The Internet-of-Things Meets Business Process
Management: Mutual Beneﬁts and Challenges
Christian Janiesch1, Agnes Koschmider2, Massimo Mecella3, Barbara Weber4, Andrea
Burattin4, Claudio Di Ciccio5, Avigdor Gal6, Udo Kannengiesser7, Felix Mannhardt8, Jan
, Andreas Oberweis
, Manfred Reichert
, Stefanie Rinderle-Ma
, WenZhan Song
Jianwen Su12, Victoria Torres13, Matthias Weidlich14, Mathias Weske15, Liang Zhang16
The Internet of Things (IoT) refers to a network of connected devices collecting and exchanging data over the
Internet. These things can be artiﬁcial or natural, and interact as autonomous agents forming a complex system.
In turn, Business Process Management (BPM) was established to analyze, discover, design, implement, execute,
monitor and evolve collaborative business processes within and across organizations. While the IoT and BPM
have been regarded as separate topics in research and practice, we strongly believe that the management of IoT
applications will strongly beneﬁt from BPM concepts, methods and technologies on the one hand; on the other
one, the IoT poses challenges that will require enhancements and extensions of the current state-of-the-art in
the BPM ﬁeld. In this paper, we question to what extent these two paradigms can be combined and we discuss
the emerging challenges.
IoT (Internet-of-Things) — BPM (Business Process Management) — Challenges — Manifesto
1University of W¨
urzburg, Germany –christian.janiesch@uni- wuerzburg.de
2Karlsruhe Institute of Technology, Germany –firstname.lastname@example.org —email@example.com
a di Roma, Italy –firstname.lastname@example.org
4Technical University of Denmark, Denmark –email@example.com —firstname.lastname@example.org
5WU Vienna, Austria –email@example.com —firstname.lastname@example.org
6Technion – Israel Institute of Technology, Israel –email@example.com
7Metasonic GmbH, Germany –firstname.lastname@example.org
8SINTEF, Trondheim, Norway –email@example.com
9Ulm University, Germany –firstname.lastname@example.org
at Wien, Austria –email@example.com
11University of Georgia, USA –firstname.lastname@example.org
12University of California at Santa Barbara, USA –email@example.com
13Technical University of Valencia, Spain –firstname.lastname@example.org
at zu Berlin, Germany –email@example.com
15University of Potsdam, Germany –firstname.lastname@example.org
16Fudan University, China –email@example.com
2.1 Introduction to the Internet of Things . . . . . . . . . 2
2.2 Introduction to Business Process Management . 3
3.1 Overview of Challenges ..................3
3.2 Challenges in Detail .....................4
4Concluding remarks 7
Our world is increasingly linked through a large number of
connected devices, typically embedded in electrical / electron-
ical components and equipped with sensors and actuators, that
enabling sensing, (re-)acting, collecting and exchange data
via various communication networks including the Internet:
the Internet of Things (IoT). As such, it enables continuous
monitoring of phenomena based on sensing devices (wear-
able devices, beacons, smartphones, machine sensors, etc.) as
well as analytics opportunities in smart environments (smart
homes, connected cars, smart logistics, Industry 4.0, etc.) and
the possibility to actuate feedback. Therefore, the IoT con-
tributes to the recent trend known as big data, being one of
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 2/9
the three main sources besides human sourced and process
Business processes represent a speciﬁc ordering of tasks
and activities across time and place to serve a business goal.
Process analytics, execution and monitoring based on IoT data
can enable an even more comprehensive view of processes
and realize unused potential for process optimization. As an
example, in the past process analytics and in particular pro-
cess mining has been hampered by the fact that process are
often incomplete or erroneous; with the IoT producing a large
amount of data stored in the cloud, even more data become
available for analysis, possibly resolving issues of incomplete-
ness and enabling providing error correction methods based
on multiple data items .
In the literature, several works are emerging on combining
Business Process Management (BPM) and IoT, e.g., utilizing
sensor data to enable the actuation of services [
] or adapting
running business processes to continuously align them with
the state of the things (e.g., assets, humans, and machines).
Still, there are many open challenges to be tackled (cf. Figure
1). Both BPM and IoT will beneﬁt from a wider integration.
How IoT can beneﬁt from BPM?
Let us consider a com-
plex system with multiple components interacting within a
smart environment being aware of the components’ locations,
movements, and interactions. Such a system can be a smart
factory with autonomous robots, a retirement home with con-
nected residents, or, at a larger scale, a smart city. While
the parties in the system can track the movements of each
component and also relate multiple components’ behaviors to
each other, they do not know the components’ agendas. Often
their interactions are based on habits, i.e., routine low-level
processes, which represent recurring tasks. Some of these
routines are more time and cost critical than others, some
may be dangerous or endanger others, and some may just be
inefﬁcient or superﬂuous. Knowing their agendas, their goals,
and their procedures can enable a better basis for planning,
execution, and safety.
