Cluster Computing

https://doi.org/10.1007/s10586-024-04351-4 (0123456789().,-volV)(0123456789().
,- volV)

A reliable method for data aggregation on the industrial internet

of things using a hybrid optimization algorithm and density
correlation degree
Arash Heidari1 • Houshang Shishehlou2 • Mehdi Darbandi3 • Nima Jafari Navimipour4 • Senay Yalcin5

Received: 30 August 2023 / Revised: 5 February 2024 / Accepted: 7 February 2024

! The Author(s) 2024

Abstract
The Internet of Things (IoT) is a new information technology sector in which each device may receive and distribute data
across a network. Industrial IoT (IIoT) and related areas, such as Industrial Wireless Networks (IWNs), big data, and cloud
computing, have made significant strides recently. Using IIoT requires a reliable and effective data collection system, such
as a spanning tree. Many previous spanning tree algorithms ignore failure and mobility. In such cases, the spanning tree is
broken, making data delivery to the base station difficult. This study proposes an algorithm to construct an optimal
spanning tree by combining an artificial bee colony, genetic operators, and density correlation degree to make suit-
able trees. The trees’ fitness is measured using hop count distances of the devices from the base station, residual energy of
the devices, and their mobility probabilities in this technique. The simulation outcomes highlight the enhanced data
collection reliability achieved by the suggested algorithm when compared to established methods like the Reliable
Spanning Tree (RST) construction algorithm in IIoT and the Hop Count Distance (HCD) based construction algorithm.
This proposed algorithm shows improved reliability across diverse node numbers, considering key parameters including
reliability, energy consumption, displacement probability, and distance.

Keywords Internet of things ! Artificial bee colony ! Genetic operators ! Spanning tree ! Mobility ! Reliability

1 Introduction

The Internet of Things (IoT) is a new paradigm quickly

& Nima Jafari Navimipour
[email protected] gaining traction in today’s wireless communication envi-
ronment [1–5]. This concept emphasizes the pervasiveness
Arash Heidari
[email protected] of objects in our environments, from home appliances and
education to vehicles and transportation [6–9]. Radio-fre-
Mehdi Darbandi
[email protected] quency Identification (RFID) tags, actuators, sensors, and
mobile data are also included in this technology [10–12].
1
Department of Software Engineering, Haliç University, Sensing, connectivity, and item interaction are some key
Istanbul 34060, Turkey
characteristics of IoT [10, 12–14]. The Industrial IoT
2
Department of Computer Engineering, Tabriz Branch, (IIoT) incorporates current Information and Communica-
Islamic Azad University, Tabriz, Iran
tions Technology (ICT) advancements into industrial
3
Pôle Universitaire Léonard de Vinci, Paris, France manufacturing systems [15–17]. It essentially denotes the
4
Department of Computer Engineering, Kadir Has University, subsequently digitized interconnection of industrial manu-
Istanbul, Turkiye and Future Technology Research Center, facturing [18]. Data will be readily merged, correlated,
National Yunlin University of Science and Technology,
compared, and incorporated into IIoT applications like
Douliou, Yunlin 64002, Taiwan
5
smart cities, smart homes, and smart energy services,
Department of Energy System Engineering, School of
notwithstanding the heterogeneity of IIoT devices [19].
Engineering and Natural Sciences, Bahçeşehir University,
Istanbul, Turkey Base stations process data quickly. Each node with

123
 Cluster Computing

resource and energy limitations uses data aggregation from In addition, Zhang, et al. [34] suggested a learning-
different sites [20, 21]. based sparse data reconstruction approach that combines
Also, the fundamental purpose of a data aggregation compressed sensing and deep learning. They aimed to
approach is to optimize energy consumption, network life, minimize the data carried via IoT networks while main-
traffic bottlenecks, and data accuracy by aggregating and taining reconstruction accuracy. A deep CS network was
collecting data packets efficiently [22–24]. Removing created using an end-to-end learning strategy to create a
duplication would reduce energy use and boost energy measurement matrix and an efficient and high-accuracy
security [25, 26]. Data aggregation efficiency is determined reconstruction network. Mo, Ahmed, et al. [35] suggested
by the network design and the magnitude of the data an energy-efficient, secure, and data-aggregated architec-
sensed. Due to the vast amount of sensed data, heavy ture employing blockchain technology for IoT devices to
network connections must be minimized [27, 28]. Thus, a address the security and energy criteria. The proposed
severe issue in the IIoT is the aggregation problem, and it is method used blockchain technology and data correlation
an NP-hard issue [29]. So, this article proposes a method reduction to safeguard IoT networks from fraudulent
for data aggregation in IIoT that uses the artificial bee activity.
cloning algorithm, genetic operators, and density correla- Previous approaches to data aggregation within the IoT
tion degree; it is compared to the Reliable Spanning Tree domain have made commendable strides in enhancing
(RST) construction algorithm, and the results are evalu- security, maximizing energy efficiency, and optimizing
ated. Data aggregation is an essential network technique network performance. However, these methods often face
because it saves energy by reducing data transmission [30]. constraints in accurately computing trust values, adapting
The goal is to create a spanning tree from an IIoT graph swiftly to evolving network conditions, managing heavy
based on residual energy and high reliability. This traffic loads, and requiring substantial data and computa-
approach aims to increase IIoT platforms’ residual energy tional resources. In contrast, the approach proposed in this
and reliability and reduce distance and displacement paper strives to overcome these limitations by leveraging
probability. the artificial bee cloning algorithm and genetic operators.
A summary of recent research related to the planned These techniques are utilized to create and refine spanning
investigation is presented in Section II. The system model trees tailored for data transmission within the IIoT. By
is described in depth in Section 3. The study findings of the harnessing these advanced algorithms, our method
proposed approach are provided in Section 4. Finally, endeavors to elevate reliability, curtail energy consump-
Section V provides a succinct summary of the entire tion, and significantly extend the lifespan of data aggre-
assignment and a conclusion. gation within the IIoT landscape. This approach is poised
to offer a more adaptive and resource-efficient framework
that adeptly navigates the challenges posed by dynamic
2 Related work network environments, thereby fortifying the efficacy of
data aggregation in industrial IoT settings.
IoT data aggregation enables the efficient collection,
analysis, and dissemination of large amounts of data from
multiple sources. By improving data security, reducing 3 Proposed method
energy consumption, and improving network performance,
data aggregation plays a vital role in enabling informed In this section, the proposed method is described in mul-
decision-making and analysis in the IoT domain. So, tiple subsections. The following section explains the
Chandnani and Khairnar [31] presented a trust-based safe aggregation issue and its formal statement. The second
data aggregation approach and an energy-efficient secure subsection describes the system model. The DABCG-IoT
routing protocol for data aggregation and routing in IoT. algorithm, a method to solve the data aggregation problem
Sajedi, et al. [32] introduced F-LEACH, a fuzzy-based using a hybrid algorithm, is presented in the third
data aggregation strategy, for IoT-enabled healthcare subsection.
applications to maximize network longevity. Furthermore,
the membership functions of the fuzzy inference system 3.1 Problem statement
were optimized, and the average of numerous executions
was chosen as the ideal parameter by adjusting the network This section formally addresses the fundamental challenge
scenario. Also, Zhu, et al. [33] developed over-the-air of data aggregation in the IoT context by conceptualizing
computation as a task-oriented method for wireless data IoT devices as constituents of a non-directionally con-
aggregation by smoothly merging communication and nected graph G = (V, E). Here, V embodies a compre-
computing. hensive collection of network nodes, encompassing diverse

