https://doi.org/10.31449/inf.v46i3.3943                                                                                      Informatica 46 (2022) 333-342     333 
Chaotic Association Feature Extraction of Big Data Clustering Based 
on the Internet of Things  
 
Xiaoming Liu
1*
, Thipendra Pal Singh
2
, Rajeev Kumar Gupta
3
, Edeh Michael Onyema
4 
1
JingZhou Vocational College of Technology, Software Engineering Institute, Jingzhou, Hubei, 434000, China 
2
School of Computer Science, University of Petroleum and Energy Studies, Dehradun, India 
3
Pandit Deendayal Energy University, Gandhinagar, India 
4
Department of Mathematics and Computer Science,  Coal City University, Enugu, Nigeria 
Emails: xiaomingliu7@126.com, thipendra@gmail.com, rajeev.gupta@sot.pdpu.ac.in, michael.edeh@ccu.edu.ng 
 
Keywords: Internet of Things; Big data; Clustering; Chaotic correlation; Feature extraction. 
 
Received: January 26, 2022 
 
This article addresses the stabilization of chaotic characteristics in abnormal data by proposing chaotic 
correlation feature extraction of big data clustering based on the Internet of things. The chaotic features 
in big data usually show complex folding and distortion without obvious rules and order and non-
synchronization. In this article, the dimension of extracted correlation is utilized as the chaotic feature 
for the clustering of big data. The one-dimensional time series that can be extended in multi-dimensional 
space is analysed based on phase space reconstruction, to extract the chaotic correlation dimension 
(CCD) features. After the relevant experimental analysis, this paper mainly compares the energy 
consumption and processing time of the two respective algorithms. In the simulation parameter design, 
the time interval of big data packet generation is 0.1s, and the data is generated from the simulation 
time of 300s. The results obtained show that when dealing with the same amount of data, the energy 
consumption of this algorithm is significantly lower than that of the traditional algorithm. When dealing 
with the same amount of data, the time required by this algorithm is significantly lower than that of the 
traditional algorithm. This is because this algorithm is easy to implement and has good clustering 
efficiency for data, so the clustering time is short. With the gradual increase in the amount of data, the 
correlation dimension of this algorithm tends to be stable. While the correlation dimension of the 
traditional algorithm fluctuates greatly, it is revealed that the proposed approach has high data 
clustering efficiency and verifies the effectiveness of this algorithm. 
Povzetek: Za internet stvari je analizirana možnost stabilizacije nenavadnih podatkov znotraj velikih 
podatkov. 
 
 
1 Introduction  
With the rapid expansion of network technology, 
the network crime activities in the big data environment 
are gradually increasing, increasing the amount of 
abnormal data in the environment of huge data [1]. 
Therefore, seeking effective big data mining methods is 
of great consequence to ensure the security of related 
systems in a big data environment [2]. Most of the 
current big data mining methods carry out big data 
mining according to the known abnormal characteristics, 
which reduces the reliability and efficiency of big data 
mining, increases the overhead of processing big data, 
and reduces the overall availability and performance of 
big data. As revealed in Figure 1, the framework of big 
data mining and analysis platform [3]. Therefore, how to 
analyse the failure rate, probability analysis, and 
adjustment scheme of big data in different regions 
without interfering with the performance of huge data 
has an emphasis on the analysis of data mining [4]. In 
large-scale data mining, massive data brings great 
difficulties to the existing abnormal data mining 
efficiency [5]. How to design sub-region mining 
algorithms for massive data has gained attention and 
becomes a research hotspot. Due to the huge amount of 
data, to reduce the pressure of hardware, when the data 
scale exceeds the upper limit, it is necessary to partition 
big data [6]. In the distributed cluster environment 
without fault tolerance, the efficiency of big data 
partitioning is inversely proportional to the hardware 
involved in mining [7]. Therefore, anomaly data mining 
of massive data is a challenging task. The traditional 
partition mining algorithm based on mean clustering is 
affected by data similarity. This kind of partition mining 
algorithm will produce a high communication load in the 
parallel process, which is difficult to achieve a high 
degree of parallelism [8].  
There are certain research gaps in the traditional 
work like the problem of stabilization of chaotic 
characteristics in abnormal data by proposing the chaotic 
correlation feature extraction of huge data clustering 
based on the Internet of things. Also, the chaotic features 
in big data usually show complex folding and distortion 

