Semantic Data Integration for Ubiquitous Logistics
An Approach Supporting Autonomous Logistics in Urban Environments
Stefan Wellsandt, Konstantin Klein, Marco Franke, Karl Hribernik and Klaus-Dieter Thoben
BIBA – Bremer Institut für Produktion und Logistik, Hochschulring 20, Bremen, Germany
Keywords: Autonomous Logistics, Ubiquitous Computing, Semantic Data Integration, Interoperability.
Abstract: This paper introduces an approach for seamless plug & play data integration in a novel urban logistics
concept. The logistics concept is called ubiquitous logistics and contains an agent-based perspective on
shared logistics resources. Each private and public vehicle participating in the concept, as well as parcels,
features an intelligent agent that may request (or offer) information services to other agents or legacy
systems. The technical approach of this paper suggests that each of the heterogeneous data sources delivers
additional information that is used to virtually integrate the data in an automated way. This additional
information concerns, the authentication, data structure and sequence – information that have to be provided
manually nowadays. The technical approach is explained using a typical situation from the future urban
logistics concept. This situation represents an intelligent agent trying to deliver small goods along a stream
of urban commuters.
1 INTRODUCTION
The exchange of goods between companies and the
delivery of final products to consumers requires
considerable logistics capacity. The planning and
realisation of both the transport and storage of goods
need to take into account time, cost and quality
constraints. Changes in customer behaviour, such as
an increase in importance of individualised/
personalized products compared to mass-produced
ones (Kumar, 2007), challenge these activities. This
demand for individualised products, combined with
the growing popularity of online shopping and the
customers’ desire for same-day delivery cause a
reduction of delivery lot size, in turn increasing the
overall amount of deliveries (Ickert et al., 2007).
This growth of delivery volumes puts urban traffic
and logistics systems under additional stress causing
congestion, limited parking space, pollution and
infrastructure deterioration (Müller et al., 2006).
An approach to reduce these negative impacts
caused by increasing delivery volumes is the
ubiquitous logistics concept as presented in
(Wellsandt et al., 2013). Cornerstone of the
approach is to describe urban freight transport as a
system of shareable logistics resources managed by
collaborating intelligent agents. Agents continuously
request and offer information services while each
service is connected to a data source.
One of the most significant problems in this
environment is the seamless exchange of
information from distributed, heterogeneous data
sources. Intelligent agents and data sources like
traffic management systems, logistic infrastructure
and other systems have a need for actual information
about relevant system states to make a sustainable
decision. Nowadays, there are several approaches to
establish the communication between different data
sources featuring different information structures
and information meaning. One of these techniques is
the approach of the virtual data integration. The
main advantage of this approach is the delivery of
information on demand without the need for an
additional data storage device like a data base.
The paper aims to describe an approach enabling
automated virtual data integration in urban logistics.
For this purpose, section 2 will explain existing
research on the fundamental topics of this paper.
Section 3 will cover the technical approach that is
explained using a guiding example. Section 4
concludes the described approach.
2 RELATED WORK
This section covers existing work related to the core
652
Wellsandt S., Klein K., Franke M., Hribernik K. and Thoben K..
Semantic Data Integration for Ubiquitous Logistics - An Approach Supporting Autonomous Logistics in Urban Environments.
DOI: 10.5220/0004961206520656
In Proceedings of the 4th International Conference on Cloud Computing and Services Science (CLOSER-2014), pages 652-656
ISBN: 978-989-758-019-2
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
ideas of this paper. The section will explain the
concept of ubiquitous logistics and its preceding
background. Furthermore, the section will cover
aspects of data integration that are reused in the
introduced approach of this paper.
2.1 Autonomous Logistics
Interpretation of logistics as a multi-agent system
(MAS) is discussed in literature for several years
(Moore et al., 1997). In 2005, Davidsson et. al.
published a survey about the use of MAS in
transportation and traffic management (Davidsson et
al., 2005). Findings indicate that many logistics
problems have characteristics similar to those of
agent-based systems. A similar argumentation is
suggested for the collaborative research centre 637
for autonomous logistics in Bremen, Germany
(Scholz-Reiter et al., 2004). According to (Gehrke et
al., 2010), autonomous logistics can be enabled by
situation-aware agents, actively gaining information
about their surroundings. Based on an agent’s
observations, decisions can be made about logistics
processes. An example for the implementation of
situation-aware agents in logistics is the intelligent
container that is described in (Lang et al., 2011). A
comprehensive study on multi-agent coordination
enabling autonomous logistics is provided by
(Schuldt, 2012).