123
Cluster Computing

devices and pivotal base stations [36]. Correspondingly, set minimum spanning tree, ensuring efficient data aggregation
E encapsulates the intricate network of links intercon- and resource optimization in the IoT network.
necting these nodes, establishing the crucial communica-
tion framework within the IoT environment [20]. 3.2 System model
V ¼ fv1 ; v1 ; . . .; vn g ð1Þ
This section examines a four-layer IoT model [38]. This
E ¼ fe1 ; e1 ; . . .; em g ð2Þ model has a sensor, network, management, and application
In addition to the network architecture, each edge within layers. A sensor network at the lowest layer includes sen-
this graph is endowed with p distinct positive real numbers, sors, actuators, and tags. IoT devices aim to create a reli-
signifying a multitude of attributes attributed to it. These able and energy-efficient tree in this layer. An extensive
attributes serve as crucial metrics, delineating various network, a mobile communication network, Wi-Fi, Ether-
parameters such as distance, cost, and other pertinent fac- net, etc., can be observed at the network layer. Once the
tors that intricately define the relationship and character- tree is created, data aggregation will be performed using
istics of the interconnected nodes. The determination of this tree. IoT devices send data to the base station through
these attributes is systematically facilitated through the tree created, and the base station sends data to the
Eq. (3), offering a quantifiable method to discern and management layer via the Internet to make the desired
assess the diverse characteristics imbibed within the net- decisions and actions. In the management layer, some
work’s edges [37]. issues appear, such as data analysis, IoT device manage-
! " ment, and security control. The simulation and imple-
W i ¼ W 1i ; W 2i ; . . .; W pi ; i ¼ ð1; 2; . . .; mÞ ð3Þ mentation of the proposed algorithm in the paper are done
Suppose x ¼ ð x1 x2 . . .xm Þ is defined as follows: in this layer.
# The processes for allowing search and discovery for IoT
X i ¼ 1If the ei edge is elected resources, such as devices and objects, are discussed in this
ð4Þ
X i ¼ 0If the ei edge is not selected section. The main phases for developing IoT that can
Creating a Minimum Spanning Tree (MST) for IoT data search for real-time resources are divided into three phases,
aggregation is crucial. This tree optimizes connections each comprising two processes [39]. Step (1) Data prepa-
between IoT devices represented as a graph, reducing ration: Because IoT systems create datasets about objects’
overall communication costs while efficiently collecting statuses and measurements, such datasets, or data streams,
and processing data. The MST offers an organized struc- must be prepped for storage and indexing. IoT and base
ture that minimizes latency and resource usage, ensuring stations collaborate to produce datasets for fog-edge nodes,
efficient data aggregation and enhancing the overall such as cloudlets. Step (2) Indexing: This is where IoT
effectiveness of IoT networks. Therefore, the spanning tree spiders, or crawlers, scan and analyze the provided data
of graph G can be expressed by the vector x. Assuming x is stream regularly, subsequently construct indexes, and plan
the set of all vectors corresponding to the spanning trees in crawling operations. The first level (lower level indexes) is
graph G, the problem of the minimum spanning tree can be built on fog-edge nodes, while the second level (higher
expressed as follows: level indexes) is built on the cloud. Lastly, Step (3)
Xm Searching involves end-users writing queries, which are
minzðxÞ ¼ w x ðx!XÞ
i¼1 i i
ð5Þ then executed by IoT and returned in the form of prioritized
lists of retrieved resources [40].
Accordingly, the data aggregation problem aims to find
First, the construction of the random spanning tree and
the minimum spanning tree of graphs and IoT devices and
the encoding and objective function of the problem are
aggregate data using this tree. In Eq. (5), the objective
described. The steps involved in constructing a random
function zðxÞ represents the heart of the MST problem in
spanning tree entail several key processes. First, multiple
IoT data aggregation. This objective seeks to minimize the
spanning trees are generated from a given graph, and the
sum of weighted edges in the chosen spanning tree, where
tree with the lowest total edge weight is identified as the
each edge’s weight (wi ) captures attributes like distance
minimum spanning tree (MST). These trees, similar in
and cost. Minimizing overall weighted edges is crucial for
vertex count to the original graph, exhibit one less edge and
optimizing data aggregation efficiency in the IoT network.
are devoid of cycles or loops. The proposed algorithm
By constructing a minimum spanning tree, the algorithm
initiates the random generation and enhancement of these
efficiently connects IoT devices, reducing communication
trees, optimizing their topology for effective data aggre-
costs and enhancing resource usage. The objective function
gation and network enhancement in IoT environments.
zð xÞ provides a quantifiable metric, guiding the algorithm
This construction method plays a crucial role in optimizing
to select an optimal edge configuration that forms a
data aggregation and network efficiency within IoT

123
 Cluster Computing

ecosystems. The process facilitates the creation of struc- 0 s representing those outside it, helps visualize the
tured tree topologies, allowing devices to transmit sensitive topology efficiently. Simultaneously, the matrix represen-
data to adjacent nodes and base stations securely. These tation refines this by offering a more detailed depiction,
optimized tree structures substantially reduce energy con- indicating node connections and exclusions within the tree
sumption, enhance network reliability, and prolong the structure. These encoding strategies enable swift evaluation
network’s lifespan. By systematically generating and of spanning trees’ fitness based on specific criteria, such as
improving spanning trees, the algorithm identifies the most displacement probability, energy utilization, and offspring
efficient structures for IoT networks, achieving optimized count. Ultimately, this encoding mechanism aids in iden-
data aggregation and overall network performance. The tifying and selecting the most efficient tree structures in
encoding of the spanning trees using binary arrays and IoT networks, optimizing data aggregation, energy con-
matrices further aids in visualizing and evaluating the fit- sumption, reliability, and network longevity. The number
ness of these structures based on criteria like displacement of graph edges in the figure is 20; an array of 20 entries is
probability, energy utilization, and node offspring count, created, representing an edge of the graph. The elements
ultimately contributing to enhanced network longevity and inside the array entries will be 1 or 0, where 1 indicates that
reliability in IoT environments. Then, the proposed method the edge of the graph is attached to the tree, and 0 shows
for aggregating data in IoT nodes is described in some that the graph’s edge is not attached to the tree. The
steps and detail with an example. Tree structure lets spanning tree is also encoded using the matrix as follows.
devices send sensitive data to neighbors and the base sta- For instance, in Fig. 1b, node N1 is connected to node N2.
tion. We can aggregate data, reduce energy use, improve Therefore, 1 is placed in the corresponding entry in the
reliability, and extend network life with a suitable tree matrix. In contrast, node N1 is not directly related to node
structure. Several spanning trees should be created from the N4; so 0 is placed in the corresponding entry. This process
desired graph; the tree with the lowest total edge weight is is repeated for each entry until the spanning tree is dis-
the minimum spanning tree. The spanning tree can have the played as a matrix, as seen in Fig. 1c.
same number of vertices as the graph and one less edge. In graph and IoT-based networks, once the trees are
Trees should not have loops or cycles. The proposed randomly generated, they need the step numbers from each
algorithm generates and improves trees randomly. Fig- node to the base station and the number of offspring. After
ure 1a demonstrates the spanning tree T from graph G [41]. defining the above criteria for each tree, the objective
Also, generating multiple spanning trees from a graph is function calculates each tree’s fitness. It improves relia-
crucial for optimizing data aggregation and enhancing IoT bility and residual energy. The tk spanning tree will break if
network efficiency. The algorithm focuses on identifying its internal nodes do not work correctly, i.e., if the nodes
the MST among them, minimizing total edge weight, and are more mobile or their energy is dissipated. The evalu-
ensuring a loop-free structure. This approach significantly ation criteria for the fitness of the tk tree are as: Pk, Ek, and
reduces energy consumption, improves network reliability, H k.
and extends network lifespan. The encoding method, using
binary arrays and matrices, plays a vital role in visualizing 3.2.1 Pk
and evaluating spanning tree fitness. The binary array
efficiently represents tree topology, and the matrix offers a It is the total displacement probability of the tree tk and is
detailed view of node connections and exclusions. This calculated as follows [40]:
encoding strategy enables rapid evaluation based on crite- X$ %
ria like displacement probability, energy utilization, and Pk ¼ chki þ 1 pi ð6Þ
ni 2v
offspring count. In summary, generating diverse spanning
trees and employing encoding methods contribute to effi- Pi is the displacement probability of nI and a random
cient data aggregation, energy optimization, and network number in the range [0,1]. The displacement probability of
longevity in IoT environments. the base station is set to zero, which means that this node is
As illustrated in Fig. 1b, some examples of spanning assumed to have no displacement. Also, Chik represents the
trees are made of the desired graph. The encoding method number of node offspring nI in the tree tk.
creates a binary array equal to the number of graph edges. X$ %
Encoding a spanning tree using a binary array and a matrix Ek ¼ chki þ 1 ei ð7Þ
ni 2v
serves the purpose of structuring and optimizing the tree
topology within IoT networks. This encoding method In the proposed network, the optimization technique is
facilitates a systematic representation of graph edges, dis- geared towards maximizing residual energy despite the
tinguishing their attachment to the tree or exclusion from it. challenges posed by diminishing energy levels. This
The binary array, with 1 s denoting edges in the tree and

123
Cluster Computing

Fig. 1 a Constructing a random spanning tree, b The presentation and encoding method, c The matrix encoding, and d The objective function of
the proposed method

technique encompasses a range of strategies, including 3.2.2 Average residual energy Ek

energy-conscious routing to prioritize paths that minimize
energy consumption during data transmission, Calculating Ek , the average residual energy in IoT net-
equitable energy distribution to prevent individual nodes works, gauges the energy left in devices post-data trans-
from exhausting their energy, dynamic power management mission. It assesses network energy efficiency by summing
that adjusts transmission power based on real-time condi- individual node energies (ei ) in tree tk, where each ei ranges
tions, and the implementation of energy-efficient protocols from 1 to 20 J. Ek provides vital insights for optimizing
to minimize unnecessary overhead. Moreover, adaptive energy management in IoT systems.it is the average
network structures are employed to facilitate efficient data residual energy used by devices to transmit aggregated data
routing. By strategically allocating resources, regulating to the base station. Ek is the total residual energy of the tree
power usage, and optimizing network configurations, our tk and is calculated as follows [41]. Also, ei represents the
techniques aim to conserve energy, enhance efficiency, and residual energy of nI and is a random number in the range
ultimately maximize residual energy within our IoT net- [1, 20] joules. The base station energy is assumed to be
work. This ensures prolonged device operation and sustains unlimited. chik represents the number of node offspring nI
the overall network. in the tree tk. So, the parameter ei indicating the energy of