334     Informatica 46 (2022) 333-342                                                                                                                                X. Liu et al. 
without obvious rules and order and non-
synchronization. The chaotic features are very complex, 
which are described by the correlation dimension. 
Thus, this article contributes to the extraction of the 
correlation dimension as the chaotic feature of huge data 
clustering. Based on the reconstruction of phase space, 
the 1D (one dimensional) time series can be extended in 
multi-dimensional space, to extract the chaotic 
correlation dimension features. Cluster analysis of big 
data is carried out according to the extracted chaotic 
correlation dimension (CCD). Relevant experimental 
analysis is carried out in this article and the traditional 
neural network algorithm is compared in terms of the 
energy consumption and processing time of the two 
algorithms. In the simulation parameter design, the time 
interval of big data packet generation is and the data is 
generated at varying simulation times. In the experiment, 
the amount of data varied from 100MB to 1GB. The 
correlation dimension of this algorithm is observed to be 
stable, while the correlation dimension of the traditional 
algorithm fluctuates greatly, verifying the effectiveness 
of the proposed algorithm for high data clustering 
efficiency. 
The structure of this paper is arranged as: the 
literature review is provided in section 2 and the huge 
data clustering process based on chaotic correlation 
dimension (CCD) feature extraction is depicted in section 
3. The experimental outcomes are presented in section 4 
while the conclusion is presented in section 5 of this 
article. 
 
 
Figure 1: Big data mining and analysis platform 
 
2 Related work 
In this section, various state-of-the-art works in the 
field of feature extraction of big data clustering based on 
the Internet of Things are discussed.  
For this research problem, there are many research 
methods related to big data clustering of the Internet of 
things. For example, the cluster analysis method of big 
data of the Internet of things proposed by Liu et al. [9]. 
Boushaki et al. proposed a multi-view fuzzy clustering 
algorithm based on the condensed information bottleneck 
audio event clustering method and representing point 
consistency constraints [10]. Single pass Bayesian fuzzy 
clustering algorithm and dynamic optimization cellular 
genetic fuzzy clustering method proposed by Yang et al. 
[11]. RNA SEQ data clustering method proposed by Park 
and Lee [12]. Grid coupled data stream clustering 
method proposed by Cui [13].  
Roy et al. proposed an uncertain data clustering 
algorithm based on Voronoi diagram in obstacle space  
 
[14]. Fast density clustering algorithm for location big 
data proposed by Li et al. [15]. Mdfuzzyk modes 
clustering algorithm based on classification matrix object 
data proposed by Yan et al. [16]. Fast adaptive clustering 
algorithm based on representative comment scoring 
strategy and geographic spatiotemporal big data 
clustering method proposed by Chen et al. [17]. The 
clustering method of Internet of things data in the cloud 
proposed by song, t, and others has the ability to classify 
the event big data with chaotic correlation characteristics 
into their respective clustering centres, and can obtain 

Chaotic Association Feature Extraction of Big Data Clustering…                                              Informatica 46 (2022) 333-342     335 
satisfactory clustering results. However, from the actual 
clustering effect, the above traditional methods have 
some key problems to be solved, such as large time 
consumption, slow speed, low agility, low data access 
load, slow convergence, large error, low efficiency of 
load balanced collaborative filtering. Research on more 
effective Internet of things big data clustering algorithm 
based on cloud mode event chaotic correlation feature 
extraction is rare [18].  
Based on the current research, this paper presents 
the chaotic correlation feature extraction of huge data 
clustering based on the Internet of things. The chaotic 
features in big data usually show complex folding and 
distortion without obvious rules and order and non-
synchronization. The chaotic features are very complex, 
which are described through the correlation dimension. 
In this article, the dimension of extracted correlation is 
used as the chaotic characteristic of huge data clustering. 
Based on the reconstruction of phase space, time series 
of one-dimensional space can be extended in multi-
dimensional space, so as to extract the chaotic features of 
correlation dimension. Cluster analysis of big data is 
presented according to the extracted chaotic correlation 
dimension. The relevant experimental analysis depicts 
some simulation outcomes which show that the proposed 
method can accurately mine abnormal data for different 
large data sets, and has high feasibility and efficiency. 
 
3 Huge data clustering algorithm 
depending on CCD feature 
extraction  
This section includes the description of clustering 
algorithm based on chaotic correlation dimension along 
with the big data clustering implementation.  
3.1 Feature extraction and analysis of 
CCD 
The chaotic characteristics in big data are usually 
complex folding and distortion without obvious rules and 
order and non-synchronization. The chaotic 
characteristics are very complex and need to be described 
by correlation dimension [19]. 
 
