Spatial awareness of autonomous embedded systems

Clemens Holzmann
Spatial Awareness of Autonomous Embedded Systems


VIEWEG+TEUBNER RESEARCH


Clemens Holzmann

Spatial Awareness of
Autonomous Embedded
Systems

VIEWEG+TEUBNER RESEARCH


Bibliographic information published by the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at .

Dissertation Universität Linz, 2008

1st Edition 2009
All rights reserved
© Vieweg+Teubner | GWV Fachverlage GmbH, Wiesbaden 2009
Editorial Office: Christel A. Roß | Anita Wilke
Vieweg+Teubner is part of the specialist publishing group Springer Science+Business Media.
www.viewegteubner.de
No part of this publication may be reproduced, stored in a retrieval system
or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of

the copyright holder.
Registered and/or industrial names, trade names, trade descriptions etc. cited in this publication are part of the law for trade-mark protection and may not be used free in any form or by
any means even if this is not specifically marked.
Cover design: KünkelLopka Medienentwicklung, Heidelberg
Printing company: STRAUSS GMBH, Mörlenbach
Printed on acid-free paper
Printed in Germany
ISBN 978-3-8348-0798-4


To my mother and father,
who have always believed in me and supported me all the way.
Thank you.


Acknowledgements
I would like to especially thank the following people who contributed to this thesis
in various ways and without whose support it would not have been possible to
finish it in such a short time and with the present extent and quality:
• My supervisor Univ.-Prof. Mag. Dr. Alois Ferscha, for giving me the opportunity to work at his department and supporting me in every way with this
dissertation. He offered me an interesting topic to work on, provided generous financial support with regard to the professional equipment I could use
in my research as well as for being able to publish and present results at
international conferences, he was always available for discussions, helped
with useful suggestions to keep the thesis on track and in time, and allowed
me to realize and develop my ideas in an industrial research project with
Siemens AG Germany. Last but not least, he gave me all the time I needed
when actually writing the thesis.
• Dipl.-Ing. Manfred Hechinger, who implemented major parts of the software
framework presented in Chapter 6 as well as the vibro-tactile space awareness application presented in Section 7.2. Also the framework architecture
and many of the related concepts were developed in close cooperation with

him.
• Univ.-Prof. Mag. Dr. Gabriele Kotsis, for reading this thesis as the second
supervisor and being a great help particularly in its initial stage by providing
her expertise with regard to technical issues and writing a dissertation in
general.
• The project partners of Siemens AG Germany, and in particular Dipl.-Inform.
Dr. Andreas Zeidler, for their critical and useful feedback on parts of my
work as well as for sharing their knowledge on real-world requirements.
• The co-authors and reviewers of my publications, who provided helpful suggestions for their improvement and thus also contributed to a higher quality
of the respective sections in this thesis. I would like to especially thank
the reviewers of my submission to the Ubicomp 2005 doctoral colloquium,


VIII

Acknowledgements

who suggested to extend my sole focus on the spatial direction of artifacts
at that time to different types of spatial relationships and anchor the work in
real-world application domains.
• Dominik Hochreiter and Dipl.-Ing. (FH) Bernadette Emsenhuber for their
work on realizing the hardware of the vibro-tactile belt (cf. Section 7.2.1),
and in this regard Dipl.-Ing. Andreas Riener for bringing in his expertise on
haptic perception.
• My partner Mag. Anja Razenböck, for supporting me emotionally and showing understanding for the many evenings and weekends I was working on
this thesis.
• All the colleagues I have come to know since my employment at the Johannes Kepler University Linz, who contributed to an inspirational and pleasant working environment and made the time at the university enjoyable and
fruitful.

