Conference PaperPDF Available

Taking Advantage of Data Generated by Products: Trends, Opportunities and Challenges

Authors:

Abstract

Now that all kinds of products are increasingly getting connected to the Internet, it is expected that it will become easier to collect data on how they are actually used during the middle-of-life stage of their product lifecycles. At the same time, a growing number of data analytics technologies offers opportunities to transform this data into actionable knowledge. Over the years, such knowledge extracted from usage data has already become a reliable input for managing maintenance and related services, but other uses such as feedback to design – where product data management systems have started to offer support for data collection practices – and providing advice to end users are now also being considered. Most data from sensors and other product-embedded information devices are collected in batches and analyzed retrospectively. In order for companies to further benefit from data collection in terms of efficacy and acceptance in society, two key challenges are (i) finding ways to effectively use data analytics techniques – which currently do not seem to be used to their full potential, and (ii) finding a good trade-off between respecting privacy and yet producing useful knowledge.
Preprint version of 2016 ASME-CIE paper 1
Taking Advantage Of Data Generated By Products: Trends, Opportunities And Chal-
lenges
Wilhelm Frederik van der Vegte
DELFT UNIVERSITY OF TECHNOLOGY DELFT, THE NETHERLANDS
ABSTRACT
Now that all kinds of products are increasingly getting connected to the Internet, it is expected that it will be-
come easier to collect data on how they are actually used during the middle-of-life stage of their product
lifecycles. At the same time, a growing number of data analytics technologies offers opportunities to trans-
form this data into actionable knowledge. Over the years, such knowledge extracted from usage data has
already become a reliable input for managing maintenance and related services, but other uses such as
feedback to designwhere product data management systems have started to offer support for data collec-
tion practices and providing advice to end users are now also being considered. Most data from sensors
and other product-embedded information devices are collected in batches and analyzed retrospectively. In
order for companies to further benefit from data collection in terms of efficacy and acceptance in society, two
key challenges are (i) finding ways to effectively use data analytics techniqueswhich currently do not seem
to be used to their full potential, and (ii) finding a good trade-off between respecting privacy and yet produc-
ing useful knowledge.
INTRODUCTION
Analysis of how different customers use products can provide valuable insights for companies which depend
on revenues generated directly or indirectly from those products. Among these companies are manufactur-
ers, resellers and third parties such as maintenance providers and insurance companies. In addition, non-
commercial parties such as law enforcement authorities and NGOs may have interest in how particular prod-
ucts are being used.
The collection of usage data to obtain particular insights has been common for a long time in the exploitation
of websites and software, as well as hardware such as computers, smartphones and digital cameras [e.g.,
1,2,3]. Now that the Internet is evolving “from a network of interconnected computers to a network of inter-
connected objects” [4], also referred to as the Internet of Things (IoT), more and more categories of products
offer opportunities for collecting data about how they are being used. This trend is extending to product cate-
gories that are deployed to achieve mostly physical effects, which did not traditionally produce any pro-
cessable data. It is facilitated by the fact that product functionality is increasingly realized with the help of in-
formation-producing and networked solution elements such as embedded software, sensorswhich convert
measurements from the physical world to dataand actuators, which convert data to changes in the physi-
cal world. Companies can track the movements of these products and monitor interactions with them, which
inspires new business models taking advantage of these behavioral data [5]. As we will see in this survey,
the changes in business models can take various forms. Knowing how customers actually use the products
is said to enhance a company's ability to segment customers, customize products, set prices to better cap-
ture value, and extend them with value-added services [6].
OBJECTIVE, SCOPE AND METHOD
In this paper we have focused on how product-producing companies can extract and exploit knowledge from
product usage to improve their products and product-related services. It is not a paper about the Internet of
Things, as it has been defined as “the networked interconnection of everyday objects, which are often
equipped with ubiquitous intelligence” [7] or “the pervasive presence around us of a variety of things or ob-
jects which (...) are able to interact with each other and cooperate with their neighbors to reach common
goals” [8]. The IoT according to these definitions adds value through networked cooperation between prod-
ucts of various kinds, in applications such as domotics control in smart homes [9]. Instead, our focus has
been on data and knowledge collected from multiple instances of physical, tangible products of the same
kind. The IoT can be seen as a possible enabler of data collection, but we have also considered more tradi-
tional, centralized communication schemes in which each instance of the product unidirectionally sends data
Preprint version of 2016 ASME-CIE paper 2
to the company. Our focus on data to be used by the product-producing company also implies that the sur-
vey does not cover collection of data only to be used within the product itself and its local context, as is the
case with applications of control engineering and local diagnostics support e.g., distributed control of air-
craft engines [10], and user access to a vehicle’s diagnostic information [11]. Furthermore, our focus on how
the product in question is being used, implies that we have not considered data collected by the product
about (behavior of) other entities that it is monitoring, like fitness trackers, smart alarm systems, smart ener-
gy meters, etc., do [12].
This survey aims to (i) accumulate what has been done in the field of gathering and utilizing data from prod-
ucts by giving an overview of the applied technologies and approaches as well as the achieved results, and
(ii) identify unresolved (research) challenges and unexploited opportunities.
In reviewing the current state of the art, we have started from the following questions:
What kinds of products have been considered in the reviewed sources?
What have been the motivations to collect and analyze the data?
What are the technologies, infrastructures and platforms that have been considered to generate,
transmit, collect and interpret the data?
In the light of the increasing connectedness of products, we were especially interested in data collection by
products that traditionally would not be expected to generate any data about their usage, i.e. products that
are not essentially computers, but products with main functions other than receiving/collecting, processing
and providing information. For those products that are essentially computers, the so-called information-
centric productssuch as smartphones and tablets the focus has been on data other than the data han-
dled by the products information-processing processes. For instance, for a laptop we would be interested in
how users handle it mechanically rather than in how often the laptop connects to wireless networks. This
special interest was motivated by the fact that for information-handling functions the collectable data about
usage comes more or less for granted, while it might be more interesting to learn about the additional efforts
needed if that is not the case.
As a basis of the survey, scientific publications, commercial materials as well as technology news reports
collected from the Web were used. The initial, central search has been for publications and websites where
the words collect, product, usage/use, and data, with or without sensors appear together. Subsequently, oth-
er sources to which the results referred were consulted, and potentially meaningful terminology often men-
tioned in the results was also used as search terms.
The structure of the remainder of this paper is as follows: in the next section, the various products that have
been provided with usage data gathering capabilities are characterized. Secondly, the various motivations for
collecting the data are discussed: what did companies and other stakeholders aim to achieve with it? Third,
the technologies (platforms, hardware, networking, analysis techniques, etc.) are discussed. These next
three sections aim to reflect the trends in collecting and utilizing data from products. After that, issues and
challenges are identified, and the paper ends with a discussion and conclusions.
CHARACTERIZATION OF PRODUCTS THAT HAVE BEEN DEPLOYED TO COLLECT DATA ABOUT
THEIR USE
Many reports on products collecting data during middle-of-life (MOL, also known as mid-life or the aftermar-
ket, i.e. the stage of the lifecycle where a product is used) focus on a particular category of products or even
a specific product. From our investigation, it appears that the practice of equipping products with data collec-
tion capabilities and utilizing the data is currently more widespread among products of a certain (high) com-
plexity. These are products that buyers generally consider investment goods. Quite a number of sources [13-
24]1 report on data collection from automobiles. Other sources predominantly report on B2B applications in
aircraft [5,18,25], military equipment [26], industrial equipment [27-31], and infrastructure such as bridges
[32,33], street lights [34] or elevators [35]. The fact that data collection has become so widespread in cars
1 The individual contributions from the referred works will be discussed in the next subsections.
Preprint version of 2016 ASME-CIE paper 3
may be due to the fact that automobiles are not only investment goods, but contrary to the other products
mentioned above, also mass products.
One example where MOL data collection has made it to less capital-intensive products is Hewlett Packard’s
Instant Ink program for inkjet printers [36]. The few other examples of data collection from less capital-
intensive goods concern studies, where researchers have collected data from one product or a small number
of products to investigate aspects of possible future data collection and utilization at a larger scale for in-
stance fridge-freezer combinations [37], notebook computers [38], and furniture [39]. Furthermore, in 2015,
Miele completed a proof-of-concept study with data collection from connected kitchen equipment [40].
