A Holistic View of the IoT Process from Sensors to the Business Value

Ateeq Khan, Matthias Pohl, Sascha Bosse, Stefan Willi Hart and Klaus Turowski

Magdeburg Research and Competence Cluster,

Faculty of Computer Science, University of Magdeburg, Magdeburg, Germany

Keywords:

IoT, Smart Systems, Cyber-physical Systems, IoT Process, Industry 4.0, Digital Manufacturing.

Abstract:

Internet of things (IoT) is the focus of research, and industries are investing heavily due to potential beneﬁts

of IoT in various ﬁelds. This paper provides a holistic view of different phases in IoT covering all phases

from sensor data collection to the generation of business value. In this paper, we propose to use the proven

Six Sigma methodology for IoT projects. We describe each phase using a structured approach. We discuss

the consequences of each phase while selecting the phase as an entry or starting point. We use predictive

maintenance as a use case to demonstrate the practicability of our IoT process. Using these insights, IoT

project managers can identify required activities and competencies to increase success probability. In the end,

we summarise the paper ﬁndings and highlight the future work.

1 INTRODUCTION

Internet of Things (IoT), digitalization, are the terms

which got popularity due to potential beneﬁts in va-

rious ﬁelds. Daily use products or things conside-

red useless so far from digitalization perspective are

becoming valuable and are used to improve services

and to generate new offerings in those areas. Some of

the examples are from manufacturing industry (digital

manufacturing or industry 4.0 terms are used in this

context), smart city, agriculture, and livestock areas.

IoT enables organisations and customers to make im-

proved decisions based on the data gathered directly

from the end devices or ﬁelds.

Due to the popularity of IoT, organisations need to

innovate faster and have to digitalise their processes

to take the beneﬁt of digitalisation (Lucas Jr et al.,

2013). Organisations are under a constant threat to be

left behind if they do not follow or adapt the required

changes or the trend. Hence it is their focus.

On the one hand, organisations are already using sen-

sors and IoT devices in traditional products (things,

machines, and equipment), to sense the environment

for the overall beneﬁts. On the other hand, such pro-

jects generate an enormous amount of data (big data)

and can be used to improve the existing services or to

offer entirely new services or business models.

However, management of such project is difﬁcult due

to higher complexity, heterogeneity, cost explosion,

blurred boundaries between the physical and virtual

world and inter-disciplinary within and outside or-

ganisations. The majority of the projects are not

successful (other projects have a higher risk of failure

(Lee and Lee, 2015)).

Existing approaches only touch part of the problem or

speciﬁc use cases and do not provide a holistic over-

view of different phases required for the IoT project.

So, it is necessary to have a roadmap for the IoT pro-

ject with clearly deﬁned phases and its descriptions.

In this paper, we address the issue of isolated parts,

and our approach is not limited to a single case in a

speciﬁc domain. We discuss why the existing data

mining and organisational change management stra-

tegies are not fully transferable or applicable to the

IoT projects.

We provide a complete overview of IoT process co-

vering from the business value perspective for the or-

ganisation to the sensors. We describe each phase

using a pre-deﬁned structure naming, e.g., challenges

and disciplines, and the analogy of our process pha-

ses with the six Sigma sub-methodology phases. To

show the practicality of IoT process, we show how

our contribution can be used in a case study scenario

namely predictive maintenance which combines IoT

and Industry 4.0.

392

Khan, A., Pohl, M., Bosse, S., Hart, S. and Turowski, K.

A Holistic View of the IoT Process from Sensors to the Business Value.

DOI: 10.5220/0006362503920399

In Proceedings of the 2nd International Conference on Internet of Things, Big Data and Security (IoTBDS 2017), pages 392-399

ISBN: 978-989-758-245-5

2 RELATED WORK AND

DISRUPTIVE ASPECTS OF IoT

To handle an IoT project, which extends within dif-

ferent departments in an organisation and also across

organisations, a methodology, process or framework

is required to manage and coordinate throughout the

IoT landscape. There are already different methodo-

logies, frameworks, and processes discussed in the

literature that are considered as a candidate for IoT

projects. Few to name are 5S (Osada, 1991), Kai-

zen (Imai, 1986), Six Sigma methodologies (Smith,

1993), Kotter model (Kotter, 1996), SEMMA (sam-

ple, explore, modify, model, and assess) (Azevedo

and Santos, 2008), CRISP-DM (cross-industry stan-

dard process for data mining) (Shearer, 2000), and

KDD processes (Fayyad et al., 1996; Lodhi et al.,

2008).