A. Reconstruction of phase space 
The data sequence belongs to nonlinear time series 
to a great extent, and the key of nonlinear time series is 
phase space reconstruction. Phase space reconstruction 
can keep many geometric features in the original system 
unchanged, establish a bridge between the original time 
series and multi-dimensional space analysis, and 
effectively extract the chaotic correlation dimension 
(CCD) features of data in multi bit phase space. The 
phase space reconstruction method is as follows: 
assuming that the time series is { 𝑥 1
, 𝑥 2
, … , 𝑥 𝑁 }, the phase 
space reconstruction result can be described as: 
 
 
     












     
 
  
 
1 1 1 1 1
2 1
2 1
2 1
... ,
,..., ,
...
,..., ,
m M m m
K
K
K
x x x
x x x
x x x
X X X X
 (1) 
Wherein, 𝐾 = 𝑁 − ( 𝑚 − 1) 𝜏 , 𝜏  is used to describe 
the time delay; M is used to describe the embedding 
dimension. If 𝑚 ≥ 2𝑑 + 1 the geometric structure of the 
dynamic system will be completely opened, and d is used 
to describe the dimension of the chaotic attractor of the 
system. The selection of embedding dimension m and 
time delay is the key to phase space reconstruction. Only 
by selecting reasonable 𝑚 and 𝜏 can we accurately 
reconstruct the phase space reflecting the characteristics 
of the original system. The detailed selection method is 
given below. For the selection of time delay 𝜏 . This study 
considers time delay 𝜏 denoted by the abscissa when the 
mutual data of delay time takes the first minimum value 
as the finest time delay for recreating phase space [20]. 
In the interval of data distribution, the probability 
distribution curve of data is established. 𝑝 𝑖 is used to 
describe the probability that 𝑥 ( 𝑡 ) appears in the interval I 
of the data distribution curve; 𝑝 𝑖𝑗
( 𝜏 ) is used to describe 
the joint probability that 𝑥 ( 𝑡 ) appears in 𝑖 and delay 
𝑥 ( 𝑡 + 𝜏 ) after a certain amount of delay 𝜏 appears in 
region 𝑗 . Then the delay time mutual information can be 
described as: 
   
 
j i
ij
ij
ij
p p
p
p I

  ln

  
(2) 
If 𝐼 ( 𝜏 ) = 0, 𝑥 ( 𝑡 + 𝜏 ) cannot be predicted, that is, 
𝑥 ( 𝑡 ) and 𝑥 ( 𝑡 + 𝜏 ) are independent of each other, and the 
smaller 𝐼 ( 𝜏 ) is more independent 𝑥 ( 𝑡 ) and 𝑥 ( 𝑡 + 𝜏 ) . 
Therefore, when 𝐼 ( 𝜏 ) reaches the minimum, the time 
delay 𝜏 corresponding to the abscissa can be utilized as 
the finest time delay for recreating the phase space. For 
the selection of embedding dimension 𝑚 , this paper uses 
the virtual nearest neighbor algorithm for the estimation 
[21]. According to Takens theorem, the 𝑚 −
𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑎𝑙 vector formed in the 𝑚 − 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑎𝑙 
phase space can be described as: 
 
              1 ,..., ,     m n x n x n x n X (3) 
Obtaining the minimum embedding dimension of 
phase space reconstruction needs to meet the conditions 
described in equation (4). If yes, 𝑋 𝜂 ( 𝑛 )
 is called the false 
nearest neighbor of 𝑋 𝑛 . 
 
 
 
 
tol
m
n n
m n m n
R
X X
x x




 
1
2

  
 (4) 

336     Informatica 46 (2022) 333-342                                                                                                                                X. Liu et al. 
Where 𝑅 𝑡𝑜𝑙 is used to describe the threshold, usually 
𝑅 𝑡𝑜𝑙 takes 15. At this time, the proportion curve of false 
nearest neighbour points is required. If the proportion of 
false nearest neighbour points is less than 5%, it is 
considered that the obtained m is the minimum 
embedding dimension of phase space reconstruction [22]. 
 
B. Feature extraction of chaotic correlation 
dimension 
In this paper, the extracted CCD is utilized as the 
chaotic element of huge data clustering. Based on phase 
space rebuilding, 1-D time series can be stretched out in 
multi-layered space to separate chaotic element aspect 
highlights [23]. As per the procedure analyzed in earlier 
section, the recreated time series can be acquired: 
 
 
 
T
m i i i i
x x x X
  1
,..., ,
  
 (5) 
In the 𝑚 − 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑎𝑙 phase space recreated by 
the above-mentioned procedure, the focuses whose 
separation from phase point 𝑥 𝑗 to 𝑥 𝑖 additional 𝑥 𝑖 itself 
doesn't surpass r can be portrayed as: 
 
 
j i
i j
x x r H Q   


 
(6) 
Where H (*) is utilized to portray the Heaviside 
work. The idea of connection work is given here. All 
focuses that might be more modest than the given 
distance 𝑟 are comparative with one another The extent 
of the complete point logarithm is known as the 
connection work, and the equation is portrayed as 
follows: 
 
 
 
 
 
  
 


N
i
N
i j
j i N
x x r H
Q Q
r C
1 1
1
2
 
(7) 
In the equation, the numerator is 2 to wipe out 
continued counting. The distance between two-stage 
focuses can be acquired by depicting the distance 
between two-stage focuses with standard, or at least, the 
greatest contrast among two vectors: 
 
