J Future 2019 04 020

Accepted Manuscript
The role of big data analytics in industrial Internet of Things
Muhammad Habib ur Rehman, Ibrar Yaqoob, Khaled Salah,

Muhammad Imran, Prem Prakash Jayaraman, Charith Perera
PII: S0167-739X(18)31364-5
DOI: https://doi.org/10.1016/j.future.2019.04.020
Reference: FUTURE 4901
To appear in: Future Generation Computer Systems
Received date : 2 June 2018

Revised date : 11 February 2019
Accepted date : 8 April 2019
Please cite this article as: M.H. ur Rehman, I. Yaqoob, K. Salah et al., The role of big data analytics
in industrial Internet of Things, Future Generation Computer Systems (2019),
https://doi.org/10.1016/j.future.2019.04.020
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
The Role of Big Data Analytics in Industrial Internet of Things
Muhammad Habib ur Rehmana , Ibrar Yaqoobb , Khaled Salahc , Muhammad Imrand , Prem
Prakash Jayaramane , Charith Pereraf
a National University of Computer and Emerging Sciences, Lahore, Pakistan
b Department of Computer Science and Engineering, Kyung Hee University, Yongin-si 17104, South Korea
c Department of Electrical and Computer Engineering, Khalifa University, UAE
d College of Computer and Information Sciences, King Saud University, Riyadh, Saudi Arabia
e Department of CS & SE, Swinburne University of Technology, VIC 3122, Australia
f School of Computer Science and Informatics, Cardiff University, United Kingdom
Abstract
Big data production in industrial Internet of Things (IIoT) is evident due to the massive deploy-
ment of sensors and Internet of Things (IoT) devices. However, big data processing is chal-
lenging due to limited computational, networking and storage resources at IoT device-end. Big
data analytics (BDA) is expected to provide operational- and customer-level intelligence in IIoT
systems. Although numerous studies on IIoT and BDA exist, only a few studies have explored
the convergence of the two paradigms. In this study, we investigate the recent BDA technolo-
gies, algorithms and techniques that can lead to the development of intelligent IIoT systems. We
devise a taxonomy by classifying and categorising the literature on the basis of important param-
eters (e.g. data sources, analytics tools, analytics techniques, requirements, industrial analytics
applications and analytics types). We present the frameworks and case studies of the various
enterprises that have benefited from BDA. We also enumerate the considerable opportunities in-
troduced by BDA in IIoT. We identify and discuss the indispensable challenges that remain to be
addressed as future research directions as well.
Keywords: Internet of Things, cyber-physical systems, cloud computing, analytics, big data.
1 1. Introduction
2 Industrial Internet of Things (IIoT) (also known as Industry 4.0), which was initially con-
3 ceived as a vision by the German government, is currently attributed as the fourth industrial rev-
4 olution. The technology ecosystem underpinning IIoT is mainly the integration of cyberphysical
5 systems (CPS) [1], Internet of Things (IoT), cloud computing [2–4], automation (e.g. intelligent
Preprint submitted to Future Generation Computer Systems April 17, 2019

6 robots in product assembly lines) [5], Internet of services [6], wireless technologies, augmented
7 reality [7] and concentric computing [8], amongst others. Advances in such related areas as IoT,
8 big data analytics (BDA), cloud computing and CPS have fuelled the formation of IIoT activities
9 to deliver unprecedented flexibility, precision and efficiency to manufacturing processes [9, 10].
10 Given this cross-platform integration, IIoT systems need to ensure interoperability, virtualisa-
11 tion, decentralisation, real-time capability, service orientation, modularity and security across all
12 verticals [11]. However, these systems are perceived to have qualities, such as self-awareness,
13 self-prediction, self-comparison, self-configuration, self-maintenance and self-organisation [12].
14 BDA is a related area that enables IIoT systems to deliver value for data captured from cross-
15 platform integration. BDA refers to the process of collecting, managing, processing, analysing
16 and visualising continuously evolving data in terms of volume, velocity, value, variety and ve-
17 racity [13]. Big data in IIoT systems arise due to unbounded internal and external activities
18 relevant to customers, business operations, production and machines [14]. BDA processes in
19 IIoT systems manage the collected data using multiple transient and persistent storage systems
20 that provide on-board, in-memory, in-network and large-scale distributed storage facilities across
21 IIoT systems [15, 16]. The granularity of data processing facilities for BDA processes in IIoT
22 systems vary from resource-constrained IoT devices to resourceful large-scale distributed cloud
23 computing systems [17]. Similarly, analytic operations differ in terms of descriptive, prescrip-
24 tive, predictive and preventive procedures [14]. In addition, BDA processes must ensure real-time
25 knowledge visualisation across multiple IIoT systems. A proper integration of BDA processes
26 into IIoT systems is perceived to maximise value creation to evolve business models for profit
27 maximisation [14, 18].
28 1.1. Motivation
29 Although IIoT [19–24] and BDA [13, 25–31] have been widely studied separately, only a
30 few studies including [32] have explored the convergence of the two domains.
31 Big data production in IIoT is evident due to large-scale deployment of sensing devices and
32 systems in pervasive and ubiquitous industrial networks. Given that the concept of IIoT systems
33 is still evolving, complete integration and implementation of BDA processes in IIoT systems are
34 unavailable yet [32, 33]. Existing surveys on IIoT systems focus on concepts related to adoption
35 of IIoTs [34, 35], the integration of IIoTs and edge cloud computing systems [36], industrial
2
36 marketplaces for IIoTs [4], big data and virtualisation technologies for IIoT systems [37], tech-
37 nological advancements relevant to CPS in IIoT systems [38], smart manufacturing [39] and big
38 data applications for business operations [40–42]. We introduced the concept of the concentric
39 computing model (CCM) for BDA in IIoT in our previous work [32] whereby we outlined the
40 discussion on different layers of CCM and discussed the relevant research challenges that must
41 be addressed to fully enable CCM for BDA in IIoT. However, to the best of our knowledge, a
42 detailed review on BDA implementation for IIoTs is still lacking in the existing literature. Thus,
43 the current study presents the key operations of BDA for value creation in IIoT systems. On the
44 basis of BDA concepts, this study surveys earlier contributions relevant to data analysis in IIoT
45 systems.
46 1.2. Contributions
47 The main contributions of this study are listed as follows.
48 • We build a case of BDA for IIoT systems whereby the role and entire process of BDA
49 are discussed. The study sets a theoretical ground to understand modern automated data
50 pipelines for enriching intelligence in IIoT systems.
51 • We investigate existing state-of-the-art research studies on IIoT in terms of BDA. In this

52 context, we categorise and classify the literature by devising a taxonomy.
53 • We present frameworks and case studies whereby BDA processes are adopted to improve
54 the overall performance of IIoT systems.
55 • We present several research opportunities, challenges and future technologies to minimise

56 the research gaps between state of the art (i.e. proposed in the literature) and state of the
57 practice (i.e. adopted by industries in practice).
58 The rest of the paper is organised as follows. Section 2 discusses the key concepts relevant
59 to BDA in IIoT systems, followed by a detailed survey of existing technologies and algorithms
60 in Section 3. Section 4 presents the taxonomy, and Section 5 highlights a few frameworks and
61 relevant case studies. Section 6 presents the opportunities, open challenges and future directions.
62 Section 7 provides the concluding remarks.
3
63 2. BDA in IIoT Systems
64 This section presents a detailed discussion on different aspects of big data adoption in IIoT
65 systems. To this end, several design principles, which should be considered prior to configuring
66 and deploying IIoT systems, are highlighted. The role of BDA and its life cycle is discussed in
67 detail to deliver end-to-end intelligence in IIoT systems.
68 2.1. Design Principles for IIoT Systems
69 The designs of IIoT systems involve seven principles [11], as depicted in Fig. 1. Firstly,
70 interoperability must be ensured amongst different technologies, such as CPS, IoT devices and
71 concentric computing systems. Wireless data communication technologies play an unparalleled
72 role to realise an interoperable system. Secondly, virtualisation technologies at all levels must
73 be considered for efficient service provisioning and delivery across IIoT systems. Virtualisation
74 varies in terms of platforms, networks, data, operating systems and applications. Thirdly, decen-
75 tralisation must be conducted to ensure highly distributed IIoT systems. Decentralisation varies
76 in terms of system-wide data processing and data storage. Fourthly, IIoT systems must provide
77 real-time feedback to all stakeholders. Fifthly, service-orientation must be guaranteed whereby
78 all system functions are implemented in the form of service-oriented architecture (SOA). Sixthly,
79 modular approach must be adopted for system implementation. Lastly, system-wide security
80 must be considered as core principle. The BDA process for IIoT systems must be designed in
81 consideration of the above-mentioned principles.
82 2.2. Rise of Big Data in IIoT Systems
83 Big data in IIoT systems emerge from a plethora of technologies. CPS refers to the integration
84 of physical machine components with on-board computations and networking facilities [38, 43].
85 CPS and IoT devices act as the backbone of IIoT systems and thus generate massive amount of
86 raw data streams, which result in big data [44]. Therefore, real-time analysis of these data can
87 improve machine health and lead to defect-free product manufacturing [1, 34, 45].
88 IoT devices in IIoT systems refer to devices that can remotely sense and actuate in industrial
89 environments [46]. IoT devices either work as stand-alone devices that roam around industrial
90 environments or are attached with existing CPS to perform certain predefined actions [47]. The
91 on-board sensing facilities in IoT devices lead the generation of big data, which may become
4
Decentralization
Interoperability Real-time
Design
Implementation
Security Principles Service Orientation
(Industry 4.0)
Modularity Visualization
Figure 1: Seven design implementation principles for Industry 4.0 systems.
92 useful for value creation in enterprises. The integration of CPS and IoT devices results in massive
93 back-end cloud service utilisation for the execution of BDA processes [48]. To achieve massively
94 customised production, the number of cloud services can be grown immensely. Thus, BDA can
95 facilitate in-service selection, service orchestration and real-time service provisioning [49].
96 2.3. Concentric Computing Model for BDA in IIoT
97 Recent evolution in sensing and computing technologies has opened new avenues for big
98 data processing. Concentric computing refers to the large-scale highly distributed comput-
99 ing systems based on a wide range of devices and computing facilities in different form fac-
100 tors [8]. Concentric computing offers big data processing at sensors levels, endpoints in IIoT
101 systems, edge servers, and centralised and decentralised cloud computing systems, as illustrated
102 in Fig. 2 [14, 36, 50]. Despite their small size and limited computational power, sensors and
103 IoT devices have the ability to filter and reduce raw data streams by using on-board smart data
104 reduction strategies [51]. However, edge servers at gateways and centralised computing clusters
105 have the ability to distribute the computing load for BDA applications [52, 53]. Multistage exe-
5
106 cution, automating, and management of BDA processes (i.e., data engineering, data preparation
107 and data analytics) are necessary in concentric computing environments (such as sensors and
108 wearable devices as endpoints, IoT devices, edge servers, and cloud computing servers) [54].
109 2.4. Big Data Analytics for Delivering Intelligence in IIoT Systems
110 BDA processes are executed as a result of multistage highly interdependent application com-
111 ponents (Fig. 3). These components are categorised as follows.
112 2.4.1. Data Engineering