123
 Cluster Computing

each node nI significantly contributes to calculating Ek, the energy dissipation scenarios. Their calculation aids in
average residual energy in IoT networks. Ranging between understanding the tree’s performance and its ability to
1 and 20 J, ei represents the diverse energy levels of remain stable while ensuring efficient energy utilization
individual nodes post-data transmission. This range allows within IoT networks. Fitness as a linear combination of the
for a comprehensive assessment of overall energy effi- above criteria is defined to minimize Pk and H k while Ek
ciency, reflecting varying node capacities and aiding in remains as high as possible. The fitness function serves to
optimizing energy management strategies. holistically evaluate the quality of a solution within IoT
networks by combining criteria Pk , Ek , and H k and network
3.2.3 Hk reliability. Through a weighted linear combination, the
function minimizes displacement probability and hop count
Hop Count Distance (HCD) in the context of the tree tk distance while maximizing residual energy and network
refers to the total number of hops or intermediary nodes reliability, all normalized for fair comparison. This amal-
traversed by data packets as they move from individual gamation aims to assess the stability, energy efficiency, and
nodes within the IoT network to the base station. It’s a reliability of the spanning tree structure. By optimizing
measure of the distance or number of connections between these factors together, the function provides a compre-
a specific node and the base station in the given tree hensive measure of the tree’s performance, facilitating the
structure. It is the total HCD of the tree tk and is calculated identification of solutions that strike a balance between
as [41]: stability, energy efficiency, and overall reliability in IoT
X $ % networks. The fitness function in the proposed method is
Hk ¼ hk
n !v i
ð8Þ
i defined as follows:
& ' & '
hiK indicates the HCD of node nI from the base station in Pk & Pmin Ek & Emin
Fitness ¼ w1 þ w2 1 &
the tree tk. P&max & Pmin ' Emax & Emin
Hk & Hmin
þ w3 þ w4 ðReÞ
3.2.4 Reliability Hmax & Hmin
ð10Þ
In the context of IoT systems, reliability is crucial because
it ensures the stability and continuity of networks, espe- Also, in this equation, Pmin, Pmax, Emin, Emax, Hmin, and
cially when faced with challenges such as device dis- Hmax are used to normalize the expressions to keep them at
placement or energy reduction. This stability is essential the same level. The values Pmin, Pmax, Emin, Emax, Hmin, and
for maintaining uninterrupted communication and func- Hmax are integral to normalizing expressions within the
tionality within the network. This criterion is formally fitness function in the context of optimization. Pmin, Pmax
defined as follows: define the lowest and highest feasible values for displace-
DFF ment probability, while Emin, Emax establish the boundaries
Reliability ¼ ð9Þ for energy efficiency metrics. Similarly, Hmin, and Hmax
TDFF
represent the minimum and maximum achievable values
DFF shows the number of devices that fail due to dis- for hop count distance. By normalizing these expressions
placement or energy dissipation so that the spanning tree is with their respective minimum and maximum limits, the
not split; such failures always occur at leaf nodes. The fitness function ensures a consistent and standardized
TDFF criterion also indicates the total failures. The fol- evaluation across diverse scenarios or datasets within IoT
lowing algorithms investigate the obtained reliability. networks. This normalization process allows fair compar-
ison and optimization, facilitating the identification of
3.2.5 Fitness solutions that strike a balance between reliability, energy
efficiency, and performance in IoT networks. Besides, W1,
The evaluation criteria Pk , Ek , and H k play a crucial role in W2, W3, and W4 are weights related to displacement
assessing the reliability and energy efficiency of the tk probability, residual energy, and the number of steps.
spanning tree in the fitness evaluation. Pk , represents the Figure 1d shows the objective function of the problem.
total displacement probability and is calculated based on After creating the spanning tree and calculating the col-
the displacement probability of individual nodes in the tree. lected values, the criteria of displacement probability,
Ek signifies the average residual energy utilized by devices residual energy, and the distance between the number of
for transmitting aggregated data to the base station, while steps and placement are shown in the objective function.
H k measures the total HCD of the tree from the base sta- Sensing, sending and receiving, and data processing con-
tion. These metrics collectively contribute to the assess- sume energy. Sending packets uses more energy than
ment of the tree’s reliability in the face of displacement or

123
Cluster Computing

receiving. Figure 2 shows a packet-sending and receiving Management, and Application, intertwine synergistically,
radio energy model. The energy required to send a packet collectively facilitating efficient data aggregation and
containing k bits to the d distance to receive k-bit infor- management within the IoT ecosystem.
mation is given in the following equations [41]. Also, the process of searching for IoT resources signif-
ETx ðk; dÞ ¼ ETx&elec ðkÞ þ ETx&amp ðk; dÞ icantly contributes to the overall functionality of IoT sys-
ð11Þ tems by enabling efficient resource discovery and
ETx ðk; dÞ ¼ Eelec ' k þ eamp ' k ' d2
utilization. This procedure, divided into phases of data
ERxðkÞ ¼ ERx&elec ðkÞ preparation, indexing, and searching, streamlines access to
ð12Þ
ERx ðkÞ ¼ Eelec ' k real-time resources within IoT environments. Firstly,
Also, Eelec is the energy required to send or receive a bit through data preparation, datasets containing crucial
of information, and eamp is the energy required to amplify information about object statuses and measurements are
the transmitted signal over a distance. The energy models readied for storage and indexing, allowing for the effective
for packet-sending and receiving in IoT involve calculating utilization of IoT-generated datasets. Subsequently, the
the energy needed for transmission and reception of indexing phase, where IoT spiders or crawlers analyze and
packets. For transmission, it combines electrical and construct indexes from the data streams, facilitates swift
amplified energy for ’k’ bits over distance ‘d’. For the and systematic resource retrieval. These constructed
reception, the energy required for k bits is solely electrical. indexes, spanning from fog-edge nodes to higher-level
Factors like transmission distance, packet size, and specific cloud infrastructure, enhance the accessibility and organi-
energy parameters influence these energy calculations.Eelec zation of available resources. Finally, the searching phase
and eamp related to electrical transmission and signal empowers end-users to execute queries efficiently,
amplification. retrieving prioritized lists of resources, thereby optimizing
So, the four-layer IoT model constitutes the sensor, the utilization of IoT resources for various applications.
network, management, and application layers, each serving This streamlined resource discovery process enhances the
distinct purposes within the system. At the sensor layer, functionality of systems by ensuring quick, targeted access
sensors, actuators, and tags collaborate to establish an to essential resources, thereby facilitating smoother and
energy-efficient and reliable tree structure for data trans- more effective operations [42, 43]. Also, the four-layer IoT
mission. Meanwhile, the network layer oversees diverse model, comprising the sensor, network, management, and
networks such as mobile communication, Wi-Fi, and Eth- application layers, harmonizes a synchronized methodol-
ernet, pivotal in aggregating and transmitting data once the ogy to optimize data aggregation and management within
tree structure is established. IoT devices utilize this struc- the IoT ecosystem. At the sensor layer’s nucleus, the col-
ture to relay data to the base station, which is subsequently laborative synergy of sensors, actuators, and tags is direc-
forwarded to the management layer via the Internet, ted toward the establishment of an energy-efficient and
responsible for decision-making, data analysis, device dependable tree structure for data transmission. This
management, and security control. This layer also serves as structured tree, once instantiated, serves as the cornerstone
the arena for simulating and implementing proposed for efficient data aggregation in the network layer. Integral
algorithms. These layers, namely the Sensor, Network, networks like mobile communication, Wi-Fi, and Ethernet
assume a pivotal role in the aggregation and transmission
of data facilitated by the established tree structure. IoT
devices adeptly exploit this framework to convey data to
the base station, thereby instigating subsequent transmis-
sion to the management layer through the Internet. In the
management layer, the proposed algorithm undergoes rig-
orous simulation and implementation, systematically
addressing pivotal challenges encompassing data analysis,
IoT device management, and security control. This layer
orchestrates decisions and actions, forming the bedrock for
proficient data aggregation and streamlined network per-
formance. The seamless integration of these layers ensures
not only dependable data transmission but also confronts
key challenges associated with data analysis, device man-
agement, and security, thereby amplifying the overall
efficiency and resilience of the IoT ecosystem.
Fig. 2 Transmitter and receiver radio model

