Preservation and Redeployment of Sensor Acquisition Processes from a
Dam Safety Information System
Anja Bachmann
, Martin Alexander Neumann
, Hossein Miri
, Jos
e Barateiro
, Gonc¸alo Antunes
and Artur Caetano
Karlsruhe Institute of Technology (KIT), TECO, Vincenz-Prießnitz-Straße 1, 76131 Karlsruhe, Germany
orio Nacional De Engenharia Civil (LNEC), Avenida Brasil 101, 1700 Lisbon, Portugal
Instituto de Engenharia de Sistemas e Computadores - Investigac¸
ao e Desenvolvimento (INESC-ID),
Rua Alves Redol 9, 1000-029 Lisbon, Portugal
Digital Preservation, Sensor Networks, Civil Engineering.
There is a growing demand for digital preservation of, not only static objects and files, e.g. text documents
and images, but also dynamic objects and whole processes, such as interactive media and entire business
and scientific processes. This paper investigates the problem of digitally preserving monitoring processes of
a dam safety information system. Monitoring processes are a crucial element in ensuring structural safety;
the interpretation of the past data produced by such processes as well as the prediction of potential future
behaviour facilitates an earlier detection of possible dam failure. After a successful preservation performance,
relevant data can be used or re-produced without the need for the original system to still exist; merely by
re-playing the preserved information and data. This enables several possibilities in the scope of a water dam
system. The retracing of former situations and structural behaviour decades later is one of them. Furthermore,
the interpretation of past data and subsequent prediction of future behaviour that could facilitate an earlier
detection of a fault or possible dam failure. This work presents a methodology for preserving the obtained
sensor data (readings, measurements, and meta-data) from a dam safety information system, whose involved
processes include: data acquisition, the preservation process itself, and the re-playing and redeployment of the
preserved data.
Within the last few years, there has been a growing
demand for digital preservation. This is caused by the
need to protect information of enduring value in in-
terrelation with the drastically increasing number of
documents and digital objects that are being produced
every day (Conway, 1996; Borghoff et al., 2006). This
amount has not only to be managed, but also pre-
served for later use and contextual interpretation, in
consideration of the preservation of the value com-
prised in this data.
Digital preservation research is concerned with
providing long-term access to and intelligibility of
digital objects, regardless of their complexity (Lee
et al., 2002; Giaretta, 2011). The focus is on preserv-
ing digital objects along with their meta-data (con-
textual or proxy information) required to achieve this
goal (Day, 2001). In the past, static objects were con-
sidered the main focus of digital preservation. These
objects do not expose any behaviour, i.e. they do not
perform any activity or interaction with their environ-
ment that is visible externally. Such objects, which
are to be preserved digitally, are primarily text and
multimedia documents. Notably, in order to be gen-
erated and interpreted, digital objects require a tech-
nological context defined by specific software and in
some cases even by specific hardware (Chou et al.,
Recent research in the area of digital preservation
has focused on extending established preservation ap-
proaches to dynamic objects. These are objects that
actually perform behaviour over time. Examples are
video games (Guttenbrunner et al., 2010), interactive
art (McHugh et al., 2010; Becker et al., 2007) and
computational environments, such as scientific work-
flows (Roure et al., 2011). Additionally, more and
more static digital objects are replaced by dynami-
cally generated ones, e.g. generated meta-data and dy-
namic websites. Digital business processes are a spe-
Bachmann A., Neumann M., Miri H., Barateiro J., Antunes G. and Caetano A..
Preservation and Redeployment of Sensor Acquisition Processes from a Dam Safety Information System.
DOI: 10.5220/0004625404900495
In Proceedings of the International Conference on Knowledge Discovery and Information Retrieval and the International Conference on Knowledge
Management and Information Sharing (KMIS-2013), pages 490-495
ISBN: 978-989-8565-75-4
2013 SCITEPRESS (Science and Technology Publications, Lda.)
cial and more complex type of dynamic object, which
present big challenges to preservation due to the mul-
titude of dependencies that have to be kept in mind.
