Patterns for IoT-based Business Process Improvements: Developing a

Metamodel

Christoph Stoiber and Stefan Schönig

Professorship for IoT-based Information Systems, University of Regensburg, Germany

Keywords: Internet of Things, Business Process Improvement, Business Process Management, Patterns, Metamodel.

Abstract: The number of Internet of Things (IoT) devices is constantly growing across all areas of private and

professional life. Especially industrial organizations are increasingly recognizing the IoT’s disruptive

capabilities and potential benefits for business processes along all value chain activities. In this regard, the

integration of IoT technology into existing business processes enables valuable Business Process

Improvements (BPI). However, it often remains unclear which BPIs can be expected by organizations and

how the anticipated BPIs are realized in detail. Furthermore, the integration of IoT technology into existing

business processes constitutes a major challenge caused by a lack of supporting methods, models, or

guidelines. The paper at hand addresses this research gap by providing a metamodel that enables the

illustration of generic IoT-based BPI patterns. It contains all relevant elements that are comprised by IoT

applications with BPI propositions and can be used by industrial organizations as blueprints for conducting

IoT projects. The metamodel development follows fundamental principles of design science research (DSR)

and is extensively evaluated by deriving a first set of patterns from real-life IoT applications of three market-

leading corporations. In addition, an expert survey is conducted to assess the metamodel’s usefulness.

1 INTRODUCTION

Internet of Things (IoT) applications are omnipresent

and influencing all facets of everyday life by

providing disruptive technologies for private

households and businesses of all kinds. Besides

various smart home, smart grid, and smart city

applications, especially industrial organizations can

remarkably benefit from integrating IoT technologies

into their business processes. The transformation of

analog information into digital data, which can be

processed worldwide in real-time, enables new

business models, revolutionizes existing ones (Ng

and Wakenshaw, 2017), and improves organizations’

competitive advantage (Li, 2012). Moreover, the

generation and use of comprehensive process data

and the connection of process entities can be used to

improve all types of business processes and thus

optimize value creation (Del Giudice, 2016).

Therefore, the integration of IoT technology into

existing business processes can lead to beneficial

Business Process Improvements (BPI) that are highly

relevant for process-oriented organizations (Janiesch,

2020). For instance, equipping in-stock products with

simple radio-frequency identification (RFID) tags

can fundamentally enhance the traceability of

warehouse processes and enable manifold further

opportunities for improving downstream operations.

Hence, the pressure on enterprises to integrate IoT

technology into their processes is growing steadily, to

the point that enterprises which don't adopt IoT, may

not be competitive in the near future (Liu, 2017).

However, a survey of more than 500 business

executives revealed that 90% of industrial

organizations are remaining in the proof of concept or

even early-stage planning phases for IoT projects

(Bosche, 2016). Knowing about the relevance of IoT

technology integration, this seems rather surprising

and indicates the existence of severe challenges for

successfully integrating IoT technology into process

landscapes. One main reason for this lack of IoT

application maturity may be the complexity and

heterogeneity of the used systems. Here, different

technologies must be combined, e.g., various sensors

and communication protocols, and integrated into the

organization's existing information systems (Sethi

and Sarangi, 2017). Another reason is the existing

discrepancy between the organizations’ expectations

of IoT projects and the actual results (Skaržauskienė

and Kalinauskas, 2015). Decision-makers need to

have an explicit understanding of the value they can

Stoiber, C. and Schönig, S.

Patterns for IoT-based Business Process Improvements: Developing a Metamodel.

DOI: 10.5220/0011059300003179

In Proceedings of the 24th International Conference on Enterprise Information Systems (ICEIS 2022) - Volume 1, pages 655-666

ISBN: 978-989-758-569-2; ISSN: 2184-4992

655

expect and the technological aspects that are required

to achieve it (Reijers and Liman Mansar, 2005).

Finally, the "Act of Improvement", i.e., how existing

business processes are transferred to the improved

target state by integrating IoT, can often not be defined

precisely. This fact reduces the plannability and thus

the chance of a successful IoT project (Forster, 2006).

To tackle these challenges, organizations need

structured models that display and describe all relevant

components of IoT-based BPIs. These models need to

be generic enough to be applicable for similar

scenarios and detailed enough to effectively guide

organizations during the implementation of individual

IoT-based BPI projects. In this regard, we define the

term “IoT-based BPI” as the purposeful use of IoT

technology within business processes to improve the

same with respect to predefined objectives. Therefore,

we formulate the following research question (RQ):

 RQ1: How can industrial organizations be

supported at the identification and

implementation of IoT-based BPI projects?

One auspicious approach to address this RQ is the

development of generic patterns. Patterns are reusable

artifacts which address a problem within a certain

context by providing a suitable solution (Alexander,

1977). In this context, patterns can represent

templates for IoT-based BPIs and are reusable for

different kinds of industrial organizations (Forster,

2006). Using patterns can reduce the risk of IoT

projects as well as support organizations with the

identification of possible BPI potentials and the

required IoT technologies, making them

extraordinary valuable. Furthermore, all relevant

application elements such as underlying problems

and challenges, industry examples, performance

indicators, or specific characteristics of the technical

solution are provided. The prerequisite to formulate

these patterns is an appropriate metamodel that

displays basic design principles. The metamodel

ensures completeness and consistency of the pattern

descriptions and specifies their structure (Falk, 2013).

Against this background, we formulate an additional

supporting RQ:

 RQ2: Which metamodel can enable the

illustration of generic yet adoptable IoT-based

BPI patterns?

The paper at hand addresses both RQs by

proposing a metamodel which can be used to create

IoT-based BPI patterns. The metamodel design is

based on the design science research (DSR)

methodology by Peffers et al. (2007) including an

evaluation according to the Framework for

Evaluation in DSR (FEDS) of Venable et al. (2016).

The remainder of this paper is structured as

follows. Section 2 presents the theoretical

foundations of the disciplines IoT and BPI as well as

an overview over the concept of patterns and

metamodels in information systems research. In

section 3, the underlying research methodology is

described that has been applied for developing and

evaluating the metamodel. Subsequent, the design

and development phases are illustrated in section 4.

Section 5 presents the summative evaluation of the

metamodel, concluding with a summary, discussion,

and the description of limitations in section 6.