      1 1
1
max
   
 
  
k j k i
m k
j i
x x x x 
(8) 
For a vector whose distance doesn't surpass 𝑟 , it 
tends to be called a cooperative vector [24]. Expecting 
that there is n 1𝐷 estimated succession information, the 
quantity of vector focuses in stage space remaking is 
𝑁 = 𝑚 − ( 𝑚 − 1) 𝜏 . Compute the extent of the stage 
point logarithm with connection in all conceivable 
𝑁 ( 𝑁 − 1) /2 sets, which is known as the relationship 
aspect. The recipe is depicted as follows: 
 
 
 
 
  
 


N
i
N
i j
j i m
x x r H
N N
r C
1 1
1
2
 (9) 
Then the relationship aspect got above is the 
tumultuous trademark amount of large information 
grouping, and the bunching of huge information is 
acknowledged by the connection aspect. 
3.2 Big data clustering implementation 
The cluster analysis is to divide different samples 
into several categories, and make the samples of an 
aggregate class more similar than those of different 
aggregate classes [25]. In this paper, huge data is 
clustered and analysed as per the extracted CCD [26-28]. 
The flowchart of big data clustering implementation in 
this article is depicted in Figure 2 and the detailed 
implementation is provided in this section. 
 
A. Input samples and parameters 
Enter n data samples {𝑥 1
, 𝑥 2
, … , 𝑥 𝑛 }, According to 
the characteristics of chaotic correlation dimension, n 
cluster centers are selected from the above samples and 
described by {𝑍 1
, 𝑍 2
, … , 𝑍 𝑛 }. 
 
B. Divide n samples into the nearest cluster 
according to the following principles 𝜔 𝑗 
 
 
j j
z x z x    min (10) 
Where ‖𝑥 − 𝑍 𝑗 ‖ is used to describe the distance 
between 𝑥 and 𝑍 𝑗 . At the same time, it is assumed that 
there are 𝑁 𝑖 samples in 𝜔 𝑗 . 
 
 
Figure 2: Flowchart of huge data clustering 
implementation 
 
C. The cluster centre value is obtained by the 
following formula 

Chaotic Association Feature Extraction of Big Data Clustering…                                              Informatica 46 (2022) 333-342     337 
 
  r xC
N
z
j
x
m
j
j 



1
 
(11) 
If the number of iterations is odd, proceed directly 
to step (5); Otherwise, follow next step. 
 
D. Split 
Assuming 𝐿 = max( 𝑥 − 𝑍 𝑖 ), 𝑥 ∈ 𝜔 𝑗 , 𝑑 1
 is used to 
describe the splitting distance. If 𝐿 > 𝑑 1
, 𝜔 𝑗 is divided 
into two categories. At this time, the cluster center can be 
described as: 



 
 
L z z
L z z
i i
i i


2
1
 (12) 
Where 𝜆 is used to describe a constant greater than 
0. If 𝐿 < 𝑑 1
 and the last merge operation was not 
performed, proceed to step (6). 
 
 
E. Merge 
Assuming 𝑙 = ‖𝑍 𝐼 − 𝑍 𝐽 ‖ = ‖𝑍 𝑖 − 𝑍 𝑗 ‖, use 𝑑 2
 to 
describe the merge distance. If 𝑙 < 𝑑 2
, then, 𝜔 𝐼 , 𝜔 𝐽 are 
merged into one class, and the merging center can be 
described as: 
 
 
J J I I
J I
IJ
z N z N
N N
z 


1
 
(13) 
If 𝑙 < 𝑑 2
, and not classified last time, proceed to 
step (6), otherwise proceed to step (3). 
 
F. End iteration 
In this paper, the data with the same chaotic 
correlation characteristics are divided into one class 
through the above clustering analysis process, so as to 
realize the effective clustering of big data [29-31]. This 
work is also considered for the industrial applications 
and contributing towards social life with the integration 
of the Internet of Things, AI, and robotics [32-35]. 
 
4 Results and Analysis 
This section presents the result analysis obtained for 
from the proposed big data clustering algorithm and 
finally presents its discussion and summary in conclusion 
section. 
In order to validate the efficiency of the huge data 
clustering algorithm based on chaotic correlation feature 
extraction proposed in this paper, relevant experimental 
analysis is needed [36-38]. Taking the traditional neural 
network algorithm as a comparison, the energy 
consumption and processing time of the two algorithms 
are mainly compared [39-42]. In this paper, the algorithm 
is verified by simulation data. All the experimental 
programs are written in C++, which is in Ubuntu 12.04 
operating system. The experimental hardware platform is 
LenovoM4390 (i3-2100 CPU, 4UB memory, 2TB disk), 
processor Intel (R) core (TM) 2duocpu2 94GHz, 
memory: 8.00GB. In the simulation parameter design, 
the time interval of big data packet generation is 0.1s, 
and the data is generated from the simulation time of 
300s. In the experiment, the amount of data is from 
100MB to 1GB, with 100MB as the unit, the data 
increases nonlinearly, discrete scheduling and interval 
boundary approximation are carried out for big data, the 
time interval of big data feature acquisition is 0.1s, and 
the parameter configuration is listed in Table 1. 
 
