Application of Data Stream Processing Technologies in Industry 4.0:
What is Missing?
Guenter Hesse
a
, Werner Sinzig, Christoph Matthies
b
and Matthias Uflacker
Hasso Plattner Institute, University of Potsdam, August-Bebel-Str. 88, 14482 Potsdam, Germany
Keywords:
Industry 4.0, Internet of Things, Data Stream Processing, Data Integration.
Abstract:
Industry 4.0 is becoming more and more important for manufacturers as the developments in the area of
Internet of Things advance. Another technology gaining more attention is data stream processing systems.
Although such streaming frameworks seem to be a natural fit for Industry 4.0 scenarios, their application in
this context is still low. The contributions in this paper are threefold. Firstly, we present industry findings
that we derived from site inspections with a focus on Industry 4.0. Moreover, our view on Industry 4.0 and
important related aspects is elaborated. As a third contribution, we illustrate our opinion on why data stream
processing technologies could act as an enabler for Industry 4.0 and point out possible obstacles on this way.
1 INTRODUCTION
In the backlight of technological and economic devel-
opments, the term Industry 4.0 gained more and more
popularity. Technically, new Internet of Things (IoT)
technologies, such as sensors, are being created, sen-
sor accuracy increases and analytical IT systems are
being developed that allow querying huge amounts of
data within seconds, to name but a few. On the eco-
nomic side, a substantial price decrease for sensor IoT
equipment can be recognized. This trend is expected
to continue in the following years. To be more con-
crete, the price for an IoT node is expected to drop by
about 50% from 2015 to 2020 (McKinsey&Company,
2015). These developments fostered the increased
deployment of IoT technologies in companies, es-
pecially in the manufacturing sector, and thus, more
IoT data is available to companies (Weiner and Line,
2014). Monetarily expressed, the total global worth
of IoT technology is expected to reach USD 6.2 tril-
lion by 2025. One of the industry sectors investing
most on IoT is industrial manufacturing (Intel, 2014).
A related term in the context of manufacturing
that gained attention in the past years is Industry 4.0.
One reason for that is the potential that is seen in
it with respect to creating an added value for enter-
prises. A survey conducted by McKinsey in January
2016 amongst enterprises in the US, Germany, and
a
https://orcid.org/0000-0002-7634-3021
b
https://orcid.org/0000-0002-6612-5055
Japan with at least 50 employees highlights the signif-
icance of Industry 4.0. The study reveals, e.g., that the
majority of companies expect Industry 4.0 to increase
competitiveness (McKinsey&Company, 2016).
One of the identified key challenges is integrating
data from different sources to enable Industry 4.0 ap-
plications (McKinsey&Company, 2016). Especially
with the emerging significance of IoT data, the fairly
old challenge of integrating disparate data sources
gets a new flavor. Data Stream Processing Systems
(DSPSs) can be a technology suitable for tackling this
issue of data integration. Within this paper, a view on
Industry 4.0 as well as the potential of Data Stream
Processing technologies in that context is presented.
Following the introduction, industry insights re-
lated to Industry 4.0 observed through interviews and
site inspections are highlighted. In Section 3, we elab-
orate our view on Industry 4.0, i.e., our definition as
well as our view on data integration and IoT. After-
ward, Section 4 discusses DSPSs and their role in the
area of Industry 4.0, including challenges regarding
their application in Industry 4.0 settings. A section to
related work and a conclusion complete this paper.
2 OBSERVATIONS IN INDUSTRY
Beginning in 2015, we conducted interviews with
multiple enterprises with a focus on Industry 4.0 im-
plementation strategies and associated challenges and
304
Hesse, G., Sinzig, W., Matthies, C. and Uflacker, M.
Application of Data Stream Processing Technologies in Industry 4.0: What is Missing?.
DOI: 10.5220/0007950203040310
In Proceedings of the 8th International Conference on Data Science, Technology and Applications (DATA 2019), pages 304-310
ISBN: 978-989-758-377-3
Copyright
c
2019 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
solutions. In this section, we describe and contrast the
Industry 4.0 efforts of two selected companies. Both
enterprises belong to the manufacturing sector and are
comparatively large, with more than 10,000 employ-
ees and revenue of more than e 1bn each.
