implementation challenges in real-time middleware for distributed autonomous systems
DESCRIPTION
Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems. Prof. Vincenzo Liberatore. Research supported in part by NSF CCR-0329910, Department of Commerce TOP 39-60-04003, NASA NNC04AA12A, a Lockheed grant, an ABB contract, and an OhioICE training grant. Motivation. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/1.jpg)
Implementation Challenges in Implementation Challenges in Real-Time Middleware forReal-Time Middleware forDistributed Autonomous Distributed Autonomous
SystemsSystems
Prof. Vincenzo LiberatoreProf. Vincenzo Liberatore
Research supported in part by NSF CCR-0329910, Department of CommerceTOP 39-60-04003, NASA NNC04AA12A, a Lockheed grant, an ABB contract, and an OhioICE training grant.
![Page 2: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/2.jpg)
MotivationMotivation
Sustainable presence on planetary surfaceSustainable presence on planetary surface Human-robotic missionsHuman-robotic missions E.g., construction, maintenanceE.g., construction, maintenance
ConsequencesConsequences Higher performanceHigher performance
Earth tele-operation inappropriate for constructionEarth tele-operation inappropriate for construction Multiple assetsMultiple assets
Communication and coordination Communication and coordination
Autonomous Distributed SystemAutonomous Distributed System
![Page 3: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/3.jpg)
Potential Scenario (Teleops)Potential Scenario (Teleops)
Tele-operationsTele-operations Robots, roversRobots, rovers Pressurized vehiclesPressurized vehicles
RequirementsRequirements Single- or multi-hopSingle- or multi-hop End-point adaptation End-point adaptation
to network non-to network non-determinismdeterminism
Quality-of-ServiceQuality-of-Service System and control System and control
metricsmetrics
Lander (Later Habitat)
Surface Terminal
4-6 Humans on EVA
AutonomousRobot
LunarRelay
TeleoperatedRobot
PressurizedVehicle
Repeater
![Page 4: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/4.jpg)
Talk OverviewTalk Overview
Bandwidth allocationBandwidth allocation
Play-back bufferPlay-back buffer
Quality-of-Service (QoS)Quality-of-Service (QoS)
DRE implementationDRE implementation
ConclusionsConclusions
![Page 5: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/5.jpg)
Bandwidth AllocationBandwidth Allocation
Objectives:Objectives: Stability of control systemsStability of control systems Efficiency & fairnessEfficiency & fairness Fully distributed, asynchronous, & scalableFully distributed, asynchronous, & scalable Dynamic & self reconfigurableDynamic & self reconfigurable
![Page 6: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/6.jpg)
Problem FormulationProblem Formulation
Define a utility fn Define a utility fn UU((rr) ) that isthat is Monotonically increasingMonotonically increasing Strictly concaveStrictly concave Defined for Defined for rr ≥ ≥ rrminmin
Optimization formulationOptimization formulation
( )
min,
max ( )
s.t. , 1,...,
and
i ii
i li l
i i
U r
r C l L
r r
S
![Page 7: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/7.jpg)
Distributed ImplementationDistributed Implementation
Two independent algorithmsTwo independent algorithms End-systems (plants) algorithm End-systems (plants) algorithm Router algorithm (later on)Router algorithm (later on)
NCS Plant NCS ControllerRouter
max
min
1( ) 1 ' ( )r
rt tp pr h U
p p
p
![Page 8: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/8.jpg)
Determination ofDetermination of k kpp andand k kii
Stability region in the Stability region in the kkii–k–kp p planeplane Stabilizes the NCS-AQM closed-loop system for Stabilizes the NCS-AQM closed-loop system for
delays less or equal delays less or equal dd
Analysis of quasi-polynomials, Analysis of quasi-polynomials, f(s,ef(s,ess))
![Page 9: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/9.jpg)
Simulations & ResultsSimulations & Results
50 NCS Plants:
( ) ( )dx
ax t bu tdt
/ ( ) a ra bKU r e
a
min ln
ar
bK abK a
()
((
))j
ju
tK
Rx
t
[Branicky et al. 2002]
[Zhang et al. 2001]
![Page 10: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/10.jpg)
Simulations & Results (cont.)Simulations & Results (cont.)