As a result, there is a necessity to not only preserve
the object but rather the process describing the con-
text of the digital object to fulfil the aim of preserving
this object fully.
Here, important civil engineering structures like
dams are considered. The monitoring of their be-
haviour is important to guarantee is faultless function.
The data required for analyzing the behaviour is ac-
quired by a set of sensors placed along the dam. To fa-
cilitate the monitoring and guarantee its functionality
in the future, the monitoring and data acquisition pro-
cess have to be preserved. It is important to preserve
data about the water dams so that later on it is possi-
ble to retrace former situations and behaviours. Pre-
served data can be used for example to build a model
and predict future behaviour or to detect anomalies.
It also facilitates to track a possible problem source
down to specific system behaviour. To preserve data,
we need to keep both, sensor data and contextual in-
formation, in a preservable format.
The remainder of this paper is structured as fol-
lows. Section 2 gives an overview of the background
of our use case and the monitoring processes of the
water dam. In Section 3, we present a typical moni-
toring architecture for water dams and name relevant
aspects to be preserved. This is followed by our pro-
posed methodology to preserve these contextual as-
pects in Section 4. This extends our general architec-
ture for preservation of business processes which has
been previously published in (Galushka et al., 2012).
The preserved data can, then, be re-played for contex-
tual interpretation. Section 5 describes this process
in detail. Finally, Section 6 discusses the usefulness
and suitability of our architecture and model in the
context of preserving and re-playing water dam mon-
itoring processes, and also gives an outlook to future
We perceive a monitoring process of a water dam as
such a kind of process to be preserved. In this paper,
we present our perspectives on what, why and how it
is to be preserved, in the context of such water dam
monitoring processes. Our general formal model to
capture what to preserve has been presented in (Neu-
mann et al., 2012).
Here, we present our specialization of this model
to the domain of water dam monitoring processes, and
will discuss its usefulness in preserving these moni-
toring processes.
2.1 Use Case Scenario
Within the last years, the interest in the area of civil
engineering raised. The focus is on “conservation of
the environment, the welfare and safety of the individ-
ual” and there is a need to “manage the prevailing nat-
ural and manmade risks in a conscious, consistent and
rational manner” (Faber, 2007). Wieland and Mueller
(Wieland and Mueller, 2009) advise their readers of
the far-reaching consequences of a possible failure of
large civil engineering structures. Especially in sys-
tems like dams or nuclear power plants a failure could
endanger the life of human beings and the nature, but
also damage important infrastructures like electricity
or water supply. There are several preventive mea-
sures and mitigation strategies to reduce the risk. An
often applied methodology is to detect anomalies and
errors as early as possible to increase the time to react
to them and prevent from the worst consequences.
In this use case, the focus will be on large civil
engineering structures like dams. They have a crit-
ical infrastructure that can face the risks mentioned
above. Some countries established legal obligations
for national research institutions to monitor the be-
haviour of a civil engineering system. Portugal for ex-
ample enforced a strict Dam Safety Legislation which
are stating some general regulations for water dams,
among others. This legislation also instructed the
orio Nacional De Engenharia Civil (LNEC)
to monitor the structural safety of a water dam system.
The system, considered here, encompasses 117 water
dams in Portugal. Each of them is equipped with a
different number of sensors, most with less than 50,
some even with more than 1000. Overall there are
31479 sensors that are taken into account.
To preserve all data relevant for the preservation
process in the context of this use case, it is neces-
sary to gather information about the sensor, its envi-
ronment and its measurements (Kutter et al., 2002).
Such information has to be exchanged between sev-
eral programs, e.g. the data collector and the evalua-
tion system. Hence, it is necessary to select a suitable
and interoperable file format to perform this task. The
resulting tasks are therefore to acquire process infor-
mation, sensor data and environmental information,
convert the data into a preservable format and, when-
ever required by the user, replay this data for contex-
tual interpretation.