However, these methodologies have shortcomings,

for example, methodologies and frameworks like

CRISP-DM, KDD, SEMMA are more concerned

with the data mining process and do not include the

implication of business applications, business models

and scarcely describe the effect of one phase on the

other. Others, such as 5S, Kotter model, are very ab-

stract and rather suitable with the view to workplace

organisation or organisation change.

There are numerous works (Swan, 2012; Bonomi

et al., 2014) who address some parts of the IoT lands-

cape. Some of the works discuss only the IT per-

spective and others focus on the analytics perspective

and neglect other parts. There is a considerable

amount of literature which discuss IoT from the ap-

plications and future trends perspective (Curran and

Curran, 2014; Khan and Turowski, 2016b; Crowley

et al., 2014; Khan and Turowski, 2016a). In (Swan,

2012), the author reviews the eco-systems for IoT. It

covers only a part of the whole landscape. In (Bonomi

et al., 2014), authors propose a hierarchical distribu-

ted architecture and use a fog platform for analysis.

We argue that there are other factors which are inter-

related and have effects on other parts of the IoT pro-

ject or landscape. We describe these factors and pha-

ses and elaborate the whole process with the help of a

case study example.

3 AN ANALOGY OF THE IoT

PROCESS MODEL

For our IoT process, we advertise that we can use Six

Sigma methodology. Six Sigma methodology (Smith,

1993) was initially proposed to remove defects and

improve quality in 1986 but is now often used to me-

asure improvement in IT process execution and ser-

vices (George and George, 2003; Antony, 2006). Six

Sigma methodology is the best candidate to remove

defects, and we can use it for IoT project problems

described earlier. We propose to use Six Sigma sub-

methodology DMAIC because of its success in other

industries, and it has an analogy with the phases in our

process model. The sub-methodology DMAIC is an

acronym of the following steps, namely deﬁne, mea-

sure, analyse, improve, and control.

The brief description of these steps of DMAIC

methodology is as follows. In the deﬁne phase, we

deﬁne what is the problem; in the measure phase, we

ﬁnd the areas of a problem; in the analyis phase, we

analyse the problem; in the improve phase, we take

necessary steps to remove the problem; and in the

control phases we control or check whether the pro-

blem is removed.

Our IoT process model consists of the following pha-

ses: sensors, pre-processing & analysis, business ap-

plications, and business value. The detailed descrip-

tion of these phases is described in Section 4.

The methodology steps of DMAIC have an analogy

with the following IoT process phases.

Deﬁne: This phase has an analogy with the business

value phase. In this phase, we deﬁne the scope and

objective of the IoT project. General questions in the

phase consider the nature of the problem or the aims

we want to achieve with this project. As this phase is

on the strategical level, we also identify the possible

stakeholder of the project, their roles, resources, and

what are the general requirements.

Measure: The sensors phase from IoT is associated

with this phase of methodology. In this phase, data or

measurements (raw data) are collected from the sen-

sors. We decide what kind of sensors we need and

how frequently measurement should be done.

Analyse: This phase is inline with the analysis phase

in the landscape. In this phase, we perform the ana-

lyses based on the data collected in an earlier phase.

We get insights from this phase which tell about the

behaviour or exhibit speciﬁc patterns.

Improve and Control: This is associated with busi-

ness application phase in the landscape. This phase

describes the necessary steps to improve the overall

situation or how we can use the outputs of other pha-

ses to improve business applications or business pro-

cesses of an organisation are. In the control phase, we

take measure to monitor the overall progress or per-

formance of the cycle.

