Weather Echoes

1
Collaborative Adaptive Sensing of the Atmosphere
Jim Kurose Department of Computer Science Center
for Collaborative Adaptive Sensing of the
Atmosphere (CASA) University of
Massachusetts Amherst MA 01003
COMSWARE Jan. 2007
2
Overview

• CASA collaborative adaptive sensing of the
  atmosphere
• introduction, motivation
• testbeds
• selected research challenges
• system architecture
• joint sensing/communication in energy-constrained
  environments
• incorporating end-user utilities
• discussion the big picture

3
Societal need observe, understand, detect,
predict hazardous weather

• tornadoes
• floods (3000 flash floods annually)
• land-falling hurricanes
• thunderstorms
• boundary-layer winds for particle transport
• 300 lives, 13B damage

4
The grand challenge

• Revolutionize our ability to observe, understand,
  predict and respond to weather hazards, using
  sensor networks that  sample the atmosphere where
  and when end-user needs are greatest.

5
NEXRAD (current US system)

• 158 radars operated by NOAA
• 230 km Doppler mode, 460 km reflectivity-only
  mode
• surveillance mode
• sit and spin

3 km coverage floor
6
NEXRAD (current US system)
Observational Data Push
ProcessingAlgorithms
Nowcasts NWP
3 km coverage floor
7
The Sensing Gap

• Sparse, high-power radar
• sensing gap earth curvature effects prevent 72
  of the troposphere below 1 km from being observed
• coarse resolution

10,000 ft
3.05 km
snow
wind
3.05 km
tornado
earth surface
Horz. Scale 1 50 km Vert. Scale 1 -- 2 km
0
40
80
120
160
200
240
RANGE (km)

There is insufficient knowledge about what is
actually happening (or is likely to happen) at
the Earths surface where people live. NRC 1998
8
CASA collaborative adaptive sensing of the
atmosphere

• CASA dense network of low power radars
• sense lower 3 km of earths atmosphere
• collaborating radars
• improved sensing
• improved detection, prediction
• finer spatial resolution
• responsive to multiple end-user needs

Sample atmosphere when and where end-user needs
are greatest
9
CASA higher resolution, different views
Jan. 2006 storm
CASA MA1 radar
Closest WSR-88D (86 miles away)
10
CASA dense network of inexpensive, short range
radars
This
instead of this.

• finer spatial resolution
• beam focus more energy into sensed volume
• multiple looks sense volume with most
  appropriate radars

11
CASA adaptive data pull
MCC Meteorological
data
command and control
storage
query
Meteorological
interface
streaming
Detection
storage
Algorithms
Feature Repository
F
2,H2
R1
G3
C2
G3
G3
J
R1
H1
,
F1
H1
,
F1
T
2,R1
R1
G3
C2
G3
G3
K
R1
H1
T
2,H1
T
2,R1
R1
G3
G3
G3
G3
End users NWS, emergency response
SNR
policy
data
Meteorological
Task
Generation
Resource planning,
optimization
resource allocation
12
CASA End Users
(Real) end users National Weather Service,
emergency response managers, researchers
13
Oklahoma 4-Node Test Bed
14
Puerto Rico, Off-the-Grid Test Bed
modified marine radar
802.11b/g
solar panel
battery

• no reliance on infrastructure
• solar/battery-operated nodes
• multi-antenna multi-hop 802.11
• directional antenna

15
Overview

• CASA collaborative adaptive sensing of the
  atmosphere
• introduction, motivation
• testbeds
• selected research challenges
• system architecture
• joint sensing/communication in energy-constrained
  environments
• incorporating end-user utilities
• application-level protocols routing, congestion
  control
• discussion, summary

16
Architectural worldview
17
Architecture overview
1 Mbps (moment) 100 Mbps (raw)
blackboard
prediction
30 sec. heartbeat
18
Meteorological Command and Control (MCC)

• Time sensitive decouple ingest from command
  generation

Data Ingest
Retrieval, Detection
Algorithmss
repository/blackboard
Meteorological
Task
Generation
radar targeting requests
19
Meteorological Command and Control
feature detection
data ingest, storage
LB
. . .
Radar 1
LB
Radar 2
LB
. . .
radars
1
2
3
4
5
6
7
8
9
A
G3
G3
G3
G3
G3
G3
G3
G3
G3
B
G3
G3
G3
G3
G3
G3
G3
G3
G3
C
G3
G3
G3
G3
G3
G3
G3
G3
G3
D
G3
G3
G3
G3
G3
G3
G3
G3
G3
F
2,H2
R1
G3
C2
G3
G3
J
R1
H1
,
F1
H1
,
F1
T
2,R1
R1
G3
C2
G3
G3
K
R1
H1
T
2,H1
T
2,R1
R1
G3
G3
G3
G3
feature repository
end-user preferences, policy
task generation
optimization
Michael Zink, David Westbrook, Sherief Abdallah,
Bryan Horling, Vijay Lakamraju, Eric Lyons,
Victoria Manfredi, Jim Kurose, Kurt Hondl,
"Meteorological Command and Control An
End-to-end Architecture for a Hazardous Weather
Detection Sensor Network," 2005 ACM Mobisys
Workshop on End-end Sense-and-response Systems,
June 200
20
Overview