To perform this task it is necessary to install sev-
eral sensors in the dam structure. It is possible to
manually or automatically gather the required infor-
mation and later on convert this data for further anal-
ysis. Data is acquired manually, e.g. by a human op-
erator using measuring devices at different locations
within the water dam. For automatic data collection
it is advisable to establish a wireless network to di-
rectly get the data from the sensor and store it. If such
a connection is established, the sensor can commu-
nicate directly and automatically with the dam safety
information system and transfer the information.
2.2 The Need for Preservation
Monitoring a dam and understanding its structure are
inseparable from each other. It is possible to evalu-
ate and validate measurements, taken from sensors in-
stalled at strategic points within the observed system,
and interpret correlations between them. The result-
ing information can further be used for the prediction
of future behaviour, the estimation of missing values
or the detection of anomalies. This underlies the re-
quirement of reliable data and measurements. De-
vices like plumb lines and piezometers are installed
at different locations within a water dam to gather in-
formation about it and monitor its behaviour (Amer-
ican Society of Civil Engineers, 2000; International
Commission on Large Dams, 1999). The number of
used sensors can range from less than 100 to more
than 1000, depending on the dimension of the dam.
Process information, meta-data and sensor mea-
surement all this data has to be preserved. In
this way, contextual interpretation is possible even
decades after the data was gathered and even without
the existence of the original system. Occurring prob-
lems are that the former processes are not existing any
more and cannot be retraced or that former data is
missing and cannot be reproduced easily. Preserva-
tion countervails these problems.
Altogether, the motivational factors for digital
preservation of sensor data acquisition processes in
the scope of civil engineering can be summarized as
Compliance with Legal Requirements – Fulfilling
the legal obligations and the agreements estab-
lished with the owners of the structures.
Replay of Processes Enabling a later replay of
former scenarios for retracing the scenario and the
original behaviour of the underlying system. En-
suring the authenticity, correctness and validation
of the execution.
Re-use of Processes Gaining new insights by
modifying parts of the execution process. Re-
execution of former processes with modified pa-
rameters or execution of a modified process with
the original data.
Assessment of the Costs of Retention Consider-
ing whether to keep a big dataset of gathered data
or to only keep the process that generate the data
with respect to the corresponding costs.
In consideration of these factors, the preservation of
processes in the scope of civil engineering monitoring
activities is a challenge that must be addressed.
In digital preservation it is not sufficient to only store
measurements that were gathered by sensors. For a
later understanding of a situation or a scenario it is
inevitable to also store information about the process
and meta-data about the environment and interrela-
tion of sensors. This environment encompasses the
location of the sensor (position, location), but also en-
vironmental information like weather, temperature or
humidity. Additionally, there is also contextual in-
formation about the process itself, e.g. dependencies
among the software or hardware components.
The focus of this project is on process preserva-
tion. It is inevitable to preserve all this data which
incorporates information about the process or repre-
sents influencing factors. Environmental components
(temperature, location, position) influence the func-
tionality of a sensor, the measuring and monitoring
processes. Dependencies and interrelations of soft-
ware or hardware components are relevant to model
the process.
3.1 Acquisition of Sensor Data
As mentioned before, we require sensor data and
meta-data, among others. The sensors have to be ac-
cessed and the measurements have to be gathered and
afterwards stored in a database. This data consists of
the identification number of the sensor, the identifi-
cation number of the measuring device, the measured
value, information about the temperature and so on.
It is possible to acquire data by manual or automatic
measurement. For example it is possible to automat-
ically get location information by checking the GPS
signal. Values like temperature can also be measured
automatically. Information about the monitoring pro-
cess itself or dependencies will have to be added semi-
automatically in the next step.