How BPM can beneﬁt from IoT?
Let us consider a com-
plex process with multiple parties interacting in the context
of a business transaction. Such a process can be, for example,
a procurement process, where goods are ordered, delivered,
stored, and paid for. While the system can track each auto-
matically executed activity on its own, it relies on messages
from other parties and manually entered data in the case of
manual activities. If this data is not entered or entered incor-
rectly, discrepancies between the digital (i.e., computerized
representation of the) process and the real-world execution
of the process occur. Similar concerns hold if the process
participants do not obey the digital process under certain cir-
cumstances, e.g., an emergency in healthcare, or have not
entered the data yet though in the real-world process the re-
spective activity was already executed.
Such scenarios might be better manageable when closely
linking the digital process with the physical world as enabled
by the integration of IoT and BPM; e.g., the completion of
manual activities can be made observable through usage of
appropriate sensors. IoT can complete BPM with continuous
data sensing and physical actuation for improved decision
making. Decisions in processes require relevant information
as basis for making meaningful decisions. In general, it is not
sufﬁcient to retrieve this data solely from traditional reposito-
ries (e.g., databases and data warehouse) providing historical
data, but also up-to-date data are needed. Data from the IoT,
such as events, provided through in-memory databases or
complex event processing can be useful in this context. The
IoT could reduce the need to manually signify the completion
of manual tasks since sensor data is already available, leading
to more accurate data, reduced errors, and efﬁciency gains.
To ensure that both domains can mutually beneﬁt from
each other, still exist several challenges to be tackled. Particu-
larly, it has to be understood:
how processes can improve the IoT by (i) taking a
process-oriented perspective and considering the pro-
cess history to (ii) bridge the abstraction gap between
raw sensor data and higher level knowledge extracted
from this event data, and to (iii) optimize the decision-
making in the large;
how to exploit IoT for BPM by (i) considering sensor
data for automatically detecting the start and end of
activities, (ii) using event data for making decisions
in a pre-deﬁned process model, and (iii) detecting dis-
crepancies between the pre-deﬁned model and actual
enactment using event data for online process compli-
ance checking and exception management.
In the following of this paper
, taking these two general
questions as starting point, we discuss IoT and BPM back-
grounds in Section 2. Then in in Section 3, we detail the
key challenges in combining BPM and IoT and elaborate on
beneﬁts of BPM for IoT and IoT for BPM, before offering
concluding remarks in Section 4.
2.1 Introduction to the Internet of Things
The Internet of Things (IoT) [
] is the inter-networking of
physical objects (the things), being such things embedded sys-
tems with electronics hardware, software, sensors, actuators,
and network connectivity. Such connected things collect and
exchange data. Each thing is uniquely identiﬁable through
its embedded computing system and is able to interoperate
within the existing network infrastructure. While things act
local, the IoT allows things to be controlled remotely across
existing network infrastructures, including the Internet.
This paper has its roots in the Dagstuhl Seminar 16191 Fresh Approaches
to Business Process Modeling, organized by Richard Hull, Agnes Koschmider,
Hajo A. Reijers, and William Wong at the Leibniz Center for Informatics in
Germany, May 8–13, 2016, cf.
volltexte/2016/6696/, to which many authors participated.
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 3/9
Figure 1. High-level overview showing the interaction between IoT and BPM. The numbering used in the blocks will
correspond to the numbering of the different paragraphs in the following sections.
The interconnection of these smart objects/things is ex-
pected to usher in automation in nearly all ﬁelds. This creates
opportunities for more direct integration of the physical world
into computer-based and digitized systems, and results in im-
proved efﬁciency, accuracy, and economic beneﬁts besides
increased automation and reduced human intervention. Ex-
perts estimate that the IoT will consist of about 30 billion
objects by 2020 .
2.2 Introduction to Business Process Management
Business Process Management is a well-established disci-
pline that deals with the identiﬁcation, discovery, analysis,
(re-)design, implementation, execution, monitoring, and evo-
lution of business processes [
]. A business process is a col-
lection of related events, activities, and decisions that involve
a number of actors and resources and that collectively lead to
an outcome that is of value for an organization or a customer.