In the remainder of the paper, application examples will be discussed in this same order: (i) automotive, (ii)
aerospace and defense, (iii) industrial equipment and infrastructure, and (iv) non-capital-intensive products
including consumer goods. For each category, first, examples will be discussed where data collection actual-
ly has become practice, followed by smaller-scale data collection experiments and novel approaches pro-
posed by researchers.
MOTIVATIONS FOR COLLECTING AND PROCESSING DATA
A company that decides to collect data from its products or from other companies’ products does so with
a particular intention. This intention, motivation or rationale is the driver behind some form of exploitation of
the processed data. It can be anything from optimizing business processes to achieving the “changes in
business models” mentioned in the introduction. By far the largest body of literature concerns managing
maintenance of products out in the field. After the state of art in that area, other reasons why data collection
has been considered or implemented will be discussed.
Maintenance management of fielded products
According to the broad definition offered by the European Federation of National Maintenance Societies,
‘maintenance’, is the combination of all technical, administrative and managerial actions during the lifecycle
of an item intended to retain or restore it to a state in which it can perform its required function [41]. Going by
this definition, there are several differently named but similar approaches aiming to exploit data collection for
support, streamlining, or optimizing maintenance of products to reduce downtime, avoid unnecessary
maintenance activities, increase customer satisfaction and extending the use phase of the product lifecycle.
In this section, the following approaches have been grouped together: condition-based, predictive, proactive
and preventive maintenance, prognostics & health management (PHM) and through-life engineering services
(TES).
Condition-based maintenance is an established and accepted maintenance practice. It aims to derive
maintenance requirements from real-time assessment of the product2 condition obtained from embedded
sensors and/or external tests and measurements. It relies on built-in diagnostic equipment or portable diag-
nostic equipment, such as PDAs and tablets [26]. The goal of condition-based maintenance is to perform
maintenance based only upon the evidence of a need rather than any predetermined time cycle, equipment
activity count, or other engineered basis.
Proactive maintenance is an approach that uses integrated, investigative and corrective practices to signifi-
cantly extend machinery life with the goal to eliminate failures of equipment forever [31].
PHM aims to monitor life-cycle environmental and usage conditions of products or systems to assess on-
going health, provide advance warning of failure through detection of failure precursors, and provide infor-
mation to improve the design and qualification of fielded and future products [42].
TES has been defined as “a result of the application of explicit and tacit ‘service knowledge’ supported by the
use of monitoring, diagnostic, prognostic technologies and decision support systems whilst the product is in
use, and maintenance (…) functions to mitigate degradation, restore ‘as design’ functionality, maximize
product availability, thus reducing whole-life operation cost” [43]. This is achieved based on five sources of
knowledge, namely knowledge of (i) degradation and failure mechanisms, (ii) means of repair, (iii) diagnos-
tics and prognostics, (iv) use, and (v) design and function.
The approaches to manage maintenance described above are often considered to underlie so-called perfor-
mance models, which represent the transition from selling products to selling performance. They are based
on the rationale that there is no inherent benefit for the customer to actually own the product [44].
2The original publication [26] specifically uses “weapon system” where “product” is used in this survey.
Preprint version of 2016 ASME-CIE paper 4
In the automotive industry, health monitoring and fault tracing based on diagnostics data collected from field-
ed cars forms an important part of service and maintenance [16]. It increases the service technicians’ ability
to diagnose and remedy problems in the increasingly complex electronically controlled vehicles and thus im-
proves customer satisfaction. The offline retrospective readouts are also uploaded to the manufacturer's da-
tabase to analyze fault occurrences collected from multiple cars, (i) to monitor the quality of components and
subsystems, (ii) to prioritize in which order problems should be addressed and (iii) to find correlations be-
tween different faults, or between faults and the operating environment. The recent trend of offering real-time
connectivity in vehicles is mainly motivated by customers’ demand for on-board internet and on-demand en-
tertainment [45] rather than by the need to collect data.
The automobile industry has introduced data collection platforms offering support for maintenance manage-
ment. For instance, GM’s OnStar emails diagnostics reports to the dealer to facilitate scheduling of service
appointments [46]. A more futuristic proposition was proposed by Amor-Segan et al. [23]: their self-healing
vehicle concept collects data from connected automobiles and is supposed to support in-vehicle autonomous
fault management. They claim that centralized collection of data based on wireless telematics can be used to
(i) facilitate more comprehensive data analysis and diagnosis at a remote support center, (ii) receiving diag-
nostics patches to aid in-vehicle diagnostics, (iii) update diagnostic and prognostic guidance and (iv) enable
new software versions for feature enhancements, correction of design and implementation errors. In addition,
Johanson et al [16] foresee support of inspection and repair during manufacturing of automobiles based on
collected data.
Performance models have been introduced in the aerospace industry, where manufacturers of jet engines
nowadays retain ownership of their products while charging airlines for the amount of thrust used [5,43]. As
an example of predictive maintenance in industrial capital goods, Marek et al describe how this has been put
into practice for mining equipment [28]. Maintenance dates are scheduled and optimized related to the actual
load on the machines. Before a maintenance job, the machine informs the crew about the tools and consum-
ables needed, thus reducing the level of service skills required. As tools and consumables can be pre-
organized and made available, hourly-based routine maintenance can be avoided and the time involved min-
imized, thus reducing downtime and improve availability.
Aspects of predictive maintenance can also be found in less capital-intensive products as Hewlett Pack-
ard’s Instant Ink program for inkjet printers shows. It enables connected inkjet printers to arrange replace-
ment cartridges for their end users before they run out [36]. Service contracts for office equipment such as
printers and computers are often based on a performance model [44].
Other uses
Other than for maintenance management, one of the uses of collected data that has been foreseen by the
literature is providing feedback to product design. This feedback is used, for instance, to reduce future prod-
uct failures and associated services required [6] to draft better requirements based on actual usage or to re-
define the functionality of a next product design iteration based on functions and features actually used [44].
Similarly, Van Horn et al [47] have suggested that data collected from deployed products enables manufac-
turers to quickly identify and efficiently solve quality issues in specific components. In addition, product usage
data can also be used to validate warranty claims and identify warranty agreement violations [6]. Further-
more, Främling et al. [48] have suggested to collect information from connected cars to (i) proactively opti-
mize engine tuning based on factors such as location and time of day, and (ii) present comparative perfor-
mance measures affecting behavior of drivers. Al-Taee et al. have suggested collecting data from connected
cars for a completely different purpose, namely, to allow the traffic control authority to record speed limit vio-
lations [21].
Data collection schemes that have been brought to practice or have been envisaged for concrete products
give evidence of some of the above and several other motivations behind data collection. The initial goal that
Ford envisaged with collecting usage information from customers’ automobiles in the 1990s was indeed to
gain understanding of how customers actually use their vehicles and to define appropriate specifications for
development and testing. This has been considered as a critical factor supporting design and development in
delivering affordable, high-reliability, high-quality products [20]. Hilpert et al [22] presented a system for real-
time collection of CO2 emissions from an entire company fleet of transportation vehicles to assess the carbon
Preprint version of 2016 ASME-CIE paper 5
footprint of the products they are transporting. Although, strictly spoken, the application is outside the scope
of this survey, it could theoretically be used to collect emission data from all fielded cars of a certain type,
and collect usage data based on which its manufacturer could possibly reduce emissions.
Two forms of third-party use of data collected from automobiles have been reported by Chui et al [5], namely
(i) insurance companies installing location sensors in customers’ cars so that they can base the price of poli-
cies on how a car is driven as well as where it travels, and (ii) rental car companies using tracking data to
optimize each car’s use.
In the aerospace industry, important objectives other than maintenance management for data collection
have been (i) improving crew decision-making and response in complex situations (ii) maintaining aircraft
safety between major inspections; and (iii) assuring safe and effective aircraft control under hazardous condi-
tions [18] .
Dienst et al. [29] propose a knowledge-based feedback system to assist product developers in exploiting da-
ta collected during the use of industrial goods, e.g., centrifugal pumps. From the given examples, the impres-
sion emerges that application of the system is limited to redesign based on component selection and param-
eter modification, e.g., selecting a better bearing to replace a bearing that the data analysis proves to fail too
often, or selecting a different material.