Currently, the Laborat
orio Nacional De Engen-
haria Civil is handling 35 types of manual sensors
and 25 types of automatic data gathering devices. The
Figure 1: UML Class diagram describing the classes of the Sensor DSO.
processes of gathering and transforming the measure-
ments and all involved data transformations and algo-
rithms are what has to be preserved within this use
3.2 Water Dam Context Model
Several aspects in the context of a business process
have to be taken into account for digital preserva-
tion to ensure successful redeployment of that process
(Neumann et al., 2012). The authors also define two
kinds of aspects that have to be taken into account
for the preservation of a process: coarse-granular as-
pects (abstract and not only relevant for the process
but its entire domain) and fine-granular aspects (more
specific to the process itself and its sub-processes).
If such knowledge about a process is machine-
interpretable formalized and stored directly with the
digital objects, semantic reasoners can immediately
incorporate it into decision making. There is no need
to consult external systems, repositories or human
preservation operators. Furthermore, if the formal
system is not a specialized language, but a generic
knowledge representation language such as OWL
(Web Ontology Language), the preservation-relevant
advantage arises that the knowledge kept close to
the digital objects may even be object-specific. This
means that the model may be specific to a preserved
business process, but still is machine-interpretable
(Neumann et al., 2012).
There are two Context Model instances (DIO and
DSO) related to this use case which will only be
shortly described here. A domain-independent ontol-
ogy (DIO) represents a neutral, domain-independent
language that is able to represent the core concepts
of our Context Model. It is designated domain-
independent since it does not address any specific
domain-dependent concerns. A domain-specific on-
tology (DSO) represents a domain-specific language
that addresses a particular set of concerns. Our Con-
text Model will comprise a core DIO and a set of
DSOs. They represent object classes and depen-
dencies and interrelations between them in order to
describe the process to be preserved and its con-
text. Contextual information and the dependencies
between processes are extracted by using different ex-
tractors. One of the DSOs is related to this use case
the Sensor DSO. It aims at capturing particular as-
pects about sensors and other related information. It
gives an overview of the relations between acquisition
algorithms and transformations and the sensors, their
types and properties. The UML class diagram in Fig.
1 describes the different classes of the Sensor DSO.
The focus of this project is on digital process preser-
vation. To achieve this the subsequently described
methodology to preserve civil engineering processes
was performed. As explained in Section 3 the first
step is to acquire data. This consists of the acquisition
of measurements, but also (meta) information about
the sensors and the associated processes. To model a
process it is also important to detect and cover depen-
dencies between data and sub-processes. Such con-
textual information is added to the content data and
stored together for preservation issues.
In todays IT domain there is a burden of coping
with the diversity and heterogeneity of data sources,
dynamically selecting appropriate data sources, and
scaling of data from mobile, distributed data sources
(Cohen et al., 2001). This causes the need for an inter-
operable, easily manageable, flexibly extensible and
common data format that facilitates the access of and
further work on a document (Bloechle et al., 2006).
An interesting candidate is XML with a sensor data
extension, e.g. Unisens
, SenML
or SensorML
They provide various features relevant to our use case.
The presented preservation approach divides the
set of required data into three parts: sensor mea-
surements, sensor meta-data and process meta-data.
A separation of these components was chosen to al-
low working on one component without influencing
or endangering the other ones. Two different data
file types are used to preserve this data. The sen-
sor measurements and the environmental sensor meta-
data (e.g. temperature) are stored in one of the men-
tioned XML-based formats. The process meta-data
(e.g. dependencies and interrelations of processes or
sensors) is stored in the Context Model, introduced in
Subsection 3.2. A specific part of the Context Model,
the OWL file “SensorDSO”, is capturing dependen-
cies and properties of sensors. All entities, depen-
dencies and properties of the DSO can be mapped to
the acquired sensor data and meta-data. The advan-
tage: OWL is not only interoperable, it also allows
a reasoning on the data to induces findings about de-
pendencies and properties. Thereby, the incorporated
meta-data will be contextual interpretable and human
One of the main purposes of the digital preserva-
tion of business process is to enable a later contex-
tual interpretation of a former scenario and to retrace
what has happened before. For realizing this, pre-
served data is replayed. As stated by van der Ho-
even et al. the focus is not only “on the digital ob-
ject, but on the hardware and software environment
in which the object is rendered”, moreover the aim is
to “(re)creating the environment in which the digital
object was originally created” (van der Hoeven et al.,
2007). In our case, this means to preserve and replay
the monitoring process.