2.1 Company I
The first company collects sensor and log data from
two sources, its machines used for manufacturing as
well as from its sold products used by its customers.
About 250 of the vended machines were configured
to collect and send sensor and log data to an exter-
nal central cloud storage service back in late 2015.
The data is sent as a batch every 23 hours and in-
cludes several state values, such as temperature and
position information. Overall, that results in about
800GB data on a monthly basis. Another external
company is responsible for data cleansing and some
basic calculations. The results are then used by Com-
pany I. As Company I is producing the machines, they
also developed the format of the log data that is col-
lected. Over time, this format changed with different
software releases, which introduces additional com-
plexity with respect to data integration.
Regarding the machines used for manufacturing,
five machines are configured to collect sensor data.
This data is recorded every 100ms and sent every hour
to the same cloud storage service. Each batch is about
20MB big with respect to size. It contains, e.g., infor-
mation about energy consumption and position data.
As of late 2015, none of the collected data has
ever been deleted. Moreover, the stored sensor data
had not found its way into an application at that point
in time. However, Company I expected growth in its
services area. As part of that, it could imagine of-
fering several services around its products for which
the collected data would be useful. Predictive main-
tenance or remote service and support scenarios are
an example of such services. Besides, the collected
data could reveal further insights about product usage
and behavior, which could help product development.
The internally captured data could be used for, e.g.,
predictive maintenance or quality improvements sce-
narios. The knowledge about production behaviors of
previously manufactured products can be combined
and gained learnings can be used to support product
development and production planning.
2.2 Company II
The second company has several measurement sta-
tions in its production line. At these stations, cer-
tain parts of the product-to-be are gauged. Result-
ing data are mostly coordinates, e.g., borehole posi-
tions. By doing so, possible inaccuracies added in
previous production steps are identified. If an inaccu-
racy exceeds a threshold, the corresponding product
is removed from the production line and the mistakes
are corrected if possible.
Furthermore, there is a central database storing all
warning or error messages that appear in the produc-
tion line. A higher ve-digit number of messages oc-
curs on a single day on average, whereas this number
can go up to more than a million messages. Besides
the time stating when the deviation took place, the
point in time when it is remedied is stored next to fur-
ther values describing the event.
With respect to Industry 4.0 applications, the com-
pany was in the evaluation process, meaning thinking
about how the existing data could be used for such
scenarios. Back in 2015, the stored warnings and er-
rors had a documentary character rather than being
used in applications for, e.g., preventing future devi-
ations or optimizing processes. However, it was an
objective to leverage this data more in such kind of
programs. The measurement data was considered first
for this kind of evaluations.
2.3 Industry Study Conclusions
Both presented and studied companies have in com-
mon the positive view on Industry 4.0, meaning they
see it as a chance rather than a threat, which fits
the before-mentioned survey conducted by McKin-
sey (McKinsey&Company, 2015). However, neither
of the companies, which can both be considered as
leaders with respect to market share or revenue, have
been able to significantly leverage the potential of In-
dustry 4.0. None of them is using data stream pro-
cessing technologies in this domain so far. To be more
concrete, IoT data is collected but no major new ap-
plications using this data or even combining it with
business data have been introduced. That might serve
as an example for technological leaders struggling to
implement new innovations. This situation is often
referred to as the innovator’s dilemma, which elab-
orates on the challenge for successful companies to
stay innovative (Christensen, 2013).
3 INDUSTRY 4.0
In this section, we elaborate on the Industry 4.0-
related topics of data integration and the Internet of
Things. Based on that, we present our view on the
term Industry 4.0 afterward.
Application of Data Stream Processing Technologies in Industry 4.0: What is Missing?
305
3.1 Data Integration
Being able to map data from different sources belong-
ing together is crucial to get holistic pictures of pro-
cesses, entities, and relationships. The more data can
be combined, the more complete and valuable is the
created view that is needed for fact-based assessments
and decisions within enterprises. Consequently, better
data integration and so more available data can lead to
greater insights and understanding, better decisions,
and thus, to a competitive advantage.
