A Decentralized Adaptive Architecture for Ubiquitous Augmented Reality Systems

1
A Decentralized Adaptive Architecture for
Ubiquitous Augmented Reality Systems

• Dissertation

2
Summary

• Ubiquitous augmented reality convergence of
  augmented reality and ubiquitous computing
• I address software engineering challenges with
• Adaptive service dependency architecture
• Distributed middleware
• Design at run time
• Approach
• Constructive problem analysis, existing
  approaches
• Explorative show feasibility in DWARF and its
  systems

3
Ubiquitous Augmented Reality (UAR)

• Extension of augmented reality interaction into
  ubiquitous computing environments

4
Developing Software for UAR

• Basic technology available many application
  areas

• ...and more
• Health care,
• Campus,
• Rescue team

However, few real systems built to date One
reason software development challenges
5
UAR Software Development Challenges

• New technology
• Many people, many disciplines
• Users in the field

• Mobile users
• Limited sensor network range

• Low latency for immersive AR
• Scalability for ubiquitous computing

6
Related Work

• Augmented Reality and Ubiquitous Computing
  projects
• Studierstube, Tinmith, Coterie, ARVIKA,
  ImageTclAR, Aura, Ninja, Gaia, Context Toolkit,
  Fluidum, UbiCom, (many more)
• Interesting approaches data flow architectures,
  latency layering, reflection, service discovery,
  prototyping
• Current software engineering techniques
• Distributed component architectures CORBA
  Component Model, Web Services
• Agile development methods Extreme Programming,
  Scrum
• DWARF (Distributed Wearable Augmented Reality
  Framework) research project at TUM
• DWARF 2001 No development process, limited
  adaptivity, limited interaction, used for only
  one system

7
Proposed Solution

• Approaches
• Data flow architecture
• Service discovery

• Approaches
• Reflection
• Latency layering

• Approaches
• Agile development
• Prototyping

8
Adaptive Service Dependency Architecture

• New architectural style
• Loosely coupled distributed services
• Parameterized dependencies between services
• Dependencies change based on context
• Middleware maintainsadaptive graphof services
• Used to buildframeworks

9
Decentralized Middleware

• Distributed middleware maintains adaptive system
• Middlewares tasks can be decomposed
• Manage a single service
• Set up communicationbetween two services
• Adapt a serviceto its context
• Manage adaptiveservice graph
• Decentralized implementation possible
• e.g. using CORBA and SLP in DWARF
• Based on externalized service descriptions

10
Process Design at Run Time

• Continuous, incremental development process
• Users provide synchronous and asynchronous
  feedback
• Developers improve running system
• Synchronous Jam Sessions
• Asynchronous Continuous Extension
• Supported by architecture
• Supported by tools

11
Results

• Several UAR systems were built using
  implementation of architecture, middleware and
  process within DWARF and they worked.
• Component uncertainty well addressed by
  architecture
• Ill-defined requirements addressed by process
• Middleware performance (latency, scalability,
  resource use) are adequate
• Feedback from users on systems
• Well understood good for prototyping
• Industrial partners adaptivity of limited
  real-world use
• Feedback from developers on framework
• Complexity and manageability concerns
• Larger system than we could have built with X

12
Conclusion

• The software engineering challenges of UAR can be
  addressed with the Adaptive Service Dependency
  Architecture, its Middleware and Design At Run
  Time
• But
• Only been shown for prototypes
• Middleware and tools in DWARF are not of
  industrial quality (manageability and scalability
  limitations)
• Model does not handle stateful services well
• Development process is mostly conceptual
• Experimental infrastructure for experimental
  systems dangerous
• Future work empirical validation
• Harden framework and middleware implementation
• Build UAR systems for real-world use (commercial,
  industrial)

13
A Decentralized Adaptive Architecture for
Ubiquitous Augmented Reality Systems

• Dissertation