eda symposium 7-9 october 2006 2 nd event processing symposium military scenario use case 7-9...
TRANSCRIPT
EDA Symposium 7-9 October 2006
2nd Event Processing Symposium
Military Scenario Use Case
7-9 October 2006Hosted by Oracle
Dr. Opher EtzionGreg Porpora
EDA Symposium 7-9 October 2006
Use Case Detailed Description This Use Case will represent a Defense – Naval Real-time
Complex Event Processing flow across the entire Enterprise of a Naval Combatant from Sensors to Command & Control (C2) to Weapons depicting the complex flow of information and events necessary for the ship to sense, track, assess, engage and destroy an incoming missile threat. The scenario will be unclassified, simplistic and representative to only a general/notional sense of actual Naval weapon systems.
The scenario will depict and examine the interacting relationships of events and information across multiple time epochs and systems throughout our imaginary ship.
EDA Symposium 7-9 October 2006
Detailed Naval Scenario DescriptionThis scenario represents a sequence of events necessary for a Naval ship to defend its self against a missile threat. This ship has 10 seconds to react and destroy the missile, in that time our Real-Time Event Driven Architecture must assess, correlate, transform, enrich and route a multitude of complex event streams across three time epochs from sensors to command and control (C2) and finally weapons across its associated Real-time Enterprise Service Bus (ESB).
1000m = .5 NM 6000m = 3 NM
SPD : 820 mph = 1203 ft/sec
T0T0 + 10 sec
Impact point
Normal Ops
t0
Detection Correlation Classification Tracking Validation Assess Confirm Task Execute
T0 + 1 T0 + 2 T0 + 3 T0 + 4 DestroyT0 + 5 T0 + 8 T0 + 9T0 + 7T0 + 6
SENSORS COMMAND & CONTROL(C2)
WEAPONS
SPD : 1000 mph = 1467 ft/secss ms
µs
RT- ESB
RT EDA-SOA
C2
ECM
PhasedArray
RADARMissile
LauncherSilkworm
RAM
EDA Symposium 7-9 October 2006
Detailed Use Case Event FlowTime Sequence
Process Activity Event Thread Activity QoS SLA for
RT-Middleware
EDA RT- ESB Activity
Complex Event Process
Phase 1 Detection •Radar & Electronic Counter measures systems detect contact
•Send contact data to tracker
•Latency Budget (time-based, dead-line)•Ownership Strength•Reliability
•Discovery
•QoS Validation
•Monitoring
•Linkage
•Validate
•Route
Phase 2 Correlation •Combine contacts via correlation algorithm
•Compute Pd >.9
•Build Track File
•Latency Budget (time-based, dead-line)
•Discovery
•QoS Validation
•Monitoring
•Store and Hold
•Enrich
•Transform
Phase 3 Classification •Compare contact electronic signature to class library
•Compare contact SPD & BNG to Threat thresholds
•Classify as Hostile Missile and send alert for C2
•Latency Budget (time-based, dead-line)•Ownership Strength
•Discovery
•QoS Validation
•Monitoring
•Store and Hold
•Linkage
•Enrich
•Transform
•Route
Phase 4 Tracking •Tag Track file and assign highest priority
•Send to Command & Control (C2) as highest alert track
•Lower priority of other track and non-essential processes and re-task IT assets
•Latency Budget (time-based, dead-line)•Ownership Strength
•Discovery
•QoS Validation
•Monitoring
•Linkage
•Enrich
•Route
Sensors Time Epoch: 500us – 2ms
EDA Symposium 7-9 October 2006
Detailed Use Case Event FlowTime Sequence
Process Activity
Event Thread Activity QoS SLA for
RT-Middleware
EDA RT- ESB Activity
Complex Event Process
Phase 5 Validation •Interrogate Track against Identification Friend or Foe (IFF) systems
•Latency Budget (time-based, dead-line)•Ownership Strength•Durability•Reliability
•Discovery
•QoS Validation
•Monitoring
•Enrich
Phase 6 Threat Assessment
•Compare track to other high priority tracks and assign Threat Assessment as Priority 1
•Latency Budget (time-based, dead-line)•Ownership Strength
•Discovery
•QoS Validation
•Monitoring
•Store and Hold
•Enrich
•Transform
Phase 7 Confirmation •Probability of Detection (Pd) + Threat Assessment > .9