We depict our process model phases and associ-

ated DMAIC steps in Figure 1. The organisation’s

vision serve as an input for the IoT project, and after

A Holistic View of the IoT Process from Sensors to the Business Value

393

Define

Measure Analyse

Improve &

Control

Organisation

Vision

Value for

Organisation

Sensors

Analysis

Business

Applications

Business

Value

Figure 1: Analogy of IoT landscape with DMAIC process.

the successful run, the value achieved from the pro-

ject is considered as an output. Due to the nature of

IoT projects and the characteristics described earlier,

the whole process is iterative.

4 IoT PROCESS PHASES

In this section, we elaborate the different phases of

the IoT landscape. We use the following structure to

describe different phases of our IoT landscape. The

structure is quite simple, self-explanatory regarding

the names, and consists of following parts, objecti-

ves and scope, challenges and requirements, tools and

methods, disciplines, interfaces, and inﬂuences on ot-

her phases. As shown in Figure 1, it is an iterative pro-

cess, so objectives and operations can be re-aligned

iteratively to optimize what is needed or possible in

current settings. In the following, we describe the

phases of IoT process.

4.1 Sensors:

In this phase, we sense the environment and measure

the surroundings. These measurements of the sur-

rounding can be properties of technical devices, envi-

ronmental situations or even human activities which

are the main focus of monitoring. However, there are

other types of sensors that are also used in IoT en-

vironments to receive executable commands and take

actions.

Objectives & Scope: The main aim of this phase is

to collect data for analysis. Such data can be collected

from physical sensors in the environment, e.g., tempe-

rature, pressure, humidity in the environment or it can

be a virtual data, e.g., from computing resources, e.g.,

disk space, rotation, memory. Physical sensors, vir-

tual sensors, and control entities are either organized

in sensor networks with a communicating link nodes,

or they are able to send data on their own to data cen-

ter. The form of an IoT landscape of sensors depends

on a given use case and requirements.

Challenges and Requirements: The challenges and

requirements in this phase are to identify what kind of

data should be collected and how frequently it should

be gathered. This identiﬁcation depends on the sce-

narios and requirements, whether the data collected

from the sensors should be in real time, every se-

cond, or aggregated from the sensors. The design

of power-optimized micro-controllers or microcom-

puters in combination with sensors and network devi-

ces can follow with attention to IT security and safety

issues as much as legal considerations. All devices

must be part of a well-structured network to provide

the desired data for a data center or storage database.

Tools and Methods: Next to standard network con-

nections like Ethernet, Wi-Fi or GPRS there are a lot

of other network technologies coming up. ZigBee, Z-

Wave and LoRa are only some of them. Nevertheless,

RFID and NFC are suitable for short-distance soluti-

ons. In the course of optimizing the data transfer, ap-

plication layer protocols like MQTT, CoAP or XMPP

are used primarily although new network technolo-

gies bring their own protocol bundles. For the same

reason, these devices are becoming smaller. While

Arduinos and Raspberry Pis are popular for tinkering,

one can get completely operative devices by Z-Wave

or LoRa, so one is not forced to think about opera-

ting systems and applications scripting. In most ca-

ses, the linkage to a data center is not made directly.

All data streams that are sent by devices will be bun-

ched on linking platforms. Depending on the size of

a platform the data preprocessing or the data analysis

can be realised on these IoT platforms (on-premise or

cloud solutions).

Disciplines: The sensor layer is strongly characteri-

sed by the discipline of electrical engineering consi-

dering the micro controllers and sensors that are nee-

ded for IoT projects. But of no less importance is the

IoTBDS 2017 - 2nd International Conference on Internet of Things, Big Data and Security

394

Figure 2: An overview of IoT process.

experience in computer system engineering and com-

puter networking.

Interfaces: Business decisions can drive the con-

struction, the extension or the improvement of a sen-

sor network, so one sees the obvious link to the bu-

siness value layer. In some scenarios, the sensors

layer and the analytics layer are not clearly separable,

because technical components are overlapping (e.g.,

SmartHomes). However, this interface is deﬁnitely

visible, especially industrial, application systems.

