iBadge and Sylph Design of a Wearable Sensor Badge for Smart Kindergarten

1
iBadge and SylphDesign of a Wearable Sensor
Badge for Smart Kindergarten

• Alessandro Costa Pereira
• Martin Adolph
• Supervisor Dr. Waltenegus Dargie
• Date 16.01.2007

2
Outline

• Motivation
• Concept
• iBadge
• Architecture
• Main Functional Units
• Software
• Focus on Speech Recognition
• Sylph middleware
• Design Overview
• Components
• Sylph Query Language
• Related Work and Conclusion
• References

3
Motivation

• Smart Kindergarten smart problem solving
  environment for early childhood education
• Provide parents and teachers with the abilities
  to comprehensively investigate students learning
  processes.
• Capture interaction between the kids, the
  teachers, common classroom objects.
• Collect, manage and fuse the information of
  sensors in logical and user-friendly manner.
• Project of UCLA

PaLS02
4
Motivation

• Therefore
• Wearable sensing and computing architecture.
• Critical design constraints form factor, energy
  consumption, cost, design flexibility, various
  sensing functionalities.
• Light and small, operation time 4-6hrs/day.
• Wireless communication of person to physical
  world interaction.
• Strong focus on speech recognition.
• iBadge a light-weight power-aware
  sensor-instrumented badge
• Sylph a sensor middleware infrastructure

5
Concept
PaLS02
6
Concept

• iBadge Node
• Developed with COTS components
• DSP, Microcontroller, standard interfaces
• Architecture elements
• Sensing Infrastructure
• includes video cameras, microphones, motion
  detectors and iBadges. iBadges will enable higher
  layer services to discover and keep track of
  context information about the students and the
  toys.
• Network Infrastructure
• allows interconnections between sensor devices to
  the wired backbone that runs the sensor
  middleware through use of several access points
  (e.g. Bluetooth for iBadge, analog UFH/VHF for
  wireless microphones).
• Middleware infrastructure Sylph
• provides various services that process, store,
  fuse, manage and present the data that either
  have been collected or are being collected from
  Sensing Infrastructure.

7
iBadge - Architecture
PaLS02
8
iBadge Main Functional Units

• Speech Recognition Unit
• DSP (120MHz 128kB), Codec chip (16kHz
  Samples_at_16bits), Microphone, Loudspeaker
• Wireless Communication Unit
• Bluetooth module (ARM core, flash memory and
  wireless transceiver). Communicates with ATMEGA
  through its RS232 port.
• Power Management/Tracking Unit
• Has the control of Switches
• Environment Sensing Unit
• Sensors Light intensity, Humidity, Air pressure
  and Temperature.
• Temperature Sensor has proprietary Dallas bus.

9
iBadge Main Functional Units

• Localization Unit (3D)
• RFM Radio
• Orientation Tilt Sensing Unit
• 2 dual-axis accelerometers and single tri-axis
  magnetometer
• Power Supply Unit
• 3,6 Lithium-ion battery with 800mAh.
• Stable output set to 3,3 V _at_ max. 2A.

10
iBadge Main Functional Units
PaLS02
11
iBadge Software

• Connects to the middleware.
• Different applications can have different
  requirements for delivering the same sensor
  information.
• Same software structure runs on ATMEGA and DSP
  (through hardware abstract layer). Same code
  could run on both processors.

Loch02
12
iBadge Software
ATMEGA Tasks
Both tables Loch02
DSP Tasks
13
iBadge Software

• Loop execution of tasks

Loch02
14
iBadge Signal Path

• Exemplary signal path

Loch02
15
iBadge Speech Recognition

• Two speech-processing scenarios are implemented
  on iBadge
• codec software compresses the speech signal and
  iBadge sends it over the Bluetooth interface to
  the wired network infrastructure at a bit rate of
  64kBit/s. The server can use the information to
  replay speech or it can run speech recognition on
  the signal.
• the entire Front-End Processing for distributed
  speech recognition takes place on iBadge. The
  process includes segmentation of speech into
  overlapping frames utilized as a basis for the
  computation of the so-called speech feature
  vectors. Computation-intensive.

16
iBadge Speech Recognition
Loch02
17
iBadge Speech Recognition

• Test results

Loch02
18
Sylph

• Middleware, lightweight, extensible proxy service
  for handling service discovery and mediating
  access.
• Designed to provide common framework for wide
  variety of devices using strong abstraction
  layers.
• Sensor developer should remain unaware of
  high-level interfaces.
• Application should simply enlist sensors and
  submit queries.

19
Sylph - Components

• Sensor module
• Initializes and accesses sensor devices including
  iBadge
• When loading a sensor proxy core reads in an XML
  configuration file, passes a set of parameters to
  the module
• User can configure a sensor module by modifying
  simple attribute-value pairs in the configuration
  file
• common interface for receiving directives from
  the proxy core and returning data
• consists completely of commands to SET and GET
  attributes

20
Sylph - Components

• Service discovery module
• Allows Sylph to advertise the presence of managed
  devices to sets of client applications
• Sensor modules first register their sensors by
  providing information in a standard format,
  including available attributes, data types, and
  other built-in characteristics
• advertises presence of new sensors through a
  standard discovery mechanism
• allows application to send text-based queries to
  the middleware and receive data values in
  response
• uses Suns Jini discovery service

21
Sylph - Components

• Proxy core module
• forms the heart of Sylph middleware, tying
  together sensor modules and service discovery
  modules while managing system resources
• updates data structures to track sensor devices
  and user applications
• parses incoming requests to form query plans and
  processing trees
• provides an extensible framework for the addition
  of specialized features (as mobility support,
  dynamic query optimization, quality-of-service
  control)
• when sensor module registers a device, proxy core
  assigns it a unique sensor identifier
• when client issues a query, proxy core provides a
  unique request identifier

22
Sylph Query Language

• From the applications point of view, Sylph acts
  as a flexible data retrieval system
• READ ltdata tuplegtWHERE ltpredicate
  clausegtEVERY lttime intervalgtFOR
  ltdurationgt
• READ temperature, pressureEVERY 1 minute FOR 1
  hour
• READ x, y, zWHERE volume gt 70FOR 30 samples

23
Related Work

• Xerox Parc Active Badge Project IR sensors to
  locate badge's wearer in office environment
• RIT Stockholm Smart-Badge system IR
• UC Berkeley Smart dust - tiny sensors with
  micro-electro-mechanical systems constantly
  evaluate and react to changes in the environment

24
Conclusion

• iBadge combines
• Speech processing capability
• Location sensing
• Environment sensing
• Wireless radio
• Power monitoring/tracking unit
• Has the size of a small pager that can be worn by
  pre-school children.
• Sylph middleware infrastructure provides
• An abstraction layer for initializing and
  accessing raw physical sensors while employing
  means of advertising the available sensor
  services to the higher layers.

25
References

• Loch02
• Locher, I. R. Design and Implementation of
  iBadge and its Distributed Speech Processing
  Capability. Master Thesis, University of
  California, Los Angeles, 2002.
• PaLS02
• Park, S. Locher, I.R. Srivastava, M. Design of
  a Wearable Sensor Badge for Smart Kindergarten.
  6th International Symposium on Wearable Computers
  (ISWC2002), Seattle, WA, October 7-10, 2002.
• iBadge Website
• http//nesl.ee.ucla.edu/projects/ibadge/default.ht
  m