2 THEORETICAL FOUNDATION

2.1 Internet of Things Meets Business

Process Improvement

There are dozens of different approaches for defining

IoT, its components, features and capabilities, and the

things itself. The Institute of Electrical and

Electronics Engineers (IEEE) combined several

different descriptions, explanations, and

characterizations towards a universal definition.

According to the IEEE, IoT is a network that connects

uniquely identifiable things to the internet. Through

the exploitation of unique identification and sensing,

information about the thing can be collected and the

state can be changed from anywhere, anytime, by

anything (Minerva, 2015). The term thing therefore

corresponds to the idea of creating a ubiquitous

presence of objects which are equipped with sensors,

actuators, or tags. On the other side, the term internet

refers to the ability of these things to build a network

of interconnected objects based on several specific

network technologies. These two perspectives can be

complemented by a semantic view, which represents

the ability of IoT to uniquely identify things and store,

process, and exchange data (Atzori, 2010). Current

research and already implemented applications now

show that IoT technology reveals many extensive

possibilities for improving business processes

(Stoiber and Schönig, 2021). In this regard, especially

redesigning and therefore improving business

processes is a timely and relevant topic in both

research and business environment and is considered

as one of “the most important and common titles in

both literature and applications” (Coskun, 2008).

Despite IoT’s capabilities to enhance BPI and

therefore sustainably optimize the organization’s

overall performance, there is a lack of research

regarding IoT-based BPI. Among the limited number

ICEIS 2022 - 24th International Conference on Enterprise Information Systems

656

of contributions, Janiesch al. (2020) created an

overview of existing research and remaining

challenges. Here, especially the need for further

research on how to benefit from the integration of IoT

into business processes has been highlighted. This

research gap can be tackled by developing a

metamodel that enables the creation of patterns and

adds to the descriptive knowledge of IoT-based BPI.

This approach has been proven in several other

research disciplines and is well received at

organizations of all industry sectors (Winter, 2009).

2.2 Metamodels and Patterns in

Information Systems Research

Patterns, initially described by Alexander (1977),

describe a recurring problem or challenge in the real

world and the basic features of the solution to this

problem. This solution is generic enough to be

applied to many similar problems without ever being

implemented in exactly the same way. Although

Alexander (1977) created this definition in the

context of architecture, the idea of patterns is

transferable to other domains, especially information

systems research (Gamma, 1994). In the context of

enterprise and systems modeling, Fowler (1996)

described patterns as an idea that has been useful in

one practical application and is likely to be useful in

others. According to Gamma et al. (1994), patterns

consist of four essential elements. First, the pattern

must have a name for identification. Then there is a

description of the problem, i.e., in what context the

pattern might be useful. The third element is a

description of the problem solution. This must not be

done by a concrete solution, because the pattern

should be applicable to different scenarios, but by a

description of the interaction of different mechanisms

that lead to a problem solution. Finally, the

consequences of the pattern must be described, i.e.,

the positive and negative effects that can result from

the application of the pattern. Depending on the

purpose of the pattern, this basic description can be

extended by further elements. There has been

considerable research on patterns in information

systems for more than two decades leading to several

relevant approaches indispensable from a research

and practical perspective. Beyond doubt, software

development is one of the disciplines that benefited

most from the creation of patterns (Winter, 2009).

Here, patterns can support the design of individual

object-oriented software components or assist with

the composition of software components to

applications (Schmidt, 2000). As this discipline

includes complex tasks, patterns can bridge the gap

between high-level integration plans and the actual

implementation challenges by providing guidelines to

compensate the lack of experience at decision makers

(Hohpe, 2003). This leads to reduced time

consumption and cost while improving the quality of

project execution. Moreover, patterns can be used for

process-related disciplines such as Workflow

Management or Business Process Modeling (Kühn,

2005). For the discipline of BPI, the creation of

specific patterns has barely been addressed in

research. Reijers and Liman Mansar (2005) described

a set of textual Business Process Redesign (BPR) best

practices including a framework to classify them.

Forster et al. (2006) built up a framework and toolset

for creating and structuring BPI patterns while

creating a first set of patterns. Another relevant

contribution by Falk et al. (2013) proposes a

metamodel that facilitates the illustration of BPI

patterns. In this respect, patterns constitute models

that are derived from an origin metamodel.

In general, a model can not only describe objects

that exist in the real world, but also abstract

constructs. If the abstract construct described is a

model, the describing model is called metamodel

(Gonzalez-Perez, 2008). The relationship between

model and metamodel can also be referred to as a

class-instance relationship. This is an analogy to

object-oriented programming, where a class describes

the attributes and methods of the objects to be formed

from it, without itself being an object. By

instantiation, objects or instances can be formed from

the class, which in turn are mappings of real objects.

A metamodel describes the types of model building

blocks available, the types of relationships between

the model building blocks, the rules for linking

between model building blocks by relationships, and

the semantics of the model building blocks and

relationships (Ferstl, 2013). To create a metamodel, a

suitable modelling language is necessary to represent

and communicate relevant information about a

model. Modeling languages are defined by their

syntax, notation, and semantics. The syntax describes

the elements of a modeling language and how they

may be linked together, i.e., it describes the

grammatical rules. The notation describes the

symbols and characters that may be used to capture a

model. Ultimately, the semantics determines how

certain information is to be interpreted, e.g., when

ambiguities occur in the model (Kühn, 2005).

2.3 Related Work

As described in subsection 2.2, there has already been

research conducted on general BPI patterns and

Patterns for IoT-based Business Process Improvements: Developing a Metamodel

657

metamodels that do not particularly focus on IoT but

consider BPIs of any kind. Especially noteworthy is

the contribution of Falk et al. (2013), who created an

explicit metamodel that enables the creation and

formulation of BPI patterns and can be used as a

template and basis for further research.

Moreover, the concept of patterns has also been

applied to several topics related to IoT. As IoT

technology consists of different layers, comprising

perceiving, networking, or data processing

technologies, a great variety of different patterns can

be formulated that support system engineers with

integrating whole applications into business

environments. The design and architecture of IoT

systems can eminently benefit from patterns that

assist in designing scalable and replicable IoT

applications (Washizaki, 2020). Another focus within

this research area is on data exchange and network

technology patterns along multiple connected

devices, machines, or process entities (Reinfurt,

2016). However, the formulation of a metamodel for

IoT-based BPI patterns has not been addressed yet.

