Spatial awareness of autonomous embedded systems

Concepts forrecognizing, representing and reasoning about qualitative spatial relations amongautonomous artifacts and their changes over time are presented, as well as an ac-cording arch

Spatial Awareness of Autonomous Embedded Systems

Spatial Awareness of

Autonomous Embedded Systems

VIEWEG+TEUBNER RESEARCH

detailed bibliographic data are available in the Internet at http://dnb.d-nb.de

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, copying, recording, or otherwise, without the prior written permission ofthe copyright holder

photo-Registered and/or industrial names, trade names, trade descriptions etc cited in this tion are part of the law for trade-mark protection and may not be used free in any form or byany means even if this is not specifically marked

publica-Cover design: KünkelLopka Medienentwicklung, Heidelberg

Printing company: STRAUSS GMBH, Mörlenbach

Printed on acid-free paper

Printed in Germany

ISBN 978-3-8348-0798-4

Thank you.

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 tofinish 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 tunity to work at his department and supporting me in every way with thisdissertation He offered me an interesting topic to work on, provided gener-ous financial support with regard to the professional equipment I could use

oppor-in my research as well as for beoppor-ing able to publish and present results atinternational conferences, he was always available for discussions, helpedwith 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 withSiemens AG Germany Last but not least, he gave me all the time I neededwhen actually writing the thesis

• Dipl.-Ing Manfred Hechinger, who implemented major parts of the softwareframework presented in Chapter 6 as well as the vibro-tactile space aware-ness application presented in Section 7.2 Also the framework architectureand many of the related concepts were developed in close cooperation withhim

• Univ.-Prof Mag Dr Gabriele Kotsis, for reading this thesis as the secondsupervisor and being a great help particularly in its initial stage by providingher expertise with regard to technical issues and writing a dissertation ingeneral

• The project partners of Siemens AG Germany, and in particular Dipl.-Inform

Dr Andreas Zeidler, for their critical and useful feedback on parts of mywork as well as for sharing their knowledge on real-world requirements

• The co-authors and reviewers of my publications, who provided helpful gestions for their improvement and thus also contributed to a higher quality

sug-of the respective sections in this thesis I would like to especially thankthe reviewers of my submission to the Ubicomp 2005 doctoral colloquium,

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 inreal-world application domains

• Dominik Hochreiter and Dipl.-Ing (FH) Bernadette Emsenhuber for theirwork 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 onhaptic perception

• My partner Mag Anja Razenböck, for supporting me emotionally and ing understanding for the many evenings and weekends I was working onthis thesis

show-• All the colleagues I have come to know since my employment at the hannes Kepler University Linz, who contributed to an inspirational and pleas-ant working environment and made the time at the university enjoyable andfruitful

Jo-Dipl.-Ing Mag Clemens Holzmann

The invisible integration of technology into everyday objects like home appliances,cars and mobile phones, which is the declared vision and fundament of research inthe field of pervasive computing, leads to huge quantities of smart objects whichare situated in the surrounding physical space Equipped with embedded systemstechnology, they become increasingly heterogeneous and interconnected, raisingthe challenge of a semantically meaningful interplay with each other In this thesis,such physical objects with embedded computing and communication technologyare referred to as digital artifacts One approach to meet this challenge is to designand implement systems that are able to operate autonomously in the background,namely with as little human intervention as possible, and interact with humans in amore unobtrusive way In order to achieve autonomy, two aspects are particularlyimportant: (i) context-awareness, which refers to the ability of a digital artifact

to acquire environmental information in order to become aware of its situationand adapt to changing situations at runtime, and (ii) context sharing, which relates

to an artifact’s ability of exchanging context information with other artifacts incommunication range

An essential part of the context of spatially distributed objects is their tion, direction and spatial extension with respect to an external reference system

posi-or with respect to other objects The focus of this thesis is to make digital artifactsaware of such context information in order to enable their autonomous adaptation

to spatial changes in the environment, whereas our main interest is on qualitativelyabstracted spatial relations and their use at the application level Concepts forrecognizing, representing and reasoning about qualitative spatial relations amongautonomous artifacts and their changes over time are presented, as well as an ac-cording architecture which has been implemented in a flexible software frameworkthat builds upon qualitative relationship abstractions and the rule-based inference

of conclusions from them Special attention is paid to the adaptivity to differentapplication domains, which influence the semantics of spatial abstractions and thusthe behavior of digital artifacts Evaluation results show the feasibility of the pro-posed concepts for developing spatially aware applications which involve one ormore spontaneously interacting digital artifacts, and in particular that qualitativelyabstracted relations can constitute an adequate basis for it