Parameter Value (Mbps ） 
Data quantity 1000 
Number of big data distribution 
Characteristics 
5 
Load per data access system 16 
Data complexity size (GB) 2 
Data execution time delay (MS) 2400 
Maximum queue size 2200 
Table 9: Parameter configuration 
 
The algorithm in this paper and the traditional 
algorithm are used to cluster different amounts of data, 
and the clustering efficiency of the two algorithms is 
counted. The outcomes are listed in Table 2 and 
graphical represented is provided in Figure 3. 
 
Data volume 
Time required for 
the proposed 
algorithm (s) 
Time required for 
traditional 
algorithm (s) 
200 1925 5998 
400 4433 12769 
800 8343 29151 
1024 10151 35832 
Table 10: Comparison results of clustering efficiency of 
two algorithms 
 

338     Informatica 46 (2022) 333-342                                                                                                                                X. Liu et al. 
 
Figure 3: Graphical comparison of clustering efficiency 
of two algorithms 
 
It can be observed from the analysis of Table 1 and 
Figure 3 that with the gradual increase of the amount of 
data, the time for data clustering of this algorithm and the 
traditional algorithm gradually increases. While this 
improvement occurs, the processing time required by this 
algorithm has been potentially lower than that of the 
traditional algorithm, which shows that this algorithm 
has high data clustering efficiency and verifies the 
effectiveness of this algorithm. 
In order to further validate the effectiveness of this 
algorithm, this paper compares the energy consumed by 
the two algorithms to process the same amount of data. 
The results are shown in Figure 4. 
By analysing Figure 4, it can be seen that when 
processing the same amount of data, the energy 
consumption of this algorithm is significantly lower than 
that of the traditional algorithm. This is because this 
algorithm is easy to implement and has high clustering 
efficiency for data, so it consumes less energy, which 
verifies the effectiveness of this algorithm. 
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
0
50
100
150
200
250
300
350
400
Energy consumption (J)
Data volume (GB)
 Improved algorithm
 Traditional algorithm
Figure 4: Comparison results of energy consumption of 
two algorithms 
 
By analysing Figures 5, 6 and 7, it can be seen that 
when processing the same amount of data, the time 
required by the algorithm in this paper is significantly 
lower than that of the traditional algorithm. This is 
because the algorithm in this paper is easy to implement 
and has good clustering efficiency for data, so the 
clustering time is short, which further verifies the 
effectiveness of the algorithm in this paper. 
 
0 1 2 3 4 5
23
24
25
26
27
28
29
Time consuming (ms)
Data volume (GB)
 Traditional algorithm
Figure 5: Time consuming of traditional algorithm 
0 1 2 3 4 5
7
8
9
10
11
12
Time consuming (ms)
Data volume (GB)
 Improved algorithm
Figure 6: Time consuming of improved algorithm 
 
0
5000
10000
15000
20000
25000
30000
35000
40000
200 400 800 1024
Time in seconds
Data Volume
 Time required for the proposed algorithm (s)
 Time required for traditional algorithm (s)

Chaotic Association Feature Extraction of Big Data Clustering…                                              Informatica 46 (2022) 333-342     339 
0 1 2 3 4 5
5
10
15
20
25
30
Time consuming (ms)
Data volume (GB)
 Traditional algorithm
 Improved algorithm
Figure 7: Time consuming comparison results of two 
algorithms 
1 2 3 4 5
5
6
7
8
9
10
Dimension
Data volume (GB)
 Traditional algorithm
 Improved algorithm
Figure 8: Comparison results of correlation dimensions 
of two algorithms 
 
It can be seen from the analysis of Figure 8 that with 
the gradual increase of the amount of data, the 
correlation dimension of the algorithm in this paper tends 
to be stable, while the correlation dimension of the 
traditional algorithm fluctuates greatly. This fluctuation 
shows that the algorithm in this paper has high data 
clustering efficiency and verifies the effectiveness of the 
algorithm in this paper [43-44]. 
 
5 Conclusions 
This article presents the CCD feature extraction of 
huge data clustering based on the Internet of things is 
proposed. By reconstructing the phase space, a multi-
dimensional state space vector and chaotic trajectory are 
established. It was revealed that many geometric features 
in the creative scheme remain unchanged, which 
provides an effective basis for analysing the chaotic 
characteristics of the original system. The false adjacent 
neighbour procedure is used to select the finest 
embedding dimension. The extracted CD is used as the 
chaotic feature of huge data clustering, and the big data is 
clustered according to the extracted chaotic correlation 
dimension. Simulation outcomes show that the proposed 
method can accurately mine abnormal data for different 
large data sets, and has high feasibility and efficiency. At 
present, the composition structure, operation mechanism 
and relevant standards of the Internet of things in cloud 
mode have not been completely unified. This can act as 
the future research scope of this article and therefore, the 
research on big data clustering of the Internet of things 
needs to be further discussed in many aspects in the 
future part of this research work. 
 
