑠𝑖𝑛²(𝛥𝜆/2) (9)
𝑐 = 2 ⋅ 𝑎𝑡𝑎𝑛2( √𝑎, √(1 − 𝑎) ) (10)
𝑑 = 𝑅 ⋅ 𝑐 (11)

End for

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
Compute bearing θ = atan2( sin Δλ ⋅ cos φ2, cos φ1 ⋅ sin φ2 − sin φ1 ⋅ cos φ2 ⋅ cos Δλ)
(12)
Output: Distances on Equirectangular approximation and bearing

5. Process and stages of the water leakage location detection


The search process starts after location data is received and push location notification hub has sent alert to the
target. The process is as follows:

Start: IoT-edge device


Process 1: Send notification on water leakage
stage 1: Each IoT-edge device is assigned a referencedID
stage 2: IoT-edge device pick location data
stage 3: Send referencedID with location information to the office of water company
End
Start: Target office location
Process 2: Get target location (water and sewage worker)
stage 1: Each target has office targetID
stage 2: Receives referencedID
stage 3: Run the function on push notification
End

Start: Load data on targetID and referencedID


Process 3: Location mapping
stage 1: Pick data on location coordinate of targetID and referencedID on database
stage 2: Run the KSA algorithm for location mapping
stage 3: Display ranked list of all optima distances of referencedID and targetID
stage 4: Send feedback to the service initiator
End

6. Push Location Notification Hub indicating perched and flight modes parameter
The push notification hub sends alert to k number of respectively as proposed by [8]; and the parameter
targets N to obtain their office location information. to control the frequency of bobbing pa is 0.9; and
The originator of an alert is the IoT-edge device. The half-life is 0.5 [8].
reference location domain P is linked to target Initially, the location data is fed into the
domain in a vector representation 𝑁 = {𝑁1 , … , 𝑁𝑘 } algorithm as vector representation V= (long1, long2,
as expressed in equation 1 by: lat1, lat2) and a single value distance is output.
Where lat1, lat2 refers to latitudes whereas long1,
𝑁1 long2 refers to longitudinal data respectively in
𝑁2 Degrees (D).
(𝑃) ↔ 𝑁3 → (𝑃 ↔ 𝑁 ∗ ) An objective function was used to evaluate the
𝑁4 distance between the random location. Also, the
bearing of the location to each water and sewage
(𝑁5 )
(1) worker was computed. In this study, having obtained
the initial location data, we simulated the proposed
where P represent a source of IoT-edge device, N is algorithm in MATLAB and 100 iterations were
vector representation of the target in k= (1, …, 5). performed where the KSA and other comparative
The results are, P maps to the selected 𝑁 ∗ by a algorithm (BAT) generated random location data in
mapping function longitude and latitude and output a single optimal
distance which is mapped to the distance of the
𝑓(𝑥𝑖𝑘 ) as shown in Figure 2.
water and sewage worker.
Initially, the referencedID was considered at
𝑓(𝑥𝑖𝑘 ) location: lat1 = 45.421527862548828D; long1 = -
P 𝑁∗ 75.697189331054688D; lat2 = 53.64135D; long2 =
-113.59273D. The initial distance was computed as
Figure 2: Structure of mapping function
53,811,205.3525 km and 9,108,062.8679 km
respectively for the equirectangular approximation
7. Evaluation and Discussion of Results and haversine formula. The location data on
The default parameters of the Kestrel-based Search latitudes (north/south) and longitudes (east/west) of
Algorithm (KSA) are zmin=0.2, zmax=0.8

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
the device of a water and sewage worker was The haversine formula and equirectangular
generated randomly using Algorithm 5.1. Algorithm approximation distance value and the respective
5.2 reduces and maps each position to the bearing or direction are computed using each of the
referencedID. nature-inspired algorithms. Table 2 show the results
of the experiment using the KSA.

Table 2: Experimental results on optimal distance using KSA.

targetID Equirectangular Bearing targetID Haversine Bearing


location approximation (D) location distance (km) (D)
N#: distance (km) N#:
N#: 23 785,800.1252 76 N#: 08 8,274,872.6757 131
N#:10 1,631,380.2821 126 N#: 65 11,889,901.2611 316
N#:07 4,101,125.1294 151 N#: 03 9,981,830.3945 335
N#:06 311,076.581 251 N#: 10 12,380,306.9381 126
N#:02 3,871,343.4586 272 N#: 20 8,904,101.4177 359

Table 2 shows the experimental results of the 8,274,872.6757 km at targetID N#:08 and bearing
optimal distance of Equirectangular approximation 131 D. Comparatively, the equirectangular
and Haversine. The distance values in bold approximation distance has the minima distances
represents the optimal distance value that were which imply a minimal location.
produced after each iteration. Equirectangular The results based on the BAT is presented in also
approximation distance for N#:06 was 311,076.581 presented in Tables 3.
km at 251 D while the Haversine distance was

Table 3: Experimental results on optimal distance using BAT.

targetID Equirectangular Bearing targetID Haversine Bearing


location approximation (D) location distance (km) (D)
N#: distance (km) N#:
N#: 34 53,811,205.3525 219 N#: 86 53,811,205.3525 307
N#:100 33,037,103.3038 342 N#: 25 9,108,062.8679 291
N#: 05 49,451,975.0264 276 N#: 33 8,885,754.7348 310
N#: 96 45,249,749.806 181 N#: 07 3,338,038.1134 131
N#: 08 31,823,284.7827 268 N#: 15 9,108,062.8679 291