3 RESEARCH METHODOLOGY

To tackle this research gap and answer the formulated

RQs, we developed a metamodel that can be used to

create and illustrate reusable IoT-based BPI patterns.

To develop the metamodel as a DSR artifact, we

followed the process model of Peffers et al. (2007).

This proven method is based on the methodology of

Hevner et al. (2004) and provides detailed phases to

carry out DSR. It consists of six iterative phases in a

nominal sequence including i) the identification and

motivation of the underlying problem, ii) the

definition of objectives of the solution, iii) the actual

design and development, iv) the demonstration, v) an

evaluation, and vi) the communication to an

appropriate audience.

Initially, every conduction of DSR is based on a

research entry point that necessitates and justifies the

artifact development. For the paper at hand, the

existing problems and challenges that organizations

face at integrating IoT into their business processes

constitute a problem-centered research entry point.

Moreover, the lack of artifacts that support the

realization of IoT-based BPIs necessitate the creation

of a suitable DSR artifact. This research endeavor is

of special interest, as the integration and use of IoT

technology is an enabler for economic success and

becomes increasingly important. The objective of the

developed artifact is to provide a basis for the creation

of reusable patterns of IoT-based BPIs which serve as

blueprints and templates for organizations.

In contrast to creating a complete new metamodel

from the scratch, the improvement and revision of an

existing and thematically related metamodel enables

the adoption of proven concepts and ideas. Therefore,

the metamodel for BPI patterns according to Falk et

al. (2013) served as the basis for development. It is

generic enough to represent all patterns of IoT-based

BPIs since these represent a subset of BPI patterns.

However, it is not specific enough to appropriately

illuminate the aspects of the IoT domain due to its

complexity and unique features. For this reason, the

base metamodel needed to be adapted with respect to

IoT. Like in the original metamodel, a class diagram

is used for modelling as it provides sufficient

semantic expressiveness for metamodeling. To adapt

the base metamodel, we performed two development

iterations comprising methods of Grounded Theory

and a Delphi study. Figure 1 shows both iterations,

including data sources, the applied research methods,

and the resulting metamodel classes after each

iteration. To evaluate the final metamodel, we

followed the framework of Venable et al. (2016).

Within the first design iteration, an explorative

inductive approach has been selected. Hereof, an

extensive systematic literature review (SLR) was

conducted to investigate literature describing IoT

applications with BPI reference. Subsequent, the

found literature was analyzed following the

Grounded Theory and its methods of open and axial

coding (Corbin and Strauss, 1990). This enabled the

identification of indispensable aspect of IoT-based

BPIs which could be used to adapt the metamodel.

Within the author team, we applied the method of

inductive reasoning (Hempel, 1966) to critically

discuss the findings and select the most appropriate

metamodel adaptions. Within the second iteration, we

included additional expert knowledge into the

research approach. Hereof, we conducted a Delphi

study with nine experts from industry and academia

to consequently refine the metamodel. In four rounds,

the experts were asked to rate and eventually adapt

the metamodel based on their expertise of the

research area. Gradually, the metamodel has been

adapted by i) removing redundant elements, ii)

adding additional required elements, and iii) retaining

or slightly adjusting the remaining elements.

Having refined the metamodel, we performed a

summative evaluation to assess, if it adequately

addresses and solves the formulated RQs. In this

regard, we introduced the metamodel to the Linde plc

and two other multinational industrial corporations.

Seven practitioners from different departments were

ICEIS 2022 - 24th International Conference on Enterprise Information Systems

658

Figure 1: Development procedure.

asked to analyze a set of 34 IoT applications of their

different business areas to derive patterns.

Subsequently, we conducted an expert survey on

the practitioners to collect evidence and feedback. We

used the results of the survey to assess the predefined

evaluation criteria of usefulness, conciseness, and

robustness. In the following section, the initial base

model and all metamodel development iterations are

described in detail.

4 METAMODEL

DEVELOPMENT

4.1 Baseline Metamodel

The metamodel for BPI patterns of Falk et al. (2013)

is illustrated as a class diagram, whereby each

element of a pattern is represented by a specific class.

The properties of these classes are described by

attributes (Fowler, 1997), while relationships

between the classes are represented by undirected

binary associations and their multiplicity. This

multiplicity specifies the relationships between the

individual object classes. The central class of the

metamodel is BPI Pattern, which is instantiated by a

unique Name and an Example (cf. Figure 2). The

name describes the overall purpose of the pattern and

can be uniquely identified. In addition, there is the

class Problem, which is defined by the attributes

Name, Description, and the actual Consequences of

the problem for the process.

Figure 2: Base metamodel.

Each pattern addresses exactly one problem, but a

specific problem can also be solved by different

patterns. Furthermore, the Context class is directly

related to BPI Pattern. It is explained by a Name and

context-specific Characteristics and describes the

required circumstances for the pattern to be

applicable. As with the problem, each pattern exists

in exactly one context, but multiple patterns can exist

in the same context. Each pattern also contains a

Solution, which is described by a Name and the

Measures required to achieve the goal. The same

solution can again be applied to multiple patterns, but

each pattern has only one solution. Bound to the

solution are one or more Mechanisms, each defined

by a Name and precise action Instructions. In

addition, a solution can optionally contain one or

more Building Blocks. These building blocks are

predefined models that can be implemented to solve

the problem without customization. In addition, the

pattern is related to an Effect, which is defined by a

Name and the BPI dimensions Cost, Time, Quality,

and Flexibility (Dumas, 2018). Finally, each pattern

is related to one or more Performance Indicators.

These are defined by a Name and Performance

Measures that can be used to represent the

improvement after the pattern has been implemented.

4.2 First Development Iteration

To adapt the base metamodel, we first performed an

inductive development iteration. We decided to start

with this approach, as a large number of IoT

applications is available in scientific literature which

can be used to identify additional metamodel classes.

