An Adaptable Data Object Service for Pervasive Computing

1
An Adaptable Data Object Service for Pervasive
Computing

• Christopher K. Hess
• University of Illinois at Urbana-Champaign

2
Introduction

• Traditional distributed systems typically operate
  in homogeneous environment.
• Distributed systems increasingly incorporate a
  diverse set of device types.
• Disparity of device resources makes integration
  more difficult.
• Key service is access to data.
• Usually handled with file systems or databases.

3
Introduction

• Traditional file systems/databases are static and
  generally targeted at homogeneous environments.
• Some attempts to incorporate weaker devices
  WAP, TACC, etc.
• Propose adaptable data service to accommodate
  requirements of diverse set of devices.
• Service handles complex tasks ordinarily left to
  the application developer.

4
Problem Description

• Entering age of ubiquitous computing.
• Future smart cities, buildings, rooms, vehicles
  will join these devices.
• Some devices possess limited resources.
• May not be able to render data in original
  format.
• Device accesses data source in the format it
  requires.
• System transparently adapts content.

5
Architecture

• User level data service using CORBA.
• System layer and user layer.

Iterator
User
Container
CORBA
System
Managers
Containers
Core
6
Architecture

• Information delivered as data objects.
• Data sources represented as containers.
• Access to data gained via iterators.
• System sets up data flow paths.
• Nodes in flow path may alter data.

7
Container Manager

• Interface of container may be changed to produce
  different data format.
• Manager chains containers together.

Obj. ref.
Obj. ref.
Type 1
Type 1
Type 2
adaptInterface()
Container
Container
Manager
Manager
8
Listener Manager

• Allow data to be pushed to applications.
• Apps register to receive data objects.

Application Listeners
Remote Listener
Streamable Container
Container Manager
Listener Manager
Application
9
Containers

• File containers access to native operating
  system files.
• Operator containers files with dynamic
  content.
• Adaptor containers change container interface.
• Converter containers transform content.
• Device containers access data of devices.

10
User Layer

• Based on containers and iterators (STL).
• Combination of template classes and C wrappers
  provides simple interface.
• Apply generic programming to distributed objects.

CORBA Container
Container Wrapper
Data Objects
Iterator
Container Template
11
Iterators

• Retrieve remote objects incrementally and cache
  (i.e., buffer cache).
• Allows easy traversal of typed objects.
• Generic containers marshal specific types.

12
Uses of Framework

• Data storage files and directories.
• Devices - printer, whiteboard, X10.
• Shared memory - tuple spaces.
• Proxies remote transformations.
• Change interfaces of data sources whiteboard to
  mouse.
• Services news.

13
Example
scribble
Audio Container
Whiteboard Listener
MailToAudio Container
Mail Container
Grep Container
WordToAscii Container
Bitmap Container
Byte Container
PowerPoint Container
Word Container
Whiteboard Container
scribble
MPEG
Word
Whiteboard
Mail
PowerPoint
14
Conclusions

• Heterogeneous distributed systems becoming more
  pervasive.
• Some devices cannot accept original data format
  due to available resources.
• Applications open data as desired type.
• Data object service adapts content to device
  characteristics.
• Simple user interface.