113 Data engineers build computing and storage infrastructure to ingest [55], clean [56], con-
114 form [57], shape [58] and transform [59] data. IIoT systems produce and ingest big data from
115 inbound enterprise operations and outbound customer activities. The raw data at the earliest
116 stage need further processing to improve the quality and establish the relevance with IIoT ap-
117 plications. Therefore, data wrangling and cleaning methodologies help select relevant datasets
118 in case of historical data or data streams in case of streaming data. Data conformity procedures
119 are applied to ensure relevant, correctly collected big data. Data shaping and transformation
Industrial Internet of Things
Outer Edge Servers
Inner Edge Servers
Core Cloud Services
Server
Figure 2: Industrial IoTs and Multilayer Computing Resources
6
Administration
Clean Refine Build
Security
Conform Orchestrate Score
Lifecycle Management
Transform Virtualize Model
Data Provenance
Shape Blend Visualize Dynamic Data Pipeline
Ingest Prepare Analyze Monitoring
Data Engineering Data Preparation Data Analytics Automation
Big Data Applications Big Data Analytics Pipeline Managing and Automating the Pipeline
Edge Servers
Infrastructure Based Cloud
Micro Cloud
Internet
Smart Router
Communication Cloudlet
Data Collection & IIoTs Channels
Concentric Computing System: Platform for Industry 4.0 Applications
Figure 3: Multistage execution, automating, and management of BDA processes (i.e., data engineering, data preparation
and data analytics) in concentric computing environments (such as sensors and wearable devices as endpoints, IoT
devices, edge servers, and cloud computing servers) [32].
120 methodologies help improve data quality by reducing the number of attributes and converting
121 data formats for uniform data processing.
122 2.4.2. Data Preparation

123 Big data emerge in raw form with large volume and enormous speed, and data scientists spend
124 70% − 80% of their time in data preparation activities [60]. Big data are refined using statistical
125 methods to handle unstructured, unbalanced and nonstandardised data points efficiently [61]. In
126 addition, data refinement helps summarise voluminous data to reduce overall complexity. As a
127 result, the spatiotemporal attributes of big data in IIoT systems vary. Ultimately, data locality is
128 necessary to reduce in-network traffic and latency in big-data applications [61]. Location-aware
129 highly virtualised data infrastructure can address these issues. However, data blending, which is
130 the process of combining data from multiple sources, becomes complex. Accordingly, further
7
131 involvement by data scientists [62] is required to perform data cleaning and noise removal [63].
132 Detection methods for outliers and anomalies are also needed to prepare big data for further
133 analysis [64, 65].
134 2.4.3. Data Analytics

135 The analytic processes in IIoT systems are executed in multiple phases [66]. Data scientists
136 generate learning models from high-quality well-prepared data. After the model is developed,
137 model scoring operations are performed by giving sample datasets and finding and ranking the
138 attributes in datasets/data streams. The correctly tuned models are deployed in production envi-
139 ronments to find the knowledge patterns from future data.
140 2.4.4. Managing and Automating the Data Pipeline

141 Although existing literature still lacks the concept of automated data pipelines in IIoT sys-
142 tems, BDA processes are executed as a sequence of operations during data engineering, prepa-
143 ration and analytics. Therefore, a holistic approach is needed to execute and administer BDA
144 processes across all layers of concentric computing systems. Life cycle management is needed
145 for full process execution from raw data acquisition to knowledge visualisation and actuation.
146 Data provenance, that is, designating ownership of data to different stack holders, also needs
147 serious attention to ensure system-wide control on data [67]. The continuous evolution in data
148 streams results in knowledge shift that enforces data pipelines to adaptively reconfigure analytic
149 processes. The data pipelines need to be continuously monitored for change detection, and the
150 entire BDA process needs to be re-executed to produce high-quality results [68]. In security
151 perspective, the cross-platform execution of BDA processes demands secure operations at IoT
152 device, CPS and big data levels [69].
153 3. Technologies and Algorithms for BDA in IIoT systems
154 A common example of IIoT systems is the concept of a smart factory system (SFS) [36]. The
155 key attributes of SFS and its subsystems are self-awareness, self-organisation, self-maintenance,
156 self-prediction, self-configuration and self-comparison [12]. This section presents the review of
157 early studies that presented BDA in the context of SFS and IIoT systems [12] in consideration of
158 the aforementioned autonomy related attributes (Table 1).
8
159 3.1. Mass Product Customization towards IIoT Lean Manufacturing
160 Although the main objective of IIoT systems is to maximise production considering massive
161 customisation in accordance with customer requirements, the existing literature still lacks an
162 end-to-end predictive analytics framework. Computational intelligence-based methods, such as
163 self-organising map (SOM) algorithms, are used to optimise big data for feeding in the produc-
164 tion systems and enable massively customised product manufacturing [70]. The neural network-
165 based SOM algorithm effectively enables smart production cycle in SFS. The cycle is based on
166 a close loop within a sequence of operations, including smart design, manufacturing, produc-
167 tion and services whereby feedback is collected after each cycle and subsequent operations at
168 each stage are improved. Clustering-based big data optimisation is another approach whereby k-
169 means clustering algorithms are used to cluster the attributes from customer data. The produced
170 clusters are used to intelligently improve the design process in the product life cycle [71]. An-
171 other alternate for massive product customisation is the adoption of cloud-based manufacturing
172 systems whereby big data integration is performed in cloud computing systems [72]. However,
173 the resultant big data are integrated from multiple sources, such as social media data streams
174 relevant to customer behaviour and IIoT data streams from manufacturing systems. This type
175 of cloud-based manufacturing benefits from open innovation and cross-continent physically iso-
176 lated product manufacturing.
177 3.2. Industrial Time Series Modeling
178 The achievement of zero-defect in SFS is a major challenge. In SFS, all manufacturing
179 components are perceived to be highly connected to ensure high-quality production. The term
180 zero-defect refers to ensuring high-quality production during the execution of a complete manu-
181 facturing process [45]. To this end, industrial time series modelling ensures the proper monitor-
182 ing of all manufacturing components during operations. However, data collection from multiple
183 components results in high-dimensional data streams. The neo-fuzzy neuron (NFN) time series
184 modelling method is adopted by IIoT systems. NFN can collaboratively connect the input data
185 streams with the final outputs. NFN benefits from the convergence of input data, which results
186 in decreased data streams and thus less iteration for learning model generation [73].
9
187 3.3. Intelligent Shop Floor Monitoring
188 The term physical Internet (PI) refers to the integration of cloud manufacturing with wireless
189 and networking technologies. PI in IIoT systems provides the backbone to IIoTs and smart
190 manufacturing object tracking systems based on radio-frequency identification. These smart
191 manufacturing objects represent different forms of products during manufacturing after each
192 process [47]. However, IIoT systems need to track these smart objects during production to
193 ensure that analytic processes provide intelligent shop monitoring. Researchers have proposed
194 a BDA-based approach for the trajectory clustering of moving objects in shop floors. Although
195 initial findings have been previously presented, a component-based BDA architecture is still
196 necessary to develop highly optimised and intelligent smart object tracking systems for shop
197 floor monitoring [47]. Performance analysis and exception diagnosis model have been proposed
198 and tested using Petri nets and decision tree algorithms [47]. The model shows feasibility, and
199 its real implementation in IIoT systems may help correctly quantify the results.
200 3.4. Industrial Microgrids
201 Massive data production in IIoT systems is evident due to feature-rich sensory and large-scale
202 deployment of IIoTs in SFS [74]. Therefore, manufacturing and environmental data, along with
203 energy consumption data, can lead towards optimised energy utilisation in SFS. The application
204 of BDA processes on these data silos can help improve planning, managing and utilising energy.
205 Researchers have proposed BDA analytics methods for industrial-level microgrid planning in
206 SFS. However, quantifiable studies that can lead towards efficient microgrid planning in IIoT
207 systems are still required [74].
208 3.5. Monitoring Machine Health
209 Prognostic health monitoring (PHM) helps find the machine behaviour for value creation
210 during mechanical operations and facilitate machine data collection and management for the
211 early diagnoses and prediction of machine faults. Several studies have performed analysis of
212 PHM data [75–77]. In accordance with multiple International Standards Organisation and In-
213 ternational Electrotechnical Commission and Society of Automotive Engineering standards, the
214 authors of [76] analysed ontological models developed from PHM data. These ontological mod-
215 els represent the hierarchical and semantic relationships amongst different machine components.
216 The remaining useful life of machine components, faults, errors and failures during machine
10
217 operations has also been explored. Studies have also presented dependency and failure mode
218 analyses of different machine components. The analysis of PHM data helps plan and schedule
219 machinery maintenance activities, thereby supporting in finding maintainable machine compo-
220 nents before total failure. However, finding the relationship amongst different attributes and the
221 failure impact of understudied machine components on other components in large-scale manu-
222 facturing environments is a challenging task [77].
223 3.6. Intelligent Predictive and Preventive Maintenance
224 Predictive and preventive maintenance are the key requirements of large-scale IIoT sys-
225 tems [11]. The BDA process can help in off-line prediction (i.e., performing prediction on the
226 basis of historical data) and online maintenance (i.e., maintaining machines without shutting
227 down the manufacturing units). Researchers have integrated Hadoop and Storm technologies for
228 big data processing and used neural network-based methods for prediction [78]. The concept of
229 adopting BDA for intelligent predictive maintenance is novel. However, new avenues need to be
230 explored to fully realise a real-time prediction system.
231 4. Taxonomy
232 Figure 4 presents the taxonomy that is devised on the basis of data sources, analytics tools,
233 analytics techniques, requirements, industrial analytics applications and analytics types.
234 4.1. Data Sources
235 In an industrial environment, numerous sources of data production, such as sensors, enter-
236 prise resource planning (ERP) systems, manufacturing execution systems (MES), supervisory
237 control and data acquisition (SCADA) systems, customer relationship management (CRM) sys-
238 tems and machine/IoT devices. ERP systems enable organisations to employ a system that is
239 composed of multiple integrated applications for managing business needs and automating many
240 back-office functions related to technology, services and human resources. MES helps keep
241 the track record of all manufacturing information in real time and receive up-to-date data from
242 robots, machine and IoT devices [79]. SCADA systems are used to monitor and control a plant
243 or equipment in industries (e.g. telecommunications, water and waste control, energy, oil and gas
244 refining and transportation). CRM systems are commonly used to manage a businesscustomer
11
245 relationship. Machines and IoT devices are also deployed in industries to perform specific tasks,
246 which generate an enormous amount of data on a daily basis. Applying analytics solutions to the
247 collected data through all the above-mentioned systems, machines and IoT devices can extract
248 valuable information that can help in decision-making purposes.
249 4.2. Analytics Tools
250 Several analytics tools are required to gain insights into a large amount of industrial data.
251 These tools include analytics software, algorithm repository, visualisation tools, modelling tools
252 and online analytics packages. Analytics software helps make predictions about unknown events.
253 An algorithm repository is a crowd-sourced repository of algorithms that is designed by analysts
254 using a common set of languages and a common interface. Visualisation tools help present data
255 in advanced formats (e.g. infographics, dials and gauges, geographic maps, sparklines, heat
256 maps and detailed bar, pie and fever charts). Modelling tools are used to define and analyse data
257 requirements for supporting business processes within the scope of corresponding information
258 systems in industries. Online analytics packages help keep track of and analyse data about web
259 traffic.
Taxonomy of BDA in IIoT
Data Sources Analytics Tools Analytics Requirements Industrial Analytics Types