For inductive approaches, the information processing

is performed from subsystems to form a perception of

a top-level system. This aggregation of information is

suitable to analyze initially unknown data

relationships and transfer them to a generic

metamodel. We followed the recommendations of

Templier and Paré (2018) to identify a set of

Patterns for IoT-based Business Process Improvements: Developing a Metamodel

659

appropriate literature and subsequently extract

relevant data. For the identification of literature, we

performed a SLR according to the method of vom

Brocke et al. (2009). To allow a rigorous search and

improve the traceability of the literature selection

process, the Preferred Items for SLRs and Meta-

Analysis (PRISMA) statement has been applied.

Initially, the search string (“IoT” OR “CPS”) AND

(“BPI” OR “Process Improvement” OR “Process

Optimi?ation” OR “Process Automation” OR

“Application” OR “Process Improvement”) and the

written-out forms have been formulated. Figure 3

illustrates the results of the SLR as a PRISMA flow

diagram. To incorporate and consider preferably all

relevant journals and conference proceedings of the

research area, ACM Direct Library, AISeL, IEEE

Xplore, ScienceDirect, Scopus, and Springer Link

have been queried. According to the PRISMA

statement, four criteria were defined that a paper

needs to achieve to be eligible for the SLR. The

publication must i) be a peer-reviewed research paper

published in a journal or conference proceeding, ii)

propose an evaluated solution or real industry

application, iii) have clear links to BPI, and iv) be

relevant and up to date. As criteria ii) and iii) are

assessed in a rather qualitative manner, criterion iv) is

defined as a publication date after 2015 and a

minimum number of 30 citations. The literature

search and the included reference follow up resulted

in the selection of 81 eligible publications.

Having identified the eligible sample of

publications, we analyzed it and extracted relevant

data using the Grounded Theory. In this regard, we

applied the methods of open and axial coding, as

proposed by Corbin and Strauss (1990). This

approach enabled the derivation of metamodel classes

and attributes from the sample of IoT applications.

Figure 3: PRISMA flow diagram.

In the first round, each author analyzed 40

publications of the sample using open coding as an

interpretive method to analytically break down all

IoT-based BPI applications. The goal was to develop

substantiate categories that enable a description,

naming, and classifying. After this first round, we

discussed the identified categories and harmonized

the individual understanding of the main elements of

IoT-based BPIs. In the second round, we applied the

method of axial coding to relate the formulated codes

to each other. This enabled the creation of further

categories and subcategories. In a second discussion,

the results were again harmonized. In round 3, the

remaining 41 publications were coded with the

created set of categories and subcategories to test

them against data. Subsequent, we clarified and

resolved any remaining coding differences.

Following inductive reasoning according to Hempel

(1966) we extensively discussed the created

categories and subcategories to select the most potent

and relevant ones for the metamodel adaption. These

have been used to create a set of classes and related

attributes which were added to the base metamodel.

4.3 Second Development Iteration

To refine the initial metamodel draft we performed a

structured four-round Delphi study. A Delphi study is

an iterative method to solicit information about a

specific topic through the completion of several

surveys (Loo, 2002). It has been widely used to

combine expert knowledge and find group consent for

complex issues that lack empirical evidence (Loo,

2002). For this reason, Delphi studies are highly

present in the field of DSR research. The study

process included the selection of experts with

different backgrounds to minimize bias. They did not

get introduced to each other, which led to more

creative outcomes and reduced conflicts within the

group as well as group pressure. The experts were

asked to rate or validate the metamodel classes and

attributes of the first draft. After each round, the

results of all experts were consolidated and used for

refinement. We formed a panel of nine experts

including five practitioners and four researchers with

expertise in the fields of IoT and BPM. The selected

experts have working experiences ranging from four

to 21 years. All experts have at least a bachelor’s

degree and are based in Germany, the US, or the

Netherlands. Figure 4 shows the applied four-round

Delphi study including all information flows between

the authors, or facilitator, and the expert panel.

Records identified through database search

ACM Digital Library (n = 136), AISel (n = 135), IEEE Xplore (n = 1033),

ScienceDirect (n = 1102), Scopus (n = 412), Springer Link (n = 1253),

Total records (n = 4071)

Records after duplicates removed (n = 2141)

Records screened (n = 2141)

Records excluded based on

eligibility criteria (n = 1718)

Full-text publications assessed for

eligibility (n = 423)

Publications excluded with reasons

• No application described (n = 55)

• No BPI reference (n = 87)

• Redundancy (n = 220)

Eligible publications (n = 61)

Eligible publications obtained via

reference follow up (n = 20)

Publications included in qualitative analysis (n = 81)

Ident.ScreeningEligibilityIncluded

ICEIS 2022 - 24th International Conference on Enterprise Information Systems

660

Figure 4: Delphi study design.

In Round 1, the expert panel was asked to rate the

metamodel classes of the initial metamodel draft.

They could Retain, Adapt or Drop the individual

classes as well as Add further ones. The results of

round 1 were analyzed and consolidated using a

systematic decision tree which has already been used

in different Delphi studies and proved to be

appropriate (Serral, 2020). A class was only dropped,

if more than 60% of the experts agreed on this option.

No adaptions were considered, if the percentage to

retain was at least 80%, while minor adaptions were

performed for a retain rate between 60% and 80%.

Major adaptions were needed if the retain rate was

below 40% or at least 50% of the experts agreed on

the option to adapt a class. In Round 2, the experts

validated the results of the first round, followed by

another consolidation phase. In Round 3, the expert

panel was requested to rate the attributes of each

class. For new classes, they were asked to introduce

corresponding attributes. The consolidated results

were validated in Round 4. After this round, a

discussion with all experts helped to get feedback and

gain insight into the background of the individual

decisions. Having refined the classes and attributes,

we analyzed relations and subsequently added

multiplicities for all classes.

4.4 The IoT-based BPI

Metamodel

The final metamodel for IoT-based BPI patterns

consists of 11 classes and 28 attributes. During the

first development iteration, we added five classes,

namely IoT Technology, Application Area,

Interaction, Goal, and Process Specification. During

the refinement, two further classes Process

Perspective and Value Proposition could be created.

The previously added classes Goal and Process

Specification, on the other hand, were removed as a

result of the Delphi study. In addition, the class

Building Block of the base metamodel was removed.

Figure 5 shows the resulting metamodel including all

classes, attributes, and relations which will be

subsequently explained in detail.