]. Examples of business processes include order-to-cash,
procure-to-pay, application-to-approval, claim-to-settlement,
or fault-to-resolution. To support business processes at an
operational level, a BPM system (BPMS) can be used [
As opposed to data- or function-centered information systems,
a BPMS separates process logic from application code and,
thus, provides an additional architectural layer. Typically, a
BPMS provides generic services necessary for operational,
software-enabled business process support, i.e., for process
modeling, process execution, process monitoring, and user
interaction (e.g., worklist management). When using a BPMS,
software-enabled business processes are designed in a top-
down manner, i.e., process logic is explicitly described in
terms of a process model providing the schema for process
execution. The BPMS is responsible for instantiating new
process instances, for controlling their execution based on the
process model, and for completing them. The progress of a
process instance is typically monitored and traces of execution
are stored in an event log and can be used for process mining
], e.g., the discovery of a process model from the event log
or for checking the compliance of the log with a given process
So far, the predominant paradigm to develop operational
support for business processes has been based on the Model-
Enact paradigm, where the business process has been depicted
as a (graphical) process model, which then could be executed
by a BPMS. This largely follows a top-down approach and is
based on the idea of a central orchestrator that controls the exe-
cution of the business process, its data, and its resources. With
the emergence of IoT, the existing Model-Enact paradigm is
challenged by the Discover-Predict paradigm; it can be charac-
terized as a bottom-up approach where data is generated from
physical devices sensing their environment and producing raw
events. Sensor data then must be aggregated and interpreted in
order to detect activities that can be used as input for process
mining algorithms supporting decision-making .
This section describes the challenges of interaction between
the IoT and the BPM paradigm.
3.1 Overview of Challenges
The IoT has to deal with a number of challenges; this includes,
for example, technological barriers such as computational
limitations of embedded systems or the connectivity to back-
end systems, security-related issues, a lack of standards, data
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 4/9
privacy issues, untapped potential in data analytics, efﬁcient
methods for the organization of IoT systems, etc. [
principal characteristic of the IoT is the communication be-
tween loosely-coupled objects, which mostly is accomplished
asynchronously and ad-hoc.
BPM deals with the discovery of models, the analysis
of pre-deﬁned models, the adaptation of models, and the
enactment of business processes. We consider processes as
explicit process representations (pre-deﬁned models), which
later are enacted. Abstract processes can also be discovered
from log ﬁles and suitable implementations for instantiation
can be predicted.
Accordingly, sensing and perception via sensors and deci-
sion based on sensors as well as decision based on actuation
according to individual goals/strategies constitute fundamen-
tal tasks of the IoT. Thereby, sensing constitutes the input
and actuation the output of any IoT-BPM interaction (see also
Figure 1). In between, raw event data is processed by event-
based systems, transforming the input events to higher-level
knowledge. In turn, the latter may be utilized by BPM con-
cepts, methods or technologies to deal with the discovery of a
model, the analysis of a pre-deﬁned model, the adaptation of
a model and the enactment of business processes.
While the IoT generally focuses on communication and
data ﬂow, BPM approaches consider control ﬂow, big mono-
lithic process models, and synchronous interactions. In addi-
tion, most BPM approaches have trouble dealing with non-
routine, non-deterministic processes, whereas IoT applica-
tions typically involve these kind of interactions.
Against these considerations, plenty of challenges arise,
which need be considered when improving business inter-
actions as well as alignment between BPM and IoT. In the
following we introduce each challenge in detail.
3.2 Challenges in Detail
C 1 – Placing sensors in a process-aware way.
to collect all relevant data, sensors need to be carefully placed.
It constitutes already a challenge to construct sensors and
place them on agents (human or artiﬁcial) or in a smart spaces,
such that they are non-intrusive but still efﬁcient: sensors
can be battery-less tags such as RFID, battery and renewable
energy powered, or outlet-powered; and the communication
methods can be wired or wireless. It is even more challenging
to decide on the type of sensor and its placement with regard
to its function in respect to the interaction between agents [
A business process (model) may guide this placement since
it offers knowledge about resources, locations and variants
of behavior (enact), that need to be covered. As well, the
trade-off between the cost of introducing additional sensing
points and the expected increase in monitoring accuracy may
be approached based on process knowledge.
C 2 – Visualization support for managing manually ex-
ecuted, physical processes.
In many settings, BPM ap-
proaches are used to automate processes through the support
of a BPMS, in which some activities require the interplay
between human operators and software/hardware modules.
Notable examples, in addition to logistic and industrial pro-
cesses, include disaster management (e.g., a BPMS in the ﬁeld
to coordinate emergency teams [
]) and healthcare processes
], or [
] – a BPMS coordinating doctors and nurses
with vocal interfaces for human tasks). In many of these sce-
nario, there is an increasing use of mobile devices fostering
the delivery of work items to the right users .
In these settings, workers do not necessarily have to in-
teract with the BPMS while carrying out physical tasks (e.g.,
moving boxes in a warehouse): sensors, which are connected
to the BPMS, monitor whether or not such a task has started
or ended. However, appropriate mapping from process activi-
ties to the GUI and usable visualizations are needed allowing
actors (process participants) to perform their work in a natural
way, without requiring non-value adding management tasks
such as clicking on conﬁrmation buttons.