Techniques Analytics
Analytics
Sensors
Software Maturity Models Applications Descriptive
Text Analytics Analytics
Manufacturing/
Algorithms Functional Operations
ERP Systems Real-time
Repository Architectures
Machine Learning Analytics
Logistics/Supply
Visualization Infrastructure Chain
MES Systems Predictive
Tools Architectures
Data Mining Analytics
Sales and
SCADA Systems Modelling Tools Integrated Marketing Prescriptive
Statistics Analytics Analytics
Research and
Online Analytics Development
CRM Systems
Packages Natural Language
Processing
Mechanical/IoT
Devices
Figure 4: Taxonomy of BDA in IIoT

12
Table 1: BDA Implementations in IIoT Systems
Ref. Problem(s) Objective(s) Analytic Compo- Mode Strengths Limitations Potential Solutions
nent(s)
[71] Finding accurate cus- Self-prediction Genetic Algorithm Historical data Smart product development Needs to be implemented Deep Learning for
tomers’ attributes for k-means clustering Rapid response to customer for real-time, Lacks large- BDA
mass customization. needs scale validation in BDA en-
vironments
[70] Finding accurate cus- Self-prediction Self-organizing Historical data Smart product development Needs to be implemented Re-enforcement learn-
tomers’ attributes for map Rapid response to customer for real-time, Lacks large- ing algorithms
mass customization. needs scale validation in BDA
[72] Enabling product cus- Self- Highlighted, but no Streaming data An end-to-end model for No real implementation Use-case implementa-
13
tomization and person- configuration real implementa- massive production and tion
alisation Self- tion discussed personalisation
organization
[73] Achieving zero-defect Self- Neo-Fuzzy Neuron Batch data Performs industrial process Accuracy needs to be im- Using alternate ML al-
problem configuration monitoring and modelling plemented gorithms
[47] implementing Physical Self-prediction Decision trees Batch data The implementation results Performance values for dif- Using alternate ML al-
Internet concept in in better prediction rate ferent workers needs to be gorithms
manufacturing shop well-defined to control the
floors rate of overestimation
[74] Developing a proactive Self-prediction A generic frame- Batch Data An end-to-end approach for Efforts are needed to ex- BDA Platform for full
and sustainable micro- work for knowl- microgrid data analysis plore analytics for full value chain Analytics
grid edge discovery value chain level knowl-
13 edge discovery in industrial
microgrids
[78] Active preventive main- Self- Neural Networks Batch data Real-time active mainte- Need to be investigated Real-time BDA plat-
tenance maintenance nance with real-time streaming form
data
260 4.3. Analytics Techniques
261 Various analytics techniques that can help obtain value from big industrial data are available,
262 thereby leading to making fast and better decisions. These analytics techniques include text
263 analytics, machine learning, data mining and statistical and natural language processing (NLP)
264 techniques. Text analytics helps derive high-quality information by unveiling patterns and trends
265 using statistical pattern learning. Machine learning techniques enable industrial devices and
266 machines to enter into a self-learning mode without being explicitly programmed. Data mining
267 solutions enable enterprises to transform raw data into knowledge. Statistical tools help collect,
268 summarise, analyse and interpret large amounts of industrial data, which lead to knowledge
269 discovery. In an industrial environment, NLP tools are used to extract and analyse unstructured
270 data.
271 4.4. Requirements
272 Certain requirements should be incorporated whilst developing new analytics systems for
273 IIoT. These requirements include maturity models, functional architecture, infrastructure archi-
274 tecture and integrated analysis. Maturity models help measure and monitor the capabilities of
275 analytics systems. They also help measure the effort required to complete a specific development
276 stage. In summary, these models help monitor the health of an organisations big data programs.
277 Functional architecture is an architectural model that helps identify the functions of analytics
278 systems and their interactions. In addition, it defines how system functions work together to
279 perform a specified system mission. In an industrial environment, analytics systems must be
280 developed such that they can handle an enormous amount of data in real time. In this context,
281 big data infrastructure requires experienced scientists to design the infrastructure from existing
282 equipment in an industrial paradigm. One of the key requirements for analytics systems is that
283 they should support the integrated analysis of multiple types of industrial IoT data.
284 4.5. Industrial Analytics Applications
285 Typical industrial analytics applications across the industrial value chain are as follows: man-
286 ufacturing/operations, logistics/supply chain, marketing/sales and research and development.
287 The use of predictive analytics in manufacturing can lead to rescheduling a maintenance plan
288 prior to machine failure by considering past machine performance history. Moreover, it can help
289 develop decision support systems for industrial processes. The appropriate use of analytics can
14
290 play an important role in the logistics/supply chain (e.g. condition monitoring, supply chain op-
291 timisation, fleet management and strategic supplier management). Analytics can help identify
292 failing parts during product usage through sensor readings and gradually improve product char-
293 acteristics (research and development). In the marketing field, analytics tools enable enterprises
294 to predict and enhance future sales (e.g. help in determining seasonal trends that can lead to
295 developing an adaptive marketing strategy).
296 4.6. Analytics Types

297 Analytics has four types: descriptive, real-time, predictive and prescriptive analytics. De-
298 scriptive analytics helps gain insights into historical data (e.g. number of defective items in the
299 past and the reason for the defects). Meanwhile, real-time analytics enables enterprises to be-
300 come aware of current situations (e.g., current status and location of a product and detection
301 of a faulty machine). By contrast, predictive analytics helps identify potential issues that can
302 occur in advance by using statistical and machine learning techniques (e.g. expected inventory
303 levels, anticipated demand levels, and prediction of equipment failure). Lastly, prescriptive ana-
304 lytics provides advice or suggestion on the best possible action that an end user should take (e.g.
305 whether a machine is receiving the right raw materials in the correct amount).
306 5. Frameworks and Case Studies
307 Value creation is a major sustainability factor in modern enterprises whereby BDA processes
308 are becoming the primary driver in creating values for customers and enterprises [80]. IIoT
309 systems are no exception. BDA processes can facilitate the amalgamation of customer and enter-
310 prise data to ensure massively customised production with zero defects. IIoT systems essentially
311 integrate historical and real-time stored and streaming data at various levels. This multisource
312 data integration leads to highly effective designs for new business models. Enterprises focus
313 on different aspects of industry-wide value creation mechanisms, such as defining value propo-
314 sitions, value capturing mechanisms, value networks and value communication strategies for
315 internal and external stakeholders [18]. Ideally, BDA processes can facilitate enterprise-level
316 value creation whereby inbound intelligence is obtained by creating value for internal enterprise
317 operations. Alternatively, outbound intelligence leads towards value creation for customers. De-
318 spite these opportunities, unlocking the perceived value from BDA technologies is challenging.
319 The existing literature presents only a few such frameworks and use cases as follows.
15
320 5.1. SnappyData
321 SnappyData is an open-source BDA framework that integrates Apaches Spark and GemFire
322 technologies [81]. Apaches Spark is adopted for big data processing, whereas GemFire facil-
323 itates highly scalable in-memory transactional data storage. The strength of SnappyData is its
324 unified BDA engine that facilitates the performance of different types of analytical operation,
325 such as online transaction processing, online analytical processing and streaming the data an-
326 alytics of operational data. Despite its high performance, SnappyData still underperforms in
327 cases with highly streaming data, which causes a bottleneck in real-time interactive visualisation
328 performance.
329 5.2. Ipanera
330 Soilless food production systems, such as Ipanera, are being aligned with IIoT systems [82].
331 Ipanera continuously monitors water level and fertilizer quality in a field and generates insights
332 for self-configuration. Although researchers have presented the concept, the Ipanera architecture
333 involves multiple layers of physical devices and systems. It includes sensor nodes at the end
334 points that actively collect data streams and transfer them to nearby IIoT clusters. These clusters
335 are responsible for end point management, communication and configuration in a field. In addi-
336 tion, IIoT clusters provide feedback to end points to reconfigure their data collection behaviour.
337 IIoT clusters transfer data streams to distributed analytics servers that run Apaches Hadoop [83],
338 MapReduce [84] and Spark [85] technologies for data processing and BDA. Ipanera provides
339 support for streaming analytics that is used to trigger alerts for end points in case a new event is
340 detected. Persistent storage and on-the-air configurations are two innovative features of the Ipan-
341 era architecture. This architecture is currently under development; hence, the complete design of
342 the proposed architecture is still unavailable.
343 5.3. Fault Detection Classification
344 Large-scale distributed cyber manufacturing systems are based on multiple interconnected
345 but geographically dispersed manufacturing units [86]. The fault detection and classification
346 (FDC) framework finds manufacturing faults in products. The core of the FDC architecture
347 is the integration of IoT devices into CPS and cloud computing technologies. IoT devices in
348 production facilities continuously collect and analyse data streams to detect various signals that
349 are transferred to back-end cloud servers. These cloud servers execute BDA processes to detect
16
350 and classify faulty products using deep belief networks based on deep learning methods [87, 88].
351 FDC was analysed by deploying it in a car headlight manufacturing unit that produced reliable
352 results.
353 5.4. BDA Architecture for Cleaner Production