References  
[1] Bu, F. (2018). An efficient fuzzy c-means approach 
based on canonical polyadic decomposition for 
clustering big data in IoT. Future Generation 
Computer Systems, 88, 675-682.  
https://doi.org/10.1016/j.future.2018.04.045 
[2] Bu, F., Hu, C., Zhang, Q., Bai, C., Yang, L. T., & 
Baker, T. (2020). A Cloud-Edge-aided Incremental 
High-order Possibilistic c-Means Algorithm for 
Medical Data Clustering. IEEE Transactions on 
Fuzzy Systems, 29(1), 148-155.  
10.1109/TFUZZ.2020.3022080 
[3] Liu, Y ., Zhang, J., & Zhan, J. (2021). Privacy 
protection for fog computing and the internet of 
things data based on blockchain. Cluster 
Computing, 24(2), 1331-1345.  
https://doi.org/10.1007/s10586-020-03190-3 
[4] Lye, G. X., Cheng, W. K., Tan, T. B., Hung, C. W., 
& Chen, Y . L. (2020). Creating personalized 
recommendations in a smart community by 
performing user trajectory analysis through social 
internet of things deployment. Sensors, 20(7), 2098. 
https://doi.org/10.3390/s20072098 
[5] Cai, G., Fang, Y ., Chen, P., Han, G., Cai, G., & 
Song, Y . (2020). Design of an MISO-SWIPT-aided 
code-index modulated multi-carrier M-DCSK 
system for e-health IoT. IEEE Journal on Selected 
Areas in Communications, 39(2), 311-324. 
10.1109/JSAC.2020.3020603 
[6] Xia, H., Huang, W., Li, N., Zhou, J., & Zhang, D. 
(2019). PARSUC: A parallel subsampling-based 
method for clustering remote sensing big 
data. Sensors, 19(15), 3438.  
https://doi.org/10.3390/s19153438 
[7] Arora, S., Sharma, M., & Anand, P. (2020). A novel 
chaotic interior search algorithm for global 
optimization and feature selection. Applied 
Artificial Intelligence, 34(4), 292-328.  
https://doi.org/10.1080/08839514.2020.1712788 
[8] Maddumala, V . R. (2020). Big Data-Driven Feature 
Extraction and Clustering Based on Statistical 
Methods. Traitement du Signal, 37(3). 
10.18280/ts.370305 
[9] Liu, W., Wang, X., & Peng, W. (2019). Secure 
remote multi-factor authentication scheme based on 
chaotic map zero-knowledge proof for 
crowdsourcing internet of things. IEEE Access, 8, 
8754-8767.  