C 3 – Connection of analytical processes with IoT.
ing process execution, a variety of information is required to
make meaningful decisions. In turn, this information often
needs to be available not only from traditional databases/data
warehouses providing historical data, but it needs to be up-to-
date and current. In order to design systems providing such
up-to-date information, and to judge the quality of the data
analysis results from such applications, it needs to be clear
where the data stems from and where it has been used (data
provenance), as well as the overall quality of the data at-hand
needs to be ensured. This is particularly critical at the presence
of big data [
]. Generally, it becomes necessary to ﬁnd a way
to annotate the data’s origin and use this (meta-)information
in process models. So far, there is no universal method to
connect the analytic processes of observation, analysis, and
decision-making to business processes in a standardized way,
]; recent attempts include the Decision Model and No-
standard. Its focus, however, is on decision
requirements, but less on the origin and use of decision data.
Hence, it still needs to be investigated how to model quality
and provenance in order to be exploitable at the process model
Erroneous sensors, not working at all or delivering erro-
neous data, need to be discovered and excluded from any
analysis. In turn, this necessitates a reasonable judgment on
which sensor data might be erroneous. Here, the process con-
text in which these data occur might be helpful to identify
erroneous sensors as well as to cope with them.
C 4 – Integrating the IoT with process correctness checks.
Well-known techniques for analyzing the quality of process
models can contribute to improve the design of interactions
in IoT, by ﬁnding deadlocks, livelocks, or dead activities in
respect to the behavior or interactions of objects [
locks and livelocks are reasons why some processes may not
terminate in the assumed time frame or not at all. It can either
occur because certain actions are waiting for each other or
when actions change states but the process does not progress.
While a rollback is a typical service in data management, it
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 5/9
becomes much more costly and complicated when managing
processes and thus should be avoided. Dead activities do not
harm a processes execution (unless they are supposed to be
mandatory) since they will never be triggered. Yet, they rep-
resent a waste of resources as either or both, physical and/or
virtual resources may have been reserved for this activity.
Therefore, designing correct process models which specif-
ically consider the IoT nature of some components becomes
crucial, as well as the veriﬁcation of important properties.
C 5 – Dealing with unstructured environments.
fers a way to structure businesses. As such, it often assumes
a controlled environment with a managed repository of ver-
sioned processes that can be orchestrated for the purpose of a
single enterprise or be choreographed between parties in case
of cross-organizational collaborations. Orchestration denomi-
nates the execution order of the interactions from the perspec-
tive and under control of a single party, whereas choreography
describes public, i.e., globally visible, message exchanges,
interaction rules and agreements made among multiple parties.
Both concepts presume knowledge about the structure and/or
interactions of each participating process. It is questionable
whether orchestration and choreography still sufﬁce as orga-
nizational concepts in an IoT world, which is much more ad
hoc and situative (e.g., devices involved in the interaction
might fail, deliver erroneous data, new devices may have to
be ﬂexibly added, etc.).
C 6 – Managing the links between micro processes.
approach to bridge the gap between IoT data and processes,
would be to break end-to-end process models into micro pro-
cesses representing habits and arrange them in a less pre-
scriptive (control-ﬂow) way. Modeling a small and possibly
autonomous micro process does not necessarily require new
modeling constructs or methods. Yet, the organization of hun-
dreds/thousands of loosely coupled small processes is a task
that cannot be left to the forces of natural evolution in a busi-
ness environment. It may require new modeling constructs
and methods to structure and represent their non-hierarchical
interaction in human-readable form .
Data-centric process paradigms offer promising perspec-
tives in this context. For example, object-aware processes
] describe the behavior of single objects through micro
processes, whereas the dynamic construction of linked objects
as well as the their synchronized execution is described and
enforced through macro processes. However, respective ap-
proaches need to be enhanced to integrate physical objects as
well as their behavior in the overall process.
C 7 – Breaking down end-to-end processes.
For a large
class of processes (typically referred to as dynamic or
knowledge-intensive), the advent of overwhelming sensor
data and things acting in the environment without central con-
trol but according to “personal” agendas, makes it practically
impossible to deﬁne comprehensive end-to-end process mod-
els. Things will perform their own routines, so called repeated
behaviour patterns or habits [
]. Accordingly, processes
will have to be organized as event-driven micro processes to
represent these habits. Whereas the overall end-to-end busi-
ness process itself may be modeled in traditional ways, the
linking of micro-process models is far more complex; to cope
with this emerging complexity, the possible interactions be-
tween micro-process models must not be described at the low
level of message exchanges, but be put at a higher semantical
level, similar to the utilization of semantic object relations
for the purpose of object interactions in object-aware process
C 8 – Detecting new processes from data.
system in a bottom-up manner without prescriptive process
models promises more ﬂexible and inclusive processes. How-
ever, the question arises to what extent we can let the system
just evolve and be discovered. When developing support for
software-enabled business processes based on the principles
of the IoT, an evolutionary self-organising process will take
place in some respect. Thus, one must ﬁnd the appropriate
level of structuring and prescription without harming the ca-
pability to self-organize. There is a gap between IoT data
and concepts at a model level to enable behavior prediction
and to identify changes in behavior. The IoT allows deriv-
ing situational knowledge when tracking and evaluating data
streams. Situational knowledge, in turn, is input to analyze
prospective knowledge, which constitutes a dynamic task.