1 Introduction 1

1.1 Problem Statement 4

1.2 Contribution 5

1.3 Thesis Outline 7

2 Spatial Awareness 9 2.1 Context and Awareness 9

2.2 Autonomous Embedded Systems 12

2.2.1 Exchange of Self-Descriptions 15

2.3 Mechanisms of Self-Organization 16

2.3.1 Self-Organization in Space 19

2.4 Spatial Context in Time 20

2.5 An Architecture for Spatial Awareness 23

2.6 Related Work 26

2.6.1 Projects 27

2.6.2 Comparison 41

3 Representation of Space 45 3.1 Quantitative Representation 46

3.1.1 Spatial Abstraction with Zones-of-Influence 46

3.1.2 Reference Systems for Position and Direction 49

3.2 Qualitative Representation 51

3.2.1 Qualitative Abstractions of Space 51

3.2.2 Recognition and Representation of Spatial Relations 53

3.2.3 Static and Dynamic Spatial Relations 56

3.2.4 Frames of Reference 60

3.2.5 Spatiotemporal Relations 61

3.3 Structure of Self-Descriptions 66

3.3.1 Quantitative Spatial Properties 67

3.3.2 Qualitative Spatial Relations 70

3.4 Summary 72

4 Distributed Spatial Reasoning 75

4.1 Overview of Qualitative Approaches 75

4.1.1 Properties and Closures of Binary Relations 77

4.1.2 Compositional Reasoning 79

4.2 Reasoning about Positional and Directional Relations 80

4.2.1 Related Approaches 81

4.2.2 Composition of Positional Relations 82

4.3 Spatial Relationship Inference and Distribution 87

4.3.1 General Concept by Exploiting Relation Properties 87

4.3.2 Distribution Algorithm Using Compositional Inference 92

4.4 Simulation Results 96

4.5 Findings and Discussion 103

5 Rule-Based Spatial Awareness 105 5.1 Achieving Spatially Aware Behavior 106

5.1.1 Reasoning with Rules 106

5.1.2 Using a Rule Engine 108

5.2 Rule-Based Qualitative Spatial Reasoning 109

5.2.1 Inferring Relations and Application-Level Actions 112

5.3 Proof of Concept 115

5.4 Summary and Open Issues 119

6 Zones-of-Influence Framework 121 6.1 Architecture 122

6.1.1 Design Considerations 122

6.1.2 Architecture Overview 123

6.1.3 Runtime Platform 126

6.2 Components 127

6.2.1 Digital Artifact Service 127

6.2.2 Zones-of-Influence Service 130

6.2.3 Relations Service 135

6.3 Runtime Behavior 140

6.4 Discussion 143

7 Framework Evaluation 147 7.1 Development of Spatially Aware Applications 147

7.2 Application Scenarios 149

7.2.1 Vibro-Tactile Space Awareness 151

7.2.2 Focus/Nimbus Awareness 158

7.2.3 Spatiotemporal Awareness 1637.3 Comparison and Discussion 169

8.1 Summary 1798.2 Outlook 185

2.1 Components of a digital artifact 14

2.2 Exchange and comparison of self-descriptions 16

2.3 Emerging “glider” pattern in “Game of Life” 18

2.4 Static spatial properties and relations of artifacts 21

2.5 General architecture for spatial awareness 24

3.1 Zone-of-Influence for a car’s braking distance [FHR+08] 47

3.2 Types of Zones-of-Influence in 2-D 48

3.3 WGS84 spherical (ϕ,λ,h) and Cartesian (x,y,z) coordinates 50

3.4 Recognition of spatial relations by comparing self-descriptions 54

3.5 Orientation relations of nearby extended objects, after [Her94] 55

3.6 Definition of static qualitative spatial relations in 2-D 56

3.7 Cone-/projection-based orientation and qualitative distance [CFH97, RM04] 57

3.8 Definition of dynamic qualitative spatial relations in 2-D 58

3.9 Epsilon-neighborhood of a static qualitative orientation relation 59

3.10 Intrinsic and extrinsic orientation relations 60

3.11 Qualitative temporal interval relations, cf [All83] 63

3.12 Qualitative spatiotemporal relations 65

4.1 An example for compositional reasoning 80

4.2 Composition of positional relations with same (top) and similar directions 84

4.3 Distribution of the transitive relation left . 90

4.4 Distribution of the symmetric and reflexive relation near . 91

4.5 Recognition, composition and intersection of qualitative positional relations 94

4.6 Compositional inference and distribution started at artifact b . 98

4.7 Simulated topologies full binary tree, line, mesh and ring 101 4.8 Traffic induced by different algorithms depending on the topology 102 4.9 Relative accuracy of different algorithms compared with flooding 103

5.1 Rule-based reasoning [PNF+08] 108

5.2 Acquisition of position and direction from two IS-900 trackers 116

5.3 Trajectories of quantitative spatial relations 117

5.4 Recognized and inferred qualitative relations 119

6.1 Overview of the Zones-of-Influence Framework architecture 124

6.2 Layered architecture of the OSGi framework, after [All05, All07] 127 6.3 Interfaces of the Digital Artifact Service 129

6.4 Interfaces of the Zones-of-Influence Service 131

6.5 Interfaces and classes for representing Zones-of-Influence 133

6.6 Interfaces and classes of the Relations Service 137

6.7 Time for relationship recognition (top) and repository update 142

6.8 Repository update time (top) and involved number of relations 143

7.1 Vibro-tactile belt with eight vibrator elements 157

7.2 Controller of the belt 157

7.3 Person wearing the belt 157

7.4 Visualization of the 2-D scene and the belt’s vibrator elements 158

7.5 Exemplary application scenario for focus/nimbus awareness 160

7.6 Visualization and relations of the focus/nimbus application 162

7.7 Qualitative spatial relations at a point in time 164

7.8 Time series of qualitative spatial relations 165

7.9 Visualization and relations of the scenario shown in Figure 7.7 168

7.10 Visualization and relations of the scenario shown in Figure 7.8 170

2.1 Static characteristics of an artifact’s spatial situation 21

2.2 Comparison of related work 43

4.1 Extrinsic orientation and distance relations with their properties 78

4.2 Comparison of approaches for reasoning about static spatial rela-tions 81

4.3 Composition table for static orientation relations 85

4.4 Composition table for static distance relations 85

4.5 Composition table for combined orientation and distance 86

4.6 Simulation results for the topology of Figure 4.6 99

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 Relevancy of spatial abstractions for application scenarios 150

7.2 Framework aspects covered by implemented spatially aware appli-cations 171

3.1 Structure of an XML-based self-description 67

3.2 Representation of a static and fragmented Zone-of-Influence 69

3.3 Representation of a dynamic Zone-of-Influence 70

3.4 Representation of a qualitative relation 72

5.1 Maintenance rule for merging equal relations 111

5.2 Inference rule for the relation passing-by-right 114

5.3 Inference rule for the relation passing-by-right-in-vicinity. 115

6.1 Specification of sensor data within a self-description 134

7.1 Zone-of-Influence for the vibro-tactile belt’s awareness shape 153

7.2 Rule for the vibro-tactile space awareness application 155

7.3 Rule for the focus/nimbus awareness application 161

7.4 Rules for the scenario shown in Figure 7.7 167

7.5 Rules for the scenario shown in Figure 7.8 168

API Application Programming Interface

CSCW Computer Supported Cooperative Work

ECEF Earth-Centered Earth-Fixed

GSM Global System for Mobile Communications

HTTP Hypertext Transfer Protocol

IEEE Institute of Electrical and Electronics EngineersIDE Integrated Development Environment