•Send Engage Order to Weapon Systems
•Latency Budget (time-based, dead-line)•Ownership Strength•Reliability
•Discovery
•QoS Validation
•Monitoring
•Linkage
•Transform
•Route
Command & Control (C2) Time Epoch: 2ms –
250ms
EDA Symposium 7-9 October 2006
Detailed Use Case Event Flow
Time Sequence
Process Activity
Event Thread Activity QoS SLA for
RT-Middleware
EDA RT- ESB Activity
Complex Event Process
Phase 8 Tasking •Task Track to best engagement weapon (Missile System) for engagement
•Select Missile cell
•Warm-up Missile
•Task Counter-Measure system to activate
•Launch Counter-Measures
•Program Missile with Track data
•Latency Budget (time-based, dead-line)•Ownership Strength•Reliability
•Discovery
•QoS Validation
•Monitoring
•Linkage
•Enrich
•Transform
•Route
Phase 9 Execution •Activate Guidance
•Deactivate Fail-Safe
•Open Vertical Launch Hatch
•Launch Missile
•Start Engagement Sequence
•Latency Budget (time-based, dead-line)•Reliability
•Discovery
•QoS Validation
•Monitoring
•Linkage
•Enrich
•Transform
•Route
Phase 10 Destruction •Receive Missile Track Intercept data
•Receive Detonation signal from Missile
•Latency Budget (time-based, dead-line)•Reliability
•Discovery
•QoS Validation
•Monitoring
•Linkage
•Route
WeaponsTime Epoch: 1us - 500us
EDA Symposium 7-9 October 2006
High Level Missile Engagement Flow
PhasedArray
RADAR
ElectronicsCounterMeasureSystem H
igh
Pe
rfo
rma
nc
e E
SB
Sensors
ThreatEngagement
ProcessorH
igh
Pe
rfo
rma
nc
e E
SB
MissileControlSystem
VerticalMissile
Launcher
Medium Performance ESB
Command &
Control
GCS
GRS
Me
diu
m P
erf
orm
an
ce
ES
B
INTELL
AREPS
PC-IMAT
NITES II
CADRTS
Me
diu
m P
erf
orm
an
ce
ES
B
Track Manager
High Performance ESB
Weapons
Command and Control
Phase 1
Phases 2 - 7
Phases 6 - 10
Total Time Sequence ~ 750 ms
Total Time Sequence ~ 1250 ms
Total Time Sequence ~ 5000 ms
EDA Symposium 7-9 October 2006
Multiple domains of time-dependent control
Non-real time
seconds
milliseconds
microseconds Radar Weapon
Tracking
Command & Control
Logistics
EDA Symposium 7-9 October 2006
An SCA modellers view
Tracking subsystem
C2 subsystem
Sonarsubsystem
Weapons subsystem
Geospatialsubsystem
Intelligencemanagement subsystem
ECMsubsystem
Radarsubsystem
DPE1A
DPE2
DPE1U
DPE3
Track Module
Fuser AnalyserDPE4
DPE5
MissionDoctrine Module
DPE6
Threatengage WCS
Missilelauncher
DPE7 DPE8
GeospatialReplication
SITUprocessor
GeospatialCollaborator
GeospatialData
Meteo
Nav
EDA Symposium 7-9 October 2006
High Level Weapon System Context Diagram
Non-OrganicInputs
Situational AwarenessProcessor
Geo-SpatialCollaboration
Processor
Track & MissionDoctrine/Policy
Global TrackFederation and Fusion
Processor
Threat trackAnalyzer
Geo-SpatialReplication
Service
ThreatEngagement
Processor
Meteorological&
Ocean EnvironmentProcessed Data
Ships NavigationData
Environment PredictionAnalysis
For Weapons
WeaponEngagement
Manager
WeaponResource
Scheduling
WeaponSelector
MissileController
MultimodePhasedRadar
ElectronicCounter-Measure
System
UnderwaterDetection
SonarSystem
GlobalCommand
&ControlSystemMaritime
Multi-DisplaySystem
IntelligenceManagement
System Filtered Intelligence
Common Operation Picture
Air & Surface Tracks
Air & Surface Tracks
Under
water
Tra
cks
Sur
face
Tra
cks
Non-OrganicInputs
Track ClassificationDoctrine
Updated as needed
All Tracks overlayed on Geo-Spatial
Context
Track ClassificationDoctrine
Updated as needed
Integrated/FusedGeo-Spatial Content
Threat Tracks
SelectedThreat Tracks
Prioritized/SelectedThreat Tracks
EDA Symposium 7-9 October 2006
WeaponsCEP
C2ISR
Seconds : Near Real-Time
Milliseconds : Soft Real-Time
Microseconds : Hard Real-Time
The Complex Event Processing engine within an Event Driven system’s Enterprise Service Bus (ESB) must be able to recognize event patterns across the functional and technology gradients of an ecosystem to the lowest level of determinism and route the aggregated results.