Coca-Cola collected data from vending machines that allowed customers to compose their own drinks, with
the objective to automatically schedule refills, but also for marketing purposes: the purchased mixtures pro-
vided indications of how new drinks are performing on the market over time, and of differences in regional
tastes [49]. Miele’s connected kitchen equipment has been developed with the initial goal to assist end users
by providing recipes on demand, but future plans include data collection for generating status report for ma-
chines or enabling remote diagnosis of problems [40].
For the EU-funded ELIMA project, data from 28 fridge-freezer combinations was collected to record events of
door opening and using the fast-freeze feature per user over time, with the goal to obtain an impression how
useful these data would be as input for (i) design improvements, (ii) offering improved logistics and (value-
added) services and (iii) possibility of reusing components from disposed products [37]. The preliminary find-
ings indicated that some potentially useful input could be collected for design and also for the contents of the
user manual.
Gu et al. [38] collected data about handling of notebook computers to (i) get an impression of variations in
use conditions between different users and in one user over a longer time span, and (ii) verify that the test
conditions in lab tests reasonably reflect actual use. Some of their tests involved hundreds of users over
hundreds of days. They were able to point out particular use conditions that were either more critical than
assumed or were not properly reflected in lab tests.
TECHNOLOGIES, INFRASTRUCTURES AND PLATFORMS TO SUPPORT UTILIZATION OF DATA
FROM PRODUCTS
All the surveyed approaches to taking advantage of data collected by products assume a processing chain
that starts with collecting or generating the data and ends with outputting the results of data processing for
utilization and storing it for possible later use. Our goal in surveying technologies, infrastructures and plat-
forms has been to get a general overview of:
how processing chains have been implemented:
To what extent are data stored and processed in the product?
Is it a continuous stream of data or is it a list of events?
Is the data transferred by wire or wirelessly?
Is this done continuously in real time or in batches?
what kind of analysis is performed:
How has the need for collecting data affected the product, i.e., to what extent does it require addi-
tional PEIDs (product-embedded information devices, i.e., sensors, transmitters and processors)?
In the automotive industry, ‘on-board diagnostics systemor ODB is the common umbrella term used for sys-
tems collecting MOL data [13]. The ODB in automobiles physically manifests in the form of the ODB-II con-
nector which is connected to the Controller Area Network (CAN) bus. The CAN bus is in turn responsible for
the communication between the electronic control units (ECUs) of the car [15]. Diagnostic trouble codes
Preprint version of 2016 ASME-CIE paper 6
(DTCs) from ECUs are routinely being read out during service from customer vehicles using a wired connec-
tion. DTCs are stored only when a reading is out of range. Readings produced at other times are generally
not recorded. This is a missed opportunity, because these could potentially be used to gain knowledge about
usage and vehicle behavior, for instance to predict faults. By establishing a real-time connection to the ODB
these off-line retrospective readouts can however be collected and sent to a manufacturer’s database for
further statistical analysis to find correlations between detected events [16].
Connecting cars to the Internet can be achieved indirectly through a smartphone [17,22], although today’s
connected cars usually have their own direct access to the mobile phone network [19]. The increasing de-
mand for bandwidth requires implementation of multiple radio interfaces, which may incur a high cost and
thereby impede further developments [45]. In the 1990s, Ford introduced CVDAS (customer vehicle data ac-
quisition system), the first platform to wirelessly connect vehicles [20]. CVDAS uses the same SAE J1850
protocol for the vehicle data communication backbone that was prescribed for OBD. Its wireless data com-
munication is based on mobile telephony standards. A recent development in that area is the ISO 13400
standard for Diagnostics over Internet Protocol (DoIP) [16]. To be able to collect the desired usage data in
CVDAS-equipped cars, the existing ECUs have been extended with additional sensors such as an ambient
temperature sensor and a rotary position sensor. To keep hardware requirements manageable, the data are
statistically analyzed inside the car and only the results are transmitted. A drawback of this approach is that,
in order to perform the right type of analysis, a-priori knowledge about system interaction effects is needed
[20]. In the system for real-time collection of CO2 emissions presented by Hilpert et al [22], OBD data were
combined with GPS, transmitted wirelessly through the mobile phone network, and collected and processed
by ERP systems.
The term OBD is also used in the aerospace industry [18], where the proactive maintenance schemes that
have been introduced by manufacturers rely on networked sensors built into airframes that send continuous
data on product wear and tear to the manufacturers’ computers [5].
In the knowledge-based feedback system that Dienst et al. [29] proposed for industrial equipment, the col-
lected data consists of (i) sensor data, which are collected automatically, and (ii) data that have been entered
manually by maintenance engineers and customers3. The system prepared the data so that these could be
handled by a product lifecycle management (PLM) system. This is needed because, according to the au-
thors, conventional PLM systems cannot deal with multiple individual instances of products, and therefore
the systems cannot store the collected data directly. Before further processing, the collected data require
additional human intervention: a knowledge engineer aggregates the data from numerous databases and
initiates Bayesian-networks based statistical analysis and visualization techniques. With the results, design-
ers can perform what-if studies with different usage conditions, and identify weak spots in the design to be
reconsidered. A decision support module guides towards the best solution from available alternatives.
The automated maintenance planning and diagnostic fault-finding for mining equipment that Marek et al. re-
ported on, uses on-board sensors. The machines’ on-board control system processes the incoming data and
compares these with the machine manufacturers’ database to ascertain whether the values are within the
defined parameters. If not, maintenance is scheduled automatically through a wired interface with the SAPTM
ERP (enterprise resource planning) system. In addition, the ERP integration facilitates automated ordering of
the consumables needed for maintenance, and assessment of the machine’s performance in the context of
the entire mine. The sensor data themselves are stored at the mining site in an SQL database to allow fur-
ther (unspecified) post-processing and visualization [28].
The fridge-freezer combinations in the ELIMA study reported in [37] were equipped with extra sensors to log
energy consumption, door openings, power on/off cycles and temperatures every second. Several other pa-
rameters could be read from the embedded software without the need for adding additional sensors. Data
logged by a built-in custom logger were transferred to the ELIMA database by a GSM module once every
three hours. At the end of the running time of the study, the collected data were visualized in histograms,
presumably by using a spreadsheet application.
The notebook computers in the experimental setup discussed in [38] were equipped with sensors capable of
measuring temperature, humidity and vibration. Part of the collected data were visualized in graphs without
3 These data that are not generated by the product and therefore outside the scope of this paper.
Preprint version of 2016 ASME-CIE paper 7
additional prior processing, to qualitatively assess characteristic patterns of variables over time and relations
between variables (e.g. between temperature and humidity inside the notebook), other part of the data was
statistically analyzed by means of ANOVA tests. Since the investigators performed analysis based on rec-
orded history of sensor data, real-time communication of data does not seem to have played a significant
role.
Wrapping up the inventory of technology that is used to realize the data-processing chain, we can state that
the first step, collection of data, typically takes place inside products, and is typically done by sensors. Em-
bedded software can also produce valuable data, thus reducing the need for additional sensors. The subse-
quent steps may take place anywhere between ‘inside the product’ and ‘at a central location’. If data pro-
cessing is done in-product, it is typically transmitted to and stored at a central location afterwards, i.e. the
product’s manufacturer or a service provider’s site. Some basic pre-processing of the bandwidth for data
transfer, e.g., when the average is considered instead of the individual values. Details about the subsequent
processing that is performed to produce actionable knowledge are not always given. Approaches that have
been mentioned are statistical techniques such as Bayesian networks analysis, ANOVA and visualization.
Some of the reported processes involve multiple steps and in some cases human interventions by, for in-
stance, knowledge engineers. In none of the discussed implementations, continuous data transfer and real-
time knowledge conversion seem to play a role.
For storage of the data and the findings, and making these accessible and manageable, PLM systems, ERP
systems and databases such as MySQL are used. The name ‘product lifecycle management’ suggests that
PLM includes management of MOL and tracking how products are actually being used. However, several
authors have indicated that conventional PLM systems are not adequately equipped for that purpose. For a
long time, these systems have focused on processes where digital systems such as CAx traditionally pro-
duce large amounts of data to be managed [48]. Conventional PLM systems are generally not equipped to
keep records of any dynamical process after the product has left the factory [50].
As a follow-up to this conclusion, it is interesting to note that software vendor PTC has recently announced
that the latest version of its WindchillTM PLM system was designed to to support collection of PEID data dur-
ing MOL [51]. This would facilitate exploitation of usage data in predictive maintenance and MOL-
information-based design, which has also been referred to as ‘closed-loop PLM’ [52].