After giving a brief overview of the current sit-
uation in enterprises that we discovered through
conducted interviews, site inspections, and research
projects, our views on the terms horizontal and verti-
cal data integration are elaborated.
3.1.1 Current Situation in Enterprises
Business processes are central artifacts that describe
an enterprise and the infrastructure they are embed-
ded in. Business systems represent such processes
digitally, e.g., in the form of data model entities like
a customer, customer order, product, production or-
der, or journal entries. For different companies, the
semantics of these entities can vary, which can ham-
per data integration exceeding company boundaries.
Within a single company, definitions should be clear.
However, that does not necessarily represent reality.
Besides business systems, sensors or IoT-related
technologies become a greater source of data that is
describing processes and infrastructure in an enter-
prise. This information is usually connected to busi-
ness systems, such as an Enterprise Resource Plan-
ning (ERP) system or alike, via a Machine Execution
System (MES) in the manufacturing sector. Addition-
ally, there might be more systems installed underlying
the MES responsible for managing the shop floor.
A typical IT landscape comprises many different
business systems. D
¨
ohler, for instance, a company
with more than 6,000 employees from the food and
beverage industry, has more than ten business sys-
tems and supporting systems that need to be managed
and where ideally data can be exchanged amongst
each other. In addition to an ERP system, there
are, e.g., systems for customer relationship manage-
ment, extended warehouse management, and an en-
terprise portal (D
¨
ohler, 2019; SAP, 2018). Often,
fragmented IT landscapes have been developed his-
torically and complexity increased through, e.g., ac-
quisitions. Simplification is a challenge in companies,
e.g., due to the lack of knowledge about old systems
that might be still used.
But even if all business systems are from the same
vendor, entities can differ between systems. A cen-
tralization to a single ERP system is unlikely to hap-
pen for multiple reasons. Such arguments can be re-
lated to aspects like data security of sensitive data,
e.g., HR data shall be decoupled from the main ERP,
or the wish to be not dependent on a single software
vendor for economic or risk diversification reasons.
Figure 1 visualizes a very simplified IT landscape
how it can be found at companies belonging to the
manufacturing sector. It distinguishes between differ-
ent system categories and highlights the areas of hor-
izontal and vertical integration, that are explained in
Section 3.1.2 and Section 3.1.3 respectively.
3.1.2 Horizontal Data Integration
We see horizontal data integration as a holistic view
of business processes, i.e., from the beginning to the
end. Technically, that means joining database tables
stored within business systems that are involved in
the business process execution as conceptually out-
lined in Figure 1. These links can be established, e.g.,
through foreign key dependencies. The greater the
number of tables that can be connected, the more de-
tailed and valuable the resulting view on a process. As
mentioned before, enterprises generally have multiple
business systems for the elaborated reasons, which in-
creases the effort for achieving a horizontal data inte-
gration. Compared to vertical integration, horizontal
integration is further developed having relatively ad-
vanced software solutions for achieving it.
3.1.3 Vertical Data Integration
Vertical data integration describes the connection of
technical data created by IoT technologies and busi-
ness systems, including the systems in between these
two layers as depicted in Figure 1. That means, two
different kinds of data have to be combined in contrast
to integrating only business data as in horizontal data
integration. These distinct data characteristics intro-
duce new challenges.
While business data is well-structured and with a
comparatively high degree of correctness, sensor data
can be relatively unstructured and error-prone. Con-
trary to the close business process reference of busi-
ness data, sensors have a strong time and location ref-
erence. Moreover, both volume and creation speed
of IoT data is generally higher, which impacts, e.g.,
the performance requirements on IT systems handling
this kind of data (Hesse et al., 2017b).
Moreover, it is a challenge to map entities in busi-
ness processes, such as a product that is being pro-
duced, to the corresponding IoT data that has been
measured while exactly this product has been pro-
duced at the corresponding workplace. In contrast to
DATA 2019 - 8th International Conference on Data Science, Technology and Applications
306
Figure 1: Conceptual Overview of Data Integration in the Context of Industry 4.0.
integrating relatively homogeneous data among busi-
ness systems, foreign keys cannot simply be used. In-
stead, a time-based approach is often applied, which
can potentially introduce errors due to imprecise time
measurements. However, the progress of vertical data
integration we experienced in site inspections, e.g., in
the form of being able to map sensor measurements
created at production machines to the corresponding
products that were being produced, is not as advanced
as the horizontal integration. Nevertheless, vertical
integration as in the previously described scenario is
desired since it can help to get further insights about
processes and thus, support to create an added value.
