SlideShare a Scribd company logo
1 of 18
Service Composition for Mobile Ad Hoc
          Networks using Distributed Matching
                                         UCAmI 2012


                       Unai Aguilera and Diego López-de-Ipiña

          DeustoTech - Deusto Institute of Technology, University of Deusto
                                  http://www.morelab.deusto.es


                                       3 December 2012




The degeneration of rationalism                                      1/6
Problem definition


►Automatic       service composition on MANET

 ●
     User or app. → specification of required service

     ●
         Description of service inputs and outputs

     ●
         Distributed process → Composition / Graph




 UCAmI 2012                                     2/18
Architecture


►Layered      architecture → Peer-to-peer nature




 UCAmI 2012                                  3/18
Dissemination


►Service        discovery → not central directories

►Nodes         propagate service information

 ●
     Service input and outputs

►Parameter       table
     ●
         Table Message → node to node
     ●
         Maximum distance for dissemination
     ●
         Triggered by

          ●
              Neighbour dis/appearance
          ●
              Reception of new information
 UCAmI 2012                                     4/18
Dissemination


►Disseminated     information is grouped
 ●
     To reduce transmitted data

 ●
     According to a taxonomy of parameter types
     ●
       Equality, subsumption, no-relation

 ●
     Only the most generic type is propagated




 UCAmI 2012                                5/18
Service Discovery


►Uses   previously disseminated information

►Searching  node propagates an special message
 ●
     Required inputs and outputs




 UCAmI 2012                              6/18
Route Management


►Combined     with search process
 ●
     Search message → route to start node
 ●
     Response message → route to responding
     node

►Routing    table update
 ●
     Message processing
 ●
     Neighbour dis/appears

►Supported      communication
     ●
         Unicast
     ●
         Multicast

 UCAmI 2012                              7/18
Distributed Matching


►Detect          relations among compatible services

►No    centralized solution → distributed information
 ●
     No easy way to perform service matching

►  Proposed solution
 ●
   Create and maintain during topology change
     ●
       Distributed Service Graph (DSG)

 ●
     Service matching → intermediate nodes

 ●
     Detect collisions among compatible parameters

 ●
    UCAmI 2012                                  8/18
Distributed Service Graph


►Graph    distributed across the network → DSG




 UCAmI 2012                                9/18
Collision Detection


1.Intermediate nodes check propagated information
2.Match compatible I/O parameters → collision occurs




 UCAmI 2012                                 10/18
Collision Detection


3.Search for compatible services → New routes
4.Responses from services → New routes
5.DSG connection update




 UCAmI 2012                             11/18
Composition Search


►DSG     used during composition search
 ●
     Obtain a compatible work-flow
 ●
     Pre-computed paths

►Searchprocess:
 1.Create composition search special services
 2.Register special services
 3.Connection if compatibility exists
 4.Forward search




 UCAmI 2012                               12/18
Composition Search




UCAmI 2012                        13/18
Evaluation


►Network      simulator ns-2

►Extended      with AgentJ → Java code

►Simulation    parameters
 ●
     30 nodes over a 300 x 300 meters area
 ●
     100 m transmission range
 ●
     Random Waypoint → 0-5 m/s uniform dist.
 ●
     54 Mb/s transmission rate
 ●
     Max. packet size 1500 bytes
 ●
     30% of the nodes → 1 search/second


 UCAmI 2012                              14/18
Evaluation Results


►Composition     length: 3, 5, 7 services
 ●
     10 executions → average values
 ●
     Distributed Service Matching (DSM)
 ●
     Compared with Flooding Based Solution (FBS)




 UCAmI 2012                                 15/18
State of the Art


►Service      Graph has been previously applied

 ●
     Centralized solutions

 ●
     Networks with infrastructure

►Semantic      descriptions of services already used

 ●
     Service described as a whole

 ●
     Not their input and output parameters
     ●
       More adequate for functional composition

 UCAmI 2012                                 16/18
Conclusions


►Proposal
 ●
     Layered architecture: Discovery, Composition
 ●
     Response to MANET topology changes
 ●
     Avoids centralization

►Composition  search time reduction
   ●
     Pre-computed DSG
►Increases search failures
   ●
     Graph maintenance / Invalid routes

►Future    work
 ●
     Evaluate with larger networks
 ●
     Test with real devices
 UCAmI 2012                                 17/18
Thank you!




UCAmI 2012                18/18

More Related Content

Similar to Service Composition for Mobile Ad Hoc Networks using Distributed Matching

Memory Efficient Graph Convolutional Network based Distributed Link Prediction
Memory Efficient Graph Convolutional Network based Distributed Link PredictionMemory Efficient Graph Convolutional Network based Distributed Link Prediction
Memory Efficient Graph Convolutional Network based Distributed Link Predictionmiyurud
 
Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)
Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)
Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)GLC Networks
 
Empowering Real-Time Decision Making with Data Streaming
Empowering Real-Time Decision Making with Data StreamingEmpowering Real-Time Decision Making with Data Streaming
Empowering Real-Time Decision Making with Data StreamingSafe Software
 
Network LACP/Bonding/Teaming with Mikrotik
Network LACP/Bonding/Teaming with MikrotikNetwork LACP/Bonding/Teaming with Mikrotik
Network LACP/Bonding/Teaming with MikrotikGLC Networks
 
Odessa Enabling Interactive Perception Applications on Mobile Devices
Odessa Enabling Interactive Perception Applications on Mobile DevicesOdessa Enabling Interactive Perception Applications on Mobile Devices
Odessa Enabling Interactive Perception Applications on Mobile DevicesMiro Cupak