Prospective knowledge addresses long-tail information about
resources (e.g., how well is the person/thing doing? Are there
any behavior changes expected?). Moreover, data streams
from sensors need to be tracked, mapped to information en-
tities, and simulated. Additionally, the output (goal) must
be known (e.g., save time, save costs, improve health) and
its derivation as well as the reconciliation of private goals
must be mapped with organizational goals, which in turn is a
challenge of the IoT. An alignment between event-based and
process-oriented systems indispensable in this context
starting point could be to deﬁne goal-based deviation patterns
and to provide modeling techniques considering sensor-data
and event data.
C 9 – Specifying the autonomy level of IoT things.
jects in the IoT are able to react to events by executing tasks or
entire processes. The execution of the latter is typically asyn-
chronous and sometimes not explicitly started from a central
coordinator. The execution of tasks or processes may further
trigger certain reactions, for example the start of another pro-
cess to correct deviating behavior. Yet, it is unfeasible to grant
things full autonomy to decide everything without supervi-
sion. Hence, there has to be a concept of autonomy levels that
dictate if things need supervision and may be vetoed, be it
an individual or a group. Currently, there is no universal way
to talk about these levels of autonomy or to resolve conﬂicts
originating from this distinction [
]. While different con-
ceptualizations of individual and group autonomy exist, they
have not been transferred to BPM yet. Moreover, there is a
lack of understanding on how to express them in a business
Alignments between both system types were discussed at a Dagstuhl
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 6/9
process model, e.g. using patterns or further attributes.
C 10 – Specifying the “social” roles of agents.
tions aim to optimize their business processes based on orga-
nizational (i.e., group) goals. However, process participants
often follow personal, i.e., individual processes or agendas
with individual goals. The challenge is to synchronize/ recon-
cile different, possibly conﬂicting goals. These agendas are
typically mitigated through governance processes prescribing
desired behavior. The individual goals of a thing are typically
not precisely known or explicitly given. Furthermore, these
processes may be less prescriptive micro processes or habits.
Hence, holistic and prescritive governance may not be possi-
ble. Hence, it is an option to deﬁne and specify social behavior
of things (such as self-interest, helpful, cooperative [
better coordinate and govern their behavior. This becomes
even more challenging, when also considering robotics, i.e.,
the integration of human actors as well as robots in processes
(raising issues like exchangeability, co-existence of different
kinds of resources etc.).
C 11 – Concretizing abstract process models.
process models are sometimes used to model processes at
design time without providing the details necessary for exe-
cution. This is a sensible approach when dealing with very
dynamic scenarios. In these cases, it is possible to deﬁne the
process but the abstract model has to be turned into a concrete
model later before being executable, for example by discover-
ing available services as well as the conditions in which these
services may be used. Context also includes physical data
about users, e.g., location, devices the user carries with him
(e.g., smartphone), etc. For the discovery phase (see Figure 1),
the semantics related to the services (i.e., what functionality
can the service offer specially within the context of the pro-
cess) should be available and it should be possible to reason
over this for matchmaking purposes. In addition, the services’
discovery phase may lead to changes in the schema of the
original abstract process. Examples of corresponding changes
include the skipping of certain tasks initially planned in the
process or the addition of new fragments (e.g., combining two
or more services either in sequence or parallel to achieve the
task goal). Despite all ﬂexibility, this phase of instantiation
presumes a given structure in form of the abstract process
C 12 – Dealing with new situations
Individual ad hoc de-
cisions may resolve a current situation from an individual’s
or a small group’s point of view towards favorable results
for them. In a complex business environment, foresightful
and structured decision making cannot only achieve similar
results but also save costs and time, and possibly improve the
total quality. Deterministic event detection and correlation
can be very well modeled and executed with event processing
languages in complex event processing engines. However, the
ﬂexible discovery of new situations and the derivation of new
responses constitute major technological challenges whose
tackling can beneﬁt from the combination with BPM.
