Transcript MURI Agilla Presentation

Rapid Development and Flexible
Deployment of Adaptive Wireless
Sensor Network Applications
Chien-Liang Fok, Gruia-Catalin Roman, Chenyang Lu
http://mobilab.wustl.edu/projects/agilla
Mobile Computing Laboratory
Distributed Object Computing Laboratory
Department of Computer Science and Engineering
Motivation
 Software development for sensor networks is hard



Limited resources
Difficult to debug
Large & highly dynamic network
 Existing software lacks flexibility


Entire network runs just one application
Cannot adapt to changes in
•
•
the environment
user requirements
MURI 2005
2
Motivating Example
 Three applications: 1) Environmental Monitoring,
2) Fire Detection, 3) Fire Tracking
MURI 2005
3
Agilla: A Flexible Middleware for
Sensor Networks
 Sensor network as a shared computing resource

Flexible application deployment
Env. monitoring agent
Fire detection agent
Fire tracking agent
MURI 2005
4
Agilla’s System Architecture
Node @ (2,1)
Node @ (1,1)
Agents
Agents
migrate
Neighbor
List
Tuplespace
remote
access
Tuplespace
Neighbor
List
Middleware Services
Middleware Services
Agilla Middleware
Agilla Middleware
TinyOS
TinyOS
MURI 2005
5
Agilla’s Computational Model
PC
Stack
Condition
Codes
Code
Clone
or
Migrate
Heap
Two variants of each:
1) Strong (code + state)
2) Weak (code only)
MURI 2005
6
Location-Base Addressing
 Nodes are addressed by location
(3,3)
(3,1)
(3,2)
(2,2)
clone to (3,3)
clone to (3,1)
(1,1)
MURI 2005
Fire Detection
Agent
(1,3)
7
Tuple Space-Based Coordination

Content-addressable shared memory



Tuple – A set of data fields
Template – A pattern that matches particular tuples
Provides spatiotemporal decoupling
“rout”
“in”
“in”
“out”
Tuplespace
MURI 2005
Tuplespace
8
Implementation on TinyOS 1.1.13
 Agilla 1.x is available for Mica2 and MicaZ motes

4 agents/node
 Agent Injector


Written in Java
Remote Injection via RMI
 Key Challenges:

Memory:
•
•

ROM: 54.7KB of 128KB
RAM: 3.5KB of 4KB
Message loss
MURI 2005
9
Agilla 1.x Test Bed
 6x9 Mica2 Mote
Test Bed
 Multi-hop Grid
 One base station
Base Station
MURI 2005
10
Performance Evaluation:
migration vs. remote tuple space access
Migration instructions are more
reliable because of hop-by-hop
acknowledgements…
…but remote tuplespace operations
have less overhead
MURI 2005
11
Agilla Instruction Execution Times
Local
Operations
Remote
Operations
MURI 2005
12
Application Experiences
 Fire Detection & Tracking

Presented at MURI 3-Year Review and IPSN 2005
 Efficient Network Exploration


In collaboration with UCI
Presented at MURI 3-Year Review
 Intruder Detection and Tracking

Yuling Liang’s CS 521S Class Project
 Cargo Tracking

In collaboration with
 Robot Navigation Around Fires

In collaboration with the Media Machines Lab
MURI 2005
13
Intruder Detection and Tracking
Base
Station
Tracker sends back a
heartbeat. Base station
re-deploys tracker if
heartbeat goes away.
MURI 2005
14
Cargo Tracking
 7 million containers arrive annually into the US

Impossible to check every container
 Existing container security devices are limited:
•Requires line-of-sight with satellite
•Low-bandwidth (six 9B msgs/day)
•$500/device, $34.95/mo
MURI 2005
15
AgiTrack: Cargo Tracking using Agilla
•Find Cargo
•Load Manifest
•Find Intrusions
Internet
base
station
base
station
•Manifest
•Security Flags
MURI 2005
16
Dynamic Context Discovery & MultiHop Network Formation
base
station
Dock
PDA
base
station
Ship
MURI 2005
17
Agilla 2.0
 Supports multiple base stations
 Allows nodes to move (physical mobility)
 Integrates wireless sensor networks with IP networks
Agilla
Tuple Space
Limone
Tuple Space
Limone
Network
(IP)
Agilla
Network
MURI 2005
Limone
Service
Registry
Limone agent
encapsulates
Agilla agent
Agilla
Network
18
Systematic Network Traversal
MURI 2005
19
Robot Navigation
 Mobile agents guide robot safely around the fires
MURI 2005
20
Robot Navigation Videos
Without sensor network data
With sensor network data
http://www.cse.wustl.edu/~bayazit/index.php?page=sensornet
MURI 2005
21
System Challenges
 Multi-Hop Routing


No geographic routing implementation available for Mica 2
Not enough memory on Mica2 motes for Agilla & Mint
 Indoor localization


Cricket motes use a different radio packet and don’t have
enough memory to run Agilla
Use Cricket & Mica 2 platforms
 Outdoor localization


Least Squares Scaling (LSS)
UIUC
MURI 2005
22
Conclusions
 Mobile agent middleware simplifies application
deployment & increases network flexibility
 Empirical results show that deploying sensor network
applications via mobile agents is reliable and efficient
 There are many applications for mobile agents in
wireless sensor networks




Fire detection
Intruder detection
Cargo tracking
Robot navigation
MURI 2005
23
Thank you!
Agilla URL: http://mobilab.wustl.edu/projects/agilla
• Source Code
• Documentation
• Tutorials
• Experience Reports
MURI 2005
24