3.2 Internet of Things
Internet of Things is a term often used in the context
of Industry 4.0 that has versatile meanings. Originally
emerged out of the area of radio frequency identifi-
cation (RFID), where it described connected physi-
cal objects using this technology, the term IoT be-
came broader over recent years. It is not limited to
RFID technology anymore, but also comprises, e.g.,
things connected via sensors or machine-to-machine
communication. Additionally, applications leverag-
ing these technologies are referred to as IoT (Ashton
et al., 2009; Miorandi et al., 2012).
We see IoT as network-connected physical ob-
jects, whereas it does not matter which exact technol-
ogy is used for establishing a connection. Moreover,
IoT is an enabler for Industry 4.0 as it is driving verti-
cal data integration and thus, paving the way for new
business applications. Through making machines or
physical objects in general digitally accessible, new
data can be analyzed, new insights be gained and a
more holistic view of processes can be created. This
increased level of live information can lead to a com-
petitive advantage for enterprises.
3.3 Our View on Industry 4.0
We see Industry 4.0 as a term describing an advanced
way of manufacturing enabled and driven by techno-
logical progress in various areas.
These areas can be categorized into two groups,
developments with respect to IoT technologies and re-
garding IT systems. While the advances related to IoT
enable to gain new, higher volumes and more precise
measurements, the IT system development progresses
allow to analyze high volumes of data with reasonable
response times nowadays. Moreover, high volumes of
data created with a high velocity can also be handled
with the help of modern DSPSs.
These achievements lead to new opportunities in
manufacturing. New data is being generated in high
volume and velocity, which can now also be analyzed
in a reasonable amount of time with state-of-the-art
IT systems. This natural fit of two technological de-
velopments generates opportunities. Making use of
both advances in combination with full data integra-
tion, i.e., horizontally as well as vertically, raises the
level of detail and completeness enterprises can have
on their processes and entities. This information gain
leads to the enablement of better data-driven deci-
sions,
facilitates new insights into processes or entities,
creates the opportunity for new business applica-
tions, and
allows for rethinking the way of manufacturing.
Specifically, holistic data integration enables a
flexible and more customizable production, i.e., mov-
ing from a nowadays commonly existing batch-wise
production to piece-wise production while not sacri-
ficing economic performance. Although we have not
observed a batch size of one as an explicitly formu-
lated objective in our side inspections, it was consid-
Application of Data Stream Processing Technologies in Industry 4.0: What is Missing?
307
ered as a desirable situation. Generally, we got the
impressions that there are greater challenges related
to IT compared to those related to the engineering as-
pect of IoT.
4 DATA STREAM PROCESSING
SYSTEMS
In this section, our view on the potential role of
DSPSs in the context of Industry 4.0 is presented.
Moreover, the related challenges that need to be tack-
led are highlighted.
4.1 A Possible Role in Industry 4.0
Although data stream processing systems is not a new
technology, it gained more attraction in the past cou-
ple of years (Hesse and Lorenz, 2015). Reasons for
that are technological advances, e.g., with respect to
distributed systems on the one hand, and on the other
hand the grown need for such systems due to the in-
creased data masses that are being created through de-
velopments like IoT for instance.
We think that stream processing technologies have
the potential to play a central role in the context of In-
dustry 4.0. A reason for that is its suitability regarding
the data characteristics of processed data, which fit
the overall purpose behind DSPSs. Instead of issuing
a query that is executed once and returns a single re-
sult set as in a database management system (DBMS),
DSPSs execute queries on a continuous basis. Simi-
larly, IoT data is often generated on a continuous ba-
sis, which is contrary to traditional business data.
Altering requirements, e.g., due to growing data
volumes introduced by added machines or advanced
IoT technologies, can be handled as modern DSPSs
are typically scalable distributed systems. As another
consequence, high elasticity is enabled, i.e., nodes can
be added or removed from the cluster as the work-
load increases or decreases. This flexibility is ad-
vantageous from an economic perspective. Especially
manufacturers that do not produce during certain pe-
riods, generally speaking companies with large IoT
workload variations, can benefit.