Dipl.-Ing. Mag. Clemens Holzmann

Abstract
The invisible integration of technology into everyday objects like home appliances,
cars and mobile phones, which is the declared vision and fundament of research in
the field of pervasive computing, leads to huge quantities of smart objects which
are situated in the surrounding physical space. Equipped with embedded systems
technology, they become increasingly heterogeneous and interconnected, raising
the challenge of a semantically meaningful interplay with each other. In this thesis,
such physical objects with embedded computing and communication technology
are referred to as digital artifacts. One approach to meet this challenge is to design
and implement systems that are able to operate autonomously in the background,
namely with as little human intervention as possible, and interact with humans in a
more unobtrusive way. In order to achieve autonomy, two aspects are particularly
important: (i) context-awareness, which refers to the ability of a digital artifact
to acquire environmental information in order to become aware of its situation
and adapt to changing situations at runtime, and (ii) context sharing, which relates
to an artifact’s ability of exchanging context information with other artifacts in
communication range.
An essential part of the context of spatially distributed objects is their position, direction and spatial extension with respect to an external reference system
or with respect to other objects. The focus of this thesis is to make digital artifacts
aware of such context information in order to enable their autonomous adaptation
to spatial changes in the environment, whereas our main interest is on qualitatively
abstracted spatial relations and their use at the application level. Concepts for
recognizing, representing and reasoning about qualitative spatial relations among
autonomous artifacts and their changes over time are presented, as well as an according architecture which has been implemented in a flexible software framework
that builds upon qualitative relationship abstractions and the rule-based inference
of conclusions from them. Special attention is paid to the adaptivity to different
application domains, which influence the semantics of spatial abstractions and thus
the behavior of digital artifacts. Evaluation results show the feasibility of the proposed concepts for developing spatially aware applications which involve one or

more spontaneously interacting digital artifacts, and in particular that qualitatively
abstracted relations can constitute an adequate basis for it.


Contents
1

Introduction
1.1 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2

Spatial Awareness
2.1 Context and Awareness . . . . . . . .
2.2 Autonomous Embedded Systems . . .
2.2.1 Exchange of Self-Descriptions
2.3 Mechanisms of Self-Organization . .
2.3.1 Self-Organization in Space . .
2.4 Spatial Context in Time . . . . . . . .
2.5 An Architecture for Spatial Awareness
2.6 Related Work . . . . . . . . . . . . .
2.6.1 Projects . . . . . . . . . . . .
2.6.2 Comparison . . . . . . . . . .

3

1
4

5
7

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

9
9
12
15
16
19
20
23
26
27
41

Representation of Space

3.1 Quantitative Representation . . . . . . . . . . . . . . . . . .
3.1.1 Spatial Abstraction with Zones-of-Influence . . . . .
3.1.2 Reference Systems for Position and Direction . . . .
3.2 Qualitative Representation . . . . . . . . . . . . . . . . . .
3.2.1 Qualitative Abstractions of Space . . . . . . . . . .
3.2.2 Recognition and Representation of Spatial Relations
3.2.3 Static and Dynamic Spatial Relations . . . . . . . .
3.2.4 Frames of Reference . . . . . . . . . . . . . . . . .
3.2.5 Spatiotemporal Relations . . . . . . . . . . . . . . .
3.3 Structure of Self-Descriptions . . . . . . . . . . . . . . . .
3.3.1 Quantitative Spatial Properties . . . . . . . . . . . .
3.3.2 Qualitative Spatial Relations . . . . . . . . . . . . .
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

45
46
46
49

51
51
53
56
60
61
66
67
70
72

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.


XII

4

5

6

7

Contents

Distributed Spatial Reasoning
4.1 Overview of Qualitative Approaches . . . . . . . . . . . . . .
4.1.1 Properties and Closures of Binary Relations . . . . . .
4.1.2 Compositional Reasoning . . . . . . . . . . . . . . .
4.2 Reasoning about Positional and Directional Relations . . . . .