According to Falk et al. (2013), the class Building

Block can be used for result-oriented patterns, i.e.,

patterns that directly describe the target process, and

are models that can be implemented without

adjustments. In contrast, procedure-oriented patterns

only describe instructions on how to improve the

process, but no direct implementation. Since IoT

systems are very complex and cannot provide any

benefit without appropriate integration in the process,

it is assumed that patterns for process improvement

through IoT can only be procedure-oriented.

Therefore, the expert panel agreed to delete this class

from the metamodel. Also, we changed the

multiplicity of the class Mechanism. In the base

metamodel, each solution contained exactly one

mechanism. However, this is an unnecessary

restriction that makes it difficult for modelers to

create domain-specific BPI models. By removing the

restriction, it is possible to define further

implementation details of the IoT system, while the

modeler is given greater freedom. The first new class

of the extended metamodel is Interaction. As part of

the solution, it describes the Human Involvement in

the IoT system. This is an essential aspect for

describing the integration of the IoT system into the

process and has already been discussed by Patterson

et al. (2017). For example, it can describe whether a

dashboard is only available to the process owner or

whether every actor in the process is always provided

with information via wearables, smartphones, or

other devices. It comprises interfaces between the IoT

application and humans regarding data input and

output. Being a domain-specific element, it constrains

the generic BPI metamodel to an IoT-based BPI

metamodel. In particular, the information output or

the information transfer to human actors had to be

modelled previously using the class Mechanism or

could not be modelled at all. Each solution can

contain one or more Interactions as there might be

several interfaces regarding data input or output, or

different groups of persons might be affected.

However, the class is not mandatory, as highly

automated IoT systems might not have any human

involvement at all. As another new class, IoT

Technology has been added to the metamodel.

Patterns for IoT-based Business Process Improvements: Developing a Metamodel

661

Figure 5: Metamodel for IoT-based BPI patterns.

As the base metamodel could not be used to

represent these requirements, the aspects for the basic

description of the technological requirements were

combined under this generalized class. The class

comprises two attributes that explain the necessary

technological specification of the IoT application. At

first, the Smart Device type reflects technological and

architectural principles. As already described in

subsection 2.1, sensors, actuators, and network

technologies can turn conventional objects into smart

things (or devices). These aspects can be displayed

within this attribute. Kortuem et al. (2010), e.g., have

already defined three different types of smart devices.

Activity-aware devices understand events and

activities causally related to the use of the object.

Policy-aware devices can reflect whether activities

and events are compliant with organizational policies

and process-aware devices can place activities and

events in the context of processes. A more detailed

description of the required hardware, e.g., sensors and

actuators, and networking technologies would be too

concrete for the creation of generic patterns. The

second attribute Data Processing describes the basic

features of how the collected IoT data is analyzed and

eventually used to improve the underlying business

processes. With cloud computing, for instance, the

IoT device is only responsible for generating the data

and does not provide any data processing capabilities.

In contrast to centralized data processing, edge

computing involves processing and analyzing the

generated data (or at least parts of it) directly at the

edge of the network by specially designed devices.

Depending on the application and the structure of the

IoT system, hybrid approaches can be possible, too.

Directly connected to IoT Technology is the new class

Value Proposition. It describes the actual value that

the IoT technology can provide to solve the addressed

problem. It goes beyond the simple description of

technological specifications but rather outlines,

which disruptive features and capabilities the

combination of sensors, networking, and data

processing technologies enables. The first attribute

that details the class is Key Capabilities. The IoT

comprises novel and disruptive capabilities that

distinguish it from other technologies. To enable

beneficial BPIs, these capabilities must be profitably

and systematically exploited. While the combination

of these capabilities is often relevant for IoT-based

BPIs, in most cases individual key capabilities can be

identified that are particularly relevant. Examples for

such capabilities would be universal scalability,

comprehensive perception, embedded intelligence, or

interoperability. By using specific IoT technologies

and therefore exploiting a set of capabilities, the IoT

Maturity can be defined. Maturity in this case refers

to the complexity of an IoT application, how deeply

it is embedded into the process, and how value is

generated. It ranges from simple data collection and

analytics to completely automated tasks within the

process. Tai Angus Lai et al. (2018) have addressed

this topic and identified different possibilities to

define this IoT maturity. They stated situational

awareness, decision-making support, information

exchange, and autonomous systems as potential

manifestations. Finally, the class Process Perspective

was added to the metamodel. It describes the

perspectives and therefore process aspects that are

influenced most by the IoT application. This is

especially useful to illustrate, how the IoT application

affects and redesigns the process. Jablonski and

Bussler (1996) have stated six process perspectives

that can be used in this regard. The behavioral

perspective comprises elements of the right process

workflow or sequence, legal regulations such as

ICEIS 2022 - 24th International Conference on Enterprise Information Systems

662

reporting obligations, and internal requirements. The

organizational perspective focuses on the personnel

that is involved in the process execution. Its main

components are responsible process owners, admins,

and users. In addition, the underlying system is part

of this perspective and represents for example the IT

environment. The functional perspective includes the

concrete process steps, tasks, and events. Most of the

processes, especially in the industry, comprise several

machines, tools, and software applications which can

be described as the operational perspective. The data

perspective involves all data and documents that are

necessary for process execution. Finally, the

locational perspective describes the specific locations

of process entities, e.g., machines or workers.

5 EVALUATION

5.1 Evaluation Setup

To be applicable for further research or industrial use,

we applied the FEDS of Venable et al. (2016) to

formulate an evaluation goal, derive evaluation

criteria, and apply an appropriate evaluation method.

To assess the goal achievement, we chose the

evaluation criteria usefulness, conciseness, and

robustness, as described by Prat et al. (2015). First,

the metamodel must enable an appropriate derivation

and description of patterns. Hence, it must comprise

all classes and attributes that are required to illustrate

generic abstractions of IoT-based BPIs. We translated

this into the criterion usefulness. Furthermore, it must

be appropriately detailed to depict different patterns,

described as robustness. However, it also must be

concise and generic enough to be applicable for a

wide range of possible applications. The evaluation is

performed by deriving a set of IoT-based BPI patterns

and investigating if the created metamodel can

adequately illustrate them. For deriving the patterns,

we requested seven practitioners from the Linde plc

and two other corporations to analyze IoT

applications within their business areas. All

practitioners have a broad knowledge in IoT

technology and business processes in general and

have working experiences of five to 21 years. They

work as technical project managers, IT managers,

automation experts, or digitalization managers while

each of them has implemented at least two major IoT

applications in the primary value chain activities of

their corporations. In total, they identified 34

applications that were suitable for further analysis. In

a joint workshop, six different patterns could be

derived and illustrated using the provided metamodel.