Scalability can be reached by using a message bro-
ker between the sources of streaming data and the
DSPS. That is a common approach seen in many ar-
chitectures, both in industry and science (Hesse et al.,
2017b). A schematic overview of a possible architec-
ture is visualized in Figure 2.
IoT devices, such as manufacturing equipment,
can send their measurements to a message broker,
Business
IoT Device
IoT Device
DSPS
Business
System
Message
Broker
Business
MES
Figure 2: Example Industry 4.0 IT Architecture in Funda-
mental Modeling Concepts (FMC).
from which a DSPS can consume the data. Stream-
ing applications that require more than IoT data, i.e.,
programs that need vertical integration, can also be
realized using DSPSs. Corresponding data can be
consumed via established interfaces, such as JDBC,
and enrich the IoT data. If a horizontal data integra-
tion can be achieved in the business system layer. A
holistic view on entities or processes can then be cre-
ated in the DSPS where all data is brought together.
Additionally, data from MES systems or alike can be
integrated as depicted in Figure 2. That makes data
stream processing technologies a suitable framework
for developing Industry 4.0 applications whose use
case does not have further requirements that can not
be satisfied in this setting.
Summarizing, since DSPSs are capable of han-
dling the high volume and high-velocity IoT data as
mentioned previously, they can act as an enabler for
vertical integration and thus, for Industry 4.0 scenar-
ios. Data can be analyzed on the fly without the need
of storing high volumes of data in advance, which has
a positive impact economically as well as on the per-
formance side. An imaginable pre-aggregation that
would lower these effects is not needed. Moreover,
aggregation comes at the cost of data loss and thus,
sacrifices accuracy.
4.2 Challenges
Certain challenges exist that could hinder an estab-
lishment of data stream processing technologies in the
context of Industry 4.0 on a broader scale.
One reason is the existing lack of a broadly ac-
cepted abstraction layer for formulating queries or
DATA 2019 - 8th International Conference on Data Science, Technology and Applications
308
developing applications, such as SQL for DBMSs.
Similarly, Stonebreaker, C¸ etintemel and Zdonik men-
tioned the need for DSPSs to support a high-level
stream processing SQL as one of the eight require-
ments of real-time stream processing they defined
in (Stonebraker et al., 2005). The lack of such an es-
tablished abstraction layer introduces multiple chal-
lenges. It reduces flexibility for enterprises as af-
ter choosing a certain system, the boundaries to ex-
actly this system are comparatively tight. Switching
to another framework, e.g., due to altered system re-
quirements or changed performance ratios amongst
the group of existing systems, is more complex and
thus, costlier for companies. Streaming applications
need to be developed using native system APIs, which
results in high porting effort if a system is supposed
to be exchanged compared to the effort needed for
switching a DBMS. The resulting potentially needed
SQL adaptions are relatively small since the same
abstraction layer, namely SQL, is also used in the
new system in typical scenarios. There are multiple
system-specific SQL dialects developed for stream
processing frameworks, but none of them gained
broader acceptance. However, there is the open-
source project Apache Beam aiming to close this gap.
It is not a domain-specific language like SQL, but
a software development kit that allows writing pro-
grams, which can be executed on any of the supported
stream processing engines. The impact of using this
abstraction layer on selected state-of-the-art DSPSs
with respect to performance is analyzed in (Hesse
et al., 2019).
Furthermore, identifying the most suitable system
might be a challenge for enterprises. Although the
growing number of DSPSs that have been developed
in recent years is generally a good thing, the more
choice the harder to make a decision for choosing a
system. Typically, performance benchmarks are used
for this task. Similarly to the previously described cir-
cumstances regarding the abstraction layers, the situ-
ation for DBMSs is more sophisticated. While there
are many well-known and often used benchmarks for
databases, such as TPC-C, TPC-H, or TPC-DS, the
area of DSPS benchmarks is significantly less devel-
oped. Linear Road is probably the best-known bench-
mark for stream processing architectures (Arasu et al.,
2004). However, it does not reflect typical Indus-
try 4.0 scenarios in contrast to a benchmark currently
under development and proposed by (Hesse et al.,
2017a), which could close the gap of not having a
suitable benchmark for comparing different DSPSs
for use in the Industry 4.0 domain.
