TinyML: Metadata for Wireless Networks

1
TinyML Meta-data for Wireless Networks

• 2007.2.22
• ???

2
Contents

• Nathan Ota, William T.C. Kramer
• Berkeley Manufacturing Institute
• Introduction
• Markup Languages
• Sensor Markup Languages
• Overview of TinyML
• TinyML Implementation
• Summary

3
Introduction

• Embedded sensor networks are rapidly becoming a
  mainstay in many areas
• Custom assembled for specific applications
• The applications entail complicated software
• One of kind combinations of hardware and unique
  deployment details
• Two sensor systems
• Monitor and control, locks on door
• different vendors
• implemented with entirely different technology
• Tries to integrate the two systems
• Fortunately, both manufactures support TinyML
• Because TinyML is based XML
• Use TinyML for this integrating

4
Introduction

• Sensor networks become more prolific
• Need for methods to retrieve the data
• Critical for sensor networks to have simple ways
  to interface to standard web method
• Easily interface between networks
• Enable many users to access data in a sensor net
• Access should be archived or real time with the
  same methods as accessing archive data
• In the future
• It will be imperative to be able to exchange and
  manage data from multiple sensor nets
• Easily interface between networks

5
Markup Languages

• Benefits of generalizing how information is
  organized
• The Extensible Markup Language (XML)
• Design to create tree relationships with data
• Extensible and flexible
• Using standard APIs for XML
• Simple API for XML (SAX)
• Document Object Model (DOM)
• The Geographic Mark Language (GML)
• XML based schema focused on the defining
  geographic and graphical information
• This is a building block of other Markup
  Languages, including SensorML

6
Sensor Markup Languages

• SensorML
• To handle data dissemination for satellite and
  other sensor networks
• Designed for complex positioning characteristics
• Data schema
• Describe the geometric, motion characteristics
  and observations of a sensor
• Does not really have the concept of an actuator
• Rather, Sensor Planning Services
• Possible to consider using SensorML for sensor
  networks
• May be overly complex for current sensor networks
• Not yet released for general use
• Does not have the full concept of in-network
  processing
• Rather, Sensor Collection Services

7
Sensor Markup Languages

• Cougar
• Method of handling queries for sensor networks
• Query optimizer
• Efficient on particular network
• Reducing the use of network resources
• TinyDB
• Query processing architecture for TinyOS sensor
  networks
• SQL like interface

8
Overview of TinyML

• Lightweight implementation following some of the
  SensorML ideas
• Built on XML
• A few fundamental elements
• Platform
• Processor, energy source, communication device,
  sensors
• Sensor field
• Collection of platform
• Collection of sensor nodes
• Provide data to link the field
• Virtual sensors and actuators
• Query flag
• When set, request a response from the component

9
Overview of TinyML
10
Overview of TinyML
11
Overview of TinyML

• Virtual Devices
• Virtualizing physical components
• Virtual sensor or actuators could be defined
• ??? ?? ????? ?? ??? ??
• Can also be a collection of sensor outputs or
  actuators actions
• There are two major types of virtual
  sensors/actuators
• Focused on platforms
• Associated only with basic sensor
• Focused on sensor fields

12
Overview of TinyML
13
Overview of TinyML
14
TinyML Implementation

• Unfortunately, it is not possible
• Requires placing XML into the sensor network
  itself
• Each platform must have the capability to parse,
  interpret, respond and formulate XML
• Not practical with current sensor/node limitations

15
TinyML Implementation

• Network Setup and Assumptions
• The network has at least one gateway node
• The gateway node utilized a sensor network
  interface, such as TinyDB

16
TinyML Implementation

• Applying TinyML
• Core Functions
• A Query is initiated by a remote user
• First translated by the users Proxy-to-DOM
  translator into a DOM
• Exported as an XML file with an XML-to-DOM parser
  and validated with the TinyML and GMS schemas
• The gateway node follows the reverse process as
  at the remote user.

17
Summary

• The goal of this project
• To investigate the issues involved with the
  creating a general interface
• Interaction between an internet users and a set
  of sensors
• TinyML
• Created in order to explore the interface between
  sensor nets and the internet services
• GML and GML-based SensorML
• Focus for satellite-based sensor coordination for
  global location services
• Differ in the following ways from mote-centric
  wireless sensor networks
• TinyML schema
• Concept of virtual sensor and actuator
• Future work
• Full evaluate the TinyML schema
• Performance study and optimization are needed
• More accessible to non-export user and for
  archiving the data retrieved

18
?? ????
19
Paper

• Introduction
• ?? ??
• Related work
• HiFi, GSN, Ubicore, ANTS
• TinyML ??
• Middleware Architecture
• ? component ?? ??
• Caching Algorithm
• ?? ??
• Experimental Result
• ???? ??, ?? ? ??
• Conclusion and Future Work

20
Implementation

• Device Connect Layer
• Done
• Reader agent
• To do
• Sensor network agent ???? ?? ?????
• Data Translation Layer
• Done
• Message Type
• Data Aggregator
• Data Model Generator
• Data Management Layer
• Done
• Data Updater
• Data Service Layer request (real-time, history)
• To do
• ????? ?? device database
• Caching algorithm Memory DB
• XML Formulation