These patterns are Process Guidance, Derivation

Detection, Authentication & Authorization, Task

Distribution, Proactive Activity Execution, and

Activity Automation.

5.2 Pattern: Process Guidance

The first pattern Process Guidance (see Figure 6)

generically describes applications focusing on

improved user guidance. By capturing situational and

process-related data, the actual process state and

subsequent process sequences can be ascertained. The

next process tasks can then be displayed to process

participants, e.g., via wearables. This pattern mainly

affects the operational and data perspectives, as the

way of performing the process tasks is changed by

using input and output data. The used smart devices

are process-aware as they need to capture process-

related data, process it, and provide it to the process

participants with respect to the current process state.

Figure 6: Process Guidance pattern.

BPI Pattern

Effect

Problem

Context

Solution

Name: Quality.1

Performance Measures: Error

rate, Repetition loops

Name: Process Ambiguities.1

Description: For new process users

or complex processes, the correct task

performances and process sequences

are difficult to understand.

Consequences: Long processing

times, Dissatisfied users, High error

rates

Name: Context.1

Characteristics: Process sequences

depend on process-related data and

situational data.

Name: Data-based Process Guidance.1

Measures: Integrating sensors to

monitor situational data. Mapping data

with process sequences to display

guidance for tasks.

Name: Process Guidance.1

Example: While purchasing

propane cylinders at a vending

machine, the customer is guided

through the process tasks. The

customer gets visual indications

where to put empty returns and

where to collect purchased items.

The system recognizes task

performances using sensors and

guides through tasks according to

the collected data and the

underlying process model.

Name: Operational.1

Description: Guided task

execution

Name: Effect.1

Cost: (0) neutral

Time: (+) positive

Quality: (+) positive

Flexibility: (-) negative

Name: Data.1

Description: Capturing and

providing situational data

Technology

Name: Perception and Provision.1

Smart Devices: Process-aware systems

capturing process-related and situational

data via sensors. Audiovisual devices

displaying process information, e.g.,

wearables or light bars.

Data Processing: Processing situational

data in edge-devices and mapping with

expected process-data from, e.g., BPMS.

Triggering subsequent process tasks and

providing explanatory information.

Performance Indicator

Mechanism

Name: Data Collection and Provision.1

Instruction: Collecting process-related

and situational data to enable provision of

correct process sequences and task

descriptions.

Process Perspective

Name: Time.1

Performance Measures:

Processing time

Performance Indicator

Interaction

Name: Guidance.1

Human Involvement: Process users are

perceiving information on process tasks

from the IoT device. They need to

follow the provided guidance.

Value Proposition

Name: Information Exchange.1

Key Capabilities:

Comprehensive perception,

Embedded intelligence

IoT Maturity: Information

exchange

Patterns for IoT-based Business Process Improvements: Developing a Metamodel

663

Figure 7: Deviation Detection pattern.

The exemplary process is taken from the Linde plc,

where customers are guided through the purchasing

process at a gas vending machine.

This is done be capturing process-related data and

highlighting the next process steps via light bars.

Another suitable literature application is the training

of new employees in a manufacturing company

(König, 2019). Employees are guided through tasks

by tracking the current process data and visualizing

process descriptions of subsequent tasks. Other

organizations have implemented applications to

guide the employees through production or logistic

processes by capturing environmental and process

data, processing it, matching it with process models,

and providing guidance for tasks (De Vries, 2015).

5.3 Pattern: Deviation Detection

The second pattern Deviation Detection is exemplary

described using a cylinder filling process of the Linde

plc. A main challenge for organizations is the

detection of process deviations during runtime to

identify incorrect task executions and adequately

adapt the subsequent process flows. Deviations lead

to low process quality, process deadlocks, or the need

for process support. The pattern is illustrated in

Figure 7. After the filling of toxic cylinders, they must

be placed in the right areas according to the process

description. Incorrect task executions include high

risk potential. By implementing location sensors that

collect data of the task execution and collating it with

expected values from the process description,

deviations can be detected. This enables the initiation

of countermeasures and leads to an improved error

recognition rate which has a positive impact on the

overall process quality. The pattern addresses the

functional and data perspectives, as the execution of

the process task is monitored. The IoT technology

includes activity-aware smart devices that process

situational data on edge devices or (hybrid) cloud

servers. To identify deviations of any kind, the key

capability comprehensive perception must be

exploited enabling situational awareness of all

process details. Similar industrial applications ca be

found for the detection of machine failures where

sensor data is used for diagnostics and detection of

deviations, e.g., at leakage detection (Ammirato,

2019) or other anomalies (Schneider, 2019).

5.4 Expert Survey Results

After the practitioners derived six patterns from the

sample of 34 applications, they were asked to perform

an expert survey. They received a list of six

statements for which they needed to indicate their

agreement or disagreement. This followed the proven

psychometric tool of the Likert scale (Albaum, 1997).

The statements were formulated in a way that allows

conclusions to be drawn about the three evaluation

criteria. Table 1 shows all statements and the obtained

survey results. As shown, most of the practitioners

agreed or strongly agreed with all statements. Only

for the second statement, one practitioner could not

specifically state, if the metamodel’s degree of

abstraction and generalization is appropriate for the

derived patterns. Within the statements, especially the

first three refer to the criterion usefulness, the fourth

to the criterion conciseness, and the fifth and sixth to

the criterion robustness. In a subsequent discussion,

the experts stated, that the metamodel enabled an

appropriate illustration of IoT-based BPI patterns.

Also, the classes and attributes supported the analysis

of heterogenous IoT applications and the derivation

of generic patterns.

BPI Pattern

Effect

Problem

Context

Solution

Name: Quality.2

Performance Measures: Error

recognition rate, Rework rate

Name: Process Intransparencies.1

Description: The correct process task

execution is hard to verify. Incorrect

executions have a negative impact on

subsequent tasks and the whole

process. The selection of appropriate

countermeasures depends on the

deviation identification.