123
 Cluster Computing

3.3 Aggregation method choose one to exploit. The greater the fitness, the more
likely it is to be selected by the onlooker bees. Employed
The Aggregation Method, leveraging an artificial bee bees whose solutions have been abandoned become scout
cloning technique, orchestrates the selection of optimal bees. Scout bees search the entire environment randomly.
spanning trees crucial for IIoT data collection. Mirroring These procedures are carried out repeatedly until the ter-
the roles of employed, onlooker, and scout bees, the mination requirement is satisfied. Because the artificial bee
method iterates through solution exploration, information cloning algorithm offers advantages in solving optimiza-
sharing, and random exploration to pinpoint high-fitness tion problems, it is used as the basis for the proposed
solutions. Employed bees actively probe the solution space, algorithm by applying genetic operators in the second
relaying insights to onlooker bees who favor superior phase of the algorithm. For data gathering, the method
solutions based on fitness evaluations. Should solutions fall generates a set of trustworthy spanning trees. First, a tree
short, scout bees venture into random exploration. This with greater fitness is used to collect data. This tree remains
process culminates in the selection of trees whose aggre- usable until one device fails due to energy dissipation or
gated values signify their efficacy in data collection within excessive internal node displacement and failure. The tree
IIoT environments. Through fitness-guided selections, the is split in this scenario, and data gathering is stopped. As a
method ensures reliable data aggregation, seamlessly result, the subsequent high-fitness spanning tree is utilized.
transitioning to alternative high-fitness trees in the event of This process is repeated until all of the trees have been
device failure or malfunction, thereby ensuring robust and utilized. The following are the steps in the suggested
dependable data-gathering strategies. The method employs method:
an artificial bee colony approach for data aggregation in The first step involves the utilization of a clustering
IIoT, using three types of bees: employed, onlooker, and method. The data density correlation degree plays a critical
scout. Employed bees explore the solution space, extract- role in the data aggregation process for sensor nodes. This
ing insights and sharing information about food resources measure, established through two key definitions, delin-
with onlooker bees. Onlooker bees choose the best eates the core sensor nodes and quantifies the correlation
resources based on fitness evaluations and perform a between data, which is essential for effective data aggre-
neighborhood search. If employed bees fail to find new gation. Firstly, the definition of a core sensor node char-
solutions, they transform into scout bees, seeking new acterizes nodes whose data from neighboring sensors
random resources. This process iterates, combining genetic exhibit similarity within specified thresholds, ensuring a
operators to generate new solutions until termination cri- coherent representation of local data clusters. Secondly, the
teria are met. The employed bees actively probe the solu- data density correlation degree for each sensor node is
tion space, while onlooker bees select superior solutions. defined as a weighted function encompassing various fac-
Scout bees explore randomly when solutions fail, ensuring tors like the distance between data points, their distribution,
constant search and adaptation until optimal solutions for and thresholds. This degree evaluates the correlation level
data aggregation are achieved. So, algorithm adjustments within sensor data, ensuring that collected information
are needed to improve the artificial bee cloning method’s represents cohesive patterns while mitigating the influence
efficiency and use it for IIoT data aggregation challenges of irrelevant or disparate data. The mechanism behind this
[44]. Each point in the search space corresponds to a correlation degree ensures that selected sensor nodes pos-
solution that artificial bees can exploit in the cloning sess a cohesive representation of data, which is vital for
technique. The sum of the aggregated values of the tree accurate aggregation and subsequent decision-making [46].
parameters indicates the solution’s fitness. In a bee colony, The data density correlation degree clustering approach is
there are three sorts of bees: onlooker bees, employed bees, described in this part. A sensor is termed the core sensor
and scout bees. The employed bee stays in the current node if the data from a given number of neighbor nodes are
solution and collects information about neighboring solu- similar [47].
tions in its memory. The onlooker bee receives the solution
Definition 1. Core sensor node: Consider that there is a
information from the employed bee and selects one solu-
sensor node v, and it has n neighbor nodes; v1 ; v2 ; :::; vn ; the
tion; the scout bee is responsible for finding the new
data of v; v1 ; v2 ; :::; vn are presented withD; D1 ; D2 ; :::; Dn .
solution. The number of employed bees is the same as the
If there are m data in D1 ; D2 ; :::; Dn whose distances to data
number of onlooker bees, and the number of solutions is
D are less than e and minp B m B n, the sensor v is called
the same as the number of onlooker bees [45].
the core sensor node, where e and minp are the data
Employed bees use their discovered spanning trees and
threshold and amount threshold, respectively.
pass the information to bees using the bee dancing opera-
tor. Onlooker bees assess a spanning tree’s fitness and

123
Cluster Computing

Definition 2. Data density correlation degree: The data step, the solutions produced may lead to loop formation.
density correlation degree of a sensor node v is defined as So, the employed bees repeat this step until the result is the
follows [48]: output of a tree. Because the employed bees explore the
8
< 0; & ' & ' & ' m\minp
whole space of the problem in this phase, a better explo-
imðvÞ ¼ 1 dD d ration will be obtained. For this reason, in the exploration
: a1 1 & þ a2 1 & þ a3 1 & ; m ( minp
expðN & minpÞ e e
phase, employed bees play a crucial role in efficiently
ð13Þ searching the entire problem space for optimal solutions.
where d is the average distance between the m data and the Their random movement towards potential spanning trees
data D, and dD is the distance between the data center of m ensures a broad exploration, increasing the likelihood of
data and the data D. Thea1 , a2 , a3 are weight values discovering high-fitness configurations for robust data
anda1 þ a2 þ a3 ¼ 1. Suppose the data density correlation aggregation in IoT networks. To prevent the formation of
degree of sensor node v is (v) defined by Eq. 13. In that loops, a careful mechanism is in place—if employed bees
case, mðvÞ defined by Eq. 13, then we can obtain the encounter loops in the produced solutions, they repeat the
properties of SimðvÞ as (1) SimðvÞ increases with the process until a loop-free spanning tree is obtained. This
increase ofN, the number of data objects which are in the e- iterative approach ensures the algorithm’s resilience
neighborhood ofD; (2)SimðvÞ increases with the decreases against loop-related challenges, contributing to the gener-
ofdD, the distance between D and the data center of the ation of diverse and reliable spanning trees for effective
data objects which are in e-neighborhood ofD; (3) SimðvÞ data aggregation in IoT networks.
increases with the decreases ofd, the average distance The fourth step involves the exploration of neighbor-
between D and those data objects which are in the e- hoods to discover new food resources or spanning trees
neighborhood ofD; (4)SimðvÞ 2 ½0; 1*. These properties are through the application of genetic operators, searching for
consistent with our intuitiveness. In definition 2, the data neighborhoods, and producing new spanning trees by
threshold e guarantees that SimðvÞ will not be impacted by employed bees. The fourth step comprises four phases. The
unrelated data. The amount threshold minPts is the mini- first phase involves the selection of a current food resource,
mum amount for sensor node v to represent some sensor as well as two random food resources. From these options,
nodes. To illustrate the validity of the data density degree a parent chooses the best food resource, and the parents’
defined by Eq. 13, let two-dimensional data objects for food resources are matched randomly. Four food resources
sensor nodes v0 ; v1 ; v2 ; :::; vn are respec- or spanning trees are selected as parents at this step. For
tivelyD0 ; D1 ; D2 ; :::; Dn . v1 ; v2 ; :::; vn are in the e-neighbor- instance, if the initial population is 100, the first tree is
hood ofv0 . Simðv0Þ is defined by Eq. (13) [47]. selected in the first iteration, the second tree is selected in
the second iteration, and so on to create the current food
The second step consists of the generation of food resource. Also, two food resources or spanning trees will be
resources or primary spanning trees, as well as the ini- chosen randomly, and one food resource will be selected as
tialization process. At this step, food resources or primary the best food resource and the best-spanning tree of the
spanning trees are randomly generated. The initial popu- population. These four food resources or spanning trees are
lation of trees is randomly generated from the desired randomly matched as parent food resources, as shown in
graph; each member of this population and each spanning Fig. 3 [25]. The second phase of the fourth step involves
tree is a possible solution to our optimization problem. The the combination of the parents’ food resources through the
tk tree is modeled using a binary array equal to the number use of a two-point merger operator, resulting in the pro-
of graph edges. To create the tk tree, |v|-1 elements are duction of offspring. Matched food resources are combined
selected randomly from the array and are set to 1; the other using a two-point merger operator to produce up to eight
elements are set to 0. It means that links from the graph offspring’s food resources.
connected in the tree are set to 1, and links that are not As illustrated in Fig. 4, the current food resource is
connected are set to 0. Selected links may form a loop. combined with the random food resource 1; the random
Therefore, the process must be repeated until the tree is food resource 1 is combined with the random food resource
obtained. 2; the random food resource 2 is combined with the best
The third step involves the employed bee phase, during food resource, and the best food resource is combined with
which employed bees move randomly toward food the current food resource using the two-point merge
resources and search for neighborhoods. After producing operator. The merging method is that two positions on
the primary trees, the employed bees randomly move to a binary vectors are randomly selected for a two-point mer-
food resource or tree; after extracting the nectar and ger operator, and then parts of the vectors are exchanged
returning to the hive, they search for a neighbor hoping to between these positions. After the two-point merging
find a tree with higher fitness. Similar to the initializing operation, there will be a repair step in our problem since

123
 Cluster Computing

Fig. 3 Selecting the current food resource, random food resources, and the best food resource as parents

Fig. 4 Merging parents’ food resources randomly with the two-point merger operator and offspring production

the number of edges to create a tree and the number of 1 operator is pivotal for combining parent food resources,
element is essential and should be one unit less than the enhancing solution diversity, and optimizing data aggre-
number of vertices of the graph (|v|-1). If the number of 1 s gation in IoT networks. This operator employs a strategic
is more, the additional 1 s will be removed randomly, and two-point crossover, randomly exchanging segments
if the number of 1 s is less, 1 will be set equal to the between binary vectors representing spanning trees. This
required number of elements. The third phase of the fourth process generates diverse offspring, contributing to the
step involves the application of a mutation operator to the algorithm’s adaptability. The two-point merger operator’s
offspring, resulting in the generation of mutants. Also, significance lies in its ability to explore new neighborhoods
swap is used for the mutation operator. A pair of elements effectively, potentially discovering improved configura-
0 and 1 are randomly selected on the vector, and their tions for data aggregation. By merging genetic information
values are changed. Mutations in 8 offspring are performed from different parents, the algorithm can navigate the
the same way, and 8 mutant offspring are produced, as solution space more efficiently, reinforcing its adaptability
shown in Fig. 5. The fourth phase of the fourth step and overall effectiveness in optimizing IoT network data
involves the calculation of the fitness of the offspring and aggregation.
mutants’ food resources, from which the best food resource The fifth step involves the comparison of the new food
(solution) is selected among the 16 new options based on resource to the current one, with the potential for
their fitness. After applying the two-point merger operators, replacement if it is superior. Additionally, the best solution
using the swap operator, and creating 16 offspring, the is updated during this step. At this step, the best-selected
fitness of each of these offspring is determined by the neighborhood is compared to the current food resource or
population according to Fig. 6. So, in the fourth step of the spanning tree. If it fits better (less) than the current food
artificial bee cloning algorithm, the two-point merger resource, the current food resource will be removed and