354 The term cleaner production refers to ensuring reduced environmental impacts during the
355 execution of the entire product life cycle. It is based on three phases [89]. The first phase is
356 about product design and manufacturing. The second phase involves product use, service pro-
357 visioning and maintenance. The third phase is concerned with product remanufacturing, reuse
358 and recycling. Considering the importance of such clean technologies, researchers have pro-
359 posed a four-stage BDA architecture. In the first stage, the architecture considers value creation
360 objectives during a products life cycle, such as improving product designs and ensuring energy
361 efficiency, proactive maintenance and environmental efficiency. In the second stage, big data
362 acquisition and integration are performed using IoT devices. In the third stage, big data are pro-
363 cessed using Apaches Hadoop and Storm technologies. Finally, BDA processes are executed
364 in the fourth stage whereby the architecture provides clustering, classification, association rule
365 mining and prediction-related algorithms. The proposed architecture was evaluated and tested
366 on an axial compressor manufacturing unit. The annual reports of the production unit show that
367 the proposed architecture realises all the value creation objectives for cleaner production.
368 5.5. Smart Maintenance Initiative: Railway Case Study

369 Apart from SFS, Japan is attempting to upgrade its railway system to a new level by adopting
370 IIoT systems for the smart maintenance of railway tracks [90]. To achieve its smart maintenance
371 vision, Japans railway is adopting IIoT, BDA and automation technologies. The smart mainte-
372 nance vision will provide a solution to four challenges: 1) ensuring condition-based maintenance,
373 2) providing work support through artificial intelligence (AI), 3) managing railway assets and 4)
374 performing database integration. The progress details of Japans railway towards this vision are
375 available in this report [90] for interested readers.
376 6. Opportunities, Research Challenges, and Future Technologies
377 Considering the vision of IIoT systems, BDA will evidently help enterprises in the value
378 creation process. BDA processes will maximise operational efficiency, reduce product develop-
17
379 ment cost, ensure massively customised production and streamline the supply chain management.
380 However, this review shows that the existing literature is considerably lagging behind this vision.
381 Table 2 presents the summary of research challenges and their perceived solutions to fully adopt
382 IIoT systems in BDA.
383 6.1. Opportunities
384 The adoption of BDA processes in IIoT systems results in multidimensional research oppor-
385 tunities.
386 6.1.1. Automation and AI

387 The enrichment of intelligent features can lead towards highly optimised and automated in-
388 dustrial processes [91, 92]. Therefore, AI will be the core component of big data optimisation
389 and analytics, which will result in highly efficient industrial processes [93]. Future IIoT systems
390 will integrate and ingest big data from various online and off-line and inbound and outbound op-
391 erations. The integration of customer and enterprise data will result in high-dimensional, multi-
392 million variable datasets. AI methods will help optimise and analyse such big datasets [94, 95].
393 6.1.2. Human Machine Interaction

394 Wearable computing and augmented reality technologies are leading towards new human-
395 machine interaction models and interfaces [96, 97]. The enrichment of such interaction models
396 with real-time knowledge patterns from big data systems will result in highly productive and rich
397 user interfaces. In addition, robotics technologies (for physical and virtual robots) will be widely
398 adopted by future IIoT systems. Therefore, BDA processes will enrich intelligence to produce
399 highly autonomous and self-sustaining non-obtrusive systems.
400 6.1.3. Cybersecurity, Privacy, and Ethics

401 Cybersecurity will become an essential requirement due to connected intelligence in IIoT
402 systems. BDA processes will help provide real-time cyber threat intelligence by analysing se-
403 curity attacks, privacy leaks, unauthorised data access and unethical data collection [98]. In
404 addition, BDA processes will help analyse network and information security-related enterprise
405 data to find anomalies, outliers, threats, attacks and vulnerabilities across IIoT systems [99].
18
406 6.1.4. Universal Standards
407 The adoption of BDA processes is still in its initial stage; thus, existing systems may not
408 be compliant with universal standards across all or multiple industries [2, 100]. New universal
409 standards are required to define the type of big data that the industries can collect from customers,
410 determine how data should be secured, preserved and shared and identify the stakeholders who
411 will benefit from the data. In addition, standards must also ensure the perceived benefits to
412 customers in exchange for their personal data. These universal standards will help address ethical
413 issues in big data systems and create value for customers by providing personalised products and
414 services.
415 6.1.5. Protocols for Interoperability

416 Practically, multiple industries are involved in the entire processfrom customer data acqui-
417 sition to finished product/service and supply chain management [2]. Interoperability is a major
418 consideration among different industries; however, new protocols are required to realise fully
419 interoperable IIoT systems. These protocols can lead towards value creation for enterprises, al-
420 though a few questions must be addressed, such as what are the interoperability parameters, how
421 will BDA processes be executed in cross-industry systems and how will heterogeneity in data,
422 computing technologies and industrial production systems be handled. A well-defined interop-
423 erability protocol can help answer these questions.
424 6.1.6. End-to-end Industrial Analytics

425 Big data in IIoT systems evolves from multiple inbound and outbound data sources, such as
426 customer data and operational data from finance, marketing, human resources, IoT devices, CPS
427 and manufacturing systems [101]. Nevertheless, existing systems manage all these data sources
428 separately to execute BDA processes. An opportunity exists to develop an end-to-end industrial
429 analytics pipeline that can handle big data from various data sources in parallel and find highly
430 correlated knowledge patterns that emerge across entire IIoT systems [102].
431 6.1.7. Precision Manufacturing

432 BDA processes can help enrich precision manufacturing systems [103]. The classification
433 and categorisation of customers needs and behaviour-related data can lead towards innovative
434 product designs. Enterprises will be able to offer the right products and services to the right cus-
19
435 tomers. Precision manufacturing will considerably help in equal value creation for customers and
436 enterprises. Early examples of precision manufacturing systems are available in the healthcare
437 industry [104]. However, these systems should be integrated into IIoT systems [103].
438 6.2. Research Challenges and Future Technologies
439 Considering the opportunities, research efforts are required to improve the entire technology
440 ecosystem for IIoT systems.
441 6.2.1. Big Data Process Integration into IIoT Systems

442 Ideally, IIoT systems should execute real-time highly interactive big data applications. In
443 practice, however, considerable effort is required for planning, creating, deploying, maintaining
444 and continuously improving domain-specific big data processes for each industry. Future BDA
445 processes should be able to provide real-time knowledge patterns and industry-wide intelligence
446 through single dashboard applications. In this regard, all legacy and state-of-the art data sources
447 should be vertically aligned such that enterprises can easily analyse and correlate different indus-
448 trial processes and operations.
449 6.2.2. Orchestrating BDA Applications Using Concentric Computing