4.2.1 Related Approaches . . . . . . . . . . . . . . . . . .
4.2.2 Composition of Positional Relations . . . . . . . . . .
4.3 Spatial Relationship Inference and Distribution . . . . . . . .
4.3.1 General Concept by Exploiting Relation Properties . .
4.3.2 Distribution Algorithm Using Compositional Inference
4.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . .
4.5 Findings and Discussion . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

75
. 75
. 77
. 79
. 80
. 81
. 82
. 87
. 87
. 92

. 96
. 103

Rule-Based Spatial Awareness
5.1 Achieving Spatially Aware Behavior . . . . . . . . . . . .
5.1.1 Reasoning with Rules . . . . . . . . . . . . . . .
5.1.2 Using a Rule Engine . . . . . . . . . . . . . . . .
5.2 Rule-Based Qualitative Spatial Reasoning . . . . . . . . .
5.2.1 Inferring Relations and Application-Level Actions
5.3 Proof of Concept . . . . . . . . . . . . . . . . . . . . . .
5.4 Summary and Open Issues . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.

.
.
.
.
.

.
.
.
.
.
.
.

105
106
106
108
109
112
115
119

Zones-of-Influence Framework
6.1 Architecture . . . . . . . . . . . .
6.1.1 Design Considerations . .
6.1.2 Architecture Overview . .
6.1.3 Runtime Platform . . . . .
6.2 Components . . . . . . . . . . . .
6.2.1 Digital Artifact Service . .

6.2.2 Zones-of-Influence Service
6.2.3 Relations Service . . . . .
6.3 Runtime Behavior . . . . . . . . .
6.4 Discussion . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

121
122
122
123
126
127
127
130
135
140
143

Framework Evaluation

7.1 Development of Spatially Aware Applications
7.2 Application Scenarios . . . . . . . . . . . . .
7.2.1 Vibro-Tactile Space Awareness . . . .
7.2.2 Focus/Nimbus Awareness . . . . . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

147
147
149
151
158

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

Contents

7.3
8

XIII

7.2.3 Spatiotemporal Awareness . . . . . . . . . . . . . . . . . 163
Comparison and Discussion . . . . . . . . . . . . . . . . . . . . 169

Conclusion
177
8.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
8.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185

Bibliography

191


List of Figures
2.1
2.2
2.3
2.4
2.5

Components of a digital artifact. . . . . . . . . .

Exchange and comparison of self-descriptions. .
Emerging “glider” pattern in “Game of Life”. . .
Static spatial properties and relations of artifacts.
General architecture for spatial awareness. . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.

.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

Zone-of-Influence for a car’s braking distance [FHR+ 08]. . . . . .

Types of Zones-of-Influence in 2-D. . . . . . . . . . . . . . . . .
WGS84 spherical (ϕ,λ ,h) and Cartesian (x,y,z) coordinates. . . . .
Recognition of spatial relations by comparing self-descriptions. . .
Orientation relations of nearby extended objects, after [Her94]. . .
Definition of static qualitative spatial relations in 2-D. . . . . . . .
Cone-/projection-based orientation and qualitative distance [CFH97,
RM04]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.8 Definition of dynamic qualitative spatial relations in 2-D. . . . . .
3.9 Epsilon-neighborhood of a static qualitative orientation relation. .
3.10 Intrinsic and extrinsic orientation relations. . . . . . . . . . . . .
3.11 Qualitative temporal interval relations, cf. [All83]. . . . . . . . . .
3.12 Qualitative spatiotemporal relations. . . . . . . . . . . . . . . . .
3.1
3.2
3.3
3.4
3.5
3.6
3.7

4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9

14
16
18
21
24
47
48
50
54
55
56
57
58
59
60
63
65

An example for compositional reasoning. . . . . . . . . . . . . . 80
Composition of positional relations with same (top) and similar
directions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Distribution of the transitive relation left. . . . . . . . . . . . . . . 90
Distribution of the symmetric and reflexive relation near. . . . . . 91
Recognition, composition and intersection of qualitative positional
relations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Compositional inference and distribution started at artifact b. . . . 98
Simulated topologies full binary tree, line, mesh and ring. . . . . . 101
Traffic induced by different algorithms depending on the topology. 102
Relative accuracy of different algorithms compared with flooding. 103

XVI

List of Figures

5.1
5.2
5.3
5.4

Rule-based reasoning [PNF+ 08]. . . . . . . . . . . . . . . . .
Acquisition of position and direction from two IS-900 trackers.
Trajectories of quantitative spatial relations. . . . . . . . . . .
Recognized and inferred qualitative relations. . . . . . . . . .

