Article about the LOOKING project by David Geer

Page 1

he Laboratory for the Ocean

Observatory

Knowledge

Integration Grid (LOOKING)

project (http://lookingtosea.ucsd.

edu) is examining technologies to

accomplish the goals of the ORION

(the Ocean Research Interactive

Observatory Networks) Program’s

cyberinfrastructure, which will make

data from automated ocean floor

observatories available to the world.

Some observatories — which

connect groups of sensors to a single

fiber optic cable or to buoys — exist

today. The sensors they employ

include a variety of data collection

instruments such as seismometers.

LOOKING and ORION are a part

of worldwide research efforts to

make data from the ocean depths

freely and instantly available to

land-based research institutions for

grid computing.

LOOKING is the predecessor to

the ORION program and is one and a

half years into its four year funded

project. ORION will provide the

federation, data modeling, and web

services in an SOA (Service Oriented

Architecture) for what will be 11

observatories strung from the coasts

of Maine to Southern California and

further off shore.

The ORION

Cyberinfrastructure

The cyberinfrastructure is in the

early stages of design. Existing

observatories are individually making

their data available as files via FTP or

HTTP download, according to

Matthew Arrot of Calit2 and member

of the ORION cyberinfrastructure

committee.

ORION will use partners who

already have grid infrastructure, web

services, and storage, and integrate

these elements to work with the

collected data.

ORION cyberinfrastructure will

support ocean floor sensor networks

in monitoring ocean environments.

Data from various regions of the

oceans will be modeled. Data models

will be trained on new data every six

hours. Measurements are gathered,

defined by the output of the previous

measurement and weighted on

clusters of information from the

measurement, according to Arrot.

The assimilated data is used to

realign the model, then the new

information is used in the next run of

the model for the next measurement

(in this way, the model improves its

own accuracy every six hours). A new

forecast covering what is expected in

the ocean environment is made every

six hours that covers the next 72

hours. This will be provided to

researchers through the Teragrid and

LOOKING at the Other

Three-quarters of the

World, Through ORION

LOOKING at the Other

Three-quarters of the

World, Through ORION

Research aims to design cyberinfrastructure that can federate

ocean floor observatories into an integrated knowledge grid.

by David Geer

December 2006 61

Page 2

December 2006

by having individual researchers plug

their systems into ORION’s Web

Services via the Internet backbone,

according to Arrot.

According to Arrot, ORION

program researchers will interact with

the instruments through a common

instrument interface. This interface is

connected to an Enterprise Service

Bus (ESB, an integration architecture

that enables incremental integration).

This ESB architecture rides on top of

the TCP/IP (or possibly UDP — yet to

be determined) protocol that will be

used across the infrastructure for data

transport.

The selected data transport

technology will also support shore-

to-sensor

communications

for

command-and-control operations and

calibration, as well as sensor-to-sensor

communications, according to Tim

McGinnis, senior engineer at the

Applied Physics Lab, University of

Washington.

The ESB uses a messaging layer

(with messaging software designed for

large scale distributed application

environments to send data in real-

time) that sends the signal, i.e., stream-

ing data containing each new interval

of measurement data, to the archives

in the repositories

and to researchers

subscribed to the

signal. Three inde-

pendent operators

will be selected

over the next six to

12 months to

supply and take

care of the archiv-

ing repositories.

One operator each

will be selected

for the coastal,

regional, and glob-

al observatories.

Underpinning

all three operators

is a fourth — the

cyberinfrastructure

independent operator — which will

be responsible for the operating

infrastructure, says Arrot.

Around the Enterprise Service Bus

there is a set of web service

templates. The web service templates

communicate about certain behavior

characteristics coming from the

sensors. Examples include the

SensorML Schema’s measurement

characteristics (sensor metadata such

as who is operating the sensor, the

sensor ID, etc.), Plug-n-Play response

characteristics (which have several

models for looking at chemistry, radia-

tion, population, and general proper-

ties in the ocean), and Manageability

Characteristics — WSDM (Web

Services Distributed Management)

specification events — which look at

lifecycles and transitions between

different states of the data. These

were standardized in LOOKING

before ORION, according to Arrot.