450 Concentric computing systems provide computational and storage support through different
451 devices and systems [8]. Thus, massive heterogeneity should be addressed in terms of processing
452 capabilities, in-memory and disk-based storage systems, battery-powered and fully powered de-
453 vices and systems and multiple communication channels with varying bandwidth capacities [17].
454 Big data applications on top of concentric computing systems should be designed by considering
455 efficiency objectives in terms of storage, in-network data movement, energy consumption, pri-
456 vacy, security and real-time knowledge availability [105, 106]. In this regard, priority should be
457 given to devices and systems near data sources. This approach can help maximise value creation
458 for enterprises in terms of operating cost for big data systems. Given that maximum data collec-
459 tion, filtration and processing are performed before data arrive in cloud computing systems, the
460 operational costs for data storage and cloud service utilisation will therefore be minimised [80].
461 Another benefit of concentric computing systems is their ability to ensure real-time or near real-
462 time intelligence near end points, IoT devices and other data sources in IIoT systems [36].
20
463 6.2.3. Emerging and Complimentary Technologies for IIoT Systems
464 On the one hand, BDA adoption is increasing in IIoT systems. On the other hand, IIoT sys-
465 tems should address massive heterogeneity without compromising overall operational efficiency
466 due to emerging, complementary technologies, such as IoT. Considering this condition, a few
467 technologies will become integral parts of future IIoT systems.
468 Virtualisation is the essence of distributed systems, such as cloud computing systems and
469 concentric computing systems. Virtualisation is traditionally performed at multiple levels, such
470 as operating systems, networks, storage, applications and hardware. Operating system-level vir-
471 tualisation is the most common whereby operating system kernels and functions are virtualised
472 as virtual machines (VMs). However, the mobility of IoT devices requires continuous VM mi-
473 gration among different computer servers [2, 107]. Containerisation is the emerging technology
474 that is gradually replacing VMs by sharing a single kernel among different applications on the
475 same type of operating systems. Containerisation technologies offers more secure and faster pro-
476 cessing; hence, they have become highly beneficial for addressing timeliness and latency issues
477 when BDA processes are executed using VMs [108].
478 Large enterprises traditionally adopt highly coupled SOAs, which are difficult to test and re-
479 sult in high maintenance cost. Microservices are emerging alternatives to SOAs whereby highly
480 scalable and loosely coupled cloud services are orchestrated [109]. The microservice architec-
481 ture can be adopted best for BDA processes because these processes should be executed across
482 multiple platforms and devices in IIoT systems [110]. The details of microservice architectures
483 implementation are available in [111] for interested readers.
484 The multipoint, multisite and high-dimensional data production in IIoT systems results in
485 complex big datasets. Graph and network theories can help reduce this massive complexity [112].
486 Graph data structures and big graph analytics methods can be adopted to separate, map and
487 analyse big data in different graph formats. The adoption of big graph analytics can lead towards
488 efficient and highly optimised execution of BDA processes across IIoT systems.
21
Table 2: Summary of Research Challenges and their Perceived Solutions
Type Issues Causes Solutions
Cybersecurity - Internal Attacks - Security Vulnerabilities - Intelligent Monitoring Tools Needed
- External Attacks - Openness of Systems - Deployment of End-to-End Security Models is Es-
sential
- System-wide Forensic Analysis should be per-
formed periodically
Privacy - Identity Breaches - Bad Security Models - Using Data Anonymisation Protocols
- Personal Data Theft - Absence of Standard Operating Procedures - Privacy preserving interaction models for users,
- Business Data Leakage - Weak Data and Information Sharing Policies devices, and systems
Big Data Processing - Bad Data Integration Heterogeneous Data Sources - Intelligent Real time Data Fusion
22
- Missing Data Streams - Mobility and Connectivity Issues - Device-centric big data processing architectures
- High Latency - Data overloading and Bandwidth Limitations - Concentric Computing Models
Standardization - Difficulty in Interoperability and and Sys- - Absence of Global Standardization Body - Developing Local, Regional, Industry-specific,
tem Integration and Global Standards
Connectivity and Commu- - Bad and Inaccurate Data Transfer - High Mobility - Need to create always-on, ultra-high available and
nication - Data Loss - Large Data Streams reliable communication protocols
- High Latency - Congestion
Scalability - Resource Discovery - Low Processing Power at device-end - Near-device data processing, In-memory Data
- Data offloading - Massive Data Production Processing, Edge Computing
- Data Management - Realtime Actuation
System management - Difficult to deploy, configure, monitor, and - Cloud-centric - Device-centric
control large scale IIoT networks
Efficiency - High Energy Utilization 22- Always-on IIoT Devices and Systems - Enabling Energy, memory, and computation-
- Resources-constraints - Massive and Continuous Data Generation and De- efficient algorithms and processes for big data pro-
- Device-overloading vice Operations cessing, management and analytics in IIOTs
- On-device Data Management and Analytics
489 Emerging technologies, such as fog computing and blockchain, can play a pivotal role in
490 BDA for IIoT [113]. Fog computing has been widely used in IoT devices [114], particularly
491 those for IIoT and smart manufacturing, for localised and timely data processing and storage,
492 and primarily to offset long delays that can be incurred in a cloud environment [115]. Blockchain
493 is the underlying technology for bitcoins; however, it has been foreseen as a distributed ledger
494 that can provide decentralised storage for data generated by IoT devices. Data are stored in a
495 blockchain ledger with high integrity, authenticity, resiliency and trust [116]. All transactions
496 are cryptographically signed by IIoT devices and validated in a decentralised manner without
497 an intermediary. The data origin is validated before being recorded on the ledger. Moreover,
498 blockchain smart contracts can be used to provide decentralised authentication, management and
499 control access to data generated by IIoT devices. Smart contracts are basically codes that are
500 executed by all blockchain miners, and the execution outcome is verified and agreed upon by
501 all mining nodes. Furthermore, given the limited computing, networking and storage capacities
502 of IIoT devices, fog nodes are envisioned to be equipped with cloud and blockchain interfaces
503 in the future to communicate and interface with the cloud environment and the blockchain net-
504 work [116].
505 7. Conclusions
506 The vision of Industry 4.0 to connect manufacturing systems with distributors and consumers
507 can only be achieved by adopting IIoT and BDA processes as core components for value creation.
508 This paper discusses the rise of big data in IIoT systems and presents a detailed survey of related
509 technologies, algorithms, frameworks and case studies. A detailed taxonomy is provided to
510 classify the key concepts in this important research area. Several indispensable frameworks
511 and case studies are outlined and discussed. Furthermore, we present a detailed discussion of
512 future opportunities, technologies and research challenges. We conclude that the adoption of
513 BDA in IIoT systems is still in its early stage. Research on complementary components of IIoT
514 systems, such as IoT devices, augmented reality and CPS, is also in its infancy. Current BDA
515 systems provide generic frameworks for data engineering, preparation and analysis. However,
516 considerable effort is required to alter existing BDA processes to meet the demands of IIoT
517 systems. Future research should be conducted to devise new standards for interoperability among
518 cross-Industry 4.0 BDA platforms and to provide capability for end-to-end reliable application
23
519 processing by considering the anatomy of concentric computing systems.
520 Acknowledgement
521 Imran’s work is supported by the Deanship of Scientific Research at King Saud University
522 through Research group No. (RG # 1435-051).
523 References
524 [1] I. Akkaya, P. Derler, S. Emoto, E. A. Lee, Systems engineering for industrial cyber–physical systems using
525 aspects, Proceedings of the IEEE 104 (2016) 997–1012.
526 [2] M. Wollschlaeger, T. Sauter, J. Jasperneite, The future of industrial communication: Automation networks in the
527 era of the internet of things and industry 4.0, IEEE Industrial Electronics Magazine 11 (2017) 17–27.
528 [3] C. Perera, S. Y. Wakenshaw, T. Baarslag, H. Haddadi, A. K. Bandara, R. Mortier, A. Crabtree, I. C. Ng,
529 D. McAuley, J. Crowcroft, Valorising the iot databox: creating value for everyone, Transactions on Emerg-
530 ing Telecommunications Technologies 28 (2017).
531 [4] C. Perera, C. H. Liu, S. Jayawardena, M. Chen, A survey on internet of things from industrial market perspective,
532 IEEE Access 2 (2014) 1660–1679.
533 [5] M. R. Pedersen, L. Nalpantidis, R. S. Andersen, C. Schou, S. Bøgh, V. Krüger, O. Madsen, Robot skills for
534 manufacturing: From concept to industrial deployment, Robotics and Computer-Integrated Manufacturing 37
535 (2016) 282–291.
536 [6] L. Liu, Services computing: From cloud services, mobile services to internet of services, IEEE Transactions on
537 Services Computing 9 (2016) 661–663.
538 [7] S. Kim, M. A. Nussbaum, J. L. Gabbard, Augmented reality ?smart glasses? in the workplace: Industry perspec-
539 tives and challenges for worker safety and health, IIE transactions on occupational ergonomics and human factors
540 4 (2016) 253–258.
541 [8] P. Howard, Concentric analytics for iot, Online, 2017.
542 [9] I. Yaqoob, E. Ahmed, I. A. T. Hashem, A. I. A. Ahmed, A. Gani, M. Imran, M. Guizani, Internet of things
543 architecture: Recent advances, taxonomy, requirements, and open challenges, IEEE wireless communications 24
544 (2017) 10–16.
545 [10] E. Ahmed, I. Yaqoob, A. Gani, M. Imran, M. Guizani, Internet-of-things-based smart environments: state of the
546 art, taxonomy, and open research challenges, IEEE Wireless Communications 23 (2016) 10–16.
547 [11] K. Wang, Intelligent predictive maintenance (ipdm) system–industry 4.0 scenario, WIT Transactions on Engi-
548 neering Sciences 113 (2016) 259–268.
549 [12] J. Lee, Smart factory systems, Informatik-Spektrum 38 (2015) 230–235.
550 [13] M. Marjani, F. Nasaruddin, A. Gani, A. Karim, I. Hashem, A. Siddiqa, I. Yaqoob, Big iot data analytics: Archi-
551 tecture, opportunities, and open research challenges, IEEE Access (2017).
552 [14] M. H. Rehman, A. Batool, et al., Big data analytics in mobile and cloud computing environments, in: Innovative
553 Research and Applications in Next-Generation High Performance Computing, IGI Global, 2016, pp. 349–367.
24
554 [15] K. Geng, L. Liu, Research of construction and application of cloud storage in the environment of industry 4.0, in:
555 International Conference on Industrial IoT Technologies and Applications, Springer, pp. 104–113.
556 [16] M. M. Gaber, A. Aneiba, S. Basurra, O. Batty, A. M. Elmisery, Y. Kovalchuk, M. H. Rehman, Internet of things
557 and data mining: From applications to techniques and systems, Wiley Interdisciplinary Reviews: Data Mining
558 and Knowledge Discovery (2019) e1292.
559 [17] M. H. Rehman, L. C. Sun, T. Y. Wah, M. K. Khan, Towards next-generation heterogeneous mobile data stream
560 mining applications: opportunities, challenges, and future research directions, Journal of Network and Computer
561 Applications (2016).
562 [18] M. Ehret, J. Wirtz, Unlocking value from machines: business models and the industrial internet of things, Journal
563 of Marketing Management 33 (2017) 111–130.
564 [19] L. Da Xu, W. He, S. Li, Internet of things in industries: A survey, IEEE Transactions on industrial informatics 10
565 (2014) 2233–2243.
566 [20] S. Mumtaz, A. Alsohaily, Z. Pang, A. Rayes, K. F. Tsang, J. Rodriguez, Massive internet of things for industrial
567 applications: Addressing wireless iiot connectivity challenges and ecosystem fragmentation, IEEE Industrial
568 Electronics Magazine 11 (2017) 28–33.
569 [21] F. Al-Turjman, S. Alturjman, Context-sensitive access in industrial internet of things (iiot) healthcare applications,
570 IEEE Transactions on Industrial Informatics 14 (2018) 2736–2744.
571 [22] R. Chaudhary, G. S. Aujla, S. Garg, N. Kumar, J. J. Rodrigues, Sdn-enabled multi-attribute-based secure commu-
572 nication for smart grid in iiot environment, IEEE Transactions on Industrial Informatics 14 (2018) 2629–2640.
573 [23] C. Perera, C. H. L. Member, S. Jayawardena, M. Chen, Context-aware computing in the internet of things: a
574 survey on internet of things from industrial market perspective, arXiv preprint arXiv:1502.00164 (2015).
575 [24] C. Perera, C. H. Liu, S. Jayawardena, The emerging internet of things marketplace from an industrial perspective:
576 A survey, IEEE Transactions on Emerging Topics in Computing 3 (2015) 585–598.
577 [25] E. Ahmed, I. Yaqoob, I. A. T. Hashem, I. Khan, A. I. A. Ahmed, M. Imran, A. V. Vasilakos, The role of big data
578 analytics in internet of things, Computer Networks 129 (2017) 459–471.
579 [26] C.-W. Tsai, C.-F. Lai, H.-C. Chao, A. V. Vasilakos, Big data analytics: a survey, Journal of Big Data 2 (2015) 21.
580 [27] Y. Chen, H. Chen, A. Gorkhali, Y. Lu, Y. Ma, L. Li, Big data analytics and big data science: a survey, Journal of
581 Management Analytics 3 (2016) 1–42.
582 [28] M. K. Saggi, S. Jain, A survey towards an integration of big data analytics to big insights for value-creation,
583 Information Processing Management 54 (2018) 758 – 790. In (Big) Data we trust: Value creation in knowledge
584 organizations.
585 [29] J. Wang, Y. Wu, N. Yen, S. Guo, Z. Cheng, Big data analytics for emergency communication networks: A survey,
586 IEEE Communications Surveys & Tutorials 18 (2016) 1758–1778.
587 [30] H. Hu, Y. Wen, T.-S. Chua, X. Li, Toward scalable systems for big data analytics: A technology tutorial, IEEE
588 access 2 (2014) 652–687.
589 [31] B. Li, M. C. Kisacikoglu, C. Liu, N. Singh, M. Erol-Kantarci, Big data analytics for electric vehicle integration
590 in green smart cities, IEEE Communications Magazine 55 (2017) 19–25.
591 [32] M. H. Rehman, E. Ahmed, I. Yaqoob, I. A. T. Hashem, M. Imran, S. Ahmad, Big data analytics in industrial iot
592 using a concentric computing model, IEEE Communications Magazine 56 (2018) 37–43.
25
593 [33] S. Radhakrishnan, S. Kamarthi, Convergence and divergence in academic and industrial interests on iot based
594 manufacturing, in: Big Data (Big Data), 2016 IEEE International Conference on, IEEE, pp. 2051–2056.
595 [34] D. Georgakopoulos, P. P. Jayaraman, Internet of things: from internet scale sensing to smart services, Computing
596 98 (2016) 1041–1058.
597 [35] I. Lee, K. Lee, The internet of things (iot): Applications, investments, and challenges for enterprises, Business
598 Horizons 58 (2015) 431–440.
599 [36] D. Georgakopoulos, P. P. Jayaraman, M. Fazia, M. Villari, R. Ranjan, Internet of things and edge cloud computing
600 roadmap for manufacturing, IEEE Cloud Computing 3 (2016) 66–73.
601 [37] R. F. Babiceanu, R. Seker, Big data and virtualization for manufacturing cyber-physical systems: A survey of the
602 current status and future outlook, Computers in Industry 81 (2016) 128–137.
603 [38] L. Wang, M. Törngren, M. Onori, Current status and advancement of cyber-physical systems in manufacturing,
604 Journal of Manufacturing Systems 37 (2015) 517–527.
605 [39] H. S. Kang, J. Y. Lee, S. Choi, H. Kim, J. H. Park, J. Y. Son, B. H. Kim, S. Do Noh, Smart manufactur-
606 ing: Past research, present findings, and future directions, International Journal of Precision Engineering and
607 Manufacturing-Green Technology 3 (2016) 111–128.
608 [40] R. Addo-Tenkorang, P. T. Helo, Big data applications in operations/supply-chain management: A literature review,
609 Computers & Industrial Engineering 101 (2016) 528–543.
610 [41] T.-M. Choi, H. K. Chan, X. Yue, Recent development in big data analytics for business operations and risk
611 management, IEEE transactions on cybernetics 47 (2017) 81–92.
612 [42] C. Chang, S. N. Srirama, R. Buyya, Mobile cloud business process management system for the internet of things:
613 a survey, ACM Computing Surveys (CSUR) 49 (2016) 70.
614 [43] K. Zhou, T. Liu, L. Liang, From cyber-physical systems to industry 4.0: make future manufacturing become
615 possible, International Journal of Manufacturing Research 11 (2016) 167–188.
616 [44] J. Lee, H. D. Ardakani, S. Yang, B. Bagheri, Industrial big data analytics and cyber-physical systems for future
617 maintenance & service innovation, Procedia CIRP 38 (2015) 3–7.
618 [45] F.-T. Cheng, H. Tieng, H.-C. Yang, M.-H. Hung, Y.-C. Lin, C.-F. Wei, Z.-Y. Shieh, Industry 4.1 for wheel
619 machining automation, IEEE Robotics and Automation Letters 1 (2016) 332–339.
620 [46] L. Cheng, T. Wang, X. Hong, Z. Wang, J. Wang, J. Liu, A study on the architecture of manufacturing internet of
621 things, International Journal of Modelling, Identification and Control 23 (2015) 8–23.
622 [47] R. Y. Zhong, C. Xu, C. Chen, G. Q. Huang, Big data analytics for physical internet-based intelligent manufacturing
623 shop floors, International Journal of Production Research 55 (2017) 2610–2621.
624 [48] F. Tao, Y. Cheng, L. Da Xu, L. Zhang, B. H. Li, Cciot-cmfg: cloud computing and internet of things-based cloud
625 manufacturing service system, IEEE Transactions on Industrial Informatics 10 (2014) 1435–1442.
626 [49] M. S. Hossain, G. Muhammad, Cloud-assisted industrial internet of things (iiot)–enabled framework for health
627 monitoring, Computer Networks 101 (2016) 192–202.
628 [50] H. Gupta, A. V. Dastjerdi, S. K. Ghosh, R. Buyya, ifogsim: A toolkit for modeling and simulation of resource man-
629 agement techniques in internet of things, edge and fog computing environments, arXiv preprint arXiv:1606.02007
630 (2016).
631 [51] M. H. Rehman, P. P. Jayaraman, C. Perera, The emergence of edge-centric distributed iot analytics platforms, in:
26
632 Internet of Things, Chapman and Hall/CRC, 2017, pp. 213–228.
633 [52] K. Lin, F. Xia, G. Fortino, Data-driven clustering for multimedia communication in internet of vehicles, Future
634 Generation Computer Systems 94 (2019) 610–619.
635 [53] P. Pace, G. Aloi, R. Gravina, G. Caliciuri, G. Fortino, A. Liotta, An edge-based architecture to support efficient
636 applications for healthcare industry 4.0, IEEE Transactions on Industrial Informatics 15 (2019) 481–489.
637 [54] N. Hassan, S. Gillani, E. Ahmed, I. Yaqoob, M. Imran, The role of edge computing in internet of things, IEEE
638 Communications Magazine (2018) 1–6.
639 [55] C. Ji, Q. Shao, J. Sun, S. Liu, L. Pan, L. Wu, C. Yang, Device data ingestion for industrial big data platforms with
640 a case study, Sensors 16 (2016) 279.
641 [56] H. Wang, M. Li, Y. Bu, J. Li, H. Gao, J. Zhang, Cleanix: A parallel big data cleaning system, ACM SIGMOD
642 Record 44 (2016) 35–40.
643 [57] A. F. Haryadi, J. Hulstijn, A. Wahyudi, H. van der Voort, M. Janssen, Antecedents of big data quality: An
644 empirical examination in financial service organizations, in: Big Data (Big Data), 2016 IEEE International
645 Conference on, IEEE, pp. 116–121.
646 [58] J. LeFevre, R. Liu, C. Inigo, L. Paz, E. Ma, M. Castellanos, M. Hsu, Building the enterprise fabric for big data
647 with vertica and spark integration, in: Proceedings of the 2016 International Conference on Management of Data,
648 ACM, pp. 63–75.
649 [59] Y. Wang, L. Kung, W. Y. C. Wang, C. G. Cegielski, Integrated big data analytics-enabled transformation model:
650 Application to health care, Information & Management (2017).
651 [60] G. Press, Cleaning big data: Most time-consuming, least enjoyable data science task, survey says, Online, 2016.
652 [61] P. Wiener, M. Stein, D. Seebacher, J. Bruns, M. Frank, V. Simko, S. Zander, J. Nimis, Biggis: a continuous refine-
653 ment approach to master heterogeneity and uncertainty in spatio-temporal big data (vision paper), in: Proceedings
654 of the 24th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, ACM,
655 p. 8.
656 [62] B. Ong, R. Wen, A. N. Zhang, Data blending in manufacturing and supply chains, in: Big Data (Big Data), 2016
657 IEEE International Conference on, IEEE, pp. 3773–3778.
658 [63] D. Garcı́a-Gil, J. Luengo, S. Garcı́a, F. Herrera, Enabling smart data: Noise filtering in big data classification,
659 arXiv preprint arXiv:1704.01770 (2017).
660 [64] M. Bai, X. Wang, J. Xin, G. Wang, An efficient algorithm for distributed density-based outlier detection on big
661 data, Neurocomputing 181 (2016) 19–28.
662 [65] L. Stojanovic, M. Dinic, N. Stojanovic, A. Stojadinovic, Big-data-driven anomaly detection in industry (4.0):
663 An approach and a case study, in: Big Data (Big Data), 2016 IEEE International Conference on, IEEE, pp.
664 1647–1652.
665 [66] M. Niño, F. Sáenz, J. M. Blanco, A. Illarramendi, Requirements for a big data capturing and integration archi-
666 tecture in a distributed manufacturing scenario, in: Industrial Informatics (INDIN), 2016 IEEE 14th International
668 [67] R. Wang, D. Sun, G. Li, M. Atif, S. Nepal, Logprov: Logging events as provenance of big data analytics pipelines
669 with trustworthiness, in: IEEE Conference on Big Data.
670 [68] K. Yamanishi, K. Miyaguchi, Detecting gradual changes from data stream using mdl-change statistics, in: Big
27
671 Data (Big Data), 2016 IEEE International Conference on, IEEE, pp. 156–163.
672 [69] X. Cheng, L. Fang, X. Hong, L. Yang, Exploiting mobile big data: Sources, features, and applications, IEEE
673 Network 31 (2017) 72–79.
674 [70] A. A. F. Saldivar, C. Goh, W.-n. Chen, Y. Li, Self-organizing tool for smart design with predictive customer
675 needs and wants to realize industry 4.0, in: Evolutionary Computation (CEC), 2016 IEEE Congress on, IEEE, pp.
676 5317–5324.
677 [71] A. A. F. Saldivar, C. Goh, Y. Li, Y. Chen, H. Yu, Identifying smart design attributes for industry 4.0 customization
678 using a clustering genetic algorithm, in: Automation and Computing (ICAC), 2016 22nd International Conference
679 on, IEEE, pp. 408–414.
680 [72] C. Yang, S. Lan, W. Shen, G. Q. Huang, X. Wang, T. Lin, Towards product customization and personalization in
681 iot-enabled cloud manufacturing, Cluster Computing (2017) 1–14.
682 [73] D. Zurita, M. Delgado, J. A. Carino, J. A. Ortega, G. Clerc, Industrial time series modelling by means of the
683 neo-fuzzy neuron, IEEE Access 4 (2016) 6151–6160.
684 [74] C. Gamarra, J. M. Guerrero, E. Montero, A knowledge discovery in databases approach for industrial microgrid
685 planning, Renewable and Sustainable Energy Reviews 60 (2016) 615–630.
686 [75] H. Fleischmann, J. Kohl, J. Franke, A. Reidt, M. Duchon, H. Krcmar, Improving maintenance processes with
687 distributed monitoring systems, in: Industrial Informatics (INDIN), 2016 IEEE 14th International Conference on,
688 IEEE, pp. 377–382.
689 [76] D. L. Nuñez, M. Borsato, An ontology-based model for prognostics and health management of machines, Journal
690 of Industrial Information Integration (2017).
691 [77] E. G. Nabati, K.-D. Thoben, Data driven decision making in planning the maintenance activities of off-shore wind
692 energy, Procedia CIRP 59 (2017) 160–165.
693 [78] J. Wan, S. Tang, D. Li, S. Wang, C. Liu, H. Abbas, A. V. Vasilakos, A manufacturing big data solution for active
694 preventive maintenance, IEEE Transactions on Industrial Informatics (2017).
695 [79] R. A. A. Habeeb, F. Nasaruddin, A. Gani, I. A. T. Hashem, E. Ahmed, M. Imran, Real-time big data processing
696 for anomaly detection: A survey, International Journal of Information Management (2018).
697 [80] M. H. Rehman, V. Chang, A. Batool, T. Y. Wah, Big data reduction framework for value creation in sustainable
698 enterprises, International Journal of Information Management 36 (2016) 917–928.
699 [81] J. R. B. Mozafari, S. W. S. Menon, N. K. H. B. S. Chakraborty, Y. M. R. M. K. Bachhav, Snappydata: Streaming,
700 transactions, and interactive analytics in a unified engine (2016).
701 [82] P. P. De Silva, P. A. De Silva, Ipanera: An industry 4.0 based architecture for distributed soil-less food production
702 systems, in: Manufacturing & Industrial Engineering Symposium (MIES), IEEE, pp. 1–5.
703 [83] V. K. Vavilapalli, A. C. Murthy, C. Douglas, S. Agarwal, M. Konar, R. Evans, T. Graves, J. Lowe, H. Shah, S. Seth,
704 et al., Apache hadoop yarn: Yet another resource negotiator, in: Proceedings of the 4th annual Symposium on
705 Cloud Computing, ACM, p. 5.
706 [84] J. Dean, S. Ghemawat, Mapreduce: simplified data processing on large clusters, Communications of the ACM 51
707 (2008) 107–113.
708 [85] X. Meng, J. Bradley, B. Yavuz, E. Sparks, S. Venkataraman, D. Liu, J. Freeman, D. Tsai, M. Amde, S. Owen,
709 et al., Mllib: Machine learning in apache spark, Journal of Machine Learning Research 17 (2016) 1–7.
28
710 [86] H. Lee, Framework and development of fault detection classification using iot device and cloud environment,
711 Journal of Manufacturing Systems (2017).
712 [87] H. Lee, R. Grosse, R. Ranganath, A. Y. Ng, Convolutional deep belief networks for scalable unsupervised learning
713 of hierarchical representations, in: Proceedings of the 26th annual international conference on machine learning,
714 ACM, pp. 609–616.
715 [88] G. E. Hinton, S. Osindero, Y.-W. Teh, A fast learning algorithm for deep belief nets, Neural computation 18
716 (2006) 1527–1554.
717 [89] Y. Zhang, S. Ren, Y. Liu, S. Si, A big data analytics architecture for cleaner manufacturing and maintenance
718 processes of complex products, Journal of Cleaner Production 142 (2017) 626–641.
719 [90] M. Takikawa, Innovation in railway maintenance utilizing information and communication technology (smart
720 maintenance initiative), Japan Railway & Transport Review (2016).
721 [91] J. Cao, H. Cui, H. Shi, L. Jiao, Big data: A parallel particle swarm optimization-back-propagation neural network
722 algorithm based on mapreduce, PLoS One 11 (2016) e0157551.
723 [92] M. Dopico, A. Gomez, D. De la Fuente, N. Garcı́a, R. Rosillo, J. Puche, A vision of industry 4.0 from an artificial
724 intelligence point of view, in: Proceedings on the International Conference on Artificial Intelligence (ICAI), The
725 Steering Committee of The World Congress in Computer Science, Computer Engineering and Applied Computing
726 (WorldComp), p. 407.
727 [93] S. Wang, J. Wan, D. Zhang, D. Li, C. Zhang, Towards smart factory for industry 4.0: A self-organized multi-agent
728 system with big data based feedback and coordination, Computer Networks 101 (2016) 158–168.
729 [94] T. Y. Berger-Wolf, Research: Computational ecology, social network analysis, data science., Journal of Parallel
730 and Distributed Computing 93 (2016) 66–86.
731 [95] M. H. Rehman, C. S. Liew, A. Abbas, P. P. Jayaraman, T. Y. Wah, S. U. Khan, Big data reduction methods: A
732 survey, Data Science and Engineering (2016) 1–20.
733 [96] T. Töniges, S. Ötting, B. Wrede, G. W. Maier, G. Sagerer, An emerging decision authority: Adaptive cyber-
734 physical system design for fair human-machine interaction and decision processes, in: Cyber-physical systems:
735 Foundations, principles, and applications, 2017.
736 [97] S. J. Oks, A. Fritzsche, K. M. Möslein, An application map for industrial cyber-physical systems, in: Industrial
737 Internet of Things, Springer, 2017, pp. 21–46.
738 [98] M. Martellini, S. Abaimov, S. Gaycken, C. Wilson, Assessing cyberattacks against wireless networks of the next
739 global internet of things revolution: Industry 4.0, in: Information Security of Highly Critical Wireless Networks,
740 Springer, 2017, pp. 63–69.
741 [99] K. Jensen, H. T. Nguyen, T. Van Do, A. Årnes, A big data analytics approach to combat telecommunication
742 vulnerabilities, Cluster Computing (2017) 1–12.
743 [100] H. Igor, J. Bohuslava, J. Martin, Proposal of communication standardization of industrial networks in industry
744 4.0, in: Intelligent Engineering Systems (INES), 2016 IEEE 20th Jubilee International Conference on, IEEE, pp.
745 119–124.
746 [101] A. Dagnino, D. Cox, Industrial analytics to discover knowledge from instrumented networked machines., in:
747 SEKE, pp. 86–89.
748 [102] C. M. de Farias, L. Pirmez, G. Fortino, A. Guerrieri, A multi-sensor data fusion technique using data correlations
29
749 among multiple applications, Future Generation Computer Systems 92 (2019) 109–118.
750 [103] P.-Y. Wu, C.-W. Cheng, C. D. Kaddi, J. Venugopalan, R. Hoffman, M. D. Wang, -omic and electronic health record
751 big data analytics for precision medicine., IEEE transactions on bio-medical engineering 64 (2017) 263–273.
752 [104] J. Wan, S. Tang, D. Li, M. Imran, C. Zhang, C. Liu, Z. Pang, Reconfigurable smart factory for drug packing in
753 healthcare industry 4.0, IEEE transactions on industrial informatics 15 (2019) 507–516.
754 [105] I. Yaqoob, I. A. T. Hashem, A. Ahmed, S. A. Kazmi, C. S. Hong, Internet of things forensics: Recent advances,
755 taxonomy, requirements, and open challenges, Future Generation Computer Systems 92 (2019) 265 – 275.
756 [106] I. Yaqoob, E. Ahmed, M. H. ur Rehman, A. I. A. Ahmed, M. A. Al-garadi, M. Imran, M. Guizani, The rise of
757 ransomware and emerging security challenges in the internet of things, Computer Networks 129 (2017) 444–458.
758 [107] Y.-W. Ma, Y.-C. Chen, J.-L. Chen, Sdn-enabled network virtualization for industry 4.0 based on iots and cloud
759 computing, in: Advanced Communication Technology (ICACT), 2017 19th International Conference on, IEEE,
760 pp. 199–202.
761 [108] M. Gupta, N. Singla, Evolution of cloud in big data with hadoop on docker platform, in: Privacy and Security
762 Policies in Big Data, IGI Global, 2017, pp. 41–62.
763 [109] T.-H. Dang-Ha, D. Roverso, R. Olsson, Graph of virtual actors (gova): A big data analytics architecture for iot,
764 in: Big Data and Smart Computing (BigComp), 2017 IEEE International Conference on, IEEE, pp. 162–169.
765 [110] M. Fokaefs, C. Barna, R. Veleda, M. Litoiu, J. Wigglesworth, R. Mateescu, Enabling devops for containerized
766 data-intensive applications: an exploratory study, in: Proceedings of the 26th Annual International Conference on
767 Computer Science and Software Engineering, IBM Corp., pp. 138–148.
768 [111] E. Wagner, Microservices as a Manufacturing Execution System Architecture, Ph.D. thesis, Ulm University, 2016.
769 [112] S. A. Jacobs, A. Dagnino, Large-scale industrial alarm reduction and critical events mining using graph analytics
770 on spark, in: Big Data Computing Service and Applications (BigDataService), 2016 IEEE Second International
772 [113] K. Salah, M. H. U. Rehman, N. Nizamuddin, A. Al-Fuqaha, Blockchain for ai: Review and open research
773 challenges, IEEE Access 7 (2019) 10127–10149.
774 [114] N. Hassan, S. Gillani, E. Ahmed, I. Yaqoob, M. Imran, The role of edge computing in internet of things, IEEE
775 Communications Magazine (2018) 1–6.
776 [115] S. El Kafhali, K. Salah, Efficient and dynamic scaling of fog nodes for iot devices, The Journal of Supercomputing
777 73 (2017) 5261–5284.
778 [116] R. Almadhoun, M. Kadadha, M. Alhemeiri, M. Alshehhi, K. Salah, A user authentication scheme of iot de-
779 vices using blockchain-enabled fog nodes, in: Proceedings of the 15th ACS/IEEE International Conference on
780 Computer Systems and Applications (ACS/IEEE AICCSA 2018), Aqaba, Jordan, October 28, 2018.
30
Muhammad Habib ur Rehman is working on big data mining systems for Internet of Things. He
received his Ph.D. (Computer Science) degree from the University of Malaya, in 2017. His research areas
include: mobile computing, edge-cloud computing, Internet of Things, data mining, and mobile
distributed analytics. He has 27 publications on his credit including 14 ISI and 02 Scopus listed Journal
articles, 03 IEEE conference proceedings, and 02 book chapters. Overall he has more than 11 years’
experience of working in professional and research organizations. Feel free to contact for supervision,
collaboration, discussion, and sharing knowledge and information about edge computing, mobile
computing, cloud computing, and future technologies.
Ibrar Yaqoob ([email protected]) is currently working as a postdoctoral researcher at Kyung Hee