Inﬂuences: A proper deﬁnition of required sensors,

sensor networks and data streams is necessary such

that one has to ensure the quality of the sensors, an

efﬁcient network and processable data. Otherwise,

it would have an effect on the interfaces or the out-

come. For instance, short error intervals of an ob-

served machine or technical component will not be

detected with a long sampling rate of a sensor. Ho-

wever, an extremely short sampling rate will produce

a lot of data and could lead to system crashes in data

preprocessing especially in a huge sensor network.

4.2 Preprocessing & Analysis

In this phase, we describe both the preprocessing and

analysis phase of the IoT project, because preproces-

sing is a necessary step for the analysis. In preproces-

sing, we prepare the data for anlysis purpose.

Objectives & Scope: The main objective of this

phase is to obtain a data model or prediction model

for knowledge generation or decision making by data

analysis. Before performing data analysis, data pre-

paration is necessary to increase the effectiveness and

efﬁciency of the analysis. The preparation of data

spread over data selection, where one identiﬁes avai-

lable and required data and provides it in a utilisable

way, and a data cleansing, where one handles mis-

sing data, noise and outliers and also verify data re-

liability. After data quality is checked one continues

with data transformation, that includes data transfor-

mation and data dimension reduction. The separation

between important and less important data also invol-

ves algorithms for clustering, classiﬁcation or com-

ponent analysis, so one sees intersections with data

analysis methods. The ﬁnal part of this process step

is data analysis. The prepared data will be the input of

the descriptive statistical analysis or the machine lear-

ning algorithm to obtain prediction models and pres-

criptive statements.

Challenges and Requirements: A deﬁnition of pro-

cessable data sets and output variables is not required,

but a postulation could simplify the data analysis. In

consequence of the assumption that data collection

is error-prone and 40% of collected data is “dirty”

(Fayyad et al., 1996), a solution is crucial, especially

in real-time IoT cases. Due to needed context infor-

mation, human interactions are sometimes necessary

but should be minimized as much as possible to avoid

additional error sources. The main challenge at all is

the choice of the best algorithm or method for spe-

ciﬁc problems. It is not just that one is interested to

gain knowledge out of data, one is encouraged to ﬁnd

a suitable trade-off between complexity and resource

consumption or between accuracy and comprehensi-

bility. That does not only refer to data analysis met-

hods, handling huge amounts of data also requires

suitable strategies that regard data management and

data processing, e.g., store raw or aggregated data, ef-

ﬁcient storage, and parallelization.

Tools and Methods: There is a considerable amount

of tools and methods, that could be used for analy-

tics. Therefore, we can differentiate between the tools

required for data processing and data analysis pur-

pose. On the one hand Apache Storm, Samza, SAP

Smart Data Streaming are used for data processing

and preprocessing, and on the other hand, Apache

Spark Mlib, R, SAP PAL are used for data analysis

purpose. The choice of certain tools depends on the

chosen system landscape. However, the methods are

mostly similar. Data integration, feature ﬁlters, wrap-

pers, clustering, pattern recognition, principal compo-

nent analysis, descriptive and predictive statistics are

rough terms that one will use in the analytics phase.

Disciplines: The mass of machine learning methods

A Holistic View of the IoT Process from Sensors to the Business Value

395

and statistical methods require expertise in mathema-

tical statistics, algorithmics, and data science. At the

latest, when one is compelled to adjust and require de-

velopment of methods for individual use cases. Ho-

wever, data management is fundamental to handle

data streams and storage.

Interfaces: This phase has interfaces on the adjacent

phases. Data streams are obviously the connection to

sensor networks and furthermore the processed and

analysed data will be provided for business applicati-

ons.

Inﬂuences: This phase has an inﬂuence on business

applications and business value. Although, a useful

outcome of data analysis is not guaranteed. A suitable

data preprocessing can affect the task of data analysis.

However, to ensure a reliable and valuable result, one

has to verify them with appropriate testing procedu-

res. Misleading statements as an input for business

applications are certainly undesired.

