Spatial Awareness of Autonomous Embedded Systems (eBook)

Dr. Clemens Holzmann is a research and teaching assistant at the Institute for Pervasive Computing of the Johannes Kepler University of Linz (Austria). His research interests include location awareness and context awareness as well as wireless ad hoc and sensor networks.

Acknowledgements 7
Abstract 9
Contents 10
List of Figures 13
List of Tables 15
Listings 16
List of Abbreviations 17
1 Introduction 19
1.1 Problem Statement 22
1.2 Contribution 23
1.3 Thesis Outline 25
2 Spatial Awareness 27
2.1 Context and Awareness 27
2.2 Autonomous Embedded Systems 30
2.3 Mechanisms of Self-Organization 34
2.4 Spatial Context in Time 38
2.5 An Architecture for Spatial Awareness 41
2.6 RelatedWork 44
3 Representation of Space 62
3.1 Quantitative Representation 63
3.2 Qualitative Representation 68
3.3 Structure of Self-Descriptions 83
3.4 Summary 89
4 Distributed Spatial Reasoning 92
4.1 Overview of Qualitative Approaches 92
4.2 Reasoning about Positional and Directional Relations 97
4.3 Spatial Relationship Inference and Distribution 104
4.4 Simulation Results 113
4.5 Findings and Discussion 120
5 Rule-Based Spatial Awareness 122
5.1 Achieving Spatially Aware Behavior 123
5.2 Rule-Based Qualitative Spatial Reasoning 126
5.3 Proof of Concept 132
5.4 Summary and Open Issues 136
6 Zones-of-Influence Framework 138
6.1 Architecture 139
6.2 Components 144
6.3 Runtime Behavior 157
6.4 Discussion 160
7 Framework Evaluation 163
7.1 Development of Spatially Aware Applications 163
7.2 Application Scenarios 165
7.3 Comparison and Discussion 185
8 Conclusion 192
8.1 Summary 194
8.2 Outlook 200
Bibliography 205

3 Representation of Space (S. 45-46)

Simple things should be simple. Complex things should be possible. Alan Kay, 1940

In order to deal with space computationally, it has to be represented in a standardized way such that it can be processed by a computer. In this regard, we distinguish between the quantitative and qualitative representation of space, which are discussed in detail in the Sections 3.1 and 3.2, respectively. While the former deals with concrete facts about the spatial properties of an artifact using numerical values (as provided by sensors for example), the latter is concerned with symbolic and human-readable abstractions of spatial relations (i.e. symbols are used for the representation rather than numeric values) which are acquired by a pairwise comparison of quantitative spatial properties. For the representation of an artifact’s spatial properties we propose to use so-called Zones-of-In.uence, which are explicitly defined geographical regions that are relevant for the application. Our focus is on qualitative spatial relationships as well as their changes over time and the application-dependent semantics.

We consider qualitative relationship abstractions to be particularly valuable for implementing services that are distributed among multiple artifacts in physical space. Quantitative spatial properties and qualitative spatial relations – which together constitute an artifact’s spatial context – are exchanged between artifacts in range by means of XML-based self-descriptions, which enables them to reason about both self-determined and received relations, an example structure of self-descriptions, which we used for evaluation purposes in this thesis, is presented in Section 3.3. Section 3.4 eventually sums up findings concerning the quantitative and qualitative representation of spatial properties and relations with regard to their representation and exchange by autonomous embedded systems.

3.1 Quantitative Representation

3.1.1 Spatial Abstraction with Zones-of-Influence

For providing spatial awareness to digital artifacts, we propose a concept referred to as Zones-of-In.uence (ZoI). It builds on initial work published in [FHR+08], where ZoIs are three-dimensional spatial regions associated with digital artifacts, in this work, they serve as an explicit proximity model in that they are used for limiting the interaction between artifacts to those whose ZoI-geometries overlap. In contrast to proximity sensors like Bluetooth signal-strength measurements of nearby artifacts as used in [FHM+04, FHO04a] for example, a Zone-of-In.uence provides added value by allowing to explicitly de.ne the "proximity range" by its shape, the size of the shape as well as its position and direction in space relative to that of the artifact for which it is defined. A similar concept has also been used in [BMK+00, HB00, HHS+02], where geometric containment relations between regions of interest that are associated with physical objects (e.g. a person and the area in front of a screen) are used for location-aware applications (e.g. to display the person’s desktop on that screen).