123
Cluster Computing

Fig. 5 Applying mutation operator on offspring and producing a mutant generation operator and offspring production

Fig. 6 Calculating the fitness of food resources of offspring and mutants

replaced with a new food resource or spanning tree. provides the other bees with information on the food
Otherwise, the current food resource will remain unchan- resources found. It is performed by bee dancing. By
ged, and one unit will be added to the trial index. Each dancing in the dance area inside the hive, determining the
iteration will also update the population’s best solution or distance of the new food resource from the hive, and
the best-spanning tree. determining the angle of the new food source from the sun,
The sixth step involves the onlooker bee phase, during bees encourage other bees to follow them and go to the
which the movement of onlooker bees toward food valuable food resource they find and extract nectar. This
resources is facilitated through the use of a roulette wheel behavior of bees is simulated using a roulette wheel as
and a neighborhood search. After the employed bee phase follows. First, the fitness and probability of selecting each
is completed, the employed bee returns to the hive and food resource are determined using Eq. (14) and (15) [49]:

123
 Cluster Computing

8
< 1 food source. It means that the nectar and fitness of the
fit ( 0
fitnessi ¼ 1 þ fiti i ð14Þ solution are not appropriate there and are not worth
:
1 þ absðfiti Þfiti \0 searching for. So, the employed bee, whose food source is
abandoned, becomes a scout bee and randomly selects a
fitnessi
Pi ¼ PSN ð15Þ new food resource or solution, replaces the existing food
j¼1 fitnessi resource, and removes the previous one.
The tenth and final step involves the repetition of steps
In Eq. (14), the fitness of each solution is calculated; in
three through nine until the termination criteria have been
Eq. (15), the chance of selecting each food resource is
satisfied. Loop conditions can be the number of loops, the
obtained; it is a number between zero and one, and the sum
number of calls, reaching a specific answer, or the time;
of the probabilities of all the answers is one. In this
here, steps 3 to 9 will be performed as much as iterations
equation, the likelihood of selecting tree i is equal to the
require. Finally, the optimal food source or tree is selected
fitness value of tree i to the total fitness values of all trees
for aggregation, and if the tree experiences failure due to
or solutions. Given that the sum of the probabilities of the
IoT device malfunction or power loss, the next highest-
trees is equal to 1, we map the likelihood of selecting the
ranking trees are utilized as an alternative. So, the algo-
trees to a vector of 1 and finally produce a random number
rithm creates a set of reliable spanning trees for data col-
between 0 and 1. If this number is in any range, that tree is
lection. First, the data collection is built on a tree with
selected. Because our search is performed using a roulette
higher fitness. This tree remains usable until one device
wheel and selection possibilities in the onlooker bee phase,
fails due to energy dissipation or failure. In this case, the
we will have better exploitation; in fact, we will find the
tree is broken, and data collection is terminated. So, the
answer to the problem faster. The reason for using the two
next spanning tree with high fitness is used.
phases of the employed bee and onlooker bee in the arti-
ficial bee colony algorithm is to take advantage of both and
to have an interface between exploration by the employed
bee and rapid exploitation by the onlooker bee. The
4 Simulation of the proposed method
onlooker bee moves toward the current food source after
In this section, the performance of the proposed algorithm
using the roulette wheel. After collecting nectar and
for data aggregation problems is investigated in compar-
returning to the hive, it searches the neighborhood for a
ison to the RST-IoT and Spanning Tree construction
fitter tree. The only difference is that the current food
algorithm based on HCD-IoT [50].
resource is selected using a roulette wheel.
The seventh step involves waiting for information from
the employed bees regarding the nectar yield of various
4.1 Experimental data and simulation
food positions. Step seventh includes 2 phases: (1) Using
parameters
the employed bee’s information, the onlooker bees choose
MATLAB is used to simulate the proposed method. A
food sources and make use of those locations; (2) The scout
computer with Windows 7, an Intel Core i5—2.5 GHz
bees find new random food positions. The eighth step
processor, and 4 GB RAM is used for all the tests.
involves the comparison of the new food resource to the
current one, with the potential for replacement if it is
superior. Additionally, the best solution is updated during
4.2 Checking the best fitness values
this step. At this step, the best-selected neighborhood is
The dimensions of the simulation environment are
compared to the current food resource or spanning tree. If it
120 m 9 120 m, with 20 devices distributed randomly
fits better (less) than the current food resource, the current
over the monitoring area. These devices are assumed to be
food resource will be removed and replaced with a new
heterogeneous, meaning that the initial energies and the
food resource or spanning tree. Otherwise, the current food
resource will remain unchanged, and one unit will be added
to the trial index. Each iteration will also update the pop-
ulation’s best solution or the best-spanning tree. Table 1 Shows the values of the parameters used
The ninth step involves the scout bee phase, during Number of nodes 10 20 30 40
which a random initialization of food resources or spanning
trees occurs if no improvement or limit is observed. At this DABCG-IoT 4.41 7.74 17.87 37.147
step, if the employed bees and the onlooker bees cannot HCT-IoT 5.63 8.99 19.44 38.77
find a new food resource in the neighborhood after several RST-IoT 12.31 15.16 38.23 77.79
visits to it, that food resource is known as an abandoned

123
Cluster Computing

0.11 0.4

HCT_IoT 0.36 HCT_IoT

0.1
RST_IoT RST_IoT
DABCG_IoT 0.32 DABCG_IoT
0.09

0.28
0.08

0.24

Fitness
Fitness

0.07
0.2

0.06
0.16

0.05
0.12

0.04 0.08

0.03 0.04
0 20 40 60 80 100 120 140 160 180 200 0 20 40 60 80 100 120 140 160 180 200

Iteration Iteration

(a) (b)
0.3 0.7

0.28 HCT_IoT
0.6
RST_IoT
0.26 DABCG_IoT
Probability of Displacement

0.5
0.24

0.4
0.22
Fitness

0.2 0.3

0.18 HCT_IoT
0.2
RST_IoT
0.16 DABCG_IoT

0.1
0.14

0.12 0
0 20 40 60 80 100 120 140 160 180 200
0 20 40 60 80 100 120 140 160 180 200
The Number of Nodes
Iteration

(c) (d)

Fig. 7 a Investigating the fitness of a tree with n = 100, b. Investigating the fitness of a tree with n = 200, c Investigating the fitness of a tree with
n = 300, d Investigating the fitness of a tree with n = 40

probabilities of their displacement are not the same [51]. reliability, energy consumption, displacement probability,
Table 1 shows the values of the parameters used. and distance. The initial discussion encompassed the fitness
The comparison between the RST algorithm in IoT and values of constructed trees, which are pivotal for evaluating
the proposed algorithm employing artificial bee cloning algorithm effectiveness. Figure 7a–d depicts the optimum
with genetic operators indicates notable performance dif- fitness values for nodes 100, 200, 300, and 400, respec-
ferences. Conducting twenty independent runs for each tively. These figures starkly showcase the superior prob-
algorithm facilitated a comprehensive evaluation, utilizing lem-solving capabilities of the proposed algorithm,
mean values to assess their efficiency. The examination of highlighting its capacity to seek and produce enhanced
the proposed algorithm focused on multiple parameters: solutions. The anticipated trend of increased execution