JEPD Jointly Exhaustive and Pairwise Disjoint

MEMS Micro-Electro-Mechanical Systems

OSGi Open Services Gateway initiative

PDA Personal Digital Assistant

Wi-Fi Wireless Fidelity

WGS84 World Geodetic System 1984

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 ing number of real-world objects with embedded computing and communicationcapabilities This development is mainly driven by technological advances withinthe past few decades, which have made it possible to shrink sensors and actua-tors as well as processing and wireless communication technologies to a size thatenables their integration into virtually everything Things of everyday use in indus-trial settings, in manufacturing, in offices or in homes, like tools, appliances, ma-chinery, furniture or even clothing have become the constituent “devices”, throughwhich applications interact with the user, and there is strong evidence that thistrend will continue It can be observed that the “computer” is no longer a singledesktop device, but is rather associated with services originating in the “digitalworld” and perceived through the “physical world” The computer is more andmore hidden in the fabric of everyday life, invisibly networked, and accessibleeverywhere; thus, we could say that computers become invisible, while their inter-faces become omnipresent

increas-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 communi-cation technologies with the environment in such a way that the technology dis-

appears There are several terms such as calm computing, invisible computing,

hidden computing, ambient intelligence, autonomous computing or pervasive puting, which all refer to the same vision but with a different focus at a time; in the

com-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

every-where, and invisibility, which refers to the disappearance of computing

technol-ogy Invisibility can be achieved by embedding computational elements into

real-world objects, but also by their autonomous operation [ECPS02] which makesthem – due to the reduced human involvement – more or less disappear fromthe 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 – physicallocations; the range of devices contributing to the service and their available in-teraction modalities may differ widely, as well Obviously, the configuration andmanagement of such dynamic “service landscapes” should be automated to a hugeextent and not be of concern for the human user This calls for some means ofself-management and -organization, as their number, heterogeneity and complex-ity 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 ministic and context-free nature appears less appropriate as an architecture for pro-viding services by the interaction of embedded and networked devices; instead, a

deter-more autonomous system architecture [Hor01, KC03] is required According to

[FHdSR+07], two aspects of system properties contributing to their ability to erate autonomously in the background can be identified: self-management andself-organization

op-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

semanti-cally meaningful way There has been much research on this topic [CK00], whichwas mainly driven by technological advances in miniaturization of sensors and ac-tuators for acquiring information about an environment and influencing it One ofthe 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 ofnearby 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

denot-ing 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-lates to spontaneous configurations of such devices, and it stands for their ability tospontaneously (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 coordinatedactions.

Both self-management and self-organization have received much research tion in computer science over the past years [MMTZ06, SFH+03] In particular,self-organization principles as inspired by nature attracted the attention of com-puter scientists [KE01, ZGMT04] In the respective literature, self-organization

atten-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 andtime Self-organization hence is way beyond centralized coordination, and com-

plex collective behavior results from contextual local interactions between

com-ponents [SFH+03] Local interactions in turn are based on individual goals andthe perception of the respective environment The essence of self-organization is

that system structure – and thus collective behavior – often appears without

ex-plicit trigger or pressure from outside the system, but is immanent to the systemand results from interactions within it System structure can evolve in time andspace, may maintain a stable form or exhibit transient phenomena, or may grow orshrink in size, number or feature An example for that is John Conway’s cellularautomaton called “Game of Life”; it consists of a two-dimensional collection ofcells which die, survive or become to live by simple local rules, whereas complexpatterns emerge depending on an initial configuration of the automaton [Gar70].The focus in this work is on autonomously operating real-world objects whichare equipped with sensors, actuators, as well as with computing and wireless com-munication technology to support ad-hoc networking These devices can havevarious different kinds of appearance (like shape, size, mobility, etc.) and embed-ded 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 reasonabout 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

prop-erties (e.g position and physical shape) – and in particular spatial relationships

between them (e.g distance and orientation) – are valuable context information

and constitute a distinguishing feature of our work Actually, most of the knownphenomena of self-organization and -adaptation in nature are phenomena of self-organization in space [MZ05], and [ZM04] identifies the concept of space and theawareness of distributed components of their surrounding to play an important rolefor mechanisms of self-organization, as for example the coordination of activities

by exploiting their spatial structure In this regard, spatial abstractions are

con-sidered to be important means for implementing services which are distributed inphysical space [Leo98, ZM04]

Self-organization is based on (direct or indirect) contextual local interactionsbetween the components of a system; for this reason, both the inference of high-level contextual information from spatial relationships, as well as standardizedmeans for exchanging spatial information between the components are issues ofresearch [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

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 reasonabout the perceived context and share this perception with others in communica-tion range Sharing spatial context information is a key property of autonomousartifacts, which allows them to adapt to changing spatial situations at runtime andthus 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 plication and sensors is problematic due to the fact that it forces the programmer

ap-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 level information like rooms and buildings instead of geographical coordinates[FHO04a] Generally speaking, abstracting context information is about separat-

high-ing details which are not relevant for a certain application Such symbolic tions of locations have been addressed by several researchers [Leo98, Sch95], withthe aim to provide location information in a sensor-independent and more naturalway that is closer to human concepts of space Abstracted relations between thelocations of devices are used in [HKG+05, KKG05], which are represented with

abstrac-meaningful names such as left or near.

In this work, we go beyond utilizing just location information; in addition, thedirection and spatial extension of mobile artifacts with respect to a global referencesystem and with respect to each other are taken into account Our primary focus

is on the investigation and development of concepts for recognizing and

repre-senting 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 areconcerned with the abstraction of continuous properties of the physical world andinferring knowledge from the respective qualitative representations, has alreadyattracted much interest by researchers [CH01] Compared with quantitative ap-proaches, qualitative ones have clear advantages whenever the spatial cognition

of humans is involved [Mus00] or systems with limited computing resources areconcerned [Fre92b], among others

Summing up, the focus of this thesis can be stated as follows:

What are concepts and architectures for making autonomous ded systems aware of qualitatively abstracted spatial relations overtime and using them in spatially aware applications?

embed-The thesis, however, does not address spatial representation and reasoning ingeneral, but rather concentrates on methods for providing qualitative spatial rela-tionship information to autonomous digital artifacts at application level Central isthe development of an architecture for enabling spatial awareness of autonomousartifacts without any kind of centralized instance, as well as its application toreal-world scenarios This includes the issues of recognizing, representing andreasoning about qualitative spatial relations both at certain points in time and byconsidering their changes over time