.
.
.
.

108
116
117
119

6.1
6.2
6.3
6.4
6.5

6.6
6.7
6.8

Overview of the Zones-of-Influence Framework architecture. . . .
Layered architecture of the OSGi framework, after [All05, All07].
Interfaces of the Digital Artifact Service. . . . . . . . . . . . . . .
Interfaces of the Zones-of-Influence Service. . . . . . . . . . . . .
Interfaces and classes for representing Zones-of-Influence. . . . .
Interfaces and classes of the Relations Service. . . . . . . . . . .
Time for relationship recognition (top) and repository update. . . .
Repository update time (top) and involved number of relations. . .

124
127
129
131
133
137
142
143

7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8

7.9
7.10

Vibro-tactile belt with eight vibrator elements. . . . . . . . . .
Controller of the belt. . . . . . . . . . . . . . . . . . . . . . .
Person wearing the belt. . . . . . . . . . . . . . . . . . . . . .
Visualization of the 2-D scene and the belt’s vibrator elements.
Exemplary application scenario for focus/nimbus awareness. .
Visualization and relations of the focus/nimbus application. . .
Qualitative spatial relations at a point in time. . . . . . . . . .
Time series of qualitative spatial relations. . . . . . . . . . . .
Visualization and relations of the scenario shown in Figure 7.7.
Visualization and relations of the scenario shown in Figure 7.8.

157
157
157
158
160
162
164
165
168
170

.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.


List of Tables
2.1
2.2

Static characteristics of an artifact’s spatial situation. . . . . . . .
Comparison of related work. . . . . . . . . . . . . . . . . . . . .

21
43

4.1
4.2

78

4.3
4.4
4.5
4.6

Extrinsic orientation and distance relations with their properties. .
Comparison of approaches for reasoning about static spatial relations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Composition table for static orientation relations. . . . . . . . . .
Composition table for static distance relations. . . . . . . . . . . .
Composition table for combined orientation and distance. . . . . .
Simulation results for the topology of Figure 4.6. . . . . . . . . .

5.1

Combinations of qualitative spatial and temporal relations. . . . . 113

6.1

Time in [s] for adding new artifacts/updating their self-descriptions. 141

7.1
7.2

Relevancy of spatial abstractions for application scenarios. . . . . 150
Framework aspects covered by implemented spatially aware applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

81
85
85
86
99


Listings
3.1
3.2
3.3
3.4

Structure of an XML-based self-description. . . . . . . . . .
Representation of a static and fragmented Zone-of-Influence.
Representation of a dynamic Zone-of-Influence. . . . . . . .
Representation of a qualitative relation. . . . . . . . . . . .

.
.
.
.

.
.
.

.

5.1
5.2
5.3

Maintenance rule for merging equal relations. . . . . . . . . . . . 111
Inference rule for the relation passing-by-right. . . . . . . . . . . 114
Inference rule for the relation passing-by-right-in-vicinity. . . . . 115

6.1

Specification of sensor data within a self-description. . . . . . . . 134

7.1
7.2
7.3
7.4
7.5

Zone-of-Influence for the vibro-tactile belt’s awareness shape.
Rule for the vibro-tactile space awareness application. . . . . .
Rule for the focus/nimbus awareness application. . . . . . . .
Rules for the scenario shown in Figure 7.7. . . . . . . . . . .
Rules for the scenario shown in Figure 7.8. . . . . . . . . . .

.
.
.
.

.

.
.
.
.

.
.
.
.
.