123
 Cluster Computing

Table 2 The values of the execution time of the algorithms the pivotal role of reliability, wherein DABCG-IoT emer-
Parameter Value
ges as a frontrunner, offering amplified reliability rates that
can significantly bolster data collection processes.
Network dimensions 120 m 9 120 m Figure 8c provides an insightful view of the reliability
Transmission range 20 m trends within the considered algorithms amidst changes in
Number of employed bees 100 the number of nodes. The impact of IoT network expansion
Number of onlooker bees 100 on reliability is visibly depicted, indicating a reduction in
Number of scout bees 100 dependability with network enlargement. Notably, the
Population 100 analysis indicates that DABCG-IoT showcases superior
W1–W2–W3 0.45–0.45–0.1 reliability, enhancing this aspect by 8% and 49% in com-
Residual energy [1–20] j parison to the RST-IoT and HCT-IoT algorithms, respec-
Displacement probability [0–1]% tively. This underlines the algorithm’s robustness in varied
network scales and reinforces its reliability compared to
other models. In Fig. 8d, the likelihood of displacement
rate concerning the node count is examined. This depiction
time with the expansion of nodes was observed for the accentuates a proportional increase in the displacement rate
desired algorithm. However, the proposed algorithm with the augmentation of nodes within the network.
maintained an acceptable execution time despite this Interestingly, the outcomes emphasize the RST-IoT algo-
escalation. Detailed execution times for both algorithms rithm’s dominance over the proposed approach in this
are documented in Table 2, further illustrating the effi- aspect while concurrently showcasing the proposed meth-
ciency of the proposed algorithm. Further analysis exam- od’s outperformance against the HCT-IoT algorithm. This
ined the average residual energy within an IoT system, nuanced analysis signifies varying performance strengths
quantified as the cumulative sum of all devices’ residual across algorithms concerning displacement rates based on
energy. Figure 8a demonstrates the average residual energy network size, underscoring distinctive capabilities among
for 200 nodes. Remarkably, the suggested approach sur- the considered models. So, the heightened reliability of the
passes the RST-IoT and HCT-IoT algorithms, exhibiting DABCG-IoT algorithm in Fig. 8c can be attributed to its
higher average residual energy, thus implying superior adaptive design, featuring dynamic adjustments to chang-
energy retention capabilities. In conclusion, the extensive ing network sizes. Key to its superior performance is a
evaluation showcases the proposed algorithm’s superiority sophisticated redundancy strategy that mitigates the impact
across diverse parameters, emphasizing its efficiency in of node expansion, ensuring sustained dependability.
generating enhanced solutions and preserving energy Additionally, the algorithm employs intelligent data
within IoT systems. These findings endorse the proposed aggregation, minimizes communication errors, and show-
algorithm’s potential for widespread applications within cases an 8% and 49% improvement over RST-IoT and
IoT frameworks. HCT-IoT, respectively. DABCG-IoT’s ability to maintain
In the realm of IoT, ensuring data collection reliability high reliability across diverse network scales highlights its
stands as a pivotal concern. Through the conducted simu- adaptability and efficiency, making it a robust choice for
lations, a meticulous examination was undertaken, IoT applications.
encompassing DABCG-IoT, RST-IoT, and HCT-IoT, Table 3 provides a detailed exploration of the intricate
specifically focusing on their reliability metrics. Figure 8b relationship between communication delay and energy
was instrumental in illustrating the comparative analysis, consumption in the DABCG-IoT algorithm, assuming a
accentuating changes in the number of failures from 10 to fixed number of nodes (50) and an initial energy con-
50 while considering 20 nodes. This rigorous investigation sumption of 182.0 J for the first run. This initial energy
into reliability revealed compelling insights. The outcomes value establishes a baseline for consumption, and the
distinctly portrayed DABCG-IoT as a superior performer subsequent runs reveal a consistent and progressive
among the assessed algorithms in terms of reliability. It increase in energy requirements as the delay parameter is
showcased a notable enhancement, surpassing the RST-IoT elevated. Such insights are crucial for understanding how
and HCT-IoT algorithms by 6% and 28%, respectively. the algorithm responds to varying communication delays
This improvement in reliability substantiates the robustness and for optimizing its performance in real-world IoT
and dependability of the DABCG-IoT algorithm, posi- deployments. In the early runs (1–5), characterized by
tioning it as a promising solution for ensuring steadfast and relatively lower delays ranging from 5 to 25 ms, the energy
consistent data collection within IoT frameworks. These consumption experiences a gradual and moderate increase.
findings significantly contribute to the understanding and This indicates a discernible sensitivity to lower delays,
evaluation of algorithms in IoT environments, highlighting suggesting that the algorithm is responsive to the initial

123
Cluster Computing

1600 1

1400 HCT_IoT
RST_IoT
DABCG_IoT 0.9
1200
Average Residual Energy (J)

Average Reliability (%)

1000
0.8

800

0.7
600

400
0.6
HCT_IoT
200 RST_IoT
DABCG_IoT

0 0.5
0 20 40 60 80 100 120 140 160 180 200 5 10 15 20 25 30 35 40 45 50
The Number of Nodes The Number of Faild Devices

(a) (b)

1 0.36

0.32 HCT_IoT
0.9
RST_IoT
DABCG_IoT
0.28
0.8
Average Reliability (%)

0.24
0.7
Fitness

0.2

0.6
0.16

0.5
0.12

HCT_IoT
0.4
RST_IoT 0.08
DABCG_IoT

0.3 0.04
0 20 40 60 80 100 120 140 160 180 200
0 20 40 60 80 100 120 140 160 180 200
The Number of Nodes
Iteration
(c) (d)
Fig. 8 a Investigation of the average residual energy against the nodes. d Investigating the displacement rate in the suggested
number of nodes. b. Investigating reliability against the number of approach and comparing it to the other two methods
broken nodes. c Investigating the reliability against the number of

stages of communication latency. As the delay surpasses optimization in real-world IoT deployments. The observed
25 ms (runs 6–15), the rate of increase in energy con- trends in the table underscore the importance of finding a
sumption becomes more pronounced, highlighting the balance between communication delays and energy con-
algorithm’s heightened sensitivity to moderate to high sumption in the design and deployment of the DABCG-IoT
communication delays. This underscores the critical trade- algorithm. Designers and implementers must consider
off between communication delay and energy consump- these trade-offs when aiming to achieve efficient data
tion, emphasizing the need for careful consideration and aggregation while navigating the constraints imposed by

123
 Cluster Computing

Table 3 The correlation between delay and energy use consumption to encompass factors like reliability and
Run Delay (ms) Energy consumption (J)
execution time in dynamic IoT environments.

1 5 182
2 10 202.2 5 Conclusion and future work
3 15 225.4
4 20 252 In this paper, we proposed a method for data aggregation in
5 25 282.2 IIoT that involves the use of the artificial bee cloning
6 30 316.3 algorithm and genetic operators to generate a set of reliable
7 35 354.1 spanning trees for data gathering. The method involves
8 40 396.7 several steps, including clustering nodes based on the data
9 45 444.1 density correlation degree, initializing and producing pri-
10 50 496.5 mary spanning trees, and employing bees to search for
11 55 554 neighborhoods and find new spanning trees using genetic
12 60 616.7 operators. Onlooker bees use a roulette wheel to select food
13 65 684.9 resources and search for neighborhoods, and scout bees are
14 70 758.5 used to find new random food resources if no improvement
15 75 837.4 is found. The process is repeated until termination criteria
are met, and the best-spanning tree is selected for data
aggregation. If the tree fails due to device failure or power
dissipation, the next best tree is used. This method is
designed to improve the efficiency of data aggregation in
communication delays. Future work could explore opti- IIoT by using the artificial bee cloning algorithm and
mization strategies, such as adaptive algorithms or dynamic genetic operators to generate and improve spanning trees. It
parameter adjustments, to enhance the algorithm’s resi- aims to improve reliability, reduce energy consumption,
lience in varying network conditions and maintain energy and extend the network lifespan by creating a suitable tree
efficiency. It is important to note that while the table pro- structure for data transmission. Some benefits of the pro-
vides valuable insights, a more rigorous analysis involving posed method include its ability to find good solutions to
a larger dataset and statistical measures would further the optimization problem efficiently and effectively, its
enhance the robustness of the findings. Additionally, con- ability to search for and explore a wide range of possible
siderations for real-world deployment should take into solutions, and its ability to adapt and improve solutions
account other factors, such as reliability and execution over time through the use of genetic operators. The pro-
time, providing a more comprehensive evaluation of the posed method’s effectiveness depends on termination cri-
algorithm’s overall performance. In conclusion, the teria and static network assumptions, limiting adaptability
detailed exploration of delay and energy consumption and scalability. Reliance on immediate backup trees affects
contributes essential knowledge for the continued devel- fault tolerance, while algorithm choice and parameter
opment and optimization of IoT algorithms, ensuring their tuning impact robustness across IoT scenarios. Integrating
effectiveness in dynamic and diverse IoT environments. learning techniques lacks exploration of complexities like
So, Table 3 details the nuanced relationship between model compatibility and data issues.
communication delay and energy consumption in the As a future challenge, this issue can be solved by other
DABCG-IoT algorithm. The observed progressive increase evolutionary algorithms, such as particle swarm optimiza-
in energy consumption with rising communication delays, tion [52] and multi-objective bat algorithm [53] with the
especially in the moderate to high range (runs 6–15), genetic or colonial competition algorithm. One potential
emphasizes the algorithm’s sensitivity. Early runs (1–5) avenue for further research would be to integrate the pro-
show responsiveness to lower delays, unveiling a crucial posed method with machine/deep learning techniques [54]
trade-off between communication delay and energy effi- to enhance its effectiveness in aggregating data in IoT
ciency. Optimization for real-world IoT deployments networks.
necessitates a delicate balance. Future exploration into
adaptive algorithms or dynamic parameter adjustments
may enhance adaptability to variable network conditions. Author contributions Arash Heidari: Conceptualization, Visualiza-
While insightful, the findings could benefit from a more tion, Funding acquisition, Formal analysis, Methodology, Writing –
original draft, Writing – review & editing. Houshang Shishehlou:
extensive dataset and statistical scrutiny. A holistic evalu-
Conceptualization, Visualization, Funding acquisition, Formal anal-
ation should extend considerations beyond energy ysis, Methodology, Writing – original draft, Writing – review &