1.2 Contribution

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

au-information have been investigated and developed In order to implement tial awareness of artifacts among each other, quantitative and qualitative repre-sentations of spatial aspects are used For the former, a concept referred to asZones-of-Influence has been developed, which builds upon initial work published

spa-in [FHR+08] Zones-of-Influence encode spatial properties of artifacts – in ticular their absolute and relative position, direction and spatial extension – withnumerical values that may be provided by sensors They are explicitly defined two-

par-or three-dimensional shapes of a certain size which are positioned and directed inphysical space, thus representing geographic regions that are of relevance for theartifacts 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 edge which is distributed across them in physical space In order to enable arti-facts to autonomously recognize spatial relations to others around, they share theirZones-of-Influence through an exchange of generic, structured self-descriptions.This allows for determining spatial relations between their Zones-of-Influence, in-cluding distance, orientation or topological relations between the zones’ spatialextensions In this regard, a data format for representing both spatial propertiesand determined relations within self-descriptions has been developed

knowl-A central point is the abstraction of spatial relations in a qualitative way, ing on the application domain which influences their semantics As the relations anartifact is aware of can also be included in its self-description and exchanged withothers, digital artifacts are able to reason about both self-determined and receivedrelationships A literature survey has been conducted in order to find suitable qual-itative abstractions of relations between Zones-of-Influence on the one hand, and

depend-to identify approaches for inferring new relations on the other hand We have veloped a spatial calculus for compositional reasoning about positional relations

de-as well de-as an algorithm which makes use of it for inferring and distributing spatialrelations among autonomous digital artifacts over multiple hops, without the needfor exchanging quantitative spatial properties between those that are out of com-munication range In this regard, effects on the induced network traffic as well asthe achieved accuracy have been evaluated by simulation means

A reference architecture for the spatial awareness of digital artifacts has been veloped, demonstrating how the above concepts for recognizing, representing andreasoning about spatial relations among spontaneously interacting artifacts can berealized A middleware framework has been prototypically implemented, with theaim to support the development of spatially aware applications for autonomousdigital artifacts For reasoning about spatial relations over time, in order to infernew relations or trigger appropriate application-level actions, this so-called Zones-of-Influence Framework makes use of rules All relations are associated with time

de-intervals in which they exist, and new high-level relations are inferred by ing such intervals, for which reason inferred relations can in turn be combined withothers The qualitative relationship abstractions are encapsulated in componentsthat can easily be exchanged at runtime, enabling artifacts to adapt the availability

combin-of spatial relations and their semantics to the current application demands Theproposed framework with its underlying concepts provides powerful means for fa-cilitating the development of applications which take into account the contextualrelations among autonomous digital artifacts An evaluation with multiple appli-cation scenarios showing the relevance of spatial relations for developing spatiallyaware applications as well as the technical feasibility and quality of the Zones-of-Influence Framework has been conducted

1.3 Thesis Outline

The thesis is structured as follows Chapter 2 gives an introduction in awareness of autonomous embedded systems, with which physical objects are aug-mented and to which we refer to as digital artifacts then, motivates its significancefor mechanisms of self-organization, discusses the role of spatial context with re-gard to their autonomy and semantically meaningful interaction among each other,and presents a general architecture for providing spatial awareness to artifacts Atthe end of this chapter, a comprehensive survey and comparison of related workare given Chapter 3 elaborates on quantitative and qualitative representations ofspatial aspects, which are the basis for representing spatial properties of artifactswith so-called Zones-of-Influence as well as for recognizing and using spatial re-lations between them In this regard, the exchange of such spatial contexts amongartifacts with structured self-descriptions is discussed Chapter 4 builds upon thequalitative abstractions of spatial relations and is concerned with the inference

context-of knowledge from them In this chapter, fundamentals context-of qualitative reasoningapproaches are presented, and an algorithm for inferring and distributing quali-tative spatial relationships among digital artifacts is proposed and evaluated InChapter 5, a rule-based approach for reasoning about qualitative spatial relationsover time is presented, which is used for inferring new relations and triggeringapplication-level actions upon observing certain patterns on the stored relations.The above concepts of exploiting spatial abstractions have been implemented inthe flexible and modular Zones-of-Influence Framework presented in Chapter 6,which enables digital artifacts to maintain and use a spatial model of their environ-ment An overview of the architecture is given in this chapter, its components forsupporting spatial awareness are described and implementation details as well as

open issues are discussed Chapter 7 eventually evaluates the implemented dleware framework by means of multiple real-world applications scenarios whichcover the concepts presented in the preceding chapters In Chapter 8, the thesis issummarized, the main contributions are discussed and an overview of open issuesfor possible future work is given.

mid-The real power of the concept comes not from any one of these devices; it emerges from the interaction of all of them.

Mark Weiser, 1952-1999

This chapter introduces spatial awareness of autonomous embedded systems,which is the main subject of this work In Section 2.1, we first give an overview oncontext-awareness, which denotes a system’s property to acquire and use informa-tion about its environment, as well as on the abstraction of context for developingcontext-aware applications Afterwards, Section 2.2 presents the components ofautonomous embedded systems which are able to sense their context, process andreason about it and share this knowledge with others in communication range.These abilities contribute to their autonomous operation in the background andthus to a reduction of human involvement for their management and configura-tion Mechanisms of self-management of individual devices and self-organization

of spontaneous configurations of multiple devices are discussed in Section 2.3,which rely on the perception of context information and its local exchange amongdevices in range In this regard, our focus is on the relevance of spatial information,

as we are dealing with real-world objects which are distributed in physical space.The subsequent Section 2.4 thus deals with spatial context information, whereas

a classification is given and our focus on spatial relations as well as their changesover time is emphasized Afterwards, an architecture for spatial awareness is pre-sented in Section 2.5, and Section 2.6 finally provides a comprehensive surveyand comparison of related work addressing the issue of how to make autonomousembedded systems aware about abstracted spatial relations over time