ISSUES AND CHALLENGES
The sources that we consulted pointed out several issues and challenges related to collection of MOL data
from products. In addition to these sources, we reviewed several issues and challenges that were identified
in works related to the IoT [53-57] to check whether these would also apply to data collection within the
scope of our survey. The following issues and challenges were brought forward by two or more sources:
limitations of the current internet [e.g., 45,53];
privacy, trust and security [e.g., 4,37,44,53,54];
conversion of data to knowledge [e.g., 53,54,55,56];
achieving standardization and overcoming heterogeneity [e.g., 4,44,53];
energy efficiency [e.g., 56,57].
Below these issues will be addressed more specifically in the context of data collection and utilization during
MOL.
Limitations of the current internet
The current Internet architecture is limited in terms of mobility, availability, manageability and scalability [53].
This may give rise to problems if data are collected to provoke immediate action on critical events [45] or if
the quality and/or completeness is crucial for achieving the objectives of collecting and processing the data.
Privacy, trust and security
The data collected by products during MOL hold information or knowledge about product usage, and thus
also about the users. Social acceptance of data collection and utilization is expected to strongly depend on
the respect for privacy that is being observed, and the protection of personal data. [4].
Preprint version of 2016 ASME-CIE paper 8
The privacy concerns arising from the collection of usage data from tangible products are perhaps best illus-
trated by what is known from the car industry, which is obviously a prominent domain where data collection
has become common practice.
In 2015, researchers from the General German Automobile Club ADAC were commissioned by the Interna-
tional Automobile Federation FIA to investigate data collection by cars with wireless connection capabilities.
They examined two cars from one manufacturer one with combustion engine and one electrical car. The
goal was to uncover (i) what data these cars collect and make available to the manufacturer and/or the work-
shop, (ii) how long these are stored inside the car and (iii) on what occasions the data are transferred [14].
Since data collection and transfer is based on closed-source mechanisms devised by the manufacturer, the
investigators had to reverse-engineer ODB information and signals transferred by the built-in wireless com-
munication means. For the same reason, the manufacturer’s motivation behind collection of the data could
not be determined.
Of the dozens of information items that they identified to be stored and/or retrieved during workshop visits
and/or wirelessly transmitted, several were labeled potentially privacy-sensitive. Among these are preferred
seat positions, telephone contacts and numbers of drives covering particular distance ranges. In the electri-
cal car they even found that, each time the ignition is turned off and the car is locked, it transmits data such
as GPS location of the parking spot, previous charging stations, recent destinations entered in the navigation
system as well as at least 25 other items. Based on these and similar results, the FIA has demanded new
legislation to ensure that car manufacturers (i) reveal what they collect, (ii) give customers access to the col-
lected data and (iii) offer an opt-out from data collection.
Findings from [37] suggest that, especially if it is used to improve service or recycling processes, most con-
sumers (~70%) would accept recording of technical data, provided that not too much of actual usage is re-
vealed. Furthermore, they appeared to accept data collection at end-of-life more easily than continuous col-
lection over the Internet. Besides, it has been suggested that the IoT and other recent ICT developments are
affecting the way privacy is understood, particularly among younger generations [4]. In that context, future
users might be more willing to accept collection of data by products.
There are strategies that can be applied to reduce the privacy sensitivity of transmitted datafor instance,
limiting the data transmission or reducing the quality or fidelity of the transmitted data. However, there is a
trade-off in applying these: it is considered unavoidable that these approaches compromise the quality of the
extracted knowledge and thus the user’s trust in it [58,59].
Besides privacy, security of information is considered a major concern when data are collected that can re-
veal insights on users [4,60]. Industrial espionage can be a threat for business data [61], and hackers can be
a threat for both business data and potentially privacy-sensitive data of users [62]. Since this is a whole field
of research in itself, it will not be elaborated here; the reader is invited to refer to the many surveys on this
specific topic, for instance [63-66].
Converting data to knowledge
The whole point of collecting data is to transform these into actionable knowledge [55]. In the context of this
survey, ‘actionable’ means that knowledge satisfies the motivation behind the data gathering (e.g., service
management or design improvement). It is however somewhat disappointing that our sources hardly provide
details on how the data were analyzed. In most cases, sources state that ‘statistical analysis’ or even just
‘data analysis’ was performed. Only in a few cases, more specifics were given, such as Bayesian Networks
[29] and ANOVA [38].
Statistical analysis is just one of the more traditional forms of data analysis, and there is a large collection of
other techniques available, including various data mining and discovery techniques, prediction techniques
and simulation techniques using real-time acquired data [43,54,67]. Developing methods to select the best
out of many analysis techniques given the characteristics of the available data and the motivation that is to
be satisfied, still seems to be a challenge. A possible reason why the industry does not seem to explore po-
tentially more advanced analysis techniques is given in the next subsection.
Standardization and homogeneity
Most implementations of data collection and utilization have been developed in closed innovation processes
[44], which gives rise to the problem that components (including networks and software) from different com-
Preprint version of 2016 ASME-CIE paper 9
panies have to work together, yet cannot be integrated or run on a common operating system [68]. Managing
heterogeneous applications, environments and devices constitute a major challenge [53]. Consequently, in
practice, the collected data are mainly used for anomaly detection and control, but not for more sophisticated
forms of analysis such as optimization, prediction or discovery [67].
Energy efficiency
Collecting, processing and transferring data consumes electrical power. Especially, the power required by 3G
and Wi-Fi connectivity is relatively high. Problems may arise when a user is responsible for maintaining the
battery and other connectivity aspects of the product [59]. Energy supply is also an issue for products that
are traditionally not powered and need to be powered just for data collection, such as furniture [39].
DISCUSSION AND CONCLUSIONS
From the inventory, the impression emerges that, apart from products that are essentially computers, MOL
data collection and analysis has mainly been deployed in the context of capital-intensive goods, such as au-
tomobiles, airplanes, professional equipment and manufacturing equipment. With the exception of automo-
biles and a few other products where data collection has been studied in small-scale experimental setups,
these products are typically deployed in a B2B context. However, data collection from consumer products
seems to be on the rise, as most examples in that area appear to be recent. This trend can perhaps be ac-
counted to the fact that the contact between companies and consumers is more anonymous than between
companies and corporate customers in a B2B context. Collecting data about these previously anonymous
consumers would offer a good opportunity to get to know them better.
In the majority of the cases, the rationale behind collecting and processing usage data is to manage mainte-
nance activities. Other purposes to which data collection has been exploited most prominently include feed-
back to design, for future products, and providing advice to end users. Furthermore, analysis results were
used for diverse purposes such as marketing, tracking and classifying users and environmental impact as-
sessment.
Currently, most data are collected at intervals and analyzed retrospectively. Real-time monitoring does not
seem to be much needed, except for condition-based maintenance. In some cases, products such as
smartphones play an intermediary role in collecting the data.
For the manufacturer, the collected data can generally be characterized as a contribution to management of
the product lifecycle. Recently, vendors of product lifecycle management software appear to have recognized
the potential, and have started offering functionality to collect data from fielded products. Among the parties
taking advantage of the data are not only the manufacturers of the products, but also resellers and third par-
ties such as maintenance providers and insurance companies. In addition, non-commercial parties such as
law enforcement authorities have shown interest in how particular products are being used. It is not surpris-
ing that all this interest in usage data might cause privacy concerns among end users especially in cases
where they do not seem to benefit from it (e.g., validation of warranty claims). Offering the possibility to opt
out from data collection seems to be a good solution to this problem. Data security is a related issue; offering
solutions for secure data handling is, however, a discipline of its own.
For the actual analysis of the data, a wide range of techniques are available including solutions from machine
learning, statistics, pattern recognition, simulation and combinations thereof. However, in most cases of ac-
tual data collection, no further analysis tools than basic statistics are being applied. One of the biggest unre-
solved challenges is to match the characteristics of the available data to those analytics tools that best sup-
port the extraction of the sought-after knowledge. A first step towards achieving this would be to conduct fur-
ther research to characterize and classify (i) all types of data that can possibly collected from products on
how they are used, (ii) motivations of stakeholders for collecting the data in terms of possible analysis results
(i.e., the sought-after knowledge: answers to questions/queries about the data), and (iii) data analytics tech-
niques, their data requirements and their knowledge extraction capabilities. In addition, the development of
knowledge extraction approaches would benefit from standardization among the involved applications, envi-
ronments and devices.