4.3 Business Applications

Information systems are used to support business.

An information system (IS) “consists of people and

machines that generates and/or uses information and

which are interconnected by communications. An in-

formation system in the narrow sense is an application

system or application software for performing opera-

tional tasks” (Gabriel, 2012). In general information

systems can be classiﬁed based on different criteria.

One possible classiﬁcation for IoT applications can

be made by means of the business intelligence con-

cept because Business Intelligence solutions play an

important role in the area of planning, analysis and

help in decision making. Gabriel deﬁnes the busi-

ness intelligence system as: “Applications based on

data warehousing, OLAP and data mining concepts,

as well as a modern reporting and portal system with

which a diverse and powerful IS can be used in vari-

ous application areas” (Gabriel, 2012).

Objectives & Scope: The objective of an IoT appli-

cation is to improve the decision quality, optimise the

business processes, and support the user in their daily

operations. This objective is achieved by utilising the

outcomes of the project and using them for analysis

and reporting.

Challenges and Requirements: IoT applications are

highly complex and interconnected as described in

(Vermesan, 2014). The single truth of information is

missing because of information silos, and correct de-

cisions can not be made on such data. Another chal-

lenge is an exponential growth of data in data silos

which lead to inconsistent and redundant data. The

key requirements in this phase are a high informative

value (usability of information, e.g., within a business

process), availability (important reports receive more

resources and available when required), and stability

(low failure rate). Another challenge in the data ware-

house architecture is to link data that historically grew

into separate data silos.

Tools and Methods: There are various tools and met-

hodologies available for this phase. From the integra-

tion of applications perspective, integration tools and

platforms are available, e.g., SAP Process Integration,

integration services to integrate the data. Change ma-

nagement and conﬁguration management methodolo-

gies can be used to manage the change and conﬁgure

the system.

Disciplines: In this phase, following disciplines from

business economics, and roles are determining factors

in the implementation and development of such IoT

applications. For example the domain expert provi-

des the technical input for the business application,

the business analyst covers the business perspective,

and the IT consultant provides the necessary link bet-

ween business and IT. The IT consultant also provides

systematically and formally presents the requirements

for the IT system so that they can be implemented in

design and implementation.

Interfaces: The information generated in the applica-

tion can serve then as input for other applications or

further analysis (Vermesan, 2014). Although an in-

terface or integration is required to access or provide

the results to other applications.Ideally, the output of

the application is a standardized format, which can be

used in other applications and systems.

Inﬂuences: The application has a direct inﬂuence on

the business value. A well-designed application can

clearly visualize the measured or analyzed data for

the end user. A well-designed application is an ap-

plication which has a good usability, availability or

actuality.

4.4 Business Value

The concept of business models and generating busi-

ness value is the focus of management in the research

literature (DaSilva and Trkman, 2014; Fielt, 2014; Pe-

trikina et al., 2014). Here we provide two deﬁnitions

of business models. A business model is deﬁned as “A

business model describes the rationale of how an or-

ganisation creates, delivers, and captures value” (Os-

terwalder and Pigneur, 2010). In (Teece, 2010), it is

deﬁned as “The essence of a business model is in de-

ﬁning the manner by which the enterprise delivers va-

lue to customers, entices customers to pay for value,

and converts those payments to proﬁt”. As in the case

of IoT projects, where customers or different orga-

IoTBDS 2017 - 2nd International Conference on Internet of Things, Big Data and Security

396

nisations are involved, business value can be deﬁned

from the project’s and participant’s perspective. This

outcome can be a new business model or improve the

existing business processes, offering new services to

the customers.

Objective and Scope: From the organisational per-

spective, it is the important phase of the IoT project.

In this phase, organisations deﬁne the objective of the

IoT project. The objectives of the project can gene-

rate new business models to be offer in a new market,

improving an existing process, understanding custo-

mer needs and requirements where the product is in

operation, solving existing problems or offering com-

pletely new business models. The scope of the project

not only deﬁnes items included or targets of the itera-

tion, but also the business value that a project brings

to the organisation as well as the problems the aims

of an organisation.