340     Informatica 46 (2022) 333-342                                                                                                                                X. Liu et al. 
10.1109/ACCESS.2019.2962912 
[10] Boushaki, S. I., Kamel, N., & Bendjeghaba, O. 
(2018). A new quantum chaotic cuckoo search 
algorithm for data clustering. Expert Systems with 
Applications, 96, 358-372.  
https://doi.org/10.1016/j.eswa.2017.12.001 
[11] Yang, Q., Ruan, J., Zhuang, Z., & Huang, D. 
(2019). Chaotic analysis and feature extraction of 
vibration signals from power circuit breakers. IEEE 
Transactions on Power Delivery, 35(3), 1124-1135. 
10.1109/TPWRD.2019.2934123 
[12] Park, S. W., & Lee, I. Y . (2019). Enhanced signature 
RTD transaction scheme based on Chebyshev 
polynomial for mobile payments service in IoT 
device environment. The Journal of 
Supercomputing, 75(8), 4617-4637.  
https://doi.org/10.1007/s11227-018-2546-8 
[13] Cui, Y . (2018). Application of the improved chaotic 
self-adapting monkey algorithm into radar systems 
of internet of things. IEEE Access, 6, 54270-54281. 
10.1109/ACCESS.2018.2869632 
[14] Roy, S., Chatterjee, S., Das, A. K., Chattopadhyay, 
S., Kumari, S., & Jo, M. (2017). Chaotic map-based 
anonymous user authentication scheme with user 
biometrics and fuzzy extractor for crowdsourcing 
Internet of Things. IEEE Internet of Things 
Journal, 5(4), 2884-2895.  
10.1109/JIOT.2017.2714179 
[15] Li, L., Wen, G., Wang, Z., & Yang, Y . (2019). 
Efficient and secure image communication system 
based on compressed sensing for IoT monitoring 
applications. IEEE Transactions on 
Multimedia, 22(1), 82-95.  
10.1109/TMM.2019.2923111 
[16] Yan, Z., Liu, J., Vasilakos, A. V ., & Yang, L. T. 
(2015). Trustworthy data fusion and mining in 
Internet of Things. Future Generation Computer 
Systems, 49(C), 45-46.  
https://doi.org/10.1016/j.future.2015.04.001 
[17] Chen, F., Li, Q., Li, M., Huang, F., Zhang, H., 
Kang, J., & Wang, P. (2021). Unclonable 
fluorescence behaviors of perovskite quantum 
dots/chaotic metasurfaces hybrid nanostructures for 
versatile security primitive. Chemical Engineering 
Journal, 411, 128350. 
https://doi.org/10.1016/j.cej.2020.128350 
[18] Song, T., Li, R., Mei, B., Yu, J., Xing, X., & Cheng, 
X. (2017). A privacy preserving communication 
protocol for IoT applications in smart homes. IEEE 
Internet of Things Journal, 4(6), 1844-1852.  
10.1109/JIOT.2017.2707489 
[19] Alarifi, A., Sankar, S., Altameem, T., Jithin, K. C., 
Amoon, M., & El-Shafai, W. (2020). A novel hybrid 
cryptosystem for secure streaming of high 
efficiency H. 265 compressed videos in IoT 
multimedia applications. IEEE Access, 8, 128548-
128573. 
10.1109/ACCESS.2020.3008644 
[20] Niu, Z., Zheng, M., Zhang, Y., & Wang, T. (2019). 
A new asymmetrical encryption algorithm based on 
semitensor compressed sensing in WBANs. IEEE 
Internet of Things Journal, 7(1), 734-750.  
10.1109/JIOT.2019.2953519 
[21] Li, L., Liu, L., Peng, H., Yang, Y ., & Cheng, S. 
(2018). Flexible and secure data transmission 
system based on semitensor compressive sensing in 
wireless body area networks. IEEE Internet of 
Things Journal, 6(2), 3212-3227. 
10.1109/JIOT.2018.2881129 
[22] Yung, C., Chen, C. C., Yuan, Y . L., & Li, C. (2019). 
A Systematic Model of Big Data Analytics for 
Clustering Browsing Records into Sessions Based 
on Web Log Data. J. Comput., 14(2), 125-133.  
10.17706/jcp.14.2.125-133 
[23] Jang, S. W., & Kim, G. Y . (2017). A monitoring 
method of semiconductor manufacturing processes 
using Internet of Things–based big data 
analysis. International Journal of Distributed 
Sensor Networks, 13(7), 1550147717721810. 
https://doi.org/10.1177/1550147717721810 
[24] Gong, X., Liu, L., Fong, S., Xu, Q., Wen, T., & Liu, 
Z. (2019). Comparative research of swarm 
intelligence clustering algorithms for analyzing 
medical data. IEEE Access, 7, 137560-137569. 
10.1109/ACCESS.2018.2881020 
[25] Lee, Y . C., Huang, S. C., Huang, C. H., & Wu, H. 
H. (2016). A new approach to identify high burnout 
medical staffs by kernel k-means cluster analysis in 
a regional teaching hospital in Taiwan. Inquiry: The 
Journal of Health Care Organization, Provision, 
and Financing, 53, 0046958016679306. 
https://doi.org/10.1177/0046958016679306  
[26] Shabaz, M., Sharma, A., Al Ajrawi, S., & Estrela, V . 
V . (2022). Multimedia-based emerging technologies 
and data analytics for Neuroscience as a Service 
(NaaS). Neuroscience Informatics, 2(3), 100067. 
https://doi.org/10.1016/j.neuri.2022.100067 
[27] Poongodi, M., Hamdi, M., Malviya, M., Sharma, 
A., Dhiman, G., & Vimal, S. (2022). Diagnosis and 
combating COVID-19 using wearable Oura smart 
ring with deep learning methods. Personal and 
ubiquitous computing, 26(1), 25-35.  
https://doi.org/10.1007/s00779-021-01541-4 
[28] Kumbinarasaiah, S., & Raghunatha, K. R. (2021). A 
novel approach on micropolar fluid flow in a porous 
channel with high mass transfer via wavelet 
frames. Nonlinear Engineering, 10(1), 39-45. 
https://doi.org/10.1515/nleng-2021-0004 
[29] Wang, H., Sharma, A., & Shabaz, M. (2022). 
Research on digital media animation control 
technology based on recurrent neural network using 
speech technology. International Journal of System 
Assurance Engineering and Management, 13(1), 
564-575. 
https://doi.org/10.1007/s13198-021-01540-x 
[30] Ting, L., Khan, M., Sharma, A., & Ansari, M. D. 
(2022). A secure framework for IoT-based smart 
climate agriculture system: Toward blockchain and 