Another challenge we recognized in site inspec-
tions is the identification of Industry 4.0 scenarios that
possibly create an added value. Although this situa-
tion is not directly linked to DSPSs, thoughts about
Industry 4.0 and technologies that can be used barely
include stream processing frameworks and their ca-
pabilities based on our industry experiences. This
lack of awareness of streaming technologies results
in not considering it for new application scenarios.
Moreover, when taken into account, there are often
reservations, such as that there is no or only little
knowledge about these technology amongst the em-
ployees. Another fear is that modern DSPSs are very
complex systems, which are hard to maintain and dif-
ficult to use for application development. However,
these points could be eliminated automatically in near
future if development efforts and improvements of
DSPSs stay as high as they are at the moment.
5 RELATED WORK
A recent work developed a framework called Produc-
tion Assessment 4.0, which aims to support enter-
prises developing Industry 4.0 use cases. For doing
so, they made use of the design thinking approach.
After elaborating on the framework and its processes,
a section about its evaluation is presented. Production
Assessment 4.0 was evaluated in several consulting
projects with enterprises. However, no details about,
e.g., their data characteristics or their state of Industry
4.0 adoption progress are given (Bauer et al., 2018).
With respect to Industry 4.0, there are many ex-
isting definitions and views published. An overview
of selected perceptions of Industry 4.0 is presented
in (Mrugalska and Wyrwicka, 2017). Moreover, it
also states that there is no generally accepted defini-
tion for the term Industry 4.0.
The Association of German Engineers (VDI) pub-
lished an architecture reference model for Industry
4.0 named RAMI 4.0 (Hankel and Rexroth, 2015).
It comprises three different dimensions, namely hier-
archy levels in factories, the product life cycle value
stream, and an architecture dimension containing sev-
eral layers from physical things up to the organization
and business processes.
Another work presents design principles for In-
dustry 4.0 that are derived through text analysis and
literature studies (Hermann et al., 2016). Thereby, it
is aimed to help both, the scientific community and
practitioners with this result. In total, four design
principles were identified, namely technical assis-
tance, interconnection, decentralized decisions, and
information transparency.
With regard to challenges, there is the previ-
ously mentioned work by Stonebreaker, C¸ etintemel
Application of Data Stream Processing Technologies in Industry 4.0: What is Missing?
309
and Zdonik that defines eight requirements for real-
time stream processing. Particularly, these are the
need (1) to keep the data moving, (2) for a stream-
ing SQL language as highlighted before, (3) for the
ability to handle stream imperfections, (4) to generate
predictable outcomes, and (5) to integrate stored as
well as streaming data, which fits to the Industry 4.0
scenarios where business (stored) and IoT (streaming)
data are integrated. Furthermore, the requirement (6)
to ensure data safety and availability, (7) to automati-
cally scale and partition programs, and (8) to process
and respond immediately are highlighted.
6 CONCLUSION
The present paper pictures a point of view on Indus-
try 4.0 and on data stream processing systems in its
context. Thereby, contributions are threefold. First,
we present insights about current situations and opin-
ions at two selected companies with respect to Indus-
try 4.0. This includes information about data char-
acteristics and Industry 4.0 applications. All findings
were derived from site inspections and alike.
Secondly, a viewpoint on Industry 4.0 as well as
on further important and closely related aspects is
given. Among others, it ensures a common under-
standing needed for the third contribution.
This third part is about data stream processing sys-
tems. Particularly, it is about why and how this tech-
nology could become an enabler for Industry 4.0. A
possible architecture for Industry 4.0 scenarios is pro-
posed and obstacles hindering DSPSs from being ap-
plied more in this context are pointed out.
REFERENCES
Arasu, A., Cherniack, M., Galvez, E. F., Maier, D., Maskey,
A., Ryvkina, E., Stonebraker, M., and Tibbetts, R.