123
Cluster Computing

editing.Mehdi Darbandi: Simulation, Visualization, Formal analysis, 10. Li, K., Ni, W., Noor, A., Guizani, M.: Employing Intelligent
Methodology, Writing – original draft, Writing – review & edit- aerial data aggregators for the internet of things: challenges and
ing.Nima Jafari Navimipour: Conceptualization, Visualization, solutions. IEEE Int. Things Magaz. 5(1), 136–141 (2022)
Funding acquisition, Formal analysis, Methodology, Writing – orig- 11. Xu, X., Liu, W., Yu, L.: Trajectory prediction for heterogeneous
inal draft, Writing – review & editing.Senay Yalcin: Conceptualiza- traffic-agents using knowledge correction data-driven model.
tion, Visualization, Funding acquisition, Formal analysis, Inform. Sci. 608, 375–391 (2022). https://doi.org/10.1016/j.ins.
Methodology, Writing – original draft, Writing – review & editing. 2022.06.073
12. Cao, B., Zhao, J., Lv, Z., Yang, P.: Diversified personalized
Funding Open access funding provided by the Scientific and Tech- recommendation optimization based on mobile data. IEEE Trans.
nological Research Council of Türkiye (TÜBİTAK). Intell. Transp. Syst. 22(4), 2133–2139 (2020)
13. Ding, Y., Zhang, W., Zhou, X., Liao, Q., Luo, Q., Ni, L.M.:
FraudTrip: taxi fraudulent trip detection from corresponding
Declarations trajectories. IEEE Int. Things J. 8(16), 12505–12517 (2020)
14. Dai, W., Zhou, X., Li, D., Zhu, S., Wang, X.: Hybrid parallel
stochastic configuration networks for industrial data analytics.
Competing interests The authors declare no competing interests.
IEEE Trans. Industr. Inf. 18(4), 2331–2341 (2022). https://doi.
org/10.1109/TII.2021.3096840
Open Access This article is licensed under a Creative Commons
15. Kalapaaking, A.P., Khalil, I., Rahman, M.S., Atiquzzaman, M.,
Attribution 4.0 International License, which permits use, sharing,
Yi, X., Almashor, M.: Blockchain-based federated learning with
adaptation, distribution and reproduction in any medium or format, as
secure aggregation in trusted execution environment for internet-
long as you give appropriate credit to the original author(s) and the
of-things. IEEE Trans. Ind. Inform. 19(2), 1703–1714 (2022)
source, provide a link to the Creative Commons licence, and indicate
16. Dai, X., et al.: Task co-offloading for D2D-Assisted mobile edge
if changes were made. The images or other third party material in this
computing in industrial internet of things. IEEE Trans. Industr.
article are included in the article’s Creative Commons licence, unless
Inf. 19(1), 480–490 (2023). https://doi.org/10.1109/TII.2022.
indicated otherwise in a credit line to the material. If material is not
3158974
included in the article’s Creative Commons licence and your intended
17. Cao, B., Zhao, J., Gu, Y., Fan, S., Yang, P.: Security-aware
use is not permitted by statutory regulation or exceeds the permitted
industrial wireless sensor network deployment optimization.
use, you will need to obtain permission directly from the copyright
IEEE Trans. Industr. Inf. 16(8), 5309–5316 (2019)
holder. To view a copy of this licence, visit http://creativecommons.
18. Wang, J., Chen, J., Xiong, N., Alfarraj, O., Tolba, A., Ren, Y.:
org/licenses/by/4.0/.
S-BDS: An effective blockchain-based data storage scheme in
zero-trust IoT. ACM Trans. Int. Technol. 23(3), 1–23 (2022)
19. Cassel, G.A.S., Rodrigues, V.F., da Rosa Righi, R., Bez, M.R.,
References Nepomuceno, A.C., da Costa, C.A.: Serverless computing for
internet of things: a systematic literature review. Future Gener
1. Zou, W., et al.: Limited sensing and deep data mining: A new Comput Syst 128, 299–316 (2022)
exploration of developing city-wide parking guidance systems. 20. Wang, X., Garg, S., Lin, H., Kaddoum, G., Hu, J., Alhamid, M.F.:
IEEE Intell. Transp. Syst. Mag. 14(1), 198–215 (2020) An intelligent uav based data aggregation algorithm for 5g-en-
2. Cheng, B., Wang, M., Zhao, S., Zhai, Z., Zhu, D., Chen, J.: abled internet of things. Comput. Netw. 185, 107628 (2021)
Situation-aware dynamic service coordination in an IoT envi- 21. Sun, G., Li, Y., Liao, D., Chang, V.: Service function chain
ronment. IEEE/ACM Trans. Netw. 25(4), 2082–2095 (2017) orchestration across multiple domains: a full mesh aggregation
3. Lyu, T., Xu, H., Zhang, L., Han, Z.: Source selection and approach. IEEE Trans. Netw. Serv. Manage. 15(3), 1175–1191
resource allocation in wireless powered relay networks: an (2018). https://doi.org/10.1109/TNSM.2018.2861717
adaptive dynamic programming based approach. IEEE Int. 22. Dan, J., Zheng, Y., Hu, J.: Research on sports training model
Things J. 11(5), 8973–8988 (2023) based on intelligent data aggregation processing in internet of
4. Jiang, Y., Li, X.: Broadband cancellation method in an adaptive things. Clust. Comput. 25(1), 727–734 (2021)
co-site interference cancellation system. Int. J. Electron. 109(5), 23. R. K. Verma, S. Bharti, and K. K. Pattanaik, ‘‘GDA: Gravita-
854–874 (2022) tional data aggregation mechanism for periodic wireless sensor
5. Cao, B., et al.: Multiobjective 3-D topology optimization of next- networks,’’ in 2018 IEEE sensors, 2018: IEEE, pp. 1–4.
generation wireless data center network. IEEE Trans. Industr. Inf. 24. Zhang, H., Luo, G., Li, J., Wang, F.Y.: C2FDA: coarse-to-fine
16(5), 3597–3605 (2019) domain adaptation for traffic object detection. IEEE Trans. Intell.
6. Sun, G., Sheng, L., Luo, L., Yu, H.: Game theoretic approach for Transp. Syst. 23(8), 12633–12647 (2022). https://doi.org/10.
multipriority data transmission in 5G vehicular networks. IEEE 1109/TITS.2021.3115823
Trans. Intell. Transp. Syst. 23(12), 24672–24685 (2022). https:// 25. Dhand, G., Tyagi, S.S.: Data aggregation techniques in WSN:-
doi.org/10.1109/TITS.2022.3198046 survey. Procedia Comput. Sci. 92, 378–384 (2016). https://doi.
7. Sun, G., Zhang, Y., Liao, D., Yu, H., Du, X., Guizani, M.: Bus- org/10.1016/j.procs.2016.07.393
trajectory-based street-centric routing for message delivery in 26. Mohapatra, H., Mohanta, B.K., Nikoo, M.R., Daneshmand, M.,
urban vehicular Ad Hoc networks. IEEE Trans. Veh. Technol. Gandomi, A.H.: MCDM-based routing for IoT-enabled smart
67(8), 7550–7563 (2018). https://doi.org/10.1109/TVT.2018. water distribution network. IEEE Int. Things J. 10(5), 4271–4280
2828651 (2022)
8. Luo, J., Zhao, C., Chen, Q., Li, G.: Using deep belief network to 27. Xie, F., Ye, X.H.: Endada: an efficient network design algorithm
construct the agricultural information system based on Internet of based on weighted graph for data aggregation in internet of things
Things. J. Supercomput. 78(1), 379–405 (2022) on marine ships. Appl. Mech. Mater. 740, 648–651 (2015).
9. Lu, J., Osorio, C.: On the analytical probabilistic modeling of https://doi.org/10.4028/www.scientific.net/AMM.740.648
flow transmission across nodes in transportation networks. 28. Mohapatra, H., Rath, A.K.: A fault tolerant routing scheme for
Transp. Res. Rec. 2676(12), 209–225 (2022) advanced metering infrastructure: an approach towards smart
grid. Clust. Comput. 24(3), 2193–2211 (2021)

