What constitutes a useful experimental result Kamin Whitehouse
- Slides: 18
What constitutes a useful experimental result? Kamin Whitehouse University of Virginia Wireless Networking Workshop May 20, 2008
A double standard n n This experiment must use a radio like 802. 11 This experiment applies only to one radio We need to bridge the gap between hardware/protocols Hardware-specific protocols have little value
Question: n What constitutes a useful experimental result? Answer: n They are always useful!
New question: n What is the lifetime of an experimental result? Answer: n Can affect parameters/optimization n Can affect theoretical limits n Can affect future hardware design
Parameter/Optimization n Characterize bit-error probability q Changes phy-layer coding techniques Hardware
Parameter/Optimization n Characterize Rx/Tx switch times q Change CSMA backoff times Hardware
Validate/Invalidate Assumptions n Characterize link quality distribution q Change network layer choices Hardware
Validate/Invalidate Assumptions
Influencing hardware design
Collision Detection and Recovery A Pream Sync Data Crc Detection Recovery n n B Recovers strongerlast packets Identifies collision Pream Sync Data Crc
Stronger-First Collision Detection A Pream Sync Data Detection B Pream Sync Data Term
Effect on Protocols Revisit fundamental protocol design decisions NACK n n NACKs Broadcast: reliability C A B
Effect on Protocols Revisit fundamental protocol design decisions n n n NACKs Broadcast: reliability TDMA: association
Effect on Protocols Revisit fundamental protocol design decisions n n n NACKs Broadcast: reliability TDMA: association CSMA: spatial reuse RTS/CTS: spatial reuse
Effect on Protocols Revisit fundamental protocol design decisions n n n n NACKs Broadcast: reliability TDMA: association CSMA: spatial reuse RTS/CTS: spatial reuse CSMA/CD: aggressiveness Rapid Flooding: latency
Measurements n We need to evaluate effectiveness q q q 15% of receptions were in a collision Up to 40 -50% transmissions collided somewhere 92% (100%) of stronger-last collisions can be recovered 84% of stronger-last collisions can be detected Both txr identifies can be recovered in 80% of detections
Changing hardware design n Wireless collision detection q q New capabilities Protocol requirements affect hardware design Hardware
Changing hardware design n This is possible q q 802. 11 generations ~12 yrs Laptop lifetime ~2 -3 yrs
Kamin v american express
Mae khamin waterfall
Whitehouse.gov
Woodrow the white house mouse
Http://www.whitehouse.gov/administration/cabinet
Http://www.whitehouse.gov/administration/cabinet
Pugyel
Univariate descriptive design
Experimental vs non experimental research
Experimental vs non experimental
Experimental vs non experimental
Experimental vs non experimental
Pedigree chart symbols
The place of useful learning
Forensic odontology would not be useful for _____.
Making data useful
Fixed assets property plant equipment
Tightly attached ceiling
The material used for cutting tools can be