Challenges and Requirements: For the IoT project

in this phase, higher management commitment played

an important role and is mentioned as a challenge in

literature (Sundmaeker et al., 2010; Khan and Turow-

ski, 2016b). IoT projects involve other departments

of an organisation, without a clear directive from hig-

her management, departments or individuals may not

be interested in participating in the project. Simi-

larly, the decision maker’s vision is also important to

prepare the organisation for the future. Tools and

Methods: There are various tools and methodologies

available for business value. From a methodological

perspective, following methods are commonly used

e.g., business canvas, design thinking workshops, and

strategic information management techniques.

Disciplines: The involved disciplines are business

economics and project management. The overall un-

derstanding of business and project management ex-

perience is vital for a successful completion of a pro-

ject. From a role perspective, project managers and

an overall understanding of the whole process is a re-

quirement.

Interfaces: The outcome of the project can be used in

business applications, offering new services, or busi-

ness models to generate revenue directly from the cu-

stomers. In the case of business applications, overall

business processes are improved, e.g., in CRM, Sales.

New offerings are made to the customer to generate

the revenue for the organisation.

Inﬂuences: This phase inﬂuences different phases of

the process. Depending on the objectives, require-

ments in other ﬁeld are decided, e.g., what kind of

data we have to measure using the sensors, business

processes or applications which we can improve, or

the whole process.

Production

Test

Proof of

Concept

Feasibility

study

Figure 3: A cycle of an IoT project.

5 DISCUSSION

For an IoT project, an entry point can be any of the

shown ﬁve phases although, the considerations are

different for each entry point. Generally, require-

ments for different phases come from a business per-

spective. Then the business case is deﬁned, and requi-

rements for each phase are decomposed. We show an

example in our case study for such entry point. Alt-

hough, in the ideal case, it is suggested to start from

the business value phase. However, in practice, some

of the organisational settings, devices already recor-

ded data (historical) in different locations, e.g., pres-

sure, temperature. Organisations start from the exis-

ting data and then enhance the scenarios or combine

the existing data with other data. An organisation may

take sensors as a starting point for business value or

better business decisions.

Different perspectives also play an important role

for the selection of an entry point. In (Khan and Tu-

rowski, 2016a), the authors discuss two inside-out and

outside-in perspectives.

There are numerous challenges which cross cut

the phases in IoT landscape. One of the challenges is

the security. The security of sensors means that sen-

sors are generating and providing real data and it is

not breached or corrupted. Another aspect concerns

the ownership of the data. Regarding this, it is impor-

tant to know who is responsible- the organisation that

collects the data or the one that generates the data?

In Figure 3, we show the lifecycle of an IoT project.

The project usually starts with a feasibility study. Af-

terwards, proof of concept is made to show the practi-

cality of the project. After an initial test, the actual

project will be rolled out in the real production envi-

ronment.

A Holistic View of the IoT Process from Sensors to the Business Value

397

6 PREDICTIVE MAINTENANCE

We use predictive maintenance as a use case to show

the importance and implication of the IoT landscape

process. This example also shows how a change in

one phase will have an impact on the whole lands-

cape. Generally, requirements come from a business

strategy or business level. Then these requirements

are further analysed or classiﬁed in each phase accor-

dingly. All phases contribute in reaching the objective

of the project.

First of all, we describe the associated IoT landscape

process with this case.

For predictive or preventive monitoring, data is

collected from numerous kind of sensors, e.g., from

pressure, temperature, humidity, and motion sensors.

These data are then processed for further analysis. It

results in solution where the company can avoid fai-

lures and avoid cases where break down occurs in the

organisation. Using sensors in the machine or at loca-

tion helps to sense the situation early, and actions can

be performed rapidly or reported at a higher level in

real time. Such data is further analysed to diagnose

potential threats, and necessary actions can be taken

before equipment breakdown or inefﬁciency occurs.

Services processes can be constructed in such a way

that error-prone parts can be replaced before they ac-

tually break.