University, South Korea. He received his Ph.D. (Computer Sciences) from the University of Malaya,
Malaysia, in 2017. Prior to that, he received his BS. IT (Hons.) degree from University of the Punjab
(Gujranwala Campus), Pakistan, in 2012. His works have been cited more than 2000 times. His total
impact factor is more than 140. He worked as a lead researcher at the Centre for Mobile Cloud
Computing Research (C4MCCR), University of Malaya. His research has appeared in several prestigious
journals, such as IEEE Wireless Communications, IEEE Communications Magazine, Information
Systems, Computer Networks, to name a few. His numerous research articles are very famous and among
the most downloaded in top journals. He has reviewed more than 150 articles for the top ISI- Indexed
journals and conferences. He is serving/served as a technical program committee member for various
international conferences, which were held in Spain, Portugal, United Kingdom, Italy, Germany, and
China. Dr. Ibrar is a working as a guest/associate editor in various Journals. Dr. Ibrar has won several
awards during his stay at the University of Malaya. His research interests include big data, edge
computing, mobile cloud computing, the Internet of Things, Computer networks, and Software-Defined
Networks.
Khaled Salah is a full professor at the Department of Electrical and Computer Engineering, Khalifa
University, UAE. He received the B.S. degree in Computer Engineering with a minor in Computer
Science from Iowa State University, USA, in 1990, the M.S. degree in Computer Systems Engineering
from Illinois Institute of Technology, USA, in 1994, and the Ph.D. degree in Computer Science from the
same institution in 2000. He joined Khalifa University in August 2010, and is teaching graduate and
undergraduate courses in the areas of cloud computing, computer and network security, computer
networks, operating systems, and performance modeling and analysis. Prior to joining Khalifa University,
Khaled worked for ten years at the department of Information and Computer Science, King Fahd
University of Petroleum and Minerals (KFUPM), KSA. Khaled has over 170 publications and 2 patents,
has been giving a number of international keynote speeches, invited talks, tutorials, and research seminars
on the subjects of Blockchain, IoT, Fog and Cloud Computing, and Cybersecurity.
Muhammad Imran ([email protected]) is an assistant professor in the College of Computer and

