Download Sensornet 2.0 Challenges

Sensornet 2.0 Challenges
Tarek Abdelzaher
Department of Computer Science
University of Illinois at Urbana Champaign
Copyright ©: Tarek Abdelzaher
#3: Human-Centric Applications
Static Sensornets  Mobile Consumer Sensor Networks
The master
controller
Nike -iPod
http://www.sensatex.com
Wii
Spot
Technology introduced by 2007
#2: A Network Paradigm Shift
Communication  Information Distillation
The data fire-hose effect
Present Internet
Goal:
Future Embedded Networks
Goal:


Communication
Information Distillation

Connecting people

Reducing data from sensing
inputs to actionable information



Protocols geared primarily for
point-to-point communication
Data consumption
applications:
e-mail, VoIP, Web, P2P, …


Human is out of the main data
reduction loop
Protocols geared for data filtering,
aggregation, mining, …
Data reduction applications:
Emergency detection/response,
statistical analysis, …
#1: Layering In-network Computation
Protocol Stack  In-network Computation “Stack”
Sensor network
as a Database
A
Event based
Paradigms
B
Present sensor networks:

Different macro-programming
paradigms for different applications

Each supported by a black-box
runtime system with a paradigmspecific API

Can’t mix and match “black boxes”
Layer 5
Node-based
Programming
Contrast with general
networking
 Layered abstractions
 Different programmers
can operate together
at different layers
 APIs exported from
multiple layers
Object-based Programming
Layer 4
Layer 3
Layer 2
Layer 1
Backup Slide:
Other Challenges









Privacy!!
Mobility (opportunistic, symbiotic, parasitic, …)
Security (control of critical infrastructure?)
Storage, indexing, archiving, …
Data analysis, statistical learning, data mining, …
Design for integration (large scale of programmable
components breeds integration problems)
Robustness (bugs will exist, but networks must
prevail)
Performance analysis tools (capacity, availability, …)
Debugging and troubleshooting tools