The argumentation of autonomous agents in
logistics offers far reaching possibilities for future
urban freight transport. In order to utilize
possibilities, the focus of planning and control of
urban transport needs to shift from a centralized
perspective towards a decentralized one. At the same
time, efficiency in urban freight transport is
increased by taking into account all available
logistics resources an urban environment provides.
2.2 Ubiquitous Logistics
In the ubiquitous logistics concept, public and
private transportation means that are able to
transport goods (parcels) within a city are considered
as logistics resources of that same city (i.e. private
vehicles, trucks, trams, buses, etc.). Primary
objective of the concept is to avoid additional traffic
for freight transport by utilizing existing logistics
resources best.
In order to manage the multitude of logistics
resources within a city, transported parcels feature
an autonomous, situation-aware agent that can be
human or artificial. The agent collects information
and takes decisions required to complete the
scheduled transportation task for a certain parcel.
The decision process each agent has to perform is
based on a multi-criteria problem. The problem
concerns the selection and realization of an optimum
route from the current location to a target destination
within a dynamic environment. Potential routes are
created and destroyed dynamically based on the
availability of local logistics resources. A situation-
aware agent picks logistics resources in order to
make progress against its scheduled transportation
task. Completion of a transportation task might
require taking different logistics resources (i.e.
vehicles or transportation modes).
The assignment of parcel and resource depends
on criteria like transportation time, load efficiency,
service level, stock level, as well as transportation
and storage cost (Nyhuis and Wiendahl, 2013).
Furthermore, environmental dynamics need to be
taken into account including for instance traffic
condition, delivery order changes, weather
conditions, and availability of logistics resources.
The decision process requires large amounts of
most recent information from distributed sources
like other agents, traffic management systems and
logistics providers. Therefore, the underlying
information infrastructure needs to be capable of
handling several thousands of decentralized
information requests and offers at the same time. In
this infrastructure, each information offer is
represented by a specific service. Services can be
provided by different organizations with different
data source structures. Complex decisions require
that several information services are requested and
integrated.
2.3 Data Integration
Information services are based on data sources.
From a technical point of view, significant data
sources in logistics are data processing systems, data
storage systems and sensors deployed in the field.
The data processing systems are decision making
agents, infrastructure management systems and other
necessary systems to handle processes in logistics.
Data storage systems are used to aggregate data in
logistics. This includes for instance generic data
bases, data warehouse systems, warehouse
management systems and product lifecycle
management systems. Sensors are typically used to
collect field data. This data may contain
measurements like temperature, light intensity and
mechanical forces. The described data is allocated in
different ways, for instance as a CSV file, text file or
database. It can be accessed by using HTTP (hyper
SemanticDataIntegrationforUbiquitousLogistics-AnApproachSupportingAutonomousLogisticsinUrban
Environments
653
text transfer protocol), web services or industrial
standards like EDIFACT or EANCOM. The data is
transmitted as a stream or in a structured way by, for
instance services, cloud applications, data gathering
modules and data warehouses.
The concept of the semantic mediator developed
to fetch data on demand based on the mentioned
virtual data integration approach is used to retrieve
data from sensors and data storage systems
(Hribernik et al., 2010). This concept abrogates the
persistence of the data from each data source in one
huge data warehouse. The implementation of the
virtual data integration concept retrieves only the
needed data on demand. The binding module to
resolve a query is implemented by a wrapper,
depicted in Figure 1.
Figure 1: Integration module concept (semantic mediator).
The wrapper contains the communication interface
between the data sources and a semantic mediator.
Nowadays, the wrapper modules have to be adapted
for new data sources manually. The mediated
schema remains untouched by this adaption. The
wrapper modules abstract the data sources and all
users query the needed information over the
wrapper. Here, the data from the particular data
sources will be transformed into a global
representation. The data sources are represented in
the figure by SQL, HTTP and CSV modules.