We also discuss a few challenges and future sce-

narios in this use case. In some systems, machine

condition data is stored in summarized form in row-

oriented database systems. Such summarization can

hide vital information regarding the machine state.

Analysing raw data over time or correlation of such

data with other attributes (like temperature, vibration,

quality of operation performed on material) will help

to ﬁnd new insight information on machine conditi-

ons in which they are operated and will increase the

reliability of a machine while predictive maintenance

can be performed. In future, a huge amount of data

can be stored without summarization and such data

can be analysed for future scenarios.

Predictive maintenance can lead to savings in

maintenance costs (up to 30%) and up to 75% fewer

failures, inducing an increase in productivity of up to

25% (Smith, 2008) (relevance proved by a study of

the US energy department). Now we discuss the vari-

ous entry points of our phases for this scenario.

Business value: Costs of machinery downtime can be

very severe to a business’ performance (e.g., loss of

reputation, costs for additional taff, replacement ma-

chines, additional stock capacities, opportunity costs)

. In order to optimize cost-effectiveness of e.g., pro-

duction processes, predictive maintenance initiative

may be introduced by the management.

Business Applications: A predictive maintenance

initiative may also be started from an application

point of view. An example could be the improvement

of supplier relations management in order to mini-

mize replacement time for equipment. In addition, the

success control for a predictive maintenance initiative

would be done on an application level (cost-beneﬁt

analysis). Thus, an identiﬁed need for further impro-

vements in existing predictive maintenance contexts

states another entry point. Also optimisation attempts

for existing processes may lead to a PM initiative.

Preprocessing and Analysis: An already performed

analysis can serve as an entry point, too. Besides

an existing, but maybe not very effective predictive

maintenance analysis (e.g., alarms could be too late,

root cause analysis does not lead to improvements),

also other analyses that utilize sensor data from equip-

ment can be used in the predictive maintenance con-

text.

Sensors: The equipment already in use may pro-

vide information about maintenance-relevant statis-

tics that are not used in a predictive maintenance con-

text. From this starting point, processing opportuni-

ties may be discussed. Another entry point could be

the availability of some sensors for an existing pre-

dictive maintenance initiative and the planned intro-

duction of new sensors to improve its quality.

7 SUMMARY AND OUTLOOK

This paper deals with the IoT process. We provide a

holistic view of different phases in IoT projects cove-

ring from sensor perspective of data collection to the

generation of business value for an organisation. We

emphasise to use proven Six Sigma methodology for

IoT projects. We discuss the consequences of each

phase while selecting the phase as an entry or starting

point. We use predictive maintenance as a use case

to demonstrate the practicability of our IoT process.

Using these insights, IoT project managers can iden-

tify required activities and competencies to increase

success probability. In the end, we summarise the pa-

per ﬁndings and highlight the future work.

REFERENCES

Antony, J. (2006). Six sigma for service processes. Business

Process Management Journal, 12(2):234–248.

Azevedo, A. I. R. L. and Santos, M. F. (2008). Kdd, semma

crisp-dm: a parallel overviee. IADS-DM.

IoTBDS 2017 - 2nd International Conference on Internet of Things, Big Data and Security

398

Bonomi, F., Milito, R., Natarajan, P., and Zhu, J. (2014).

Fog Computing: A Platform for Internet of Things and

Analytics, pages 169–186. Springer International Pu-

blishing, Cham.

Crowley, D. N., Curry, E., and Breslin, J. G. (2014). Lever-

aging Social Media and IoT to Bootstrap Smart Envi-

ronments, pages 379–399. Springer International Pu-

blishing, Cham.

Curran, K. and Curran, N. (2014). Social Networking Ana-

lysis, pages 367–378. Springer International Publis-

hing, Cham.

DaSilva, C. M. and Trkman, P. (2014). Business model:

What it is and what it is not. Long Range Planning,

47(6):379–389.

Fayyad, U., Piatetsky-Shapiro, G., and Smyth, P. (1996).

From data mining to knowledge discovery in databa-

ses. AI magazine, 17(3):37.