Another important issue, especially when it comes to societal acceptance of data collection practices, is find-
ing a way to deal with the trade-offs that arise between respecting privacy of individual end users and striving
Preprint version of 2016 ASME-CIE paper 10
to get the highest-quality knowledge out of the collected data. This, however, is a problem that is also being
dealt with in related other fields in particular analysis of website statistics.
ACKNOWLEDGMENTS
Part of this research has been funded under the EC Horizon 2020 Programme, in the context of the FALCON
project (http://www.falcon-h2020.eu/). The author wishes to acknowledge the Commission and all the FAL-
CON project partners for the fruitful collaboration. In particular, he would like to thank Karl A. Hribernik of BI-
BA - Bremer Institut für Produktion und Logistik GmbH for his valuable suggestions and corrections.
REFERENCES
[1] Atterer, R., Wnuk, M., and Schmidt, A., (2006), "Knowing the user's every move: user activity tracking
for website usability evaluation and implicit interaction", Proceedings International Conference on
World Wide Web, ACM, pp. 203-212.
[2] Kanai, S., Higuchi, T., and Kikuta, Y., (2009), "3D digital prototyping and usability enhancement of
information appliances based on UsiXML", International Journal on Interactive Design and
Manufacturing (IJIDeM), Vol. 3 (3), pp. 201-222.
[3] Bollen, L., Giemza, A., and Hoppe, H.U., (2008), "Flexible analysis of user actions in heterogeneous
distributed learning environments", in: Times of convergence. Technologies across learning contexts,
Springer, pp. 62-73.
[4] Commission Of The European Communities, (2009), "Internet of Things - An action plan for Europe",
Brussels, 2009.
[5] Chui, M., Löffler, M., and Roberts, R., (2010), "The internet of things", McKinsey Quarterly, Vol. 2
(2010), pp. 1-9.
[6] Porter, M.E., and Heppelmann, J.E., (2014), "How smart, connected products are transforming
competition", Harvard Business Review, Vol. 92 (11), pp. 11-64.
[7] Xia, F., Yang, L.T., Wang, L., and Vinel, A., (2012), "Internet of things", International Journal of
Communication Systems, Vol. 25 (9), p. 1101.
[8] Atzori, L., Iera, A., and Morabito, G., (2010), "The internet of things: A survey", Computer networks,
Vol. 54 (15), pp. 2787-2805.
[9] Jimeno, R., Salvador, Z., Lafuente, A., Larrea, M., and Uribarren, A., (2004), "An architecture for the
personalized control of domotic resources", Proceedings Proceedings of the 2nd European Union
symposium on Ambient intelligence, ACM, pp. 51-54.
[10] Tulpule, B., Behbahani, A., and Millar, R., (2007), "Vision for next generation modular adaptive generic
integrated controls (MAGIC) for military/commercial turbine engines", American Institute of Aeronautics
and Astronautics, JPC.
[11] Yun, D.S., Lee, J.W., Lee, S.K., and Kwon, O.-C., (2011), "Development of Mobile Common
Component for providing vehicle information on mobile device", Proceedings Computer Sciences and
Convergence Information Technology (ICCIT), 2011 6th International Conference on, IEEE, pp. 809-
812.
[12] Koreshoff, T.L., Robertson, T., and Leong, T.W., (2013), "Internet of things: a review of literature and
products", Proceedings Australian Computer-Human Interaction Conference: Augmentation,
Application, Innovation, Collaboration, ACM, pp. 335-344.
[13] Society of Automotive Engineers, (1988), SAE On-board Diagnostics for Light and Medium Duty
Vehicles Standards Manual, Society of Automotive Engineers, Warrendale, PA.
[14] FIA, (2015), My car, my data - technical study, Fédération Internationale de l'Automobile, Region I,
Brussels.
[15] Johanson, M., (2011), Information and Communication Support for Automotive Testing and Validation,
INTECH Open Access Publisher.
[16] Johanson, M., Dahle, P., and Söderberg, A., (2011), "Remote Vehicle Diagnostics over the Internet
using the DoIP Protocol", Proceedings Systems and Networks Communications (ICSNC), pp. 226-231.
[17] Tahat, A., Said, A., Jaouni, F., and Qadamani, W., (2012), "Android-based universal vehicle diagnostic
and tracking system", Proceedings Consumer Electronics (ISCE), 2012 IEEE 16th International
Symposium on, IEEE, pp. 137-143.
Preprint version of 2016 ASME-CIE paper 11
[18] Hunter, G.W., Ross, R.W., Berger, D.E., Lekki, J.D., Mah, R.W., Perey, D.F., Schuet, S.R., Simon,
D.L., and Smith, S.W., (2013), A Concept of Operations for an Integrated Vehicle Health Assurance
System, NASA Center for Aerospace Information, Hanover, MD.
[19] Phalake, M., and Bhalerao, D., (2011), "Vehicle telematics system using GPRS", International Journal
of Computer Technology and Applications, Vol. 2 (1).
[20] Hung, S.T., (1998), "CVDAS-a data acquisition/retrieval architecture for statistical characterization of
vehicle usage", Proceedings AUTOTESTCON Systems Readiness Technology Conference, IEEE, pp.
239-246.
[21] Al-Taee, M., Khader, O.B., and Al-Saber, N., (2007), "Remote monitoring of vehicle diagnostics and
location using a smart box with Global Positioning System and General Packet Radio Service",
Proceedings ACS International Conference on Computer Systems and Applications ( AICCSA), IEEE,
pp. 385-388.
[22] Hilpert, H., Thoroe, L., and Schumann, M., (2011), "Real-time data collection for product carbon
footprints in transportation processes based on OBD2 and smartphones", Proceedings 44th Hawaii
International Conference on System Sciences (HICSS), IEEE, pp. 1-10.
[23] Amor-Segan, M., McMurran, R., Dhadyalla, G., and Jones, R., (2007), "Towards the Self Healing
Vehicle", Proceedings Institution of Engineering and Technology Conference on Automotive
Electronics, IET, pp. 1-7.
[24] Barabba, V., Huber, C., Cooke, F., Pudar, N., Smith, J., and Paich, M., (2002), "A multimethod
approach for creating new business models: The General Motors OnStar project", Interfaces, Vol. 32
(1), pp. 20-34.
[25] Benedettini, O., Baines, T., Lightfoot, H., and Greenough, R., (2009), "State-of-the-art in integrated
vehicle health management", Proceedings of the Institution of Mechanical Engineers, Part G: Journal
of Aerospace Engineering, Vol. 223 (2), pp. 157-170.
[26] Cutter, D.M., and Thompson, O.R., (2005), "Condition-Based Maintenance Plus Select Program
Survey", http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA430668, Accessed February 15, 2016.
[27] Koç, M., and Lee, J., (2002), "E-manufacturing-fundamentals, requirements and expected impacts",
Proceedings International Conference on Responsive Manufacturing (ICRM), Turkey.
[28] Marek, A., Thorley, S., and Haubmann, H., (2012), "The age of intelligence: online diagnostic and
automated maintenance planning", Proceedings International Platinum Conference, The Southern
African Institute of Mining and Metallurgy, Sun City, pp. 91-100.
[29] Dienst, S., Fathi, M., Abramovici, M., and Lindner, A., (2014), "Development of a knowledge-based
feedback assistance system of product use information for product improvement", International
Journal of Product Development, Vol. 19 (4), pp. 191-210.
[30] Ellen Macarthur Foundation, (2016), "Enit Systems: The Enit Agent", in: Intelligent Assets: Unlocking
the Circular Economy Potential. Appendix: Selected Case Studies, Ellen Macarthur Foundation,
Cowes, p. 15.
[31] Lee, J., (1995), "Machine performance monitoring and proactive maintenance in computer-integrated
manufacturing: review and perspective", International Journal of Computer Integrated Manufacturing,
Vol. 8 (5), pp. 370-380.
[32] Ellen Macarthur Foundation, (2016), "Arup: IoT in Construction and Infrastructure", in: Intelligent
Assets: Unlocking the Circular Economy Potential. Appendix: Selected Case Studies, Ellen Macarthur
Foundation, Cowes, p. 9.