BPM methodologies and technologies can support the
identiﬁcation and selection of appropriate responses by rec-
ommending tasks, triggering tasks or whole processes, and
automating as well as monitoring their execution. These re-
actions can be pre-deﬁned using existing BPM technologies
and learning can be based on the analysis of historic traces
to identify beneﬁcial habits from a higher level perspective.
Furthermore, reference models can help to identify state-of-
the-art industry blueprints, which can be contextualized and
instantiated to ﬁnd a proper reaction for the context and the
history of the situation. The capability of IoT sensing can be
of additional beneﬁt here.
C 13 – Bridging the gap between event-based and pro-
A challenge is to bridge the gap be-
tween clouds of sensor data and event logs for process mining.
Events captured by sensors are available in high volume, ve-
locity, and variety. They are often affected by noise and errors.
Process knowledge can be employed to support the identi-
ﬁcation of events from raw event data and in a subsequent
step entire processes including their activities from event data.
This is a non-trivial problem since event data belonging to
different activities can be interleaving. Moreover, event data
can belong to or be relevant for several activities, so that com-
plex n:m relations between events and activities have to be
considered. The question of mapping start and end of event
(streams) to the start and end of a process is closely related.
Once the activities have been discovered, the next challenge is
to discover the corresponding processes, i.e., to correlate the
activities with the corresponding process instances. Process
knowledge and BPM methodologies can support the discovery
of these habits, the identiﬁcation of their underlying interac-
tions as processes as well as the optimization of these habits to
reduce the waste of time and resources and increase the safety
of all involved agents. Process mining techniques provide
promising ex post perspectives in this respect, but require the
presence of an event log that organizes the events in terms of
traces representing the execution of a process instance [
Similarly, but in an on-line fashion, complex event process-
ing can be used to derive higher level knowledge from raw
events to provide an ex nunc perspective [
]. Here, the timely
provisioning of events is crucial.
C 14 – Improving online conformance checking.
tailed earlier, conformance checking is a process mining tech-
nique that compares an existing process model with an event
log of the same process. It can be used to check if the re-
ality of process execution, as recorded in the log, conforms
to the model and vice versa. Online conformance checking
takes as input the context data and performs the comparison
online. This requires high quality data and almost complete
information. Again, the IoT as a data source and data man-
agement technology can play a major role and might improve
the conformance checking of the actual physical execution
with the execution order as recorded by the BPMS based on
a secondary log of sensor data. Similarly, the checking and
monitoring of compliance rules to be obeyed during process
execution might beneﬁt from IoT data sources and data man-
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 7/9
C 15 – Improving resource utilization optimization.
can provide a governance structure for an organization, be it
physical or virtual. BPM initiatives break up traditional func-
tional silos and introduce process managers being responsible
for processes across departments. While complex systems
and the IoT is centered around situations to react to, BPM ini-
tiatives are organized around processes. This entails that some
coordination instance responsible for priorities and resource
provisioning can monitor and intervene with additional knowl-
edge if necessary. In a pure IoT paradigm, there is the danger
that decision will only produce local optima. Research has
shown, for example, that situational decisions about resource
provisioning as it is common in virtualized environments,
e.g., CPU usage, can be optimized from a process perspective
through BPM knowledge when executing known procedures
or behavior patterns [
]. Vice versa, research has also
shown that process knowledge can optimize event processing
]. The coordinating unit responsible for resource provi-
sioning has advanced knowledge about the future behavior
of agents since they have to follow their process models and,
thus, can provide resources (e.g., computing power, network
bandwidth, or things) with greater accuracy reducing pro-
cessing time and thus increasing the throughput of a process.
It also helps to reduce communication time-outs and thus,
rollbacks, or abnormal process terminations.
C 16 – Improving resource monitoring and quality of task
The execution of tasks in a business process con-
sumes resources. These can be IT, such as storage capacity for
process data, computing power for calculations in scientiﬁc
workﬂows, artiﬁcial agents, such as as robots automatically
executing manual tasks, or human beings entering or analyz-
ing data or performing manual tasks. Also machines, e.g.,
packing drugs, can be considered as resources (e.g., predictive
monitoring, i.e., when does the machine have to be maintained
taking its usage as well as historical data into account).
All these resources migth suffer from issues, which hinder
optimal working conditions such as over- or under-utilization
or even damage/ illness. IoT-based sensors can pick up these
issues by measuring machine-behavior or human stress lev-
] and suggest changes to process execution to alleviate
these effects. Furthermore, the IoT can support the execution
of (knowledge-intensive) tasks in a process through context-
speciﬁc knowledge provisioning, e.g., in terms of instructions
or training materials on how to execute the task, or regula-
tions that are relevant for the user’s particular context. Sensor
data can be leveraged to determine the actual context [
and to identify information needs (e.g., detection of cognitive
overload or stress).