The replay of a former scenario is important to
retrace what had happened before or to make further
use of this historic data. There are several alternatives
to perform an emulation of past sensor data.
One method: as input data, a certain time and date
combination and the ID of a sensor are given. Af-
terwards it is possible to track down this one sen-
sor to the time back then and retrieve the past mea-
A second way: again, time and date for the
emulation are given. This time, a whole water
dam/sensor system is considered and not a sin-
gle sensor. Then, by using the past measurements
and the contextual information, it is possible to get
an overview of the former behaviour of this water
It is also possible to apply machine learning tech-
niques to the sensor data and contextual informa-
tion. A model is trained and learned. Later on
this model is able to reproduce information about
former behaviour based on given input.
Based on the concrete application and its com-
plexity, a different method has to be chosen. Prob-
ably, there are even more possible replaying methods,
the presented ones are only giving a first insight. A
preserved process has been redeployed successfully
if the replayed process shows the same behaviour as
the original one.
It is not only possible to use the preserved data
to replay former scenarios and interpret them. The
data can also be used as historic input data for sev-
eral stream mining algorithms. It is possible to pre-
dict the future behaviour of a water dam based on this
data. Moreover it is possible to learn a model to de-
tect anomalies. It is also possible to compare different
The relevance of digital preservation in areas such as
civil engineering is rising due to the evolvement of big
data and the increasing importance of safety civil in-
frastructures. This paper underlined the significance
of digital preservation of monitoring such an infras-
tructures. It further presented a methodology for pre-
serving complex dynamic processes of a water dam
system. Still, there are open tasks and starting points
for further investigations.
Digital preservation is challenging. Its aim is not
only to store cultural heritage but also to enable con-
textual interpretation and retracing of the past. We are
investigating a use case situated in the area of civil en-
gineering. One aim of civil engineering is to increase
safety and reduce the risks of a failure by a better
monitoring and by future anomaly detection. Here,
the focus is on one step before: the preservation of
data gathering and monitoring processes, containing
information of what already happened and is happen-
ing. We presented a methodology for this preserva-
tion, bringing civil engineering one first step closer to
risk mitigation and failure avoidance.
The considerations presented in this paper are a
good basis, but requesting further research. First of
all it is inevitable to develop a conversion tool for
preservation. This tool is supposed to take sensor
and contextual information as input and convert them
into a preservable format. Of course, it is necessary
to check whether the preserved data contains all rel-
evant information that was given by the input data.
The preserved data will be replayed some time later
on. To perform this action a replaying algorithm is
required. This algorithm has to be examined regard-
ing its replay accuracy and usefulness. It is important
to check whether the replayed data behaves like the
original data did. As mentioned before in Subsection
4 there are additional possibilities to work with pre-
served data in the future than just replay it. There is
also the option to use the preserved, historic data to
predict future behaviour, e.g. by performing stream
mining algorithms. Especially in the framework of
civil engineering, where the focus is often on enhanc-
ing the safety and reducing a risk by avoiding failures,
the early detection errors or uncommon behaviour is
important. This can be achieved for example by learn-
ing a model by applying machine learning techniques
on the data. It is considerable to simulate emergency
situations, e.g. the bursting or the failure of a water
dam, by altering some parameters. The task for the
developed system is to detect this anomaly as soon as
possible to raise an alarm. The question is whether
the system is able to detect such a failure in time.
The authors would like to acknowledge the funding
by the European Commission under the ICT project
“TIMBUS” (Project No. 269940, FP7-ICT-2009-6)
within the 7th Framework Programme. This work
was also supported by national funds through FCT
ao para a Ci
encia e a Tecnologia, under project
American Society of Civil Engineers (2000). Guidelines
for instrumentation and measurements for monitoring
dam performance.