[33] Ellen Macarthur Foundation, (2016), "Cisco: Europe's first SmartROAD", in: Intelligent Assets:
Unlocking the Circular Economy Potential. Appendix: Selected Case Studies, Ellen Macarthur
Foundation, Cowes, p. 17.
[34] Ellen Macarthur Foundation, (2016), "Philips City Touch: Public Lighting Services", in: Intelligent
Assets: Unlocking the Circular Economy Potential. Appendix: Selected Case Studies, Ellen Macarthur
Foundation, Cowes, p. 19.
[35] Ellen Macarthur Foundation, (2016), "Delta Development: Park 20|20 and Schiphol Trade Park", in:
Intelligent Assets: Unlocking the Circular Economy Potential. Appendix: Selected Case Studies, Ellen
Macarthur Foundation, Cowes, p. 14.
Preprint version of 2016 ASME-CIE paper 12
[36] Ellen Macarthur Foundation, (2016), "HP: Instant Ink", in: Intelligent Assets: Unlocking the Circular
Economy Potential. Appendix: Selected Case Studies, Ellen Macarthur Foundation, Cowes, p. 16.
[37] Bodenhoefer, K., Schneider, A., Cock, T., Brooks, A., Sands, G., Allman, L., Simon, M., Chong, S.,
Yang, X., and Delannoy, O., (2004), "Environmental life cycle information management and
acquisitionfirst experiences and results from field trials", Proceedings Electronics Goes Green, pp. 5-
8.
[38] Gu, J., Vichare, N.M., Tinsley, E.C., and Pecht, M.G., (2009), "Computer usage monitoring for design
and reliability tests", Components and Packaging Technologies, IEEE Transactions on, Vol. 32 (3), pp.
550-556.
[39] Bleda, A.L., Maestre, R., Santa, G., Jara, A.J., and Skarmeta, A.G., (2012), "Web of Things as a
product improvement tool: Furniture as case study", Proceedings Innovative Mobile and Internet
Services in Ubiquitous Computing (IMIS), , IEEE, pp. 846-851.
[40] Wilson, M., (2015), "Microsoft and Miele team up to cook up an IoT revolution" Betanews,
http://betanews.com/2015/04/12/microsoft-and-miele-team-up-to-cook-up-an-iot-revolution/, Accessed
February 11, 2016.
[41] European Federation of National Maintenance Societies vzw, What does EFNMS stand for?,
http://www.efnms.org/What-EFNMS-stands-for/m13l2/What-EFNMS-stands-for.html, Accessed
February 16, 2016.
[42] Vichare, N., Rodgers, P., Eveloy, V., and Pecht, M., (2007), "Environment and usage monitoring of
electronic products for health assessment and product design", Quality Technology & Quantitative
Management, Vol. 4 (2), pp. 235-250.
[43] Redding, L., (2015), "Through-Life Engineering Services: Definition and Scope: A Perspective from the
Literature", in: Through-life Engineering Services, Springer, pp. 13-28.
[44] Ellen Macarthur Foundation, (2016), Intelligent Assets: Unlocking the Circular Economy Potential,
Ellen Macarthur Foundation, Cowes.
[45] Lu, N., Cheng, N., Zhang, N., Shen, X., and Mark, J.W., (2014), "Connected vehicles: solutions and
challenges", Internet of Things Journal, IEEE, Vol. 1 (4), pp. 289-299.
[46] General Motors Company, OnStar website, https://www.onstar.com, Accessed February 15, 2016.
[47] Van Horn, D., Olewnik, A., and Lewis, K., (2012), "Design analytics: capturing, understanding, and
meeting Customer needs using big data", Proceedings International Design Engineering Technical
Conferences and Computers and Information in Engineering Conference (IDETC), ASME, pp. 863-
875.
[48] Främling, K., Holmström, J., Loukkola, J., Nyman, J., and Kaustell, A., (2013), "Sustainable PLM
through intelligent products", Engineering Applications of Artificial Intelligence, Vol. 26 (2), pp. 789-
799.
[49] Elliott, T., (2009), Drink Dispenser Analytics: Coca-Cola Goes Freestyle, With Help from SAP B,
http://timoelliott.com/blog/2009/06/drink-dispenser-analytics-coca-cola-goes-freestyle-with-help-from-
sap-bi.html, Accessed February 11, 2016.
[50] Heppelmann, J.E., (2015),"PTC Windchill 11 telecast", http://www.ptc.com/product-lifecycle-
management/windchill/11/telecast, Accessed December 28, 2015.
[51] Frost & Sullivan, (2015), "PTC - 2105 Global IoT PLM Technology leadership Award", New York,
http://www.ptc.com/~/media/Files/PDFs/PLM/Windchill/PTC-Inc-Award-Write-Up.ashx?la=en,
Accessed April 22, 2016.
[52] Kiritsis, D., (2011), "Closed-loop PLM for intelligent products in the era of the Internet of things",
Computer-Aided Design, Vol. 43 (5), pp. 479-501.
[53] Bandyopadhyay, D., and Sen, J., (2011), "Internet of things: Applications and challenges in technology
and standardization", Wireless Personal Communications, Vol. 58 (1), pp. 49-69.
[54] Chen, M., Mao, S., and Liu, Y., (2014), "Big data: A survey", Mobile Networks and Applications, Vol.
19 (2), pp. 171-209.
[55] Barnaghi, P., Sheth, A., and Henson, C., (2013), "From Data to Actionable Knowledge: Big Data
Challenges in the Web of Things [Guest Editors' Introduction]", Intelligent Systems, IEEE, Vol. 28 (6),
pp. 6-11.
Preprint version of 2016 ASME-CIE paper 13
[56] Chen, Y.-K., (2012), "Challenges and opportunities of internet of things", Proceedings Asia and South
PacificDesign Automation Conference (ASP-DAC), IEEE, pp. 383-388.
[57] Miorandi, D., Sicari, S., De Pellegrini, F., and Chlamtac, I., (2012), "Internet of things: Vision,
applications and research challenges", Ad Hoc Networks, Vol. 10 (7), pp. 1497-1516.
[58] Aggarwal, C.C., Ashish, N., and Sheth, A.P., (2013), "The Internet of Things: A Survey from the Data-
Centric Perspective", 2013.
[59] Aggarwal, C.C., (2008), "On unifying privacy and uncertain data models", Proceedings Data
Engineering, 2008. ICDE 2008. IEEE 24th International Conference on, IEEE, pp. 386-395.
[60] Hoffman, D.L., Novak, T.P., and Peralta, M., (1999), "Building consumer trust online", Communications
of the ACM, Vol. 42 (4), pp. 80-85.
[61] Podbregar, I., (2006), "Some patterns of industrial espionage", Journal of Criminal Justice and
Security, Vol. 8 (3,4), pp. 323-331.
[62] Matwyshyn, A.M., (2005), "Material Vulnerabilities: Data privacy, corporate information security, and
securities regulation", Berkeley Bus. LJ, Vol. 3, p. 129.
[63] Zhao, K., and Ge, L., (2013), "A survey on the internet of things security", Proceedings Computational
Intelligence and Security (CIS), 2013 9th International Conference on, IEEE, pp. 663-667.
[64] Botta, A., de Donato, W., Persico, V., and Pescapé, A., (2016), "Integration of cloud computing and
internet of things: a survey", Future Generation Computer Systems, Vol. 56, pp. 684-700.
[65] Xiaohui, X., (2013), "Study on security problems and key technologies of the internet of things",
Proceedings Computational and Information Sciences (ICCIS), 2013 Fifth International Conference on,
IEEE, pp. 407-410.
[66] Gang, G., Zeyong, L., and Jun, J., (2011), "Internet of things security analysis", Proceedings
International Conference on Internet Technology and Applications (iTAP), IEEE, pp. 1-4.
[67] Groopman, J., (2014), "Interoperability: the challenge facing the Internet of Things",
http://www.altimetergroup.com/2014/02/interoperability-the-challenge-facing-the-internet-of-things/,
Accessed February 16, 2016.
[68] IBM Institute for Business Value, (2015), "Device Democracy - Saving the future of the Internet of
Things", http://www-01.ibm.com/common/ssi/cgi-
bin/ssialias?infotype=PM&subtype=XB&htmlfid=GBE03620USEN, Accessed February 16, 2016.