4. Concluding remarks
The IoT provides many opportunities for industry as well as
for personal use through the meaningful, yet dynamic inter-
action of humans, software, machines, and things. BPM is
a well established discipline that deals with the discovery,
analysis, (re-)design, implementation, execution, monitoring,
controlling and evolution of business processes.
So far, both areas have been considered separately. In this
we have formulated a number of challenges for the
amalgamation of the IoT and BPM, which we deem important
to be tackled in the near future in order for the IoT to beneﬁt
from business processes and vice-versa.
Before concluding, we would like to highlight a cross-
issue, i.e., dealing with security and, in particular, privacy
issues. For example, privacy levels that exist at the sensors
level might be different with respect to those at the BPM
side. A full-disclosure approach should be avoided, espe-
cially in contexts where sensitive (i.e., personal) information
is collected. The most relevant challenge, in this case, is
the communication between the two worlds, each of them
with corresponding privacy/security levels and policies. The
layer in charge of integrating these two sides should be de-
signed according to the principles of privacy by design [
“identify and examine possible data protection problems when
designing new technology and to incorporate privacy protec-
tion into the overall design, instead of having to come up
with laborious and time-consuming “patches” later on” [
This issues can also be seen as a “non-functional requirement”
referring to C1, C3, C4, C6, C8, C13, and C14, but also other
challenges might be affected. Fibally, partially related to the
previous point, are ethical aspects of the integration of IoT
and BPM: the introduction of raw events paves the way to
a whole new set of analyses and explorations. On the one
hand, these analyses must preserve the privacy of the individ-
ual (privacy is recognized as a fundamental right
.). At the
same time, the analyses should not be unfair and should not
provide unequal treatment of people based on membership to
a category or a minority. This problem is typically referred to
as “discrimination-aware data mining” [
]. More generally,
the literature also talks about “privacy-preserving data mining”
]. There are several challenges that are directly affected
by that such as C2-C6 and C13-C15. This is due to the set
of analyses that the integration of IoT and BPM will make
This paper is a living document. All persons willing to provide feedbacks
and improvements are welcome to contact the authors.
Cf. Article 8 of “European Convention on Human Rights”
and Article 12 of the “Universal Declaration of Human Rights”
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 8/9
M. T. P. Adam, H. Gimpel, A. Maedche, and R. Riedl.
Design blueprint for stress-sensitive adaptive enterprise
systems. Business & Information Systems Engineering,
pages 1–15, 2016.
R. Agrawal and R. Srikant. Privacy-preserving data min-
ing. SIGMOD Rec., 29(2):439–450, May 2000.
K. Ashton. That ‘Internet of Things’ Thing. RFiD
C. Batini and M. Scannapieco. Data and Information
Quality - Dimensions, Principles and Techniques. Data-
Centric Systems and Applications. Springer, 2016.
T. Catarci, M. de Leoni, A. Marrella, M. Mecella, B. Sal-
vatore, G. Vetere, S. Dustdar, L. Juszczyk, A. Manzoor,
and H. L. Truong. Pervasive software environments for
supporting disaster responses. IEEE Internet Computing,
F. Cossu, A. Marrella, M. Mecella, A. Russo, S. Kimani,
G. Bertazzoni, A. Colabianchi, A. Corona, A. De Luise,
F. Grasso, and M. Suppa. Supporting doctors through
mobile multimodal interaction and process-aware exe-
cution of clinical guidelines. In 7th IEEE International
Conference on Service-Oriented Computing and Appli-
cations, SOCA 2014, Matsue, Japan, November 17-19,
2014, pages 183–190. IEEE Computer Society, 2014.
M. Dumas, M. La Rosa, J. Mendling, and H. A. Rei-
jers. Fundamentals of Business Process Management.
Springer Publishing Company, Incorporated, 2013.
S. Euting, C. Janiesch, R. Fischer, S. Tai, and I. Weber.
Scalable business process execution in the cloud. In 2nd
IEEE Conference on Cloud Engineering (IC2E), pages
175–184. IEEE, 2014.
D. Firmani, M. Mecella, M. Scannapieco, and C. Ba-
tini. On the meaningfulness of “big data quality”. Data
Science and Engineering, 1(1):6–20, 2016.
G. Grambow, R. Oberhauser, and M. Reichert. Context-
aware and process-centric knowledge provisioning: An
example from the software development domain. In
L. Razmerita, G. E. Phillips-Wren, and L. C. Jain, editors,
Innovations in Knowledge Management - The Impact of
Social Media, Semantic Web and Cloud Computing, vol-
ume 95 of Intelligent Systems Reference Library, pages
179–209. Springer, 2016.