123
 Cluster Computing

29. Fu, X., Yang, Y.: Modeling and analyzing cascading failures for 48. Y. Gao, X. Li, J. Li, and Y. Gao, ‘‘A Trustworthy Data Aggre-
Internet of Things. Inf. Sci. 545, 753–770 (2021) gation Model Based on Context and Data Density Correlation
30. Harb, H., Makhoul, A., Laiymani, D., Jaber, A.: A distance-based Degree,’’ presented at the Proceedings of the 21st ACM Inter-
data aggregation technique for periodic sensor networks. ACM national Conference on Modeling, Analysis and Simulation of
Trans. Sens. Netw. (TOSN) 13(4), 1–40 (2017) Wireless and Mobile Systems, Montreal, QC, Canada, 2018.
31. Chandnani, N., Khairnar, C.N.: Bio-Inspired multilevel security [Online]. Available: https://doi.org/10.1145/3242102.3242127.
protocol for data aggregation and routing in IoT WSNs. Mobile 49. Ozturk, C., Hancer, E., Karaboga, D.: A novel binary artificial
Netw. Appl. 27(3), 1030–1049 (2022) bee colony algorithm based on genetic operators. Inf. Sci. 297,
32. Sajedi, S.N., Maadani, M., Nesari Moghadam, M.: F-LEACH: a 154–170 (2015)
fuzzy-based data aggregation scheme for healthcare IoT systems. 50. Najjar-Ghabel, S., Yousefi, S., Farzinvash, L.: Reliable data
J. Supercomput. 78(1), 1030–1047 (2022) gathering in the Internet of Things using artificial bee colony.
33. Zhu, G., Xu, J., Huang, K., Cui, S.: Over-the-air computing for Turk. J. Electr. Eng. Comput. Sci. 26(4), 1710–1723 (2018)
wireless data aggregation in massive IoT. IEEE Wirel. Commun. 51. Chen, Y., Zhu, L., Hu, Z., Chen, S., Zheng, X.: Risk propagation
28(4), 57–65 (2021) in multilayer heterogeneous network of coupled system of large
34. Zhang, M., Zhang, H., Yuan, D., Zhang, M.: Learning-based engineering project. J. Manag. Eng. 38(3), 04022003 (2022)
sparse data reconstruction for compressed data aggregation in IoT 52. Xu, X., Lin, Z., Li, X., Shang, C., Shen, Q.: Multi-objective
networks. IEEE Int. Things J. 8(14), 11732–11742 (2021) robust optimisation model for MDVRPLS in refined oil distri-
35. Ahmed, A., Abdullah, S., Bukhsh, M., Ahmad, I., Mushtaq, Z.: bution. Int. J. Prod. Res. 60(22), 6772–6792 (2022). https://doi.
An energy-efficient data aggregation mechanism for IoT secured org/10.1080/00207543.2021.1887534
by blockchain. IEEE Access 10, 11404–11419 (2022) 53. Xiao, Z., et al.: Multi-objective parallel task offloading and
36. Jabraeil Jamali, M.A., et al.: The IoT Landscape. In: Jamali, content caching in D2D-aided MEC networks. IEEE Trans. Mob.
M.A.J., Bahrami, B., Heidari, A., Allahverdizadeh, P., Norouzi, Comput. 22(11), 6599–6615 (2023). https://doi.org/10.1109/
F. (eds.) Towards the Internet of Things: Architectures, Security, TMC.2022.3199876
and Applications. Springer, Cham (2020) 54. Wang, Q., Dai, W., Zhang, C., Zhu, J., Ma, X.: A compact
37. Krishnasamy, L., Dhanaraj, R.K., Ganesh Gopal, D., Reddy constraint incremental method for random weight networks and
Gadekallu, T., Aboudaif, M.K., Abouel Nasr, E.: A heuristic its application. IEEE Trans. Neural Netw. Learn. Syst. (2023).
angular clustering framework for secured statistical data aggre- https://doi.org/10.1109/TNNLS.2023.3289798
gation in sensor networks. Sensors 20(17), 4937 (2020)
38. Da Xu, L., He, W., Li, S.: Internet of things in industries: a Publisher’s Note Springer Nature remains neutral with regard to
survey. IEEE Trans. Industr. Inf. 10(4), 2233–2243 (2014) jurisdictional claims in published maps and institutional affiliations.
39. Sivakumar, N.R., Nagarajan, S.M., Devarajan, G.G., Pullagura,
L., Mahapatra, R.P.: Enhancing network lifespan in wireless
sensor networks using deep learning based Graph Neural Net-
work. Phys. Commun. 59, 102076 (2023) Arash Heidari is a visiting
40. Younan, M., Elhoseny, M., Ali, A.E.-M.A., Houssein, E.H.: Data researcher at Halic University in
Reduction model for balancing indexing and securing resources Turkey, specializing in Com-
in the internet-of-things applications. IEEE Int. Things J. 8(7), puter Engineering. With a
5953–5972 (2020) strong academic foundation and
41. Sreedevi, P., Venkateswarlu, S.: An Efficient Intra-Cluster Data a passion for research, he has
Aggregation and finding the Best Sink location in WSN using made significant contributions
EEC-MA-PSOGA approach. Int. J. Commun. Syst. 35(8), e5110 as an author and co-author, with
(2022) numerous publications in
42. Selvaraj, A., Patan, R., Gandomi, A.H., Deverajan, G.G., Push- respected technical journals,
paraj, M.: Optimal virtual machine selection for anomaly detec- conferences, and authoritative
tion using a swarm intelligence approach. Appl. Soft Comput. 84, technical books. His commit-
105686 (2019) ment to advancing research is
43. Y. Mao, Y. Zhu, Z. Tang, and Z. Chen, ‘‘A Novel Airspace evident through his roles as a
Planning Algorithm for Cooperative Target Localization,’’ guest editor for prestigious
Electronics, vol. 11, no. 18, p. 2950, 2022. [Online]. Available: journals like ‘‘Sustainability’’ and ‘‘Symmetry’’ by MDPI, as well as
https://www.mdpi.com/2079-9292/11/18/2950. ‘‘Cluster Computing’’ and ‘‘Grid Computing’’ by Springer. He also
44. Gopal, D.G., Saravanan, R.: Selfish node detection based on serves as an associate editor for ‘‘SageOpen,’’ where he actively
evidence by trust authority and selfish replica allocation in contributes to shaping the landscape of cutting-edge research. In
DANET. Int. J. Inf. Commun. Technol. 9(4), 473–491 (2016) addition to his academic pursuits, his entrepreneurial spirit has led to
45. Manuel, A.J., Deverajan, G.G., Patan, R., Gandomi, A.H.: Opti- the successful establishment and co-founding of several thriving
mization of routing-based clustering approaches in wireless Iranian start-ups, showcasing his innovative abilities and leadership.
sensor network: review and open research issues. Electronics He actively participates in the academic community as a diligent
9(10), 1630 (2020) reviewer for esteemed journals, ensuring the quality and rigor of
46. Zheng, C., et al.: Knowledge-based engineering approach for research in his field. His research interests encompass a wide range of
defining robotic manufacturing system architectures. Int. J. Prod. areas, including IoT, Computer Vision, Distributed Computing, Deep
Res. 61(5), 1436–1454 (2023). https://doi.org/10.1080/00207543. Learning, Explainable AI, Blockchain, and NLP. He is a proud
2022.2037025 member of professional organizations like IEEE and ACM, under-
47. Yuan, F., Zhan, Y., Wang, Y.: Data density correlation degree scoring his unwavering commitment to the field of computer engi-
clustering method for data aggregation in WSN. IEEE Sens. J. neering. His affiliation with Iran’s National Elites Foundation (INEF)
14(4), 1089–1098 (2014). https://doi.org/10.1109/JSEN.2013. reflects his dedication to nurturing and supporting exceptionally tal-
2293093 ented individuals across various domains. He has also demonstrated
exceptional leadership in fostering interdisciplinary research

123
Cluster Computing

collaboration. He has played a pivotal role in establishing and leading ranked journals. He has been giving invited tutorials/talks in IEEE
interdisciplinary research teams, bridging diverse academic back- conferences and has been invited to give lectures in different uni-
grounds. His well-rounded expertise, spanning academia, versities. He also won the Publons Top Peer Review Awards in 2018
entrepreneurship, and active community engagement, positions him and 2019. He has been featured among the World’s Top 2% Scientists
as a valuable contributor to the field of computer engineering and its List, according to a conducted study by US-based Stanford University
diverse interdisciplinary applications. in 2020. He is also a senior member of IEEE, IEEE Communications
Society and IEEE Young Professionals. His research interests include
Houshang Shishehlou received green computing, cloud and distributed computing, Internet of Things
his B.S. in computer software (IoT), Software-Defined Networking (SDN), information systems,
engineering, from Applied Sci- computational intelligence, evolutionary computing, and quantum
ence University, Tabriz, Iran, in computing. He has published many papers in various journals and
2013. The M.S. in information conference proceedings, as well as supervising/co-supervising several
technology, computer networks, Ph.D. and Master’s students in these research areas.
from Tabriz Branch, Islamic
Azad University, Tabriz, Iran, in Senay Yalcin is a professor at
2019. The title of his thesis for Bahçeşehir University, Istanbul,
the Master of Science degree Turkey. He received his under-
was ‘‘A method for data aggre- graduate degree from the
gation in the internet of things Department of Physics and
using artificial bee colony and Mathematics at Ege University.
genetic operators.’’ He is work- He completed his master’s
ing in web programming, SEO degree in Nuclear Sciences and
and digital marketing now. His research interests include cloud his doctorate in Nuclear Sci-
computing, Internet of Things (IoT), information systems, artificial ences at Ege University. He was
intelligence and wireless sensor network. a postdoctoral research fellow in
the field of hydrogen energy at
Mehdi Darbandi got his M.Sc. in the University of Miami Clean
Electrical Engineering from Energy Center. He worked as a
Eastern Mediterranean Univer- researcher on photoelectron
sity in 2019. At the moment he spectrography at Montana State University. He served as the Head of
is studying M.Sc. of Interna- the Department of Mathematics and Computer Sciences at Bahçeşehir
tional Business at Pôle Univer- University. His research focuses on nuclear energy, geometric aug-
sitaire Léonard de Vinci. His mented product codes, separable codes, soft calculations, Monte
research interests include Cloud Carlo Computation in nuclear reactors and neuro-fuzzy interference
Computing, Estimation and system. He is the editor of various journals.
Prediction, and IoT.

Nima Jafari Navimipour received

his B.Sc., M.Sc., and Ph.D.
degrees in computer engineer-
ing from IAU, Iran, in 2008,
2009, and 2014, respectively.
He is a technical committee
member, guest editor, and
associate editor of some high-
ranked journals such as IET
Quantum Communication,
Optik, Journal of Management
& Organization, Computer
Communication, Cluster Com-
puting, and Kybernetes. Fur-
thermore, he is a chair member
of many prestigious conferences and a reviewer of several high-

123