Information Science, King Saud University. His research interests include mobile ad hoc and sensor
networks, WBANs, IoT, M2M, multihop wireless networks, and fault-tolerant computing. He has
published a number of research papers in peer reviewed international journals and conferences. His
research is financially supported by several grants. He is serving as a Co-Editor-in-Chief for EAI
Transactions on Pervasive Health and Technology. He also serves as an Associate Editor for the Wireless
Communication and Mobile Computing Journal (Wiley), the Inderscience International Journal of
Autonomous and Adaptive Communications Systems, Wireless Sensor Systems (IET), and the
International Journal of Information Technology and Electrical Engineering. He has served/serves as a
Guest Editor for IEEE Communications Magazine, IJAACS, and the International Journal of Distributed
Sensor Networks. He has been involved in a number of conferences and workshops in various capacities
such as a Program Co-Chair, Track Chair/Co-Chair, and Technical Program Committee member. These
include IEEE GLOBECOM, ICC, AINA, LCN, IWCMC, IFIP WWIC, and BWCCA. He has received a
number of awards such as an Asia Pacific Advanced Network fellowship.
Prem Prakash Jayaraman works in the area of distributed systems in particular Internet of Things (IoT),
Cloud and Mobile Computing. He has published over 70 papers including 28 Journal articles in the
related areas of his research. Prior to joining Swinburne, Dr. Jayaraman was a Post-Doctoral Research
Scientist (2012 - 2015) in the Digital Productivity and Services Flagship of Commonwealth Scientific and
Industrial Research Organization (CSIRO) – Australian Government’s Premier Research Agency. Prior to
that, he worked as a Research Fellow and Lecturer at the Centre for Distributed Systems and Software
Engineering, Monash University, Melbourne, Australia (2010 – 2011). Dr. Jayaraman obtained his PhD
(2011) from Monash University where he worked on developing a context-aware middleware for
collecting data from wireless sensors networks using mobile smart phones in smart city environments.
Charith Perera received his BSc (Hons.) in Computer Science in 2009 from Staffordshire University,
Stoke-on-Trent, United Kingdom and MBA in Business Administration in 2012 from University of Wales,
Cardiff, United Kingdom. He is completed his PhD in Computer Science from The Australian National
University, Canberra, Australia. He worked at Information Engineering Laboratory, ICT Centre, CSIRO
and involved in OpenIoT Project (Open source blueprint for large scale self-organizing cloud
environments for IoT applications), which is co-funded by the European Commission under seventh
framework program. His research interests include Internet of Things, pervasive and ubiquitous
computing with a focus on sensor networks, middleware, context aware computing, mobile computing
and semantic technologies. He is a member of the Association for Computing Machinery (ACM) and the
Institute of Electrical and Electronics Engineers (IEEE).
Highlights
 We investigate the state-of-the-art research studies that exist on IIoT in terms of big data analytics.
 We build a case of big data analytics for IIoT systems, and classify the literature by devising a taxonomy.
 We present the frameworks and case studies whereby the big data analytics processes were adopted to
improve overall performance of the IIoT systems.
 We present several research opportunities, challenges, and future technologies.

