NeTS NOSS: Networking of Sensor Systems NSF 05-505

1
NeTS NOSS Networking of Sensor SystemsNSF
05-505

• Guru Parulkar
• Computer and Network Systems Division
• National Science Foundationgparulka_at_nsf.gov

2
Exec Summary

• NSF NOSS focus area emphasis
• common architecture
• reusable systems and science
• To enable plug and play sensor network substrate
  
• High impact projects preferred overflawless
  incremental projects
• Beware of a hammer looking for a nail trap
• Abstract problem gt Methodology gt Solution

3
Agenda

• Program context and goals
• FY04 competition
• FY05 focus and non-focus
• Proposals and review process
• Summary

4
Need Plug and Play Sensor Network Substrate
5
Focus of the Focus Area

• Monitoring Managing Spaces and Things

Applications
NetworkProgramming
HW/SWSystems
ProtocolsAlgorithms
PrivacySecurity
NetworkArchitecture
Store
Comm.
uRobots actuate
MEMS sensing
Proc
Power
technology
Miniature Connections to Physical World
6
Type of Solutions Sought
CommercialSolutions
DeployedInfrastructureApplications
ReusableSystems Science
Accelerate Progress Realize Full Potential
Close Coupling Required
FoundationsResearch
ExperimentalSystems
7
Expected Results of Program

• Foundations research
• Theoretical, algorithms and systems
• Sensor network architecture
• Systems
• Several networkable sensor platforms range of
  capabilities
• OS and protocols stacks
• Network programming environments, in-network data
  processing
• Many local testbeds with applications
• Education and Training
• New graduate and undergraduate courses with
  experimental projects
• Many graduate students with hands-on experience
• Identifiable distinct community with members from
  
• sensors, embedded hardware, os, networking,
  middleware, applications

8
Agenda

• Program context and goals
• FY04 competition
• FY05 focus and non-focus
• Proposals and review process
• Summary

9
NOSS FY-04

• 130 projects
• 15 projects selected for awards
• 11.6M
• Four panels
• Models
• Algorithms and protocols
• Architectures and systems
• Network programming

10
NOSS FY-04 Awards

• Models
• Funneling Impulses in Sensor Networks, Columbia
• Communication Patterns for Collaborative
  Reasoning in Sensor Networks, Stanford
• Algorithms and Protocols
• Semantic Internetworking of Sensor Systems for
  Efficient In-Network Information Processing,
  Boston Univ
• Exploring the Design Space of Sensor Networks
  Using Route-aware MAC Protocols, NC State

11
NOSS FY-04 Awards

• Architectures and Systems
• Creating A Wireless Sensor Net Architecture, UC
  Berkeley
• Collaborative Multiscale Processing Arch for
  Sensor Net, Rice
• Ultra Low-Power Self-Configuring Wireless Sensor
  Net, Cornell
• Lightweight and Flexible Sensor Network
  Management, Univ of Michigan -- Management
• PARIS A Framework for Privacy Augmented
  Relaying of Information from Sensors, Rutgers --
  Privacy
• Network Programming
• Programming Language and Middleware Support for
  Sensor Network Applications, UC Davis
• Sensor Coordination using Active Dataspaces, SRI
• High-Level and Efficient Sensor Network Programs,
  UCLA
• Data-Centric Active Querying in Sensor Networks,
  USC

12
Agenda

• Program context and goals
• FY04 competition
• FY05 focus and non-focus
• Proposals and review process
• Summary

13
FY-05 Solicitation Essentially Same -- Some
Difference in Mindset
14
Tremendous Pull Developing Too Quickly
Applications
Solutions fromResearch Community
Agencies
Sensors
Startups EagerStandards Body

• If we dont deliver
• Our impact will be minimized
• Too many ad-hoc solutions will create a mess

15
Something To Worry About?

• Too many point problems and point solutions
• Too many application specific solutions
• Claim that every application is unique
• There are hundreds of such unique applications
• Minimal efforts towards plug and play
• No consensus on
• Network architecture
• Software framework --- layers of software and
  APIs
• Protocol functions or layers

16
Sensor Networking Today
Appln
EnviroTrack
Hood
TinyDB
Regions
FTSP
Dir.Diffusion
SPIN
Transport
TTDD
Trickle
Deluge
Drip
MMRP
Arrive
Routing
TORA
Ascent
MintRoute
CGSR
AODV
GPSR
ARA
DSR
GSR
GRAD
DBF
DSDV
TBRPF
Scheduling
Resynch
SPAN
FPS
GAF
ReORg
Topology
PC
Yao
SMAC
WooMac
PAMAS
BMAC
TMAC
WiseMAC
Link
Pico
802.15.4
Bluetooth
Phy
eyes
RadioMetrix
CC1000
nordic
RFM
Thanks to David Culler
17
Challenge of the Moment

• How can we, the research community, lead the
  sensor network revolution by
• Championing an architecture
• Realizing the architecture in hardware and
  software
• Building testbeds and applications
• Building on each others work
• Standardizing protocols and their implementations
• to enable a programmable, robust, secured,
  manageable sensor network substrate