J. Gubbia, R. Buyyab, S. Marusica, and M. Palaniswamia.
Internet of things (iot): A vision, architectural elements,
and future directions. Future Generation Computer Sys-
tems, 29(7):1645–1660, 2013.
oller, V. Tsiatsis, C. Mulligan, S. Avesand, and
D. Boyle. From Machine-to-Machine to the Internet
of Things - Introduction to a New Age of Intelligence.
Academic Press, 2014.
R. Hull, A. Koschmider, H. A. Reijers, and W. Wong.
Fresh Approaches to Business Process Modeling
(Dagstuhl Seminar 16191). Dagstuhl Reports, 6(5):1–
C. Janiesch and J. Diebold. Conceptual modeling of event
processing networks. In 24th European Conference on
Information Systems (ECIS), pages 1–15. AIS, 2016.
C. Janiesch, M. Matzner, and O. M
uller. Beyond process
monitoring: A proof-of-concept of event-driven business
activity management. Business Process Management
Journal, 18(4):625–643, 2012.
C. Janiesch, I. Weber, M. Menzel, and J. Kuhlenkamp.
Optimizing the performance of automated business pro-
cesses executed on virtualized infrastructure. In 47th
Hawaii International Conference on System Sciences
(HICSS), pages 3818–3826. IEEE, 2014.
S. Kalenka and N. Jennings. Socially responsible decision
making by autonomous agents. In 5th International Col-
loquium on Cognitive Science, pages 135–149. Springer,
A. Kokkonen and W. Bandara. Expertise in business pro-
cess management. In J. vom Brocke and M. Rosemann,
editors, Handbook on Business Process Management,
volume 2, pages 517–546. Springer, Berlin, 2010.
unzle and M. Reichert. Philharmonicﬂows: towards
a framework for object-aware process management. Jour-
nal of Software Maintenance, 23(4):205–244, 2011.
M. Langheinrich. Privacy by design - principles of
privacy-aware ubiquitous systems. In Proceedings of
the 3rd International Conference on Ubiquitous Com-
puting, UbiComp ’01, pages 273–291. Springer-Verlag,
M. Lincoln and A. Gal. Searching business process repos-
itories using operational similarity. In On the Move to
Meaningful Internet Systems: OTM 2011 - Confeder-
ated International Conferences: CoopIS, DOA-SVI, and
ODBASE 2011, Hersonissos, Crete, Greece, October 17-
21, 2011, Proceedings, Part I, pages 2–19, 2011.
A. Nordrum. Popular Internet of Things Fore-
cast of 50 Billion Devices by 2020 Is Outdated.
Spectrum Tech Talk, 18 August 2016.
R. Parasuraman, T. Sheridan, and C. Wickens. A model
for types and levels of human interaction with automation.
IEEE Transactions on Systems, Man and Cybernetics -
Part A: Systems and Humans, 30(3):286–297, 2000.
D. Pedreshi, S. Ruggieri, and F. Turini. Discrimination-
aware data mining. In Proceedings of the 14th ACM
SIGKDD International Conference on Knowledge Discov-
ery and Data Mining, KDD ’08, pages 560–568. ACM,
The Internet-of-Things Meets Business Process Management: Mutual Beneﬁts and Challenges — 9/9
R. Pryss, N. Mundbrod, D. Langer, and M. Reichert.
Supporting medical ward rounds through mobile task
and process management. Information Systems and e-
Business Management, 13(1):107–146, 2015.
R. Pryss and M. Reichert. Robust execution of mobile
activities in process-aware information systems. IJISMD,
M. Reichert and B. Weber. Enabling Flexibility in
Process-Aware Information Systems - Challenges, Meth-
ods, Technologies. Springer, 2012.
P. Schaar. Privacy by design. Identity in the Information
Society, 3(2):267–274, Aug 2010.
M. Schillo and K. Fischer. A taxonomy of autonomy in
multiagent organisation. In Agents and Computational
Autonomy. LNCS, volume 2969, pages 68–82. Springer,
New York, NY, 2004.
F. Stertz, J. Mangler, and S. Rinderle-Ma. NFC-Based
Task Enactment for Automatic Documentation of Treat-
ment Processes. In Enterprise, Business-Process and
Information Systems Modeling - 18th International Con-
ference, BPMDS 2017, 22nd International Conference,
EMMSAD 2017, Essen, Germany, June 12-13, 2017, Pro-
ceedings, pages 34–48, 2017.
J. Vaidya, Y. Zhu, and C. W. Clifton. Privacy Preserving
Data Mining, volume 19 of Advances in Information
Security. Springer, 2006.
W. M. P. van der Aalst. Process Mining: Discovery,
Conformance and Enhancement of Business Processes.
Springer Publishing Company, Incorporated, 1st edition,
M. Weidlich, H. Ziekow, A. Gal, J. Mendling, and
M. Weske. Optimizing event pattern matching using
business process models. IEEE Trans. Knowl. Data Eng.,