(2004). Linear Road: A Stream Data Management
Benchmark. In (e)Proc. International Conference on
Very Large Data Bases, pages 480–491.
Ashton, K. et al. (2009). That ’Internet of Things’ Thing.
RFID journal, 22(7):97–114.
Bauer, W., Pokorni, B., and Findeisen, S. (2018). Produc-
tion Assessment 4.0 Methods for the Development
and Evaluation of Industry 4.0 Use Cases. In Inter-
national Conference on Applied Human Factors and
Ergonomics, pages 501–510. Springer.
Christensen, C. M. (2013). THE INNOVATOR’S DILEMMA
- WHEN NEW TECHNOLOGIES CAUSE GREAT
FIRMS TO FAIL//. Harvard Business Review Press.
D
¨
ohler (2019). About D
¨
ohler — Who we are. https://www.
doehler.com/en/about-doehler/who-we-are.html. Ac-
cessed: 2019-04-03.
Hankel, M. and Rexroth, B. (2015). Industrie 4.0: The Ref-
erence Architectural Model Industrie 4.0 (RAMI 4.0).
ZVEI, 2:2.
Hermann, M., Pentek, T., and Otto, B. (2016). Design Prin-
ciples for Industrie 4.0 Scenarios. In Hawaii Interna-
tional Conference on System Sciences, HICSS, pages
3928–3937.
Hesse, G. and Lorenz, M. (2015). Conceptual Survey on
Data Stream Processing Systems. In IEEE Interna-
tional Conference on Parallel and Distributed Sys-
tems, ICPADS 2015, pages 797–802.
Hesse, G., Matthies, C., Glass, K., Huegle, J., and Uflacker,
M. (2019). Quantitative Impact Evaluation of an Ab-
straction Layer for Data Stream Processing Systems.
In International Conference on Distributed Comput-
ing Systems, ICDCS.
Hesse, G., Matthies, C., Reissaus, B., and Uflacker, M.
(2017a). A New Application Benchmark for Data
Stream Processing Architectures in an Enterprise
Context: Doctoral Symposium. In ACM International
Conference on Distributed and Event-based Systems
(DEBS), pages 359–362.
Hesse, G., Reissaus, B., Matthies, C., Lorenz, M., Kraus,
M., and Uflacker, M. (2017b). Senska - Towards an
Enterprise Streaming Benchmark. In TPC Technology
Conference, TPCTC, pages 25–40.
Intel (2014). A GUIDE TO THE INTERNET OF
THINGS. https://www.intel.com/content/www/us/
en/internet-of-things/infographics/guide-to-iot.html.
Accessed: 2019-04-03.
McKinsey&Company (2015). Industry 4.0 - How
to navigate digitization of the manufacturing sec-
tor. http://www.forschungsnetzwerk.at/downloadpub/
mck\ industry\ 40\ report.pdf. Accessed: 2019-04-
02.
McKinsey&Company (2016). Industry 4.0 after the ini-
tial hype - Where manufacturers are finding value and
how they can best capture it.
Miorandi, D., Sicari, S., Pellegrini, F. D., and Chlamtac, I.
(2012). Internet of things: Vision, applications and
research challenges. Ad Hoc Networks, 10(7):1497–
1516.
Mrugalska, B. and Wyrwicka, M. K. (2017). Towards Lean
Production in Industry 4.0. Procedia Engineering,
182:466 – 473.
SAP (2018). Cloud Experts Tell It Like It Is - D
¨
ohler’s
Best Practices on Setting Up a Hybrid IT Land-
scape. https://d.dam.sap.com/m/xmxUAO/Cloud
Experts Doehler Interview Brochure-V03.pdf. Ac-
cessed: 2019-04-03.
Stonebraker, M., C¸ etintemel, U., and Zdonik, S. B. (2005).
The 8 Requirements of Real-Time Stream Processing.
SIGMOD Record, 34(4):42–47.
Weiner, S. and Line, D. (2014). Manufacturing and the data
conundrum - Too much? Too little? Or just right?
https://www.eiuperspectives.economist.com/sites/
default/files/Manufacturing Data Conundrum Jul14.
pdf. Accessed: 2019-04-02.
DATA 2019 - 8th International Conference on Data Science, Technology and Applications
310