2.1 Context and Awareness

Subject of the field of pervasive computing is the integration of technology intoeveryday objects With computing power invisibly integrated in a huge number

of networked objects it becomes necessary that they operate more autonomously,

as the embedding of digital technology leads to constrained human-computer terfaces on the one hand, and the ubiquity of such technology-enriched physicalobjects makes it impossible for humans to individually manage their operation on

in-the oin-ther hand A contributor to in-the autonomy of embedded systems is

context-awareness [SAW94, Sch95], which refers to the ability of a system to acquire

environmental information in order to understand its situation or context – e.g itslocation or spatial relations to other systems – and adapt to changing contexts atruntime Context-awareness is considered important especially for applicationswhere the environment changes frequently [Dey00, SAW94], which is commonlythe case in mobile and pervasive computing and requires an adaptation to the cur-rent context of use A survey on the foundations of context-aware computing andfirst research projects is given in [CK00]

Several definitions of context and awareness can be found in literature In

the field of computer supported cooperative work (CSCW), [DB92] considers anawareness of the others’ activities necessary for successful collaboration, and de-

be-ing, the more specific term context-awareness refers to a system’s awareness of the context in which it is used It has been introduced in [SAW94], where context-

awareness is defined as a system’s ability to adapt according to

the location of use, the collection of nearby people, hosts, and sible devices, as well as to changes of such things over time.

acces-Three important aspects of context pointed out in [SAW94] are where you are,

who you are with and what resources are nearby A later definition of context is

given in [BBC97], where it is also defined by means of examples, including theuser’s location, the time of the day and the temperature From these definitions,

it can be observed that location and spatial proximity are considered important

context information; in fact, [CK00] states that few contexts other than location

have been used in actual applications Our literature survey conducted in Section

2.6 leads to a similar result, namely that location and proximity are the most ten used contexts in spatially aware applications (i.e applications that are aware

of-of spatial context information, cf Section 2.5) Time is also considered importantfor context-aware applications [CK00, Dey00], and it allows for reasoning about

both contexts and their changes over time However, there are many other types ofcontext information, which [SBG99] classifies in (i) human factors (i.e informa-tion on the user, his social environment and his tasks) and (ii) information on thephysical environment (i.e location, infrastructure and physical conditions).

A more general definition of context can be found in [Dey00]:

Context is any information than can be used to characterize the ation of an entity An entity is a person, place, or object that is con- sidered relevant to the interaction between a user and an application, including the user and application themselves.

situ-According to this – probably most often used – definition, it depends on theparticular application scenario what kinds of information are context in that theyhave an influence on the application We adopt this definition for our work, and

consider context as any information than can be used to characterize the situation

of a person or an object which is considered relevant for an application.

In order to be able to use context information – i.e information about a

sys-tem’s physical environment – in an application, it has to be acquired with sensors

first Research on context-awareness is thus driven by technological advances to

a certain extent, and many projects had a focus on technologies for sensing ious physical contexts like location [WHFG92], direction [SBG99], light, noise,movement, skin conductance and temperature [BGS01, GSB02, HMS+01] or rela-tive orientation and distance [HKG+05] With emerging micro-electro-mechanicalsystems (MEMS) technology, smaller, more powerful and cheaper sensors are be-coming available [GSB02] (e.g MEMS-based accelerometers and gyroscopes forsensing linear and angular acceleration, respectively), which facilitates their inte-gration into everyday objects at low cost If the application has to have an impact

var-on the physical world, actuators are required for changing the envirvar-onment’s state.

An actuator is any output to the physical world (cf [Dey00]), as for example amechanical switch, a light-emitting diode or a computer display According to[Fer03], sensors and actuators can be considered as the input/output interface be-tween the physical and the virtual world

For using sensed context information at the application level however, it must

be abstracted to decouple sensor details that are not relevant for the application

and therewith allow application programmers to use context at a higher level ofabstraction [Dey01, Sch95], namely without having to consider low-level aspects

of certain sensor technologies This is a necessary step as sensors often do notprovide context information in the form required by an application [SDA99] Withregard to the spatial focus of this thesis, sensor data has been abstracted with sym-bolic locations [Leo98] and qualitative spatial relations [KKG05], representing the

respective spatial context at a high level and more naturally by using semanticallymeaningful names As will be seen in Section 2.6, abstractions of sensors areactually employed in most projects which make use of context information.The acquisition, abstraction and use of context information are essential com-ponents of context-aware architectures A more detailed view on a possible archi-tecture of a context-aware system is given in [Fer03, FHO04a], where the follow-

ing layers are in-between those for sensing and actuating: (i) context

transforma-tion (i.e aggregatransforma-tion and interpretatransforma-tion of the low-level contexts into high-level

information which is meaningful for the application), (ii) context representation

(i.e data structures representing the abstract context information of the previous

layer), (iii) context prediction (i.e inference of future from past contexts) and (iv) context triggering (i.e rules for triggering actuators upon observing certain

context conditions) Similar layered architectures are common in context-awaresystems (e.g in [Dey00, GSB02, HKG+05, LM98, SGKK04]), as is the use ofrules for defining context-dependent application behavior (e.g [BCFM03, GSB02,SAW94, SGKK04, TTH+04]) In our framework outlined in Section 2.5 and re-fined later in Chapter 6, sensed context goes through similar layers of abstrac-tion until reaching the application, and a rule-based approach has been chosen fordefining application-level behavior

2.2 Autonomous Embedded Systems

As stated in the previous section, context-awareness is a crucial contributor to theautonomy of embedded systems, which are considered as computer systems thatare parts of larger systems and realize dedicated functions In this section, the

components of autonomous embedded systems are discussed in more detail and

with respect to existing research, as well as their mechanisms for sharing information with other systems in order to allow for spontaneous contextual inter-action with them The following explanations are in large parts limited to aspectswhich are relevant regarding the conceptual and architectural contents in the re-mainder of this thesis

context-Since Mark Weiser’s seminal paper [Wei91], there has been a considerableamount of research on the embedding of computing technology in the physical

environment An example are wireless sensor networks [ASSC02], which

typi-cally consist of a huge number of sensor nodes deployed in an environment inorder to sense certain phenomena The nodes perform local processing on thesensed data and cooperate with other nodes in routing it back to a sink by means

of multi-hop wireless communication Another related field of research is tangible

user interfaces [Fis04, IU97], which deals with user interfaces that augment the real physical world by coupling digital information to everyday physical objects and environments [IU97] We also conducted research in this area by studying the

use of physical objects with embedded direction sensing for the purpose of devicecontrol (cf [FHR06b, FRH06, FRHR05, HRLF06])