67
69
70
72

153
155
161
167
168


List of Abbreviations
API
CSCW
CSV
DOF

ECA
ECEF
EPC
FoV
GIS
GML
GPS
GSM
GUI
HTTP
IEEE
IDE
IR
JEPD
KML
MEMS
NED
OSI
OSGi
PDA
PML
P2P
RCC
SQL
TOTA
URL
Wi-Fi

Application Programming Interface
Computer Supported Cooperative Work

Comma Separated Value
Degrees of Freedom
Event-Condition-Action
Earth-Centered Earth-Fixed
Electronic Product Code
Field of View
Geographic Information Systems
Geography Markup Language
Global Positioning System
Global System for Mobile Communications
Graphical User Interface
Hypertext Transfer Protocol
Institute of Electrical and Electronics Engineers
Integrated Development Environment
Infrared
Jointly Exhaustive and Pairwise Disjoint
Keyhole Markup Language
Micro-Electro-Mechanical Systems
North-East-Down
Open Systems Interconnection
Open Services Gateway initiative
Personal Digital Assistant
Physical Markup Language
Peer-to-Peer
Region Connection Calculus
Structured Query Language
Tuples on the Air
Uniform Resource Locator
Wireless Fidelity

XXII

WGS84
XML
ZoI

List of Abbreviations

World Geodetic System 1984
Extensible Markup Language
Zone-of-Influence


1 Introduction
Time and space are modes by which we
think and not conditions in which we live.
Albert Einstein, 1879-1955

At the time of this writing, we are facing a world with a huge, and ever increasing number of real-world objects with embedded computing and communication
capabilities. This development is mainly driven by technological advances within
the past few decades, which have made it possible to shrink sensors and actuators as well as processing and wireless communication technologies to a size that
enables their integration into virtually everything. Things of everyday use in industrial settings, in manufacturing, in offices or in homes, like tools, appliances, machinery, furniture or even clothing have become the constituent “devices”, through
which applications interact with the user, and there is strong evidence that this
trend will continue. It can be observed that the “computer” is no longer a single
desktop device, but is rather associated with services originating in the “digital
world” and perceived through the “physical world”. The computer is more and
more hidden in the fabric of everyday life, invisibly networked, and accessible
everywhere; thus, we could say that computers become invisible, while their interfaces become omnipresent.
A field of research tackling related challenges is ubiquitous computing, which

was coined by Mark Weiser in a Scientific American article in 1991 [Wei91].
The vision of ubiquitous computing is the fusion of computing and communication technologies with the environment in such a way that the technology disappears. There are several terms such as calm computing, invisible computing,
hidden computing, ambient intelligence, autonomous computing or pervasive computing, which all refer to the same vision but with a different focus at a time; in the
context of this thesis we use the term pervasive computing to stress the pervasion
of real-world objects with computing technology.
Two key objectives of ubiquitous computing are identified in [Wei91]: ubiquity,
namely the availability of computation and communication anytime and everywhere, and invisibility, which refers to the disappearance of computing technology. Invisibility can be achieved by embedding computational elements into real-


2

Introduction

world objects, but also by their autonomous operation [ECPS02] which makes
them – due to the reduced human involvement – more or less disappear from
the users’ consciousness. Pervasive computing is an inherently multidisciplinary
topic, as many research fields including embedded systems, distributed systems,
mobile computing, human/computer interaction and artificial intelligence [Sat01]
address specific aspects for making the vision of “computing beyond the desktop”
become reality.
A typical service may consist of a large number of sub-services that themselves
can be deployed on various devices at different – possibly changing – physical
locations; the range of devices contributing to the service and their available interaction modalities may differ widely, as well. Obviously, the configuration and
management of such dynamic “service landscapes” should be automated to a huge
extent and not be of concern for the human user. This calls for some means of
self-management and -organization, as their number, heterogeneity and complexity will sooner or later exceed the limits of human capability [HG03].
Moreover, interaction within such service environments will have to be more
implicit (i.e. at the periphery of human attention) rather than explicit (i.e. at the
focus of attention), so that it becomes very important that the individual devices
manage themselves autonomously, with as little human intervention as possible.

