The Thinking The Thinking Telescopes Project, Telescopes Project,

RAPTOR,RAPTOR,and the TALONS and the TALONS Communication Communication

SystemSystem

Machine LearningGENIE,

ML Classifiers,Anomaly Detection

Context KnowledgeRecord of

Sky variability (Virtual Observatories),

Massive Distributed Disk Array

Robotic HardwareRobotic HardwareWide-Field Sky Wide-Field Sky

MonitoringMonitoring Rapid Response Rapid Response

Telescopes,Telescopes,Real Time Real Time

PipelinePipeline

Thinking TelescopesAn Engine for Discovery

in the Time Domain

Goal is to Integrate Three Components

System Adaptability: Querying the System Adaptability: Querying the SkySky

Traditional ApproachTraditional Approach““Hard Wired” to find specific Hard Wired” to find specific

artifacts and phenomenaartifacts and phenomena For example---not in For example---not in

previous frame, not in previous frame, not in sky catalog, and no sky catalog, and no parallaxparallax

Thinking TelescopeThinking Telescope Monitoring of persistent Monitoring of persistent

sources for important sources for important changes in real timechanges in real time

Adaptive processingAdaptive processing Machine learningMachine learning Anomaly detection and Anomaly detection and

automated classificationautomated classification ““find more like this”find more like this”

Machine LearningMachine Learning

Automated identification of artifacts and Automated identification of artifacts and transients in direct and difference images.transients in direct and difference images.

Automated classification of celestial objects Automated classification of celestial objects based on temporal and spectral properties.based on temporal and spectral properties.

Real time recognition of important deviations Real time recognition of important deviations from normal behavior for persistent sources.from normal behavior for persistent sources.

Memory and Context

http://skydot.lanl.govhttp://skydot.lanl.gov

Raptor: Sky Monitoring with Both Eyes Raptor: Sky Monitoring with Both Eyes OpenOpen

Wide-field imaging system Wide-field imaging system monitors ~1300 square-deg with monitors ~1300 square-deg with resolution ~35 arcsec and limiting resolution ~35 arcsec and limiting magnitude of R~13magnitude of R~13thth in 60 seconds. in 60 seconds. ( ( like the rod cells of the retina like the rod cells of the retina ))

Each array has a “fovea” Each array has a “fovea” telescope with limiting magnitude telescope with limiting magnitude of R~16.5 (60 sec), resolution of of R~16.5 (60 sec), resolution of ~7 arcsec and Gunn g (or r) filter. ~7 arcsec and Gunn g (or r) filter. Provides color, better resolution, Provides color, better resolution, and faster cadence light curves and faster cadence light curves ((cone cells of foveacone cells of fovea))

Rapidly slewing mount places the Rapidly slewing mount places the “fovea” anywhere in the field in <3 “fovea” anywhere in the field in <3 seconds. (seconds. (rapid eye movementrapid eye movement).).

Two identical arrays are separated Two identical arrays are separated by ~38 km. (by ~38 km. (stereoscopic visionstereoscopic vision) )

Best Solution – A Best Solution – A Distributed Sensor NetworkDistributed Sensor Network

Initial work done by military and contractors to Initial work done by military and contractors to support war fighters.support war fighters.

Sensor elements are self sustaining and autonomous.Sensor elements are self sustaining and autonomous. Sensor elements gather data on environment Sensor elements gather data on environment

independently.independently. Data is communicated back to a central location and Data is communicated back to a central location and

collaboratively processed.collaboratively processed. Working together these elements should provide a Working together these elements should provide a

better overall picture of the environment, than single-better overall picture of the environment, than single-point sensors.point sensors.

Ultimate Goal –Ultimate Goal – To make decisions or gain knowledge based on To make decisions or gain knowledge based on information fused from distributed inputsinformation fused from distributed inputs

Collaborative Signal Processing

Signal Processing

Sensing Modalities

event

data

nodes

…

DataGathering

Event Detection

DecisionMaking

ENVIRONMENT

GENERAL CONCEPT FOR DISTRIBUTED SENSOR NETWORK

Signal Processing

Sensing Modalities

data

event

The Distributed Sensor Network The Distributed Sensor Network Idea Applied to the RAPTOR Idea Applied to the RAPTOR

SystemSystem

event

data

nodes

…

DataGathering

Event Detection

DecisionMaking

data

event

DSN Qualifications DSN Qualifications

In General In General Applied to an Astronomical Applied to an Astronomical SystemSystem

Full scalability.Full scalability. Any number of systems coming and Any number of systems coming and goinggoing

Fault tolerance.Fault tolerance. System dropouts, weather, System dropouts, weather, instrument failure, instrument failure, etc. etc.