The starting point of our work are physical artifacts augmented with autonomous

embedded systems, which – in contrast to the two above examples – comprise

sensing, actuating, computing and wireless communication capabilities, and allowfor the autonomous provision of contextualized services by sharing informationamong each other without relying on an external infrastructure The general aim

of augmenting artifacts that are not computers themselves with computing nology is stated in [BGS01]:

tech-Computer-augmentation of artefacts will not be geared toward ing them more computer-alike but at preserving their individual pur- poses and uses while enabling added value through digital informa- tion processing.

mak-The embedding of computing capabilities in everyday artifacts, and in particulartheir context-awareness and inter-networking capabilities, were subject of manyprojects surveyed in Section 2.6, as for example [BGS01, DBK+04, FHdSR+07,HMS+01, SGKK04]

We adopt the term digital artifact (often simply referred to as artifact in the

following) for computer-augmented physical artifacts, which has been defined in[BGS01] as follows:

A digital artefact is an everyday artefact augmented with computing and communication, enabling it to establish and exchange informa- tion about itself with other digital artefacts and/or computer applica- tions.

The concepts and solutions for spatial awareness developed in this thesis are

based on digital artifacts They consist of the following components also shown in

Figure 2.1, which are either part of the augmented artifact’s original components

or have been added by embedding computing and communication technology:

• Sensors: acquisition of context information from the environment, e.g using

location, motion or optical sensors (cf [SBG99])

• Actuators: output to the physical world, e.g by creating mechanical motion

or with light-emitting diodes

Figure 2.1: Components of a digital artifact.

• Communication facilities: ad-hoc exchange of data with other artifacts and

IT-systems in range, e.g using IEEE 802.11 or IEEE 802.15 wireless munication

com-• Runtime system: representation and reasoning about context information,

which is either acquired with local sensors or received from other artifacts,and triggering respective application actions based on the current context

• External services: provision of contextualized functionality to the artifact’s

surroundings, as for example to users or other artifacts

A digital artifact is typically a mobile device with a small form factor, strained computing capabilities and limited energy resources due to battery opera-tion, it is context-aware because of its sensors and operates autonomously (i.e with-out centralized control), and it acts according to certain defined goals – like forexample the provision of a certain service to the user – which may be achieved incooperation with other artifacts As this thesis is primarily concerned with soft-ware concepts for spatial awareness at the application level, hardware aspects re-garding the embedding of miniaturized computing technology in everyday objectsare beyond its scope For evaluation purposes, notebook computers with attachedsensors and actuators have been used A possible platform for digital artifacts arePeer-its [FHdSR+07], which comprise the components listed above and are capa-

con-ble of discovering and spontaneously interacting with other such artifacts (cf tion 2.6.1 for more details).

Sec-2.2.1 Exchange of Self-Descriptions

Clearly, information sharing is a key property for the collaboration among facts and their autonomous adaptation to changing conditions at runtime Within

arti-the scope of this arti-thesis, our focus is on sharing context information among digital

artifacts, which we consider fundamental for a semantically meaningful, tual interaction among them and thus beneficial to their autonomous operation inthe background According to [BGS01], context-awareness and ad-hoc informa-tion sharing are essential properties of digital artifacts, and they are enabler for

contex-emerging functionality (cf Section 2.3) Ad-hoc context sharing has been the

ba-sis for interaction in several of the projects presented in Section 2.6.1; for example,[BGS01] employs message-based communication for broadcasting context infor-mation of artifacts via infrared transceivers mounted on the ceiling, [HMS+01]uses wireless communication for sharing contexts like e.g movement patternsamong artifacts in communication range, and XML-based documents containinginformation about the connections and spatial directions of physically connectedartifacts are exchanged via a wired network in [KOY+05]

We employ peer-to-peer concepts for the interaction between digital artifacts

in a spontaneous ad-hoc networking fashion, which appear to be more suitablethan client-server architectures due to the decentralized nature of networked digital

artifacts In order to share context information, so-called self-descriptions are used

for representing the contexts of artifacts and exchanged among them upon cominginto communication range; each artifact then performs a pairwise comparison ofits own with the received self-descriptions and triggers certain actions (cf Figure2.2) Hence, self-descriptions build up a standardized interface for the sharing ofcontext information, and their ad-hoc exchange provides mutual awareness aboute.g physical properties and contexts to artifacts, which facilitates their interaction

with regard to an open-world assumption (i.e the interacting artifacts do not know

each other in advance)

This approach builds on previous work presented in [FHdSR+07, FHM+04,FHR+06a], where the authors suggest to use XML-based documents representinginformation about an artifact and exchanging them with other artifacts in range;upon sufficiently high similarity between two such documents at a time, the respec-

tive artifacts start to interact Similarly, self-descriptions in this thesis are

struc-tured documents based on XML which contain information that may be relevant

for the interaction among artifacts, as for example an artifact’s physical properties

Figure 2.2: Exchange and comparison of self-descriptions.

like its size or weight, its capabilities in terms of provided services or spatial

con-text information like position and direction It enables artifacts to autonomously

become aware of the contexts of others in a decentralized way, but has the

disad-vantages that just contexts of artifacts in communication range can be accessed andeach artifact has to store and maintain all the information it may require The scope

of the present thesis is on the exchange of spatial information, whereas a detailedexplanation of our representation of spatial contexts in self-descriptions, which wekept as simple as possible for the purpose of realizing and evaluating the proposedconcepts, is given in Section 3.3 However, issues of how self-descriptions are ex-changed from a technological viewpoint, under which circumstances (e.g takinginto account energy constraints) and among which artifacts (e.g with respect tocertain communication policies) are not considered

2.3 Mechanisms of Self-Organization

In the introduction, we argued that the growing number of computationally hanced and increasingly interconnected, heterogeneous and complex everyday ar-tifacts – be it home appliances, cars or mobile phones – calls for mechanisms ofself-management and -organization which allow them to operate autonomously inthe background, with no or little human interaction The reason for that is thatsystems have become too complex and heterogeneous for humans to be able touse them without difficulties [HG03, KC03, MMTZ06] According to [Fer07b,FHdSR+07], self-management stands for the ability of individual artifacts to oper- ate autonomously by adapting to changing contexts at runtime, while self-organi-

en-zation stands for the spontaneous contextual interaction of multiple devices in

or-der to achieve collective behavior One approach to self-management is IBM’s

autonomic computing vision [KC03], in which systems are able to self-configure,

-heal, -optimize and -protect and thus function more or less independently from

human supervision, just based on high-level administrative goals and constraints.Self-organization goes further by not only autonomously controlling a system’sbehavior, but creating its own structure which is not explicitly defined by a pro-grammer.