Clearly, the traditional approach of instructive systems [Wan04] with their deterministic and context-free nature appears less appropriate as an architecture for providing services by the interaction of embedded and networked devices; instead, a
more autonomous system architecture [Hor01, KC03] is required. According to
[FHdSR+ 07], two aspects of system properties contributing to their ability to operate autonomously in the background can be identified: self-management and
self-organization.
First, self-management relates to the individual devices. It stands for the ability
of a single device to acquire information that can help to understand its situation or
context, and to automatically adapt to changing contexts at runtime in a semantically meaningful way. There has been much research on this topic [CK00], which
was mainly driven by technological advances in miniaturization of sensors and actuators for acquiring information about an environment and influencing it. One of
the first definitions of context with regard to pervasive computing systems is given
in [SAW94], where context is considered as the location of use, the collection of
nearby people, hosts, and accessible devices, as well as the changes of such things
over time. A more formal definition is given in [Dey00], defining context as any
information than can be used to characterize the situation of an entity and denoting a system context-aware if it uses context to provide relevant information and/or
services to the user. The second system property is self-organization, which re-


Introduction

3

lates to spontaneous configurations of such devices, and it stands for their ability to
spontaneously (e.g. upon service requests or detection of certain spatial situations)
join into ad-hoc “service ensembles” to e.g. negotiate and achieve ensemble goals
– as for example the provision of a certain service to the user – through coordinated
actions.
Both self-management and self-organization have received much research attention in computer science over the past years [MMTZ06, SFH+ 03]. In particular,
self-organization principles as inspired by nature attracted the attention of computer scientists [KE01, ZGMT04]. In the respective literature, self-organization
is defined as a process in which pattern at the global level of a system emerge
solely from numerous interactions among lower-level components of the system
[CFS+ 01], where “pattern” refers to structure and organization in both space and

time. Self-organization hence is way beyond centralized coordination, and complex collective behavior results from contextual local interactions between components [SFH+ 03]. Local interactions in turn are based on individual goals and
the perception of the respective environment. The essence of self-organization is
that system structure – and thus collective behavior – often appears without explicit trigger or pressure from outside the system, but is immanent to the system
and results from interactions within it. System structure can evolve in time and
space, may maintain a stable form or exhibit transient phenomena, or may grow or
shrink in size, number or feature. An example for that is John Conway’s cellular
automaton called “Game of Life”; it consists of a two-dimensional collection of
cells which die, survive or become to live by simple local rules, whereas complex
patterns emerge depending on an initial configuration of the automaton [Gar70].
The focus in this work is on autonomously operating real-world objects which
are equipped with sensors, actuators, as well as with computing and wireless communication technology to support ad-hoc networking. These devices can have
various different kinds of appearance (like shape, size, mobility, etc.) and embedded digital technology (e.g. mobile phones, smart appliances, smart rooms, etc.),
and we refer to them as digital artifacts. A digital artifact has to have the ability
to sense its context, and it must possess reasonable means to process and reason
about the perceived context as well as the possibility to share its perceptions with
others in range. Obviously, this is a fundamental requirement for such autonomous
embedded systems, as sharing information about their context is a key property for
collaboration and autonomous adaptation in order to reach ensemble goals based
on local information gathering.
As digital artifacts are by nature situated in physical space, their spatial properties (e.g. position and physical shape) – and in particular spatial relationships
between them (e.g. distance and orientation) – are valuable context information


4

Introduction

and constitute a distinguishing feature of our work. Actually, most of the known
phenomena of self-organization and -adaptation in nature are phenomena of selforganization in space [MZ05], and [ZM04] identifies the concept of space and the
awareness of distributed components of their surrounding to play an important role

for mechanisms of self-organization, as for example the coordination of activities
by exploiting their spatial structure. In this regard, spatial abstractions are considered to be important means for implementing services which are distributed in
physical space [Leo98, ZM04].
Self-organization is based on (direct or indirect) contextual local interactions
between the components of a system; for this reason, both the inference of highlevel contextual information from spatial relationships, as well as standardized
means for exchanging spatial information between the components are issues of
research [HE03, MZ03]. This thesis is primarily concerned with techniques for
making digital artifacts spatially aware (i.e. aware of spatial context information),
which facilitates the semantically meaningful interaction among them.