... The timing of this work aligns with the recent emergence of the internet of things where information can now be acquired and utilized directly from product operation [49][50][51]. However, acquiring and utilizing data generated by products is restricted to capital intensive products such as automobiles, aircrafts, etc. [51]. ...
... The timing of this work aligns with the recent emergence of the internet of things where information can now be acquired and utilized directly from product operation [49][50][51]. However, acquiring and utilizing data generated by products is restricted to capital intensive products such as automobiles, aircrafts, etc. [51]. Also, use of such data are restricted to predictive maintenance [51][52][53][54], marketing [49][50][51], and environmental impact assessment [51,55]. ...
... However, acquiring and utilizing data generated by products is restricted to capital intensive products such as automobiles, aircrafts, etc. [51]. Also, use of such data are restricted to predictive maintenance [51][52][53][54], marketing [49][50][51], and environmental impact assessment [51,55]. In addition, there is a lack of research on utilizing data generated by products as feedback to a design process [51,56]. ...
Article
Background: A critical task in product design is mapping information from consumer to design space. Currently, this process largely depends on designers identifying and mapping psychological and consumer level factors to engineered attributes. In this way, current methodologies lack provision to test a designer's cognitive reasoning and could introduce bias when mapping from consumer to design space. In addition, current dominant frameworks do not include user-product interaction data in design decision making, nor do they assist designers in understanding why a consumer has a particular perception about a product. Method of approach: This paper proposes a new framework - Cyber-Empathic Design - where user-product interaction data is acquired via embedded sensors. To understand the motivations behind consumer perceptions, a network of latent constructs forms a causal model framework. Structural Equation Modeling (SEM) is used as the parameter estimation and hypothesis testing technique, making the framework falsifiable in nature. Results: To demonstrate the framework, a case study of sensor-integrated shoes is presented, where two models are compared - one Survey-based and one using the Cyber-Empathic framework model. Two methods are used to estimate the parameters and the fit indices - Covariance based SEM and Partial Least Square SEM. It is shown that the Cyber-Empathic framework results in improved fit using both estimation techniques over survey-only SEM. Conclusion: This work demonstrates how low level user-product interaction data can be used to understand and model user perceptions in a way that can support falsifiable design inference.
... The relationship between a manufacturer and its customers changes because smart products establish a new communication channel that makes the interaction between both stakeholders continuous and open-ended, thus leading to new responsibilities [6]. However, the data generated by smart products or services can be very useful for product design if companies are able to better understand how customers really use their products [1,7]. ...
... In marketing data analytics can help to better understand customers and address them with suitable advertising or services [5]. Also, in product development use phase data allows to detect the weak spots of a product and avoid them in the future [7]. ...
... Based on findings in literature, engineering companies that already collect data are mainly from the automotive, aerospace, or industrial and military equipment sector [7]. In general, it seems that use phase data is especially collected for products with a distinct level of complexity and for those that require a higher capital investment. ...
... IoT solutions, for instance, deliver value to both the provider and the user by offering at the same time several benefits. Additionally, the value therefore produced is greater than monetary compensation [138,139]. According to [140], when the IoT promotes ongoing client contacts, it creates a situation where both users and providers are influenced. ...
Article
Full-text available
Edge–fog computing and IoT have the ability to revolutionize businesses across all sectors and functions, from customer engagement to manufacturing, which is what makes them so fascinating and emerging. On the basis of research methodology by Webster and Watson (2020), 124 peer-reviewed articles were discussed. According to the literature, these technologies lead to reduced latency, costs, bandwidth, and disruption, but at the same time, they improved response time, compliance, security and greater autonomy. The results of this review revealed the open issues and topics which call for further research/examination in order for edge–fog computing to unveil new business value streams along with IoT capabilities for the organizations. Only by adopting and implementing precisely these revolutionary will new solutions organizations succeed in the digital transformation of the modern era. Despite the fact that they are cutting-edge solutions to business operations and knowledge creation, there are still practical implementation issues to be dealt with and a lack of experience in the strategic integration of the variable architectures, which hinder efforts to generate business value.
... The access to information via the Internet has turned customers into generators of structured, semi-structured and unstructured data (Erevelles, Fukawa, & Swayne, 2016). In addition, products are becoming complex systems that are successively equipped with more software and sensors (Dawid et al., 2016), offering opportunities for collecting data about how products are used in the customer environment, the analysis of which can produce valuable insights for companies (Van der Vegte, 2016). ...
Thesis
To develop products through a customer-centric strategy, early stages of product development such as target setting play an important role. In the target setting stage Customer Needs (CN) are gathered and translated into Design Requirements (DR) in order to subsequently set product targets that fit cost constraints and at the same time result in high Customer Satisfaction (CS). Continuous advances in information technology create new opportunities for companies to gather information about the customer, for example, for marketing purposes, or to assess customer reactions after the launch of new products. In addition, products are becoming complex systems that are successively equipped with more software and sensors offering opportunities for collecting data on how they are used. Knowing how customers use the product enhances a company’s ability to segment customers and customize products. Despite customer information availability from different sources (sensors, social media, etc.), surveys and focus groups are considered today as the main data source to derive the set of CN statements during target setting. Further, the team’s interpretation of CNs, which are often described in abstract language, must be translated into DRs, which are described in a more technical language. Hence, the translation process of CNs into DRs is said to be subjective. To set product targets, CS sensitivity to changes in DR levels is also considered. Surveys and benchmarking data containing customer perceptions on competitors’ performance are often the main customer data input into the process. While insightful information may be obtained, surveys are costly and time consuming and only encompass a small part of the market population. The research presented in this doctoral thesis explores how customer information obtained from sensors (e.g. product usage data) and text data (e.g. from websites, open-survey questionnaires) can be factored in the target setting process before concept generation to enhance customer focus without compromising product development time. The aim is to increase designers’ awareness of target population and in turn increase the quality of the design decisions on product targets. For this purpose, a customer-focused data-driven target setting methodology is proposed. The presented methodology changes the actual target setting methodology by means of indicators and autonomous activities on those parts of the process where marketing or design decisions are needed. The proposed methodology gives the incentive for a more integrated product development where marketing and designers need to work closely. This further allows a sustainable customer information gathering strategy that strives for missing customer information that is required for setting product targets. The indicators act as feedback channels for continuous product improvement. The use of such indicators and autonomous activities highlights the potential of a more efficient, less subjective and higher-quality target setting process.
... Feeding structured MoL data and use patterns back to product designers is an insufficiently addressed issue [4]. The key challenge is to find ways of using data analytics techniques effectively in purposeful combinations, depending on the application contexts and specific objectives of product designers [5]. ...
Conference Paper
Full-text available
Companies are getting increasingly interested in learning how different customers use their products. Collecting data about the use of products provides useful insights and facilitates design enhancements. Effective data analytics needs dedicated tools. In this paper, we summarize the results of our literature research done with special attention to existing tools. We observed that everything is changing rapidly and getting more complex in terms of data and processing methods and tools. While remarkable attention has been paid to processing big data, much less is being devoted to effective semantic progressing of middle-of-life (MoL) data. One of our findings is that commercialized data analytics tools have not addressed extraction, aggregation, and handling genuine MoL data adequately. Another one is that the currently available tools are in the lack of the capability to adapt themselves to designers needs and to produce results that could be reused in multiple design tasks. Nowadays products are equipped with smart capabilities and this offers new opportunities for exploiting middle-of-life data. The knowledge aggregated in this study will be used in the development of a sophisticated toolbox. This will: (i) integrate various tools under a unified interface, (ii) implement various semantics orientated and smart reasoning-based functions, and (iii) facilitate data transformations by practicing designers in contexts.
... The access to information via the Internet has turned customers into generators of structured, semi-structured and unstructured data (Erevelles, Fukawa, & Swayne, 2016). In addition, products are becoming complex systems that are successively equipped with more software and sensors (Dawid et al., 2016), offering opportunities for collecting data about how products are used in the customer environment, the analysis of which can produce valuable insights for companies (Van der Vegte, 2016). ...