The following information in the data integration
domain need to be implemented manually: (1) the
access method; (2) the sequencing of the data access
and exchange; (3) the adaption to the structure of the
data sources. Related to the seamless integration of
information service in ubiquitous logistics, there is a
high demand for the automated handling in the
mentioned three points. Otherwise, the dynamic
integration of, for instance logistics resources (and
their related agents), is too time and cost intensive.
3 APPROACH
The technical approach of the proposed solution
aims for seamless (plug & play) integration of data
from different sources, in order to facilitate a
collaborative service infrastructure in urban
logistics. For this purpose, automated binding of
heterogeneous data sources that leaves only a few
activities up to the user is targeted in this paper. The
technical foundation of the approach is explained in
the next section, followed by a guiding example
consisting of a small situation.
3.1 Technical Foundation
In order to realize an automated binding of different
data sources, each source is able to describe itself.
Key information concerns the authentication,
sequence and structural characteristics of a data
source. Each of these three descriptors is explained
briefly in the following:
Authentication – concerned with the security
aspects of the access process of the particular
service. The information ensures subsequent
access authentication to eligible machines,
services and persons.
The authentication is typically based on
username and password. More sophisticated
methods are also possible. Moreover, the
protocol for the authentication can differ from
the essential data exchange protocol. It can
happen in one or more steps.
Sequence – typically, available data exchange
processes are conducted in multiple steps. Up to
now, implemented approaches don’t consider
information flows in more than one step.
Especially the data access over web services
proceeds during several steps. Moreover, the
most data accessing procedures happen in one
session with a recurrent sequential pattern of
“data access” and “data processing”. This pattern
is common in authenticated sessions in which the
subsequent data query has to carry the secure
token.
Data structure – describes what information and
what kind of data model the data source uses.
Furthermore, it contains information how certain
integration problems related to the data structure
can be solved. The data structure problems that
need to be solve are: (a) different names for
attributes with the same meaning; (b) different
data representation (describes how data is
conceived, manipulated and recorded); (c) use of
different data model; (d) data sources cover
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
654
different attributes.
The data sources send the information about their
authentication process, sequencing information and
the data structure information to the virtual data
integration module (semantic mediator) described in
section 2.3. The authentication and data structure
information relies on ontology models, located in the
mediated schema, describing the needed information
semantically. Ontologies lack the ability to describe
sequencing. According to this fact, the description of
the sequencing information is modelled by means of
sequence diagrams which are part of the Unified
Modelling Language (UML).
3.2 Guiding Example
The following example (see Figure 2) will be used to
explain how the integration of data might work in
the ubiquitous logistics concept.
Figure 2: Illustration of the guiding example for data
integration in ubiquitous logistics.
A hypothetical stream of daily commuters
(logistics resources) will be exploited to transport
small sized goods within a city. In order to keep the
delivery time and cost low, delivery paths (routes)
are decided by an intelligent. The agent utilizes
based the information extracted from heterogeneous
data sources. These sources concern for instance
weather and traffic data from the field (sensor data).
In addition, information about the delivery order
needs to be taken into account. These information
concern for instance the delivery destination
(address), expected delivery day and pricing
information. Each commuter’s agent offers a service
for its location and delivery status. Due to the
underlying agent-based network in ubiquitous
logistics, each parcel’s agent has to integrate the
distributed information for decision-making.
The necessary data exchange process for the
decision making happens through the proposed
approach. According the described situation the
following data integration process flow is
conceivable:
1. An intelligent agent needs additional traffic
information because a scheduled delivery route
collapsed. This might happen, for instance if a
commuter leaves her daily travel route and
another logistics resource has to complete the
delivery.
2. The intelligent agent sends a request for
additional routing information. The integration
module (e.g. semantic mediator) looks up for
the particular service(s) which can deliver the
requested information from a data source.
3. The integration module asks for the particular
authentication data and sequence models the
intelligent agent can use. After that the
authentication process between the requested
service and the intelligent agent proceeds.
4. The next step is the exchange of the existing
data structure model and the sequencing model
for the data exchange. After that, the data
exchange process proceeds.
5. The final logout process is part of the
authentication process. The necessary
information for this step has been exchanged in
the third step.