• CASA collaborative adaptive sensing of the
  atmosphere
• introduction, motivation
• testbeds
• selected research challenges
• system architecture
• joint sensing/communication in energy-constrained
  environments
• incorporating end-user utilities
• application-level protocols routing, congestion
  control
• discussion, summary

21
Optimal joint sensing/routing in energy
constrained environments

• energy expenditures sensing, send/receive data
• each node must determine
• si sensing (data generation) rate,
• Xij how to route sensed data towards sink,
• subject to power constraints
• node decision affects others sensed data must be
  sent

22
Distributed optimization

• Algorithm
• receive marginal utility information from
  downstream nodes

• si

i

• change flow rates to downstream to balance
  marginal utility
• compute own marginal utility wrt upstream flow,
  send upstream

• convergence proof, step-size requirements,
  evaluation

C. Zhang, J. Kurose, Y. Liu, D. Towsley, An
Optimal Distributed Algorithm for Joint Resource
Allocation and Routing in Node-based Wireless
Networks, IEEE Int. Conference on Network
Protocols, Nov. 2006.
23
Optimal joint sensing/routing many open
questions!

• in-network computation (data fusion)
• data flow no longer conserved!
• considering battery recharge/drain
• implementation, measurement
• point-point directional links
• end-end system

24
On-going research efforts (computing, networking)

• application-level congestion control (CSU)
• background data transfer
• TCP-nice
• distributed, application routed,
  data-file-transfer (many-to-1)
• architecture separation of control and data
• optimization planning, prediction (Kalman
  filtering) for radar targeting

25
Overview

• CASA collaborative adaptive sensing of the
  atmosphere
• introduction, motivation
• testbeds
• selected research challenges
• system architecture
• joint sensing/communication in energy-constrained
  environments
• incorporating end-user utilities
• discussion, summary

26
Incorporating end-user utilities
1 Mbps (moment) 100 Mbps (raw)
blackboard
30 sec. heartbeat
27
Where to point?
Optimizing radar scans incorporating end user
considerations
28
Optimizing radar scans architecture!

• separation of how important, U(t,k), from how
  good,Q(t,C)
• U(t,k,Q(t,C)) would have been possible but
• complex to solve
• complex to specify and update U(t,k,Q(t,C))
• stovepipe design

29
How to define how important Ug(t,k)

• user values for detected weather features

30
How to define how important Ug(t,k)

• naturally group-sensitive utility for each
  feature (tornado, wind shear, hail core) scanned
• and the survey says..

• User feedback
• NWS want mental movie scanning areas of
  interest at regular intervals
• feature-based too jumpy
• need context scan areas around features (storm
  cell)
• dynamic data requests

31
User Utility Rules (revised)

• interval-based preferences do X every Y time
  units

32
Overview

• CASA collaborative adaptive sensing of the
  atmosphere
• introduction, motivation
• testbeds
• selected research challenges
• system architecture
• joint sensing/communication in energy-constrained
  environments
• incorporating end-user utilities
• discussion, summary

33
A big picture question

• importance of end-user

34
Seamless Mobility Functional Architecture
Its the end-user, stupid
Management
Security
User Experience
Intelligent User Interaction
Experience Architecture
Sensing, Monitoring, Control
Content Handling
Real-time Communications (Media-based Collaborati
on)
Application Domains
Connectivity Architecture
Seamless Sessions
Network
Heterogeneous Network Access
Access

• Set of solutions for being connected anywhere,
  anytime, anything with any service -Seamless
  continuity of experiences across domains, devices
  and networks

35
A big picture question

• importance of user requirements

• architecture (as opposed to stovepipe) for
  embedding user requirements into network?
• sensor networks
• content distribution
• special-purpose overlays

36
Architecture stovepipes or layers?
habitat sensing net
atmospheric sensing nets
37
Architecture stovepipes or layers?
applications
habitat sensing
atmosp. sensing
geo sensing
habitat sensing net
physical
atmospheric sensing nets
38
Wide range of sensor nets
habitat monitoring
microclimate monitoring
animal tracking
vehicle tracking in sensor field
radar/weather
video surveillance
auto traffic monitoring
satellite observation (EODIS)
underwater sensing
network traffic monitoring
in spite of differences, commonalities as well!
39

• Thanks!
• acknowledgements T. Banka, J. Brotzge, V.
  Chandrasekar, S. Cruz Pol, K. Brewster, V.
  Bringi, J. Colom, A. Defonso, B. Dolan, B.
  Donovan, K. Droegemeier, S. Fraiser, K. Hondl, E.
  Insanic, A. Jayasumana, F. Junyent, E. Knapp, V.
  Lesser, E. Lyons, D. McLaughlin, B. Philips, M.
  Preston, S. Rutledge, P. Shenoy, D. Towsley, D.
  Westbrook, L. White, M. Zink and many others!
• and to COMSWARE 2007 for the opportunity to
  speak
• slides available http//gaia.cs.umass.edu/kurose/
  talks/

40
?? / /