1.1 Problem Statement
The focus of this thesis is on mechanisms to provide digital artifacts, which are
spatially distributed real-world objects with embedded digital technology, with
an awareness about their spatial contexts. In this regard, an essential part of the
information describing the spatial context of an artifact is its position, direction
and spatial extension. In order to achieve such spatial awareness, it must have
the ability to acquire its spatial context with sensors, possess means to reason
about the perceived context and share this perception with others in communication range. Sharing spatial context information is a key property of autonomous
artifacts, which allows them to adapt to changing spatial situations at runtime and
thus contributes to their semantically meaningful, contextual interaction in space.
As artifacts may have to coordinate their actions for providing certain services, we
consider especially spatial relations among them as well as relationship changes
to be of particular relevance.
When developing context-aware applications, a tight coupling between the application and sensors is problematic due to the fact that it forces the programmer
to deal with sensor details. Hence, low-level context information provided by
sensors must be abstracted to be used by context-aware applications [Dey00]; for
example, an application using location information may only be interested in highlevel information like rooms and buildings instead of geographical coordinates
[FHO04a]. Generally speaking, abstracting context information is about separat-

1.2 Contribution

5

ing details which are not relevant for a certain application. Such symbolic abstractions of locations have been addressed by several researchers [Leo98, Sch95], with
the aim to provide location information in a sensor-independent and more natural
way that is closer to human concepts of space. Abstracted relations between the
locations of devices are used in [HKG+ 05, KKG05], which are represented with
meaningful names such as left or near.
In this work, we go beyond utilizing just location information; in addition, the
direction and spatial extension of mobile artifacts with respect to a global reference
system and with respect to each other are taken into account. Our primary focus
is on the investigation and development of concepts for recognizing and representing spatial relations between autonomous digital artifacts by using qualitative
abstractions, as well as the inference of high-level context information out of it.
The representation of and reasoning about qualitative spatial relations, which are
concerned with the abstraction of continuous properties of the physical world and
inferring knowledge from the respective qualitative representations, has already
attracted much interest by researchers [CH01]. Compared with quantitative approaches, qualitative ones have clear advantages whenever the spatial cognition
of humans is involved [Mus00] or systems with limited computing resources are
concerned [Fre92b], among others.
Summing up, the focus of this thesis can be stated as follows:
What are concepts and architectures for making autonomous embedded systems aware of qualitatively abstracted spatial relations over
time and using them in spatially aware applications?
The thesis, however, does not address spatial representation and reasoning in
general, but rather concentrates on methods for providing qualitative spatial relationship information to autonomous digital artifacts at application level. Central is
the development of an architecture for enabling spatial awareness of autonomous
artifacts without any kind of centralized instance, as well as its application to
real-world scenarios. This includes the issues of recognizing, representing and
reasoning about qualitative spatial relations both at certain points in time and by
considering their changes over time.

1.2 Contribution
Within the scope of this thesis, application-independent concepts that allow autonomously operating digital artifacts to become aware of and use spatial context