There are web services for

interacting with the metadata catalog

and the repository. There are web

services for dealing with storage; this

allows operators at the observatory

level to pull their own storage facilities

into the operating environment in a

standard manner. They will be able to

reach into the local storage facilities

for the individual observatories and

aggregate the catalog and guarantee

the replication of the data for

redundancy, according to Arrot. The

web services will also enable the

cyberinfrastructure to outsource some

storage requirements to supercomput-

ing centers, explains Arrot.

Researchers and entities outside

of ORION will be able to connect

through portals and web service

interfaces to interact with the reposi-

tory or subscribe to real-time data.

“There will be one or a family

of web services associated with

workflow. There are services for time

and location focused back at the

observatories and the instruments to

give them a common basis for time,”

says Arrot.

The cyberinfrastructure will use a

proprietary protocol to poll for

sensor metadata such as sensor type,

manufacturer, last calibration date and

results, sample rate, resolution, con-

version factors to engineering units,

and 3D location. NEPTUNE (North

East Pacific Time-Series Undersea

Networked Experiments) will either

use a broad, established standard or

develop its own, according to

McGinnis; if a broad, established

standard that fits the bill is not already

available, they will have to develop

their own. The cyberinfrastructure will

be complete in three or four years,

according to McGinnis.

The Observatories

The observatories will allow

researchers to collect, process, and

interact in real time with this

data, which now takes over a year to

collect.

There are three kinds of ocean

observatories, according to John

Orcutt, professor of geophysics at

Scripps Institution of Oceanography,

University of California, San Diego,

and the Principal Investigator for

LOOKING for the NSF cooperative

at UCSD.

• Global observatories — These are all

buoyed observatories. They are

anchored to the seafloor so they don’t

move around.

• Regional observatories — Mostly

cabled with instruments floated from

the seafloor with subsurface and

surface buoys attached.

FIGURE 1. This shows buoyed unmanned

undersea observatories with sensing

instruments attached to fiber optic

cable networks which send collected

data back through the observatory and

on to the rest of the world for study and

collaborative research.

Page 3

Figure 1 shows both subsurface

and surface moorings as part of the

Dynamics of Earth and Ocean

Systems (DEOS) for the regional

observatories. “The mooring is

generally taut so the buoy doesn’t

move laterally a great deal. The

mooring lines carry both power and

communications to the instruments

supported by the buoys. The picture is

similar for the regional cabled

observatory except the large, central

buoy is not included and seafloor

cables carry data and power into the

site,” says Orcutt.

• Coastal observatories — Mostly

buoyed with some cabling proposed.

These geographical classifications

(global, regional, and coastal) are char-

acterized by their implementations

(buoyed or cabled) out of practical

necessity. According to Orcutt, for

global observatories, it’s generally too

expensive to run a 2,000-2,500 km

seafloor cable to a single site. The

costs would consume all the funds

available. The only practical way

to do this is using a buoy with satellite

communications capabilities on the

surface.

For the regional cabled observa-

tory, the costs of laying cable

consume a large portion (maybe 40%)

of the funds available to ORION, “so

there is only one of these planned for

now,” says Orcutt. Coastal observato-

ries require only short cable runs of

about 100 km; “either buoys or cables

(in a small number of instances) are

possible for these,” says Orcutt.

Some early observatories are

being constructed off the US and

Canadian western coasts. According

to Alan D. Chave, senior scientist at

the Woods Hole Oceanographic

Institution, there will be observatories

on the east coast as part of the OOI

(Ocean Observatories Initiative, part

of the ORION program, which will

capitalize on new technical capabili-

ties provided by the OOI. The LOOK-

ING project is intended to serve the

OOI by providing the cyberinfrastruc-

ture design.). The Regional Cabled

Observatory (which is the first of its

kind) will be off the Pacific Northwest.

The existing observatories on the east

coast are coastal, not regional.

Major observatories now in

service include the Martha’s Vineyard

Coastal Observatory, the Long-Term

Ecosystem Observatory (LEO15) off

New Jersey run by Rutgers, and the

Victoria Experimental Network Under

the Sea (Venus) off the coast of

Canada, according to Chave. These

are all coastal observatories. These

three observatories represent three

instances of undersea sensor

equipment

connected

through

fiber-optic communication cables to

onshore ocean research institutions

that are connected to the high-

performance research networks

normally associated with “the Grid.”

