Mobile Agent Applications in Sensor Networks - PowerPoint PPT Presentation

1 / 12
About This Presentation
Title:

Mobile Agent Applications in Sensor Networks

Description:

Same memory can be used by multiple ... Mote. Fire Detection. 1. 2. 3. 4. Active Data Aggregation. s1. s2. s3. s1. s2. s3. s1. s2. s3. Active Tracking ... – PowerPoint PPT presentation

Number of Views:102
Avg rating:3.0/5.0
Slides: 13
Provided by: formal
Category:

less

Transcript and Presenter's Notes

Title: Mobile Agent Applications in Sensor Networks


1
Mobile Agent Applications in Sensor Networks
  • Chien-Liang Fok
  • Gruia-Catalin Roman
  • Mobile Computing Lab
  • Washington University in St. Louis

2
Wireless Sensor Networks
3
Mobile Agents
Mobile Agent Mobile Code State Intelligence
  • Not to be confused with
  • Remote evaluation
  • Code on demand
  • Two types of state mobility
  • Strong migrate data thread state
  • Weak migrate data state only

4
Why Mobile Agents in S.N.?
  • Generic network
  • Reprogram a network after deployment
  • Asynchronous operation
  • Agents operate independently of the creator
  • Memory usage
  • Same memory can be used by multiple applications
    (at different times)
  • Active networks
  • Intelligent migration of code and state based on
    the situation

5
Mobile Agent Middleware
  • Mobile agent support should be implemented in
    middleware
  • What features should the mobile agent middleware
    provide?
  • Operation primitives
  • Data and instruction stores (?)
  • How can the middleware simplify application
    development?

6
Primitive Operations
Mobile Agent
Mote
TupleSpace
Agent Lifecycle Inject(Program, Parameters,
Destination) Clone(Parameters, Destination) Move(D
estination) Expire() Agent Communication / Data
Ops. Send(Destination, Data, TimeToLive) Send(Agen
t, Data, TimeToLive)
Agent / Data Query findAgents(Region,
Attributes) findNodes(Region) findData(Region,
Attributes) grabData(Region, Attributes)
7
Fire Detection
2
1
3
4
8
Active Data Aggregation
s1
s2
s3
s1
s2
s1
s3
s2
s3
9
Active Tracking
10
Sweep-Line Search
11
Sensing Coverage
12
Research Directions
  • Refinement of primitives
  • Are they sufficiently expressive and powerful?
  • Need a formal specification of their semantics
  • Implementation
  • Integration within a virtual machine (prototype)
  • Evaluation of mobile agent benefits
  • Agent evolution
  • Split each agents code into fragments, mix up
    the fragments to create new mutant agents
Write a Comment
User Comments (0)
About PowerShow.com