Consequences: Low process quality,

Process deadlocks, Scrap and rework

Name: Context.2

Characteristics: Poorly or incorrectly

executed tasks must be identified.

Appropriate countermeasures must be

initiated.

Name: Process Data Reconcilitaion.1

Measures: Integrating sensors to

collect and analyze process task data.

Mapping collected data with expected

data to find deviations. Initiating

countermeasures, if required.

Name: Deviation Detection.1

Example: After the filling of toxic

gas cylinders, the worker must place

the cylinders in the right area for

subsequent tasks. If the cylinders are

placed in the wrong area, this

involves high risk potential, e.g., for

placinginfoodorchemicalgasareas.

Sensors collect locational data of the

toxic cylinders. This data is

processed, analyzed, and mapped

with expected data of the process

task. If the perceived data does not

comply with the expected locational

data, a deviation is detected. This

information can be used to prevent a

wrong placement or to initiate a

repositioning of the toxic cylinders

and a scrapping of the food or

medical gases.

Name: Functional.1

Description: Process task

monitoring

Name: Effect.2

Cost: (0) neutral

Time: (0) neutral

Quality: (+) positive

Flexibility: (0) neutral

Name: Data.2

Description: Collection and

reconciliation of data

Technology

Name: Sensing.1

Smart Devices: Activity-aware

systems sensing and collecting

data.

Data Processing: Processing

situational data in edge-devices or

(cloud-) servers. Comparing data

with defined values/thresholds.

Performance Indicator

Mechanism

Name: Data Collection and

Reconciliation.1

Instruction: Sensing, collecting,

and analyzing task data. Mapping

with expected data and initiating

countermeasures, if required.

Process Perspective

Value Proposition

Name: Situational Awareness.1

Key Capabilities:

Comprehensive perception

IoT Maturity:

Situational awareness

ICEIS 2022 - 24th International Conference on Enterprise Information Systems

664

Table 1: Expert survey results.

No. Statement

Strongly

agree

Agree Neutral Disagree

Strongly

disagree

The metamodel’s classes and attributes enable an appropriate

illustration of IoT-

ased BPI patterns.

29% 71% 0% 0% 0%

The classes and attributes allow an appropriate degree of

abstraction and generalization.

29% 57% 14% 0% 0%

The created patterns can illustrate and describe generic business

process problems and potential solutions provided by IoT

technology.

14% 86% 0% 0% 0%

Extending the metamodel would contradict its generic design,

limit its generality, and decrease the number of applications that

can be covered by a pattern.

29% 42% 29% 0% 0%

Removing classes and attributes would reduce the

expressiveness of the patterns.

86% 14% 0% 0% 0%

The classes and attributes allow a sufficient differentiation of the

represented patterns.

71% 29% 0% 0% 0%

6 CONCLUSION

The contribution of this paper is a metamodel to

illustrate generic IoT-based BPI patterns as an

extension of the metamodel for BPI patterns of Falk

et al. (2013). In two development iterations,

additional classes and attributes were discovered and

irrelevant ones were dropped. At first, we performed

an inductive development iteration including an SLR

followed by open and axial coding. Based on the

results, additional classes and attributes could be

derived and added to the existing metamodel. The

first metamodel draft was then refined by conducting

a Delphi study with nine experts from industry and

academia. To evaluate the final metamodel, seven

practitioners from the Linde plc and two other

corporations analyzed a set of 34 real-life IoT

application of their business areas. Eventually, they

derived six IoT-based BPI patterns and illustrated

them using the metamodel. In a subsequent survey the

experts assessed the metamodel according to the

predefined evaluation criteria usefulness,

conciseness, and robustness. The survey showed that

the metamodel sufficiently meets these criteria.

Despite the rigorous research methodology, the

contribution is not without limitations due to the

nature of DSR. Following an inductive approach for

metamodeling is a proven concept that provides

several advantages arising from building up on actual

observations. However, the underlying SLR cannot

cover all existing data of the phenomenon under

investigation. The identification of literature is

limited to the incorporated databases and formulated

queries. To mitigate this subjectivity, we conducted a

subsequent Delphi study. This enabled both a

formative evaluation of the first metamodel draft and

the inclusion of broad expert knowledge.

Further research should be conducted in various

directions. Having evaluated the usefulness,

conciseness, and robustness of the metamodel, the

actual applicability of the generated patterns must be

assessed. In this regard, we plan to create and

introduce patterns to industrial organizations. This

enables a further evaluation, if the patterns can be

effectively used to realize IoT-based BPI

applications. Moreover, the creation of a

comprehensive pattern catalogue would provide

additional benefit and validate the metamodel itself.

REFERENCES

Albaum, G., 1997. The Likert Scale Revisited. Int. J. Mark.

Res. 39(2).

Alexander, C., 1977. A pattern language: towns, buildings,

construction. Oxford University Press, New York City.

Ammirato, A., Sofo, F., Felicetti, A. M., Raso, C. 2019. The

potential of IoT in redesigning the bank branch

protection system: An Italian case study. Bus. Process

Manag. J. 26(7).

Atzori, L., Ier,a A., Morabito, G., 2010. The Internet of

Things: A survey. Comput. Netw. 54(15).

Bosche, A., Crawford, D., et al., 2016. How providers can

succeed in the Internet of Things. Bain & Company Inc.

https://www.bain.com/insights/how-providers-can-

succeed-in-the-internet-of-things/.

Corbin, J. M., Strauss, A., 1990. Grounded theory research:

Procedures, canons, and evaluative criteria. Qual.

Sociol. 13.

Coskun, S., Basligil, H., Baracli, H., 2008. A weakness

determination and analysis model for business process

improvement. Bus. Process Manag. J. 14(2).

De Vries, J., de Koster, R., Stam, D., 2015. Exploring the

role of picker personality in predicting picking

Patterns for IoT-based Business Process Improvements: Developing a Metamodel

665

performance with pick by voice, pick to light and RF-

terminal picking. Int. J. Prod. Res. 54(8).

Del Giudice, M., 2016. Discovering the Internet of Things

(IoT) within the business process management: a

literature review on technological revitalization. Bus.

Process Manag. J. 22(2).

Dumas, M., La Rosa, M., Mendling, M., Reijers, H. A.,

2018. Fundamentals of Business Process Management.

Springer Verlag, Berlin, Heidelberg.