Mosaic coverage.Mosaic coverage. L Large area combined imaginingarge area combined imagining Depth of data. Depth of data. Multiple instruments on same object Multiple instruments on same object

and/or and/or a variety of instrument sensitivitiesa variety of instrument sensitivities Temporal coverage.Temporal coverage. Data covering continuous Data covering continuous

observationsobservations

TALONS TALONS ComponentsComponents

Monitor: Monitor: run from any subscriber’s run from any subscriber’s computercomputer

Shows real-time activity of client Shows real-time activity of client systems and Centralsystems and Central

Central:Central: Provides connection point for clients to Provides connection point for clients to

transmit and/or receive. transmit and/or receive. Provides security for connections.Provides security for connections. Provides cooperative analysis .Provides cooperative analysis . Relay from outside networks to clients.Relay from outside networks to clients. Logs activity.Logs activity. Filters information to and from clients.Filters information to and from clients. Issues activity requests or alerts via Issues activity requests or alerts via

sockets and e-mails.sockets and e-mails.

Client: Client: resides on each client resides on each client computercomputer

Provides connections back to Provides connections back to server receive and/or transmit.server receive and/or transmit.

Monitors connections and Monitors connections and repairs or notifies as necessary.repairs or notifies as necessary.

Filters incoming information Filters incoming information based on previousbased on previous

activity, interests, operational activity, interests, operational capability of client system.capability of client system.

Logs activity on clientLogs activity on client Prepares data for transmissionPrepares data for transmission Decodes data for responseDecodes data for response

Injector: Injector: accessed with clientaccessed with client Provides a method for manual Provides a method for manual

alert generationalert generation Provides method for manual Provides method for manual

response follow-upresponse follow-up

TALONS TALONS MonitorMonitor

TALONS TALONS InjectorInjector

The Communication The Communication PacketsPackets All Packets All Packets Packet SizePacket Size

Identifier byteIdentifier byte What type, who sent What type, who sent 1 int1 int

Data BytesData Bytes How much data to follow 1 How much data to follow 1 intint

GCNGCN Header packet (as above)Header packet (as above) GCN Alert Data All data (as per GCN packet info) 40 GCN Alert Data All data (as per GCN packet info) 40

intint TALONSTALONS

Header Packet (as above)Header Packet (as above) TALONS Data – TALONS Data – 9 int 9 int

Target Follow-up Requests (Alerts)Target Follow-up Requests (Alerts) Target of Opportunity RequestsTarget of Opportunity Requests Follow-up responsesFollow-up responses

StatusStatus Header Packet (as above)Header Packet (as above) Status Data – Instrument or Observatory StatusStatus Data – Instrument or Observatory Status 5 5

int int

NOTE: All int values are 32 bit

Breakdown of Packet Breakdown of Packet InformationInformationGCN Packets – Information varies per satellite.

Packet details at http://gcn.gsfc.nasa.gov/sock_pkt_def_doc.htmlTALONS Packets

TID –Trigger ID (Bit field definitions, encoded & decoded at client) Imagine Instrument or Observatory systems ID Follow-up or Initial spotting Either Suspected or Confirmed target type (Nova, SN, etc.) Known or new target object

1 TrigNO2 TID3 TOO4 Time5 RA6 DEC7 Mag8 MagErr9 EOF

TrigNo - Trigger (or Alert) Number

For Initial Spotting, returns a 0 to Central and Central assigns a new value.

For follow-up observations, Trigger number is passed as the event identifier

TOO – Target of Opportunity(Bit field definitions, encoded & decoded at client)Works with TID above to define type details Defines whether this packet is for a TOO Identifies if the event is transient (approaching, or receding from event. Requested Observation Type (Spectra, Photon Counting, Any, All, etc) Requested Observation Range (FIR, Gamma, Radio, etc.). Requested imaging durations (How many seconds?)Time - Time imaged expressed in TJDRA - Target CoordinateDEC - Target CoordinateMag - Magnitude of targetMagErr - Error in Magnitude measureNOTE: RA and DEC errors are to come

}

Summary Summary TALONS has been in operation now for three TALONS has been in operation now for three

years, supporting the RAPTOR systemyears, supporting the RAPTOR system TALONS can easily grow to support any TALONS can easily grow to support any

number of additional robotic and manual number of additional robotic and manual telescopic systemstelescopic systems

TALONS works well in concert with GCN. No TALONS works well in concert with GCN. No additional coding necessary to receive GCN.additional coding necessary to receive GCN.

The TALONS Client library can quickly and The TALONS Client library can quickly and easily be added to any existing telescope easily be added to any existing telescope operation software and can quickly be operation software and can quickly be configured to support the interests of the configured to support the interests of the user.user.

The information packets are flexible and can The information packets are flexible and can be changed to suit needs or additional packet be changed to suit needs or additional packet types can be added as needed.types can be added as needed.