Self-organization has received much research attention in computer science over

the past years, as can be seen from several recent survey papers [Mil07, MMTZ06,SFH+03] In particular, mechanisms of self-organization which are inspired by na-ture (e.g ants food foraging or nest building [MMTZ06]) have attracted the atten-tion of computer scientists [KE01, MMTZ06, ZGMT04] In the context of patternformation in biological systems, [CFS+01] defines self-organization as follows:

Self-organization is a process in which pattern at the global level of a system emerges solely from numerous interactions among the lower- level components of the system Moreover, the rules specifying inter- actions among the system’s components are executed using only local information, without reference to the global pattern.

According to [CFS+01], patterns – i.e organized arrangements of objects in

space or time – are emergent properties, which means that they are produced

with-out external input but rather result from interactions among system componentsthat are based on local information only A prominent example is the cellular au-tomaton “Game of Life” [Gar70]; it consists of a two-dimensional orthogonal grid

of cells which are either white (“dead” cells) or black (“alive” cells) The cells teract with their eight neighbors at a time only, and may change – simultaneously,

in-at discrete points in time – by using three simple rules: (i) each white cell with actly three black neighbors becomes black, (ii) each black cell with more than three

ex-or fewer than two black neighbex-ors becomes white and (iii) each black cell with two

or three black neighbors remains black Interestingly, although just simple rulesabout a cell’s local neighborhood are used, complex patterns such as the so-called

“glider” shown in Figure 2.3 may emerge depending on the automaton’s initial

configuration A related approach is amorphous computing [AAC+00], which fers from cellular automata in that the “agents” are not regularly placed on a gridand that they do not operate synchronously [JBL+06]

dif-A more general definition is given in [WH04], where the authors define

self-organization as a dynamical and adaptive process where systems acquire and

maintain structure themselves, without external control In this regard, “structure”

can be a spatial, temporal or functional structure, and “without external control”refers to the absence of any control instructions from outside the system, whichdoes not exclude data inputs however We adopt this definition for our work, andconsider systems as self-organizing in some respect if they spontaneously, and de-

Figure 2.3: Emerging “glider” pattern in “Game of Life”.

pending on their determined context, interact with other systems in order to achievecollective behavior

Following the considerations in [BGS01, MZ05], we identify two enabling

prop-erties for the self-organization of autonomous embedded systems: (i)

context-awareness (e.g using local sensors for acquiring physical contexts, cf Section

2.1) and (ii) ad-hoc context sharing (e.g via an exchange of self-descriptions,

cf Section 2.2) These two properties in combination allow for developing plications whose functionality not only depends on the individual contexts of au-tonomous embedded systems, but on the co-existence of other such systems andtheir contextual relations to one another In this regard, the present thesis coversself-organization insofar as it provides enabling concepts and solutions regardingthe awareness of digital artifacts about spatial properties and relations among eachother, as well as the exchange of such spatial contexts with other artifacts in range

ap-In the pervasive computing literature, a number of projects can be found whichexhibit some forms of self-organization So as to clarify what we mean by self-organization in the context of this thesis, a few examples – taken from projectswhich are explained in more detail in Section 2.6 – will be given in the follow-ing In [KOY+05] for example, so-called u-Textures (i.e digital artifacts equippedwith displays and direction sensors) can be assembled together, whereas (i) each

of them autonomously adapts its behavior according to the current shape of theoverall structure and its spatial relation to it, and (ii) the assembled u-Textures as

a whole cooperatively run a corresponding application Another example are diacups [BGS01, HB00] (i.e coffee cups with embedded computing, sensing andcommunication capabilities) that have been used for a smart doorplate application,

Me-in which an Me-interactive display (the doorplate) adapts its content based on both thespatial co-location of Mediacups and their temperature, which are context informa-tion that is shared among those digital artifacts and thus enables the emergence offunctionality by spontaneous interaction The TOTA (“Tuples on the Air”) middle-ware [MZ04, MZ05] is an approach for building distributed data structures throughmulti-hop propagation of tuples containing both content and propagation rules, bymeans of which self-organizing applications e.g for distributed motion coordina-

tion can be realized Further projects which show mechanisms of self-organizationcan be found in [BIK+04, KKP99, LSBP02, ZGMT04], among others.

2.3.1 Self-Organization in Space

We consider spatial context information valuable for mechanisms of tion among digital artifacts, as they are by nature situated in physical space andthus spatial properties as well as spatial relations between them, which are par-ticularly important due to the fact that interaction is based on local information,are crucial for characterizing an artifact’s situation From the examples above, itbecomes apparent that self-organization is often related to spatial aspects like theposition of artifacts or their distance- and orientation-relations to each other, and

self-organiza-[ZM04] actually states that a vast majority of known phenomena of

self-organizati-on and self-adaptatiself-organizati-on in nature [ ] are actually phenomena of self-organizatiself-organizati-on

in space [ ].

In [Zam04], it is argued that spatial concepts play an important role in tributed computing scenarios like sensor networks, mobile ad-hoc networks andpeer-to-peer (P2P) networks, and that such distributed systems require some forms

dis-of self-organization which can be based on a spatial computing model According

to [MZ05], spatial self-organization can be found at different levels in distributed

systems: at (i) the physical level which enables spontaneous interaction between system components (e.g radio broadcasts), (ii) the structure level at which spatial structures emerge (e.g beacon-based self-localization), (iii) the navigation level

which exploits available spatial structures (e.g for geographical routing) and at

(iv) the application level where the underlying navigation mechanisms are used

(e.g pattern formation)

As the basis of a spatial computing model, [Zam04] proposes the exploitation

of spatial abstractions, meaning that the activities of application components are

abstracted as taking place in some sort of abstract metric space, and for which self-organization derives from the autonomous capability of components of sens- ing, acting in, and navigating that space Thus, instead of addressing entities by

their names, their positions in space are used for interaction purposes In a similarspirit, we propose the use of spatial abstractions for the implementation of spa-tially aware applications in this thesis, but with a focus on physical objects andtheir relations among each other We therefore distinguish the following two types

of spatial abstractions, which are in detail discussed in Chapter 3: (i) quantitative

abstractions by means of one or more spatial zones representing an artifact’s

po-sition, direction and spatial extension, and (ii) qualitative abstractions of relations

between such zones representing relationships between artifacts that have a certainmeaning for the application.