No observatories will be built

with LOOKING project funds, howev-

er, as the purpose of the LOOKING

project is simply to research and

design the cyberinfrastructure that will

manage the data from observatories.

According to Chave, it may be that

these observatories will use the

product from LOOKING in a few

years. These observatories were built

with NSF and WHOI funds (MVCO),

NOAA funds (LEO-15), and Canadian

Foundation for Innovation funds

(VENUS), explains Chave.

Planned US observatories include

the Monterey Accelerated Research

System (MARS) off the coast of

Monterey, California, which will be

installed later this year. According to

Chave, the installation will include the

F/O cable, which will be laid from the

shore to the node location in 1,100 m

of water and about 60 km from the

coast. The node will be installed on

the end of the cable. All of the systems

to power and communicate with the

node will be up and running and

subsequently

MARS

will

commissioned after testing and

tuning, adds Chave.

MARS will be a test bed for

Neptune US, a regional cabled obser-

vatory in the northeast Pacific Ocean.

Service-Oriented

Architecture

“Biological sensors remain a very

big challenge,” says Orcutt, explaining

that, while there are acoustical and

optical sensors that detect and

identify biology, instruments that can

actually sample and sequence DNA

are largely still laboratory-bound.

Physical measurements (current

speed and direction, ground motion

and temperature) are all easier to

make and will be used broadly in

observatories while we learn how

to make chemical and biological

measurements, adds Orcutt.

Researchers will eventually be

able to sequence DNA and see

changes in species concentrations

over time. “It’s an enabling technolo-

gy we can build on for decades,”

includes Orcutt.

Neptune Canada (under construc-

tion) is two years ahead of Neptune

US. “That will be deploying in 2008,”

says McGinnis. Neptune US, awaiting

December 2006 63

“

LOOKING and ORION are a part of worldwide research

efforts to make data from the ocean depths freely and

instantly available to land-based research institutions

for grid computing.

”

Page 4

funding, should be complete by 2010,

he adds, noting that the National

Science Foundation is confident that it

will be funded.

The cyberinfrastructure that

results from LOOKING will provide

power,

communications,

and

resources that support a service-

oriented architecture and distributed

system, according to Chave. “It will

enable a user at the University of

Kansas, for example, to share instru-

ments with a user at Scripps to put

together their own virtual network of

instruments for experimentation.”

Data transport from all observato-

ries will be transparent, he adds, so

the user doesn’t have to translate

data between different formats from

different observatories, says Chave.

Within the observatories, relative

to the instruments, there are resource

management services, the command

and control interfaces, and a whole

family of web services associated with

operations, according to Arrot.

Between the observatories, there are

three fundamental services. One is for

data products; how they are made

available after you catalog the data.

Then, the instrument services, a

fundamental innovation created for

ORION for the ubiquitous availability

of data and interaction with the

instruments. Finally, there is a

governance layer that scopes who has

access to what.

The LOOKING grid infrastructure

will let researchers detect events such

as volcanic eruptions, changes in

surface temperatures in response to El

Nino, and tsunamis in real time,

according to Orcutt. They could

even launch autonomous underwater

vehicles on missions in response to

real-time observations, he adds. This is

a big advance over visiting ocean-

sensing equipment by ship once a

year to retrieve data.

Collected in real time, researchers

can integrate the ocean’s physical

properties

into

oceanic

and

atmospheric models to predict sea

conditions that can aid in missions

such as search and rescue.

Fiber-optics offer the chance to get

images from the ocean bed that could

help, for example, predict tsunamis,

according to Mohamed A. Osman, a

professor of electrical engineering and

computer science at Washington State

University.

With some observatories up and

running, the physical infrastructure

and data collection by automated

ocean floor observatories are proven

technologies. With the completion of

the LOOKING cyberinfrastructure, we

will be set to begin a new exploration

of the other three-quarters of the

world we live in. According to Osman,

we know more about Mars and

Jupiter than our own ocean beds. “It’s

time to understand how planet Earth

works,” Osman says. And, according

to Orcutt, we can expect discoveries

that we can’t currently envision. “Until

you can monitor things day in and day

out for a period of years, you don’t

know what you can learn from it,”

Orcutt says. NV

December 2006