8
Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

Embed Size (px)

DESCRIPTION

3 Performance is Poor – Causes Are Not Understood [Phil Levis, Stanford]

Citation preview

Page 1: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

Why are real WSNs so hard for us to achieve?

Jan Beutel, ETH Zurich

Page 2: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

2

“Proof-of-Concept” Deployment Experience

[Koe

n La

ngen

doen

, TU

Delft

][G

ilman

n To

lle, U

C Be

rkel

ey]

[Pra

bal D

utta

, UC

Berk

eley

][P

raba

l Dut

ta, U

C Be

rkel

ey]

Page 3: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

3

Performance is Poor – Causes Are Not Understood

[Phi

l Lev

is, S

tanf

ord]

Page 4: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

Is it just generally painful?

Page 5: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

Are we doing things the wrong way?

Page 6: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

Were we promising too much?

Page 7: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

7

My Contribution – WSN Deployment Survey

• A survey on the success, failure and generally experience in wireless sensor network deployments.

• Online at– http://www.btnode.ethz.ch/Projects/WSNDeploymentS

urvey

• Contact– Jan Beutel, ETH Zurich – [email protected]

• Prize – Win a Free Ski Weekend in the Alps– Drawing is closing on December 15, 2007.

Page 8: Why are real WSNs so hard for us to achieve? Jan Beutel, ETH Zurich

8

Outrageous Opinions

• Why are ...– we not applying the same rigor and methods we

teach to students in WSN design/deployment practice?• Code reviews, calibration, comparable/repetitive

experiments ...

– we cramming more and more into tiny microcontrollers, operating them beyond limits, wondering why last minute quick-fixes fail?• The embedded industry designs and scales architecture

exactly for the required performance/complexity!

– so many computer scientists building (lower layer) MAC protocols and not the radio designers/manufacturers?