2.4 Spatial Context in Time

The focus of this thesis is on the awareness of digital artifacts about their spatial

contexts, which is important information characterizing the situation of artifacts

that are by nature distributed throughout physical space Similar to [CFH97], weclassify spatial context information both in terms of the artifacts’ inherent char-acteristics and with respect to other artifacts, and distinguish two corresponding

types of spatial context information: (i) the spatial properties of artifacts, which are independent of other artifacts, and (ii) spatial relations between artifacts – or

rather relations between their spatial properties Spatial properties and relationsare valuable context information for a variety of applications, and they constitute

a distinguishing feature of this work Awareness about spatial contexts is a condition for the autonomous adaptation of digital artifacts to changing spatialenvironments, and spatial relations among them – as well as relationship changesover time – are particularly relevant with regard to their spontaneous contextuallocal interaction [HHS+02, HKG+05, Sch02] In fact, location information is themost prevalent type of context in context-aware applications (cf [CK00, SBG99]),and our related work survey of Section 2.6 also shows that location and proximityare the most commonly used spatial contexts

pre-In Table 2.4, which was inspired by [CFH97, CH01], three types of spatial

prop-erties of an artifact are distinguished: (i) its position (synonymously used with the

term location in this thesis), namely where it is located in physical space, (ii) thedirection in which it is aligned and (iii) the artifact’s spatial extension comprisingits shape and size It should be noted that the terms orientation and direction arenot used consistently in literature, and sometimes inverse to our notion Corre-

spondingly, three types of spatial relations between the spatial properties of two

artifacts at a time are distinguished: (i) positional relations, which include bothorientation and distance relations between the positions of two artifacts, (ii) direc-tional relations, which are relations between two artifacts’ axes of direction, and(iii) topological relations, namely relations concerning the spatial arrangement ofextended objects which is defined by their extensions in addition to position and di-rection A simple visualization of static properties and relations with a rectangular-and a triangular-shaped artifact can be seen in Figure 2.4 The actually used re-lations depend on application demands, but also on the sensing and processingcapabilities of the involved digital artifacts; for example, the consideration of spa-

Table 2.1: Static characteristics of an artifact’s spatial situation.

Spatial properties Spatial relationsPosition geographic position orientation and distance relationsDirection intrinsic direction axis relations between direction axesExtension shape and size arrangement of extended objects

Figure 2.4: Static spatial properties and relations of artifacts

tial extensions – especially when freeform shapes are involved – requires muchcomputational power, which may exceed the capabilities of embedded systemswith limited resources

Within the scope of this thesis, spatial context can be given either absolute,

namely with respect to a fixed external frame of reference such as the earth

refer-ence frame, or relative, namely with respect to the referrefer-ence frame of an artifact.

This distinction between absolute and relative space ranges back to a historicalcorrespondence [LCA98] between Leibniz and Clarke, who wrote Newton’s let-ters, in the eighteenth century For Newton, space was an infinite and immovablethree-dimensional container with its origin at the center of the universe and whichexists whether or not anything is in it [Lev96] From their correspondence, it isobvious that an absolute space was not a meaningful concept for Leibniz; instead,

he considered space as something merely relative, as time is, namely as an order

of coexistences, as time is an order of successions [LCA98] Hence, relative space

according to Leibnitz’ view is basically a means for describing relations amongobjects We will come back to the distinction between absolute and relative space

in Section 3.2.4, where reference frames for qualitative spatial relations are cussed

dis-Another distinction of spatial contexts concerns its representation, where we

distinguish two types: quantitative and qualitative representation We use the

for-mer for representing spatial properties of artifacts by means of nufor-merical values,and according to certain reference systems in order to ensure a common under-standing between artifacts, which is a precondition for the recognition of spatialrelations out of them An example are WGS84 coordinates used by GPS, whichrepresent absolute positions with numerical longitude, latitude and altitude co-ordinates (cf Section 3.1) For the representation of spatial relations, qualitativeabstractions are used, which have advantages compared to quantitative ones when-ever the spatial cognition of humans is involved [Mus00] for example Qualitative

spatial relations are based on discrete symbols such as left or near, representing

continuous properties such as orientation or distance in an abstract way which ismeaningful for a certain application [CH01]

The six types of spatial context information shown in Table 2.4 are referred to as

static spatial context in that they describe an artifact’s spatial situation at a

partic-ular point in time According to [AEG94], the position, direction and extension of

an artifact – i.e its spatial properties – can be changed through translation, rotationand scaling respectively, determining how its spatial relations to other artifacts are

changed The latter are thus referred to as dynamic spatial context, as they describe

how its static situation is changing at a certain point in time We use a discrete timefor the sensor observations, which meets the requirements of computing systemswith discrete clock cycles [Mus00] In addition, the spatial context of an artifact

need not correspond to a certain point in time, but may also represent an artifact’s situation over a time period by comprising a time series of spatial contexts Summing up, we distinguish four categories of spatial context information with-

in the scope of this thesis:

• Absolute (i.e with respect to a fixed external frame of reference) vs relative

(i.e with respect to the reference frame of an artifact)

• Quantitative (i.e using numerical values) vs qualitative (i.e using discrete

symbols which represent intervals of quantitative values)

• Static (i.e at a certain point in time) vs dynamic (i.e the degree of change

at a point in time)

• Point in time (i.e a snapshot at a certain time) vs time series (i.e a trajectory

of spatial contexts over a sequence of points in time)

With respect to the above classification, the focus of this thesis is on (i)

quan-titative spatial properties of artifacts given with respect to an absolute frame of