PI¤
P¤
![Page 11: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/11.jpg)
Talk OverviewTalk Overview
Bandwidth allocationBandwidth allocation
Play-back bufferPlay-back buffer
Quality-of-Service (QoS)Quality-of-Service (QoS)
DRE implementationDRE implementation
ConclusionsConclusions
![Page 12: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/12.jpg)
Information FlowInformation Flow
FlowFlow Sensor dataSensor data Remote controllerRemote controller Control packetsControl packets
Timely deliveryTimely delivery StabilityStability SafetySafety PerformancePerformance
![Page 13: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/13.jpg)
Main IdeasMain Ideas
Predictable application timePredictable application time If control applied early, plant is not in the state If control applied early, plant is not in the state
for which the control was meant for which the control was meant If control applied for too long, plant no longer If control applied for too long, plant no longer
in desired statein desired state
Keep plant simpleKeep plant simple Low space requirementsLow space requirements
Integrate Playback, Sampling, and ControlIntegrate Playback, Sampling, and Control
![Page 14: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/14.jpg)
AlgorithmAlgorithm
Send regular controlSend regular control Playback timePlayback time
Late playback okayLate playback okay ExpirationExpiration
Piggyback contingency controlPiggyback contingency control
![Page 15: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/15.jpg)
Plant outputPlant output
Open Loop Play-back
![Page 16: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/16.jpg)
Packet lossesPacket losses
Figure 8
![Page 17: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/17.jpg)
Talk OverviewTalk Overview
Bandwidth allocationBandwidth allocation
Play-back bufferPlay-back buffer
Quality-of-Service (QoS)Quality-of-Service (QoS)
DRE implementationDRE implementation
ConclusionsConclusions
![Page 18: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/18.jpg)
Network Quality-of-Service (QoS)Network Quality-of-Service (QoS)
Support real-time distributed applicationsSupport real-time distributed applications Voice, videoVoice, video Networked controlNetworked control
GuaranteesGuarantees Network metricsNetwork metrics
BandwidthBandwidthDelaysDelaysDelay jitterDelay jitterLoss ratesLoss rates
End-point metricsEnd-point metricsTracking in networked controlTracking in networked control
ExampleExample Packet prioritiesPacket priorities
Current support in InternetCurrent support in Internet Significant research and developmentSignificant research and development None of the above: best-effortNone of the above: best-effort
![Page 19: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/19.jpg)
QoS and Space NetworksQoS and Space Networks
ExamplesExamples Human-robotic missions necessitate real-time communicationHuman-robotic missions necessitate real-time communication QoS no longer only for commercial satellite networkQoS no longer only for commercial satellite network
Fully Distributed QoS Fully Distributed QoS [IWQoS 2004][IWQoS 2004] Local mechanisms to protect from global congestion risksLocal mechanisms to protect from global congestion risks Addition to planned QoSAddition to planned QoS Autonomously adaptable to QoS requirements with no human Autonomously adaptable to QoS requirements with no human
supervisionsupervision Protects from error in networks configurationProtects from error in networks configuration Suitable for Distributed Autonomous systemsSuitable for Distributed Autonomous systems Higher performanceHigher performance On the flightOn the flight
![Page 20: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/20.jpg)
The following videos were made possible by NASA funds provided by GRC under Contract NNC05CB20C
Videos:Tele-Operation, Cross-Traffic and
Distributed QoS
Note: video not included in SMC-IT proceedings
![Page 21: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/21.jpg)
Distributed QoSDistributed QoS
DefinitionDefinition Local mechanisms to protect from global riskLocal mechanisms to protect from global risk
Deployment and benefitsDeployment and benefits Addition to planned QoSAddition to planned QoS Autonomously adaptable to QoS requirements with no Autonomously adaptable to QoS requirements with no
human supervisionhuman supervision Protects from error in networks configurationProtects from error in networks configuration Suitable for Distributed Autonomous systemsSuitable for Distributed Autonomous systems Higher performanceHigher performance On the flightOn the flight
![Page 22: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/22.jpg)
Talk OverviewTalk Overview
Bandwidth allocationBandwidth allocation
Play-back bufferPlay-back buffer
Quality-of-Service (QoS)Quality-of-Service (QoS)
DRE implementationDRE implementation
ConclusionsConclusions
![Page 23: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/23.jpg)
Middleware implementationMiddleware implementation
Sophisticated commercial DRESophisticated commercial DREIssuesIssues Embedded devices with limited memory, Embedded devices with limited memory,
computation, powercomputation, power Support for real-time protocolsSupport for real-time protocols Support for network QoSSupport for network QoS Incorporate research contributionsIncorporate research contributions
E.g., bandwidth allocation, buffersE.g., bandwidth allocation, buffers
On-going workOn-going work
![Page 24: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/24.jpg)
Talk OverviewTalk Overview
Bandwidth allocationBandwidth allocation
Play-back bufferPlay-back buffer
Quality-of-Service (QoS)Quality-of-Service (QoS)
DRE implementationDRE implementation
ConclusionsConclusions
![Page 25: Implementation Challenges in Real-Time Middleware for Distributed Autonomous Systems](https://reader035.vdocument.in/reader035/viewer/2022062309/56813686550346895d9e1055/html5/thumbnails/25.jpg)
ConclusionsConclusions
Sustainable presence on planetary surfaceSustainable presence on planetary surface Human-robotic missionsHuman-robotic missions E.g. construction, maintenanceE.g. construction, maintenance
NeedsNeeds Higher performanceHigher performance Multiple assetsMultiple assets
ImplicationsImplications Network researchNetwork research
Distributed QoSDistributed QoS Middleware researchMiddleware research
Resource allocationResource allocationBuffersBuffersEmbedded implementationEmbedded implementation
Middleware research and development fits between Middleware research and development fits between NetworksNetworks Intelligent systemsIntelligent systems