Fielt, E. (2014). Conceptualising business models: Deﬁni-

tions, frameworks and classiﬁcations. Journal of Bu-

siness Models, 1(1):85–105.

Gabriel, R. (2012). Informationssy-

stem. http://www.enzyklopaedie-der-

wirtschaftsinformatik.de/lexikon/uebergreifendes/

Kontext-und-Grundlagen/Informationssystem. En-

zyklop

adie der Wirtschaftsinformatik.

George, M. L. and George, M. (2003). Lean six sigma for

service. McGraw-Hill New York, NY.

Imai, M. (1986). Kaizen: The Key to Japan’s Competi-

tive Success. Random House Business Division, New

York.

Khan, A. and Turowski, K. (2016a). A Perspective on In-

dustry 4.0: From Challenges to Opportunities in Pro-

duction Systems. In Proceedings of the IOTBD, pages

441–448, Rome, Italy. Science and Technology Publi-

cations, Lda, Scitepress.

Khan, A. and Turowski, K. (2016b). A Survey of Current

Challenges in Manufacturing Industry and Prepara-

tion for Industry 4.0. In Abraham, A., Kovalev, S.,

Tarassov, V., and Sn

sel, V., editors, Proceedings of

the First International Scientiﬁc Conference (IITI’16):

Volume 1, volume 450 of Advances in Intelligent Sys-

tems and Computing, pages 15–26. Springer Interna-

tional Publishing.

Kotter, J. P. (1996). Leading Change. Professional Deve-

lopment Collection. Harvard Business School Press.

Lee, I. and Lee, K. (2015). The internet of things (iot):

Applications, investments, and challenges for enter-

prises. Business Horizons, 58(4):431–440.

Lodhi, A., Kassem, G., and Rautenstrauch, C. (2008). Ap-

plication of process mining in knowledge manage-

ment. In In Proceedings of The 3rd International Con-

ference on Interactive Mobile and Computer Aided

Learning (IMCL 2008), number ISBN: 978-3-89958-

351-9.

Lucas Jr, H. C., Agarwal, R., Clemons, E. K., El Sawy,

O. A., and Weber, B. W. (2013). Impactful research

on transformational information technology: An op-

portunity to inform new audiences. Mis Quarterly,

37(2):371–382.

Osada, T. (1991). The 5S’s: Five Keys to a Total Quality

Environment. Asian Productivity Organization.

Osterwalder, A. and Pigneur, Y. (2010). Business model ge-

neration: a handbook for visionaries, game changers,

and challengers. John Wiley & Sons.

Petrikina, J., Drews, P., Schirmer, I., and Zimmermann,

K. (2014). Integrating business models and enter-

prise architecture. In Enterprise Distributed Object

Computing Conference Workshops and Demonstrati-

ons (EDOCW), 2014 IEEE 18th International, pages

47–56. IEEE.

Shearer, C. (2000). The CRISP-DM Model: The new blu-

eprint for data mining. Journal of Data Warehousing,

5(4):13–22.

Smith, A. (2008). Next-Gen EAM Architectures In-

corporate Real-Time Asset Performance Data.

https://www.gartner.com/doc/1342596/nextgen-eam-

architectures-incorporate-realtime. AMR Research.

Smith, B. (1993). Six-sigma design (quality control). IEEE

spectrum, 30(9):43–47.

Sundmaeker, H., Guillemin, P., Friess, P., and Woelfﬂ

e, S.

(2010). Vision and challenges for realising the internet

of things. Cluster of European Research Projects on

the Internet of Things, European Commision.

Swan, M. (2012). Sensor mania! the internet of things, wea-

rable computing, objective metrics, and the quantiﬁed

self 2.0. Journal of Sensor and Actuator Networks,

1(3):217.

Teece, D. J. (2010). Business models, business strategy and

innovation. Long range planning, 43(2):172–194.

Vermesan, O. (2014). Internet of Things Applications -

From Research and Innovation to Market Deploy-

ment. River Publishers Series in Communications. Ri-

ver Publishers, Aalborg.

A Holistic View of the IoT Process from Sensors to the Business Value

399