Falk, T., Griesberger, P., Johannsen, F., Leist, S., 2013.

Patterns for Business Process Improvement - A first

approach. In 21

European Conference on Information

Systems. ECIS.

Ferstl, O. K., Sinz, E. J., 2013. Grundlagen der

Wirtschaftsinformatik. Oldenbourg Verlag, München.

Forster, F., 2006. The idea behind Business Process

Improvement: Toward a Business Process

Improvement pattern framework. BP Trends 4.

Fowler, M., 1996. Analysis Patterns: Reusable object

models. Addison-Wesley, Boston.

Fowler, M., Scott, K., 1997. UML Distilled: Applying the

standard object modelling language. Addison-Wesley

Longman, Amsterdam.

Gamma, E., Helm, R., Johnson, R., 1994. Design Patterns:

Elements of reusable object-oriented software. Prentice

Hall, Hoboken.

Gonzalez-Perez, C., Henderson-Sellers, B., 2008.

Metamodelling for software engineering. Wiley,

Hoboken.

Hempel, C. G., 1966. Philosophy of natural science.

Prentice-Hall, Upper Saddle River.

Hevner, A. R., March, S. T., Park, J., and Ram, S., 2004.

DS in information systems research. MISQ 28(1).

Hohpe, G., Woolf, B., 2003. Enterprise Integration

Patterns: Designing, building, and deploying

messaging solutions. Addison Wesley, Boston.

Jablonski, S., Bussler, C., 1996. Workflow Management:

Modeling concepts, architecture and implementation.

Cengage Learning EMEA, Andover.

Janiesch, C., Koschmider, A., et al., 2020. The Internet of

Things meets Business Process Management: A

manifesto. IEEE Trans. Syst. Man. Cybern. Syst. 6(4).

Kortuem, G., Kawsar, F., Sundramoorthy, V., Fitton, D.,

2010. Smart objects as building blocks for the Internet

of Things. IEEE Internet Comput. 14(1).

König, U. M, Röglinger, M., Urbach, N., 2019. Industrie

4.0 in kleinen und mittleren Unternehmen – Welche

Potenziale lassen sich mit smarten Geräten in der

Produktion heben? HMD der Wirtschaftsinformatik 56.

Kühn, H., Karagiannis, D., 2005. Strategie-, Prozess- und

IT-Management: Ein Pattern-orientierter

Integrationsansatz. In Ferstl, O. K., Sinz, E. J., et al.

(ed) Wirtschaftsinformatik 2005. Physica, Heidelberg.

Li, Y., Hou, M., Liu, H., Liu, Y., 2012. Towards a

theoretical framework of strategic decision, supporting

capability and information sharing under the context of

Internet of Things. J. Inf. Technol. Manag. 13.

Liu, V., 2017. Business benefits of the Internet of Things:

Gartner report. Gartner Inc. https://www.gartner.com/

en/documents/3806366/business-benefits-of-the-

internet-of-things-a-gartner-tr.

Loo, R., 2002. The Delphi method: a powerful tool for

strategic management. Policing 25(4).

Minerva, R., Biru, A., Rotondi, D., 2015. Towards a

definition of the Internet of Things (IoT). IEEE.

https://iot.ieee.org/definition.html.

Ng, I., Wakenshaw, S., 2017. The Internet-of-Things:

Review and research directions. Int. Res. Mark. 34(1).

Patterson, R. E., 2017. Intuitive cognition and models of

human-automation interaction. Hum Factors 59(1).

Peffers, K., Tuunanen, T., Rothenberger, M., Chatterjee, S.,

2007. A DSR Methodology for Information Systems

Research. J. Manag. Inf. Syst. 24(3).

Prat, N., Wattiau, I., Akoka, J., 2015. A Taxonomy of

Evaluation Methods for Information Systems Artifacts.

J. Manag. Inf. Syst. 32(3).

Reijers, H. A., Liman Mansar, S., 2005. Best practices in

business process redesign: an overview and qualitative

evaluation of successful heuristics. Omega 33(4).

Reinfurt, L., Breitenbücher, U., Falkenthal, M., et al., 2016.

Internet of Things patterns. In 21

European

Conference on Pattern Languages of Programs.

Schmidt, D. C., Stal, M., Rohnert, H., Buschmann, F., 2000.

Pattern-oriented software architecture. Wiley,

Hoboken

Schneider M., Lucke D., Adolf T., 2019. A cyber-physical

failure management system for smart factories.

Procedia CIRP 81.

Serral, E., Stede, C. V., Hasić, F., 2020. Leveraging IoT in

etail Industry: A Maturity Model. In 22

Conference

on Business Informatics.

Sethi, P., Sarangi, S. R., 2017. Internet of Things:

Architectures, protocols, and applications. Int. J. Electr.

Comput. Eng. 2017.

Skaržauskienė, A., Kalinauskas, M., 2015. The internet of

things: when reality meets expectations. Int. J. Innov.

Learn. 17(2).

Stoiber, C., Schönig, S., 2021. Process-aware Decision

Support Model for Integrating Internet of Things

Applications using AHP. In 23

International

Conference on Enterprise Information Systems. ICEIS.

Tai Angus Lai, C., Jackson, P. R., Jiang, W., 2018.

Designing service business models of the IoT: Aspects

from manufacturing firms. Am. J. Math. Ma. Sci. 3(2).

Templier, M., Paré, G., 2018. Transparency in literature

reviews: an assessment of reporting practices across

review types and genres. Eur. J. Inf. Syst 27(5).

Venable, J., Pries-Heje, J., Baskerville, R., 2016. FEDS: a

Framework for Evaluation in Design Science Research.

Eur. J. Inf. Syst. 25.

Vom Brocke, J., Niehaves, A., et al., 2009. Reconstructing

the giant: On the importance of rigour in documenting

the literature search process. In 17

European

Conference on Information Systems, ECIS.

Washizaki H., Ogata S., Hazeyama A., et al., 2020.

Landscape of architecture and design patterns for IoT

systems. IEEE Internet Things J.

Winter, R., vom Brocke, J., et al., 2009. Patterns in business

and information systems engineering. Bus. Inf. Syst.

Eng. 1(6).

ICEIS 2022 - 24th International Conference on Enterprise Information Systems

666