Finder
Shooter
Decider&
Connector
Time Epochs
EDA Symposium 7-9 October 2006
Complex Event Processing and Event-based Architecture Support
Validate
Enrich
Transform
Route
Operate
Filtering; Temporal and causality constraints
Aggregation; event-data join
Intelligent routing; content-based routing; event-driven flows
Service invocation; alert; dynamic flows
Create “complex event”
Event Source
Event Consumer
Summary of activity over 1 hour
ComplexEvent
Processor
EDA Symposium 7-9 October 2006
F1
Node: ISR1
F2
F3
F1
Node: ISRn
F2
F3
D1
Node: C21
Co
nn
ecto
r In
fras
tru
ctu
re
D1
Node: C22
CEP ActionS1
Node: S1
CEP ActionS2
Node: S2
CEP ActionS3
Node: S3
CEP ActionSn
Node: Sn
Dee
p S
enso
r an
d H
isto
rica
l d
ata
Rea
chFINDERS•Intelligence•Surveillance•Reconnaissance
DECIDERS•Command & Control
SHOOTERS•Weapons
Near Real-Time Soft Real-Time Hard Real-Time
EDA Symposium 7-9 October 2006
High level Event Processing Flow for Naval Use Case
Sensor 1
Sensor 2
Sensor 3
EP 1
TrackData
Sensor 1
Sensor 2
Sensor 3
AlertTrigger
EP 2
A1orA2orA3
E[Alert trigger (Ax)]
)321( SSSAx
)321( orAsorAsAsAT
E[Alert Track]
EP 4
Fire Intercept Missile
)(ATIFFAT
EP 3 EP 5
ASSET 1
ASSET 2
ASSET n
AssetsThat canIntercept
(FINDERS)
IFF(AT) Trigger
E[Alert Confirm]
Compute Intercept PointAssign asset
E[Alert Track]
E[Intercept Asset (Asn)]
(DECIDER) (SHOOTER)
Total Time from Detect to Engage = 10 seconds
Microsecond ESB
Millisecond ESB
Second(s) ESB
EDA Symposium 7-9 October 2006
Problem Space Inability to manage Latency and Determinism Lack of tools Large Data ingest High message traffic Large Legacy Domain Lack of temporal interlock Complexity or perception Ability to abstract Domains for determinism is a relative term Reactive versus proactive Take RT Event and compare against massive amount of history Simple patterns Making incremental change based on incremental pieces of data Must think of the problem at different levels of the stack (Vertical
versus Horizontal)
EDA Symposium 7-9 October 2006
Problem Domain at a High level - Continued With these three key parameters (Bandwidth/Communication, Memory, CPU processing) how
do we architect an EDA IT infrastructure that can meet the Deterministic latencies required to intercept the Missile. However, if we examine the end to end Missile engagement process it is a very large Stochastic
Process whereby there is only a probability of achieving RT Determinism
The most critical constraint is provision in the face of either high data ingest or faults
How do we degrade gracefully in this situation…. How do we trade off Urgency of processing versus importance
How will your system perform in the presence of overload and resource starvation ( Shed Load, lockup, etc.)
It is not good enough to just reduce priorities or make something a higher priorities, Resource management and provisioning play crucial role here this goes for memory, CPU and BW). Also not all processes can be arbitrarily stopped some provide critical admin and support services that must continue to execute otherwise we run the risk of catastrophic system collapse
This problem is not unlike a futures market whereby you ensure against risk by over provisioning
We work well on a per single node bases but when executing multiple JVM’s on a single CPU resource provision and scheduling become more complex.
EDA Symposium 7-9 October 2006
Value of Using RT Event Processing In specific Domain Spaces i.e. DoD can not meet 2010 Mission Requirements
without RT CEP
Address the latency Gap
Become more proactive to predict future events
Modelling based on historical data and current understanding of event cloud to predict near term outcomes with updating for continuous learning
Real-time is relative term depending on required predicted outcome
Extract to higher levels of abstraction for inference
EDA Symposium 7-9 October 2006
Market Drivers, Motivators, Challenges and Entry Points Drivers
Predictability
Adaptability
Dynamic
Latency and Determinism management Motivators
Tactical as well as strategic information advantage
Get inside competitive decision loop Challenges
Legacy Integration
Cost
Complexity
Entry Points Start small
EDA Symposium 7-9 October 2006
Where do we go from Here ?
Better understanding of respective Domain space requirements and constraints
Standards
Tools Modelling Development Trace and Monitoring