J Future 2019 04 020

Uploaded by

Copyright:

Available Formats

J Future 2019 04 020

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

J Future 2019 04 020

Uploaded by

Copyright:

Available Formats

Accepted Manuscript

The role of big data analytics in industrial Internet of Things

Muhammad Habib ur Rehman, Ibrar Yaqoob, Khaled Salah,

To appear in: Future Generation Computer Systems

Received date : 2 June 2018

Preprint submitted to Future Generation Computer Systems April 17, 2019

47 The main contributions of this study are listed as follows.

51 • We investigate existing state-of-the-art research studies on IIoT in terms of BDA. In this

55 • We present several research opportunities, challenges and future technologies to minimise

68 2.1. Design Principles for IIoT Systems

82 2.2. Rise of Big Data in IIoT Systems

Figure 1: Seven design implementation principles for Industry 4.0 systems.

96 2.3. Concentric Computing Model for BDA in IIoT

112 2.4.1. Data Engineering

Industrial Internet of Things

Outer Edge Servers

Inner Edge Servers

Core Cloud Services

Figure 2: Industrial IoTs and Multilayer Computing Resources

Data Engineering Data Preparation Data Analytics Automation

Concentric Computing System: Platform for Industry 4.0 Applications

122 2.4.2. Data Preparation

134 2.4.3. Data Analytics

140 2.4.4. Managing and Automating the Data Pipeline

153 3. Technologies and Algorithms for BDA in IIoT systems

177 3.2. Industrial Time Series Modeling

200 3.4. Industrial Microgrids

208 3.5. Monitoring Machine Health

223 3.6. Intelligent Predictive and Preventive Maintenance

234 4.1. Data Sources

249 4.2. Analytics Tools

Taxonomy of BDA in IIoT

Data Sources Analytics Tools Analytics Requirements Industrial Analytics Types

Figure 4: Taxonomy of BDA in IIoT

271 4.4. Requirements

284 4.5. Industrial Analytics Applications

296 4.6. Analytics Types

306 5. Frameworks and Case Studies

329 5.2. Ipanera

343 5.3. Fault Detection Classification

353 5.4. BDA Architecture for Cleaner Production

368 5.5. Smart Maintenance Initiative: Railway Case Study

376 6. Opportunities, Research Challenges, and Future Technologies

383 6.1. Opportunities

386 6.1.1. Automation and AI

393 6.1.2. Human Machine Interaction

400 6.1.3. Cybersecurity, Privacy, and Ethics

415 6.1.5. Protocols for Interoperability

424 6.1.6. End-to-end Industrial Analytics

431 6.1.7. Precision Manufacturing

438 6.2. Research Challenges and Future Technologies

441 6.2.1. Big Data Process Integration into IIoT Systems

449 6.2.2. Orchestrating BDA Applications Using Concentric Computing

Ibrar Yaqoob ([email protected]) is currently working as a postdoctoral researcher at Kyung Hee

Muhammad Imran ([email protected]) is an assistant professor in the College of Computer and

You might also like