6

Introduction

information have been investigated and developed. In order to implement spatial awareness of artifacts among each other, quantitative and qualitative representations of spatial aspects are used. For the former, a concept referred to as
Zones-of-Influence has been developed, which builds upon initial work published
in [FHR+ 08]. Zones-of-Influence encode spatial properties of artifacts – in particular their absolute and relative position, direction and spatial extension – with
numerical values that may be provided by sensors. They are explicitly defined twoor three-dimensional shapes of a certain size which are positioned and directed in
physical space, thus representing geographic regions that are of relevance for the
artifacts and their users or applications. Each digital artifact is associated with one
or more of such zones at a time, together representing the relevant spatial knowledge which is distributed across them in physical space. In order to enable artifacts to autonomously recognize spatial relations to others around, they share their
Zones-of-Influence through an exchange of generic, structured self-descriptions.
This allows for determining spatial relations between their Zones-of-Influence, including distance, orientation or topological relations between the zones’ spatial
extensions. In this regard, a data format for representing both spatial properties
and determined relations within self-descriptions has been developed.
A central point is the abstraction of spatial relations in a qualitative way, depending on the application domain which influences their semantics. As the relations an
artifact is aware of can also be included in its self-description and exchanged with
others, digital artifacts are able to reason about both self-determined and received
relationships. A literature survey has been conducted in order to find suitable qualitative abstractions of relations between Zones-of-Influence on the one hand, and
to identify approaches for inferring new relations on the other hand. We have developed a spatial calculus for compositional reasoning about positional relations
as well as an algorithm which makes use of it for inferring and distributing spatial
relations among autonomous digital artifacts over multiple hops, without the need
for exchanging quantitative spatial properties between those that are out of communication range. In this regard, effects on the induced network traffic as well as
the achieved accuracy have been evaluated by simulation means.
A reference architecture for the spatial awareness of digital artifacts has been developed, demonstrating how the above concepts for recognizing, representing and

reasoning about spatial relations among spontaneously interacting artifacts can be
realized. A middleware framework has been prototypically implemented, with the
aim to support the development of spatially aware applications for autonomous
digital artifacts. For reasoning about spatial relations over time, in order to infer
new relations or trigger appropriate application-level actions, this so-called Zonesof-Influence Framework makes use of rules. All relations are associated with time


1.3 Thesis Outline

7

intervals in which they exist, and new high-level relations are inferred by combining such intervals, for which reason inferred relations can in turn be combined with
others. The qualitative relationship abstractions are encapsulated in components
that can easily be exchanged at runtime, enabling artifacts to adapt the availability
of spatial relations and their semantics to the current application demands. The
proposed framework with its underlying concepts provides powerful means for facilitating the development of applications which take into account the contextual
relations among autonomous digital artifacts. An evaluation with multiple application scenarios showing the relevance of spatial relations for developing spatially
aware applications as well as the technical feasibility and quality of the Zones-ofInfluence Framework has been conducted.

1.3 Thesis Outline
The thesis is structured as follows. Chapter 2 gives an introduction in contextawareness of autonomous embedded systems, with which physical objects are augmented and to which we refer to as digital artifacts then, motivates its significance
for mechanisms of self-organization, discusses the role of spatial context with regard to their autonomy and semantically meaningful interaction among each other,
and presents a general architecture for providing spatial awareness to artifacts. At
the end of this chapter, a comprehensive survey and comparison of related work
are given. Chapter 3 elaborates on quantitative and qualitative representations of
spatial aspects, which are the basis for representing spatial properties of artifacts
with so-called Zones-of-Influence as well as for recognizing and using spatial relations between them. In this regard, the exchange of such spatial contexts among
artifacts with structured self-descriptions is discussed. Chapter 4 builds upon the
qualitative abstractions of spatial relations and is concerned with the inference
of knowledge from them. In this chapter, fundamentals of qualitative reasoning

approaches are presented, and an algorithm for inferring and distributing qualitative spatial relationships among digital artifacts is proposed and evaluated. In
Chapter 5, a rule-based approach for reasoning about qualitative spatial relations
over time is presented, which is used for inferring new relations and triggering
application-level actions upon observing certain patterns on the stored relations.
The above concepts of exploiting spatial abstractions have been implemented in
the flexible and modular Zones-of-Influence Framework presented in Chapter 6,
which enables digital artifacts to maintain and use a spatial model of their environment. An overview of the architecture is given in this chapter, its components for
supporting spatial awareness are described and implementation details as well as