Becker, C., Kolar, G., Kueng, J., and Rauber, A. (2007).
Preserving Interactive Multimedia Art: A Case Study
in Preservation Planning. In Proceedings of the
10th International Conference on Asian Digital Li-
braries, ICADL’07, pages 257–266, Berlin, Heidel-
berg. Springer-Verlag.
Bloechle, J. L., Rigamonti, M., Hadjar, K., Lalanne, D.,
and Ingolf, R. (2006). XCDF: A Canonical and Struc-
tured Data Format. Document Analysis Systems VII,
Borghoff, U. M., R
odig, P., and Scheffczyk, J. (2006).
Long-Term Preservation of Digital Documents, Prin-
ciples and Practices. Springer Berlin Heidelberg.
Chou, C., Dappert, A., Delve, J., and Peyrard, S. (2012).
Describing Digital Object Environments in PREMIS.
In Proceedings of the 9th International Conference on
Preservation of Digital Objects, iPRES 2012.
Cohen, N., Ourakayashta, A., Turek, J., Wong, L., and Yeh,
D. (2001). Challenges in Flexible Aggregation of Per-
vasive Data.
Conway, P. (1996). Archival Preservation in a Nationwide
Context. 2:204–222.
Day, M. (2001). Metadata for Digital Preservation: A Re-
view of Recent Developments. In Proceedings of the
5th European Conference on Research and Advanced
Technology for Digital Libraries, ECDL ’01, pages
161–172, London, UK, UK. Springer-Verlag.
Faber, M. H. (2007). Risk and safety in civil engineering.
Galushka, M., Taylor, P., Gilani, W., Thomson, J., Strodl,
S., and Neumann, M. A. (2012). Digital Preservation
Of Business Processes with TIMBUS Architecture. In
Proceedings of the 9th International Conference on
Preservation of Digital Objects, iPRES 2012.
Giaretta, D. (2011). Preservation of intelligibilityintelligi-
bility of digital objects. pages 39–166.
Guttenbrunner, M., Becker, C., and Rauber, A. (2010).
Keeping the Game Alive: Evaluating Strategies for the
Preservation of Console Video Games. International
Journal of Digital Curation, 5(1).
International Commission on Large Dams (1999). Guide-
lines for automated dam monitoring systems.
Kutter, B., Wilson, D., and Bardet, J. (2002). Metadata
Structure for Geotechnical Physical Models (and Sim-
ulations?). Technical report.
Lee, K.-H., Slattery, O., Lu, R., Tang, X., and McCrary,
V. (2002). The state of the art and practice in digital
preservation. Journal of Research-National Institute
of Standards and Technology, 107(1):93–106.
McHugh, A., Konstantelos, L., and Barr, M. (2010). Re-
flections on Preserving the State of New Media Art.
In Proceedings of the Archiving Conference.
Neumann, M. A., Miri, H., Thomson, J., Antunes, G.,
Mayer, R., and Beigl, M. (2012). Towards a Deci-
sion Support Architecture for Digital Preservation of
Business Processes. In Proceedings of the 9th Interna-
tional Conference on Preservation of Digital Objects,
iPRES 2012.
Roure, D. D., Belhajjame, K., Missier, P., Gmez-Prez, J. M.,
Palma, R., Ruiz, J. E., Hettne, K., Roos, M., Klyne,
G., and Goble, C. (2011). Towards the Preservation
of Scientific Workflows. In Proceedings of the 8th
International Conference on Preservation of Digital
Objects, iPRES 2011.
van der Hoeven, J., Lohman, B., and Verdegem, R. (2007).
Emulation for Digital Preservation in Practice: The
Results. 2:123–132.
Wieland, M. and Mueller, R. (2009). Dam safety, emer-
gency action plans and water alarm systems.