According to this brief example the data from
different sources can be integrated by the intelligent
agent automatically. The agent can request the
needed information services in order to manage the
delivery of the parcel. In case an additional service
provider (and the related data source) becomes
available, the new wrapper for the data source does
not need to be implemented manually but receives
relevant information from the data sources.
4 CONCLUSIONS
This paper presents an approach allowing intelligent
agents within an urban logistics scenario to integrate
heterogeneous information services and their related
data automatically. The paper starts with a brief
overview of autonomous and a future urban logistics
concept, as well as an overview of data integration.
In the following, the technical foundation of the
approach is outlined and explained through an
example. The example represents a typical situation
potentially occurring in ubiquitous logistics.
Benefit of the proposed approach is the reduction
of time and cost for manual adaption of data
integration tools. Additionally, it allows the
integration of data for distributed agents, reducing
SemanticDataIntegrationforUbiquitousLogistics-AnApproachSupportingAutonomousLogisticsinUrban
Environments
655
the need for maintaining local data bases. This will
become especially useful in the mentioned concept
for ubiquitous logistics that is grounded on agent-
based decision making based on numerous
distributed data sources. Dynamically changing
sources, especially related to logistics resources, can
be seamlessly integrated through the introduced
approach.
ACKNOWLEDGEMENTS
The paper is based on research that is done in the
research project “CyProS”, funded by the German
Federal Ministry of Education and Research
(BMBF) within the Framework Concept ”Research
for Tomorrow’s Production” (fund number
02PJ2461).
REFERENCES
Davidsson, P., Henesey, L., Ramstedt, L., Törnquist, J.,
Wernstedt, F., 2005. An analysis of agent-based
approaches to transport logistics. Transp. Res. Part C
Emerg. Technol. 13, 255–271.
Gehrke, J. D., Herzog, O., Langer, H., Malaka, R., Porzel,
R., Warden, T., 2010. An agent-based approach to
autonomous logistic processes. KI-Künstl. Intell. 24,
137–141.
Hribernik, K. A., Kramer, C., Hans, C., Thoben, K.-D.,
2010. A Semantic Mediator for Data Integration in
Autonomous Logistics Processes, in: Enterprise
Interoperability IV. Springer, pp. 157–167.
Ickert, L., Matthes, U., Rommerskirchen, S., Weyand, E.,
Schlesinger, M., Limbers, J., 2007. Abschätzung der
langfristigen Entwicklung des Güterverkehrs in
Deutschland bis 2050 (Study for the German Ministry
of Traffic, Construction and Urban Development).
Prognosen und Strategyberatung für Transport und
Verkehr.
Kumar, A., 2007. From mass customization to mass
personalization: a strategic transformation. Int. J. Flex.
Manuf. Syst. 19, 533–547.
Lang, W., Jedermann, R., Mrugala, D., Jabbari, A., Krieg-
Brückner, B., Schill, K., 2011. The “Intelligent
Container” — A Cognitive Sensor Network for
Transport Management. Sens. J. IEEE 11, 688–698.
Moore, M. L., Reyns, R. B., Kumara, S. R., Hummel, J.
R., 1997. Distributed intelligent agents for logistics
(DIAL), in: Systems, Man, and Cybernetics, 1997.
Computational Cybernetics and Simulation., 1997
IEEE International Conference on. IEEE, pp. 2782–
2787.
Müller, M., Görnert, S., Volkamer, A., 2006.
Gueterverkehr in der Stadt - Ein unterschätztes
Problem. Verkehrsclub Deutschland e.V.
Nyhuis, P., Wiendahl, H.-P., 2013. Logistische
Kennlinien: Grundlagen, Werkzeuge und
Anwendungen. Springer DE.
Scholz-Reiter, B., Windt, K., Freitag, M., 2004.
Autonomous logistic processes: New demands and
first approaches, in: Proceedings of the 37th CIRP
International Seminar on Manufacturing Systems. pp.
357–362.
Schuldt, A., 2012. Multiagent Coordination Enabling
Autonomous Logistics. KI - Künstl. Intell. 26, 91–94.
Wellsandt, S., Werthmann, D., Hribernik, K., Thoben, K.-
D., 2013. Ubiquitous logistics: a business and
technology concept based on shared resources. Dublin,
Ireland.
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
656