Table 3 shows the experimental results of the


optimal distance of Equirectangular approximation 8. Conclusion
and Haversine. Equirectangular approximation In this paper, we have proposed a solution for water
distance for N#:08 was 31,823,284.7827 km at 268 leakage detection using IoT-edge device and the
D while the Haversine distance was 3,338,038.1134 nature-inspired search method for optimal location
km at targetID N#:07 and bearing 131 D. identification. The results of the evaluation
Comparatively, the Haversine distance has the demonstrate the capability of KSA over the BAT
minima distance which imply a minimal location. algorithm. The KSA algorithm was seen as a perfect
The experiment outcome between KSA and model for this paper. The application of the
BAT suggests that KSA output a minimal distance algorithm is seen to avert several issues associated
value for the Equirectangular approximation with urban water leakages and unmetered water
whereas BAT output a minimum distance for the supplies to both domestic and commercial users.
Haversine. Again, comparing the output of Thus, the algorithm was able to discover the
Equirectangular approximation and Haversine geographic locations and the optimal distance of
distance value on both KSA and BAT indicates that water leakage within an urban water supply network
KSA produced the minimal value of 311,076.581 topology. Future research could look into the
Km.

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.
implementation of the real-life application for water [10] M. Bowling, “Leakage testing method for a
and sewage companies. plate heat exchanger”. U.S Patent 6,062,068,
May 16, 2000.

[11] A. S. Almazyad, Y. M. Seddiq, A. M. Alotaibi,


A. Y. Al- Nasheri, A. M. Obeid and S. M.
Reference Qasim, “A Proposed Scalable Design and
[1] I. E. Agbehadji, R. Millham, S. Fong, and H.-J. Simulation of Wireless Sensor Network-Based
Hong, "Kestrel-based Search Algorithm (KSA) Long-Distance Water Pipeline Leakage
for parameter tuning unto Long Short Term Monitoring System,” Sensors, vol. 14, no. 2,
Memory (LSTM) Network for feature selection pp. 3557-3577, 2014.
in classification of high-dimensional
bioinformatics datasets," Proceedings of the
Federated Conference on Computer Science [12] F. Karray, A. Garcia-Ortiz, M. W. Jmal, A. M.
and Information Systems pp. 15-20, 2018. Obeid and M. Abid, “EARNPIPE: A Testbed
for Smart Water Pipeline Monitoring Using
[2] C. Reynolds, "Flocks, herds, and schools: a Wireless Sensor Network,” in 20th
distributed behavioural model," Comput International Conference on Knowledge Based
Graph., vol. 21, pp. 25-34, 1987. and Intelligent Information and Engineering,
2016.
[3] N. Siddique and H. Adeli, "Nature inspired
computing: An overview and some future [13] M. Daadoo, A. Eleyan and D. Eleyan, (2017)
directions," Cogn Comput, pp. 706–714, 2015. ‘Optimization water leakage detection using
wireless sensor networks (OWLD)’, ACM
[4] I. E. Agbehadji, R. Millham, S. Fong, and H.-J. International Conference Proceeding Series,
Hong, "Wolf search algorithm for numeric Part F130522(July). doi:
association rule mining," in 2016 IEEE 10.1145/3102304.3102309.
International Conference of Cloud Computing
and Big data analytics (ICCBDA), Chengdu, [14] M. Hammoudeh, F. Al-Fayez, H. Lloyd, R.
China 2016, pp. 146-151. Newman, B. Adebisi, A. Bounceur and A.
Abuarqoub, “A Wireless Sensor Network
[5] J. J. Q. Yua and V. O. K. Li, "A Social Spider Border Monitoring System: Deployment Issues
Algorithm for Global Optimization," Preprint and Routing Protocols,” IEEE Sensors Journal,
submitted to Elsevier, pp. 1-18, 2015. vol. 17, no. 8, 2017.
[6] S. O. Frimpong, I. E. Agbehadji, R. C. Millham,
and J. J. Jung, "Nature-Inspired Search Method [15] I. E. Agbehadji, B. O. Awuzie, A. B. Ngowi, and
for Cost Optimization of Hybrid Renewable R. C. 2020, Millham. Review of Big Data
Energy Generation at the Edge " in 2020 Analytics, Artificial Intelligence and Nature-
International Conference on Artificial inspired Computing Models towards Accurate
Intelligence, Big Data, Computing and Data Detection of COVID-19 Pandemic Cases and
Communication Systems (icABCD), Durban- Contact Tracing. International Journal of
South Africa, 2020, pp. 1-6. Environmental Research and Public Health.
2020: 1-13.
[7] I. E. Agbehadji, R. Millham, T. Surendra, H.
Yang, and H. Addo, "Visualization of [16] Agbehadji, I. E., Millham, R. and Fong, S. 2016.
frequently changed patterns based on the Kestrel-based search algorithm for association
behaviour of dung beetle," presented at the rule mining and classification of frequently
Fourth International Conference on Soft changed items. In: IEEE ed. Proceedings of
Computing in Data Science 2018 (SCDS2018). 2016 8th International Conference on
Communications in Computer and Information Computational Intelligence and
Science, Chulalongkorn University, Bangkok, Communication Networks (CICN). Dehadrun,
Thailand, 2019. India, 2019/12/23. IEEE, 356-360.

[8] I. E. Agbehadji, R. Millham, S. J. Fong, and H.-


J. Hong, "Integration of Kestrel-based search
algorithm with artificial neural network for
feature subset selection " Int. J. Bio-Inspired
Computation, vol. 13, pp. 222-233, 2019.

[9] D. Misiunas, A. R. Simpson and G. Olsson,


“Burst detection and location in water
distribution networks,” 5(3–4), 71– 80.
(2005).,” Water Science and Technology:
Water Supply, vol. 5, no. 3-4, pp. 71-80, 2005.

Authorized licensed use limited to: Miami University Libraries. Downloaded on June 15,2021 at 12:26:28 UTC from IEEE Xplore. Restrictions apply.

You might also like