Conference Paper
Customer satisfaction is used by many companies as a key performance indicator and it is strategically important to be able to define design requirements that contribute to customer satisfaction when setting targets. For highly complex products such as vehicles, target setting is an evolving process based on continually changing internal and external requirements. Quality Function Deployment (QFD) is a method that provides a structured approach for incorporating customer needs into the product development process. However, in addition to product targets, product usage proficiency also contributes to customer satisfaction. Customers often do not read manuals; they learn by trying things out and sometimes the use of the product ends up outside the expected acceptable range of the designers, delivering to the customer low product performance. The intention of this article is therefore to gain a deeper understanding of the customer by analyzing customer-product interaction of customer products and integrating it into QFD to identify the most interesting design requirements to improve customer satisfaction when developing products that are comparable to the ones lunched in the market. The proposed method facilitates designer awareness of target population before re-designing an existing product and it helps designers to set a starting point to improve usage proficency for each customer by providing individualized feedback. Copyright © 2016 by ASME Country-Specific Mortality and Growth Failure in Infancy and Yound Children and Association With Material Stature Use interactive graphics and maps to view and sort country-specific infant and early dhildhood mortality and growth failure data and their association with maternal
Article
In the field of Product Lifecycle Management, the gaps in terms of product usage data collection and exploitation must be addressed. The proposal addresses the modeling of the information chain from the product to the stakeholder. This model considers the product and its context composed of the user, task, and environment. It exploits a holonic view of the product and a dynamic informational structure to store the data, information, and knowledge collected on the product and its context during various instances of usage. The proposed model is applied to the diagnosis of equipment in the field of railway transportation.
Thesis
Full-text available
Housing affordability is a growing problem in the Netherlands. Approximately 15% of all households has a housing cost overburden rate (i.e. spending more than 40% of the household’s disposable income on the total housing expenses). This research explores a new way to contribute to housing affordability. Dwelling-generated revenue is introduced as umbrella concept to catch all potential ideas that may lead to revenue generated by a dwelling or its affiliated services that can be used to reduce housing expenses. A taxonomy of dwelling-generated revenue options is presented. Revenue options can be either object (i.e.: dwelling) or subject (i.e.: resident) bound. Object related revenues are selling a surplus of energy, third party compensation and (sub)letting. Subject related revenues are the monetization of in-home generated data and monetization of recyclable household waste. Through a multi-criteria analysis, in which five housing and innovation experts judged the revenue options, the most promising options are determined. The implementation of those options in the social housing sector is discussed, since most households with payment risks are found within sector. Keywords: housing affordability, user costs, housing expenses, innovation, exploratory, dwelling-generated revenue, taxonomy, multi-criteria analysis, expert judgement, affordable housing, social housing, data monetization, surplus energy dwelling, subletting, Airbnb
Article
Full-text available
Cloud computing and Internet of Things (IoT) are two very different technologies that are both already part of our life. Their adoption and use are expected to be more and more pervasive, making them important components of the Future Internet. A novel paradigm where Cloud and IoT are merged together is foreseen as disruptive and as an enabler of a large number of application scenarios.In this paper, we focus our attention on the integration of Cloud and IoT, which is what we call the CloudIoT paradigm. Many works in literature have surveyed Cloud and IoT separately and, more precisely, their main properties, features, underlying technologies, and open issues. However, to the best of our knowledge, these works lack a detailed analysis of the new CloudIoT paradigm, which involves completely new applications, challenges, and research issues. To bridge this gap, in this paper we provide a literature survey on the integration of Cloud and IoT. Starting by analyzing the basics of both IoT and Cloud Computing, we discuss their complementarity, detailing what is currently driving to their integration. Thanks to the adoption of the CloudIoT paradigm a number of applications are gaining momentum: we provide an up-to-date picture of CloudIoT applications in literature, with a focus on their specific research challenges. These challenges are then analyzed in details to show where the main body of research is currently heading. We also discuss what is already available in terms of platforms-both proprietary and open source-and projects implementing the CloudIoT paradigm. Finally, we identify open issues and future directions in this field, which we expect to play a leading role in the landscape of the Future Internet.
Article
Full-text available
Next generation vehicles will provide powerful connectivity and telematics services, enabling many new applications of vehicle communication. We will in this paper study the opportunities of performing remote vehicle diagnostics, where the diagnostic tool (test equipment) and the vehicle are separated by an internetwork, e.g., the Internet. The development of a prototype system for remote vehicle diagnostics, based on the emerging Diagnostics over IP (DoIP) ISO standard, is presented and early usage experiments with synchronous remote diagnostic read-out and control are described. A number of safety related issues are identified that will need closer study before a broad deployment of remote diagnostics services is feasible. Furthermore, a classification of vehicle diagnostics applications is provided, which is intended to elucidate the differences between synchronous (online) and asynchronous (offline) operation in local and distributed settings.
Article
Full-text available
Providing various wireless connectivities for vehicles enables the communication between vehicles and their internal and external environments. Such a connected vehicle solution is expected to be the next frontier for automotive revolution and the key to the evolution to next generation intelligent transportation systems (ITSs). Moreover, connected vehicles are also the building blocks of emerging Internet of Vehicles (IoV). Extensive research activities and numerous industrial initiatives have paved the way for the coming era of connected vehicles. In this paper, we focus on wireless technologies and potential challenges to provide vehicle-to-x connectivity. In particular, we discuss the challenges and review the state-of-the-art wireless solutions for vehicle-to-sensor, vehicle-to-vehicle, vehicle-to-Internet, and vehicle-to-road infrastructure connectivities. We also identify future research issues for building connected vehicles.
Article
• A Concept of Operations Document for Vehicle Health Assurance has been completed. • Develop, document, and also vet with the external community within the Vehicle Health Assurance working group an integrated system concept for vehicle health assurance that fully integrates ground-based inspection and repair information with in-flight measurement data for airframe, propulsion, and avionics subsystems. • The document is the product of a multi-center team within the Aviation Safety program • The vetting process included industry and government and has been completed. The document has been modified to reflect these external comments. • The document is a NASA TM: NASA/TM - 2013-217825.
Article
The pathways through which information is gathered, stored, and dispatched in organizations are changing: the physical world itself is becoming a type of information system. In what's called the Internet of Things, sensors and actuators embedded in physical objectsfrom roadways to pacemakersare linked through wired and wireless networks, often using the same Internet Protocol (IP) that connects the Internet. These networks churn out huge volumes of data as they sense the environment and as devices communicate with one another. The data can be marshaled to aid decision makers or to respond automatically to events. These emerging information networks promise to change business models for many companies, offering new ways to interact with consumers, fine-tune processes for greater productivity, automate dangerous tasks, and better manage risk.
Chapter
Through-life Engineering Services (TES) provide product support throughout each stage of the product-lifecycle; from conception, through design, manufacture and operational life, to end of life disposal. They are seen as a natural stage in the evolution of product support and maintenance, repair and overhaul strategy. They are the sum of many diverse product support strategies which use emerging and traditional technologies, processes, and applications. Whilst there are increasing numbers of contributions to be found within the literature defining the content, scope, purpose and application of the supporting technologies one sees no definition for TES emerging. This chapter offers a definition for Through-life Engineering Services which states what the concept is. It gives dimension, application, and purpose for TES in its role as a facilitator of Technology Enabled Service Delivery Systems which support manufacturing organisations wishing to compete through the adoption of Product Service Systems. An initial taxonomy is also presented.
Article
Due to technological advancement, an increase in creating and storing digital data can be observed today, for instance throughout the product life cycle. Through especially product use, a large number of data are created. The data are mainly used during the product use phase (e.g. condition-based maintenance). However, those data can be used as feedback to generate knowledge and support the product developer through improving future product generations. This paper presents development and prototypical realisation of a feedback assistance system for the management, analysis and visualisation of feedback data. The system supports a product developer by improving the process of decision-making to discover development potentials. Thereupon, four modules have been developed to analyse related indicators statistically and to diagnose machine failures. The obtained knowledge about the current condition of machines should then be deployed to improve the next machine generations utilising the modules for decision support and prediction.
Article
In this paper, we review the background and state-of-the-art of big data. We first introduce the general background of big data and review related technologies, such as could computing, Internet of Things, data centers, and Hadoop. We then focus on the four phases of the value chain of big data, i.e., data generation, data acquisition, data storage, and data analysis. For each phase, we introduce the general background, discuss the technical challenges, and review the latest advances. We finally examine the several representative applications of big data, including enterprise management, Internet of Things, online social networks, medial applications, collective intelligence, and smart grid. These discussions aim to provide a comprehensive overview and big-picture to readers of this exciting area. This survey is concluded with a discussion of open problems and future directions.