18
UCB Proposed Sensor Net Arch
Applications Compose what they need
Tracking Application
Sensing Application
Multiple Network Layer Protocols
Thanks to David Culler
19
Towards A Common Architecture

• Use of UCB architecture
• A framework for problem/solution definition
• Reuse components if possible
• Develop and contribute ideas and software
• UCB Proposed Sensor Net Arch details
• Scott Shenker talk
• http//webs.cs.berkeley.edu/SNA/
• Just a suggestion
• Welcome to propose competing architecture
• Comprehensive and compelling

20
Other Topics to Think About

• High-rate sensor networks
• Hari Balakrishnan
• Still a network of wireless low power sensors
• Sensor networks with limited mobility
• Bill Kaiser
• Security and privacy (same as before)
• Radha Poovendran
• New platforms
• Rajit Manohar

21
Topics Outside The Scope

• Sensor development in isolation
• New applications by themselves
• A whole range of embedded systems
• System development with
• No new capabilities
• No research content
• Modeling and algorithmic efforts
• Not aimed at real systems
• Theoretical work
• Not well motivated with its real relevance
• If in doubt please ask or send email

22
A Proposal that is NOT a Match

• represents tremendous cost to all parties and
  reduces effectiveness of the entire system
• Costs include precious time and efforts of
• PIs
• Graduate students and staff
• Departmental administrative staff
• Research office staff
• NSF administrative staff
• NSF program director(s) division director
• Reviewers
• Reduces pool of reviewers

23
Agenda

• Program context and goals
• FY04 competition
• FY05 focus and non-focus
• Proposals and review process
• Summary

24
Charge to the Review Panel

• My presentation from Information Meeting
• Guidance to put emphasis on high impact
• As opposed to flawless incremental proposals
• Guidance via email and a presentation

25
NSF Funding Decisions

• Objective and fair peer review process
• Not as good in funding high impact, bold, or
  high risk, high reward projects
• Why?
• Stiff competition leads to recommendation of
  incremental flawless projects as opposed to bold
  and potentially risky projects
• Consensus building leads to conservative
  decisions
• Decision makers do not want to fail

26
Emphasis on High Impact Projects

• High impact means more than just good papers it
  has to change practice for the better!
• It is OK to fund high-risk high-reward projects
• Even if some do not succeed
• Even if they dont have all details worked out
• It is NOT OK only to fund flawless projects that
  would predictably lead to incremental results

27
Proposals

• 10-15 acceptance rate
• Too competitive and discouraging
• Note 70 acceptance rate
• Among proposals liked by two panelists
• Majority of the proposals not serious candidates
• Why?
• PIs too busy and dont put in the necessary
  efforts
• Strong candidates too busy to serve on panel

28
Issues with Proposals
Sub Area
Blue Sky Vision
Application Class
TechnologyChoices
ProposedSolutions
Well EstablishedMethodologies
Abstract ProblemFormulation

• Overall approach reasonable
• In many cases
• Abstract Problem gt Methodology gt Solution a
  hammer looking for a nail
• Link to real sensor network vision is missing

29
My Recommendations

• Slow down
• Take time and efforts to develop a compelling
  proposal
• Talk by Ray
• Dont lose sight of the big picture
• Outline expected results and their potential
  impact
• In the context of the big picture
• Outline how your results can be generalized
• For other applications
• Used by others (hardware, software, APIs,
  theoretical results)

30
Need Partnership at Every Step of The Process
Focus AreaCommitment

• Send your best ideas consistent with focus area
• Save others for appropriate solicitation
• DONT submit same proposal to multiple programs
• NSF committed to providing all help

FundingDecisions
31
Need Partnership at Every Step of The Process
Focus AreaCommitment
InformationalMeeting
ProposalSubmission
Review Panels

• Suggest strong panelists
• Volunteer to be a panelist
• NSF committed to running objectiveand thorough
  review process

FundingDecisions
CommunityBuilding
ExecuteResearch
HighlightResults
32
Need Partnership at Every Step of The Process
Focus AreaCommitment
InformationalMeeting
ProposalSubmission
Review Panels

• Committed to funding best in support of focus
  area goals
• Due consideration to high risk high reward prop
• Committed to be responsive and timely

FundingDecisions
CommunityBuilding
ExecuteResearch
HighlightResults
33
Need Partnership at Every Step of The Process
Focus AreaCommitment
InformationalMeeting
ProposalSubmission
Review Panels

• NSF funded workshops and PI meetings
• ACM/IEEE conferences, workshops, journals

FundingDecisions
CommunityBuilding
ExecuteResearch
HighlightResults
34
A Perspective on NSF Program Director Role

• What if the final symphony turns out to be a
  cacophony?
• You are not providing your input into the system
• Program director is not doing his/her job well

35
Summary

• NSF NOSS focus area emphasis
• common architecture
• reusable systems and science
• To enable plug and play sensor network substrate
  
• High impact projects preferred overflawless
  incremental projects
• Beware of a hammer looking for a nail trap
• Abstract problem gt Methodology gt Solution