Chaotic Association Feature Extraction of Big Data Clustering…                                              Informatica 46 (2022) 333-342     341 
edge computing. Journal of Intelligent 
Systems, 31(1), 221-236.  
https://doi.org/10.1515/jisys-2022-0012 
[31] Sharma, D., Kaur, R., Sandhir, M., & Sharma, H. 
(2021). Finite element method for stress and strain 
analysis of FGM hollow cylinder under effect of 
temperature profiles and inhomogeneity 
parameter. Nonlinear Engineering, 10(1), 477-487. 
https://doi.org/10.1515/nleng-2021-0039 
[32] Ren, Y ., Rubaiee, S., Ahmed, A., Othman, A. M., & 
Arora, S. K. (2022). Multi-objective optimization 
design of steel structure building energy 
consumption simulation based on genetic 
algorithm. Nonlinear Engineering, 11(1), 20-28.  
https://doi.org/10.1515/nleng-2022-0012 
[33] Singh, P. K., & Sharma, A. (2022). An intelligent 
WSN-UA V-based IoT framework for precision 
agriculture application. Computers and Electrical 
Engineering, 100, 107912. 
https://doi.org/10.1016/j.compeleceng.2022.107912 
[34] Zeng, H., Dhiman, G., Sharma, A., Sharma, A., & 
Tselykh, A. (2021). An IoT and Blockchain‐based 
approach for the smart water management system in 
agriculture. Expert Systems, e12892. 
https://doi.org/10.1111/exsy.12892 
[35] Sharma, A., & Singh, P. K. (2021). UA V‐based 
framework for effective data analysis of forest fire 
detection using 5G networks: An effective approach 
towards smart cities solutions. International 
Journal of Communication Systems, e4826. 
https://doi.org/10.1002/dac.4826 
[36] Sharma, A., Singh, P. K., & Kumar, Y . (2020). An 
integrated fire detection system using IoT and 
image processing technique for smart 
cities. Sustainable Cities and Society, 61, 102332. 
https://doi.org/10.1016/j.scs.2020.102332 
[37] Gunupudi, R. K., Nimmala, M., Gugulothu, N., & 
Gali, S. R. (2017). CLAPP: A self constructing 
feature clustering approach for anomaly 
detection. Future Generation Computer 
Systems, 74, 417-429. 
https://doi.org/10.1016/j.future.2016.12.040 
[38] Sharma, A. (2021). Integrity and Multimedia Data 
Management using Emerging Technologies in the 
Healthcare Applications-Part II. Recent Advances in 
Electrical & Electronic Engineering (Formerly 
Recent Patents on Electrical & Electronic 
Engineering), 14(7), 698-699. 
https://doi.org/10.2174/23520965140721110409193
0 
[39] Wang, Y ., Chen, Q., Kang, C., & Xia, Q. (2016). 
Clustering of electricity consumption behavior 
dynamics toward big data applications. IEEE 
transactions on smart grid, 7(5), 2437-2447.  
https://doi.org/10.1109/TSG.2016.2548565 
[40] Guo, Z., & Xiao, Z. (2021). Research on online 
calibration of lidar and camera for intelligent 
connected vehicles based on depth-edge 
matching. Nonlinear Engineering, 10(1), 469-476. 
https://doi.org/10.1515/nleng-2021-0038 
[41] Deng, Z., Hu, Y ., Zhu, M., Huang, X., & Du, B. 
(2015). A scalable and fast OPTICS for clustering 
trajectory big data. Cluster Computing, 18(2), 549-
562. 
https://doi.org/10.1007/s10586-014-0413-9 
[42] Chen, Y ., Zhang, W., Dong, L., Cengiz, K., & 
Sharma, A. (2021). Study on vibration and noise 
influence for optimization of garden 
mower. Nonlinear Engineering, 10(1), 428-435. 
https://doi.org/10.1515/nleng-2021-0034 
[43] Sharma, A., Singh, P. K., Hong, W. C., Dhiman, G., 
& Slowik, A. (2021). Introduction to the Special 
Issue on Artificial Intelligence for Smart Cities and 
Industries. Scalable Computing: Practice and 
Experience, 22(2), 89-91. 
https://doi.org/10.12694/scpe.v22i2.1939 
[44] Luna-Romera, J. M., García-Gutiérrez, J., Martínez-
Ballesteros, M., & Riquelme Santos, J. C. (2018). 
An approach to validity indices for clustering 
techniques in big data. Progress in Artificial 
Intelligence, 7(2), 81-94. 
https://doi.org/10.1007/s13748-017-0135-3 
 
 
 
 
 
 
 
 
 
 
 
 

342     Informatica 46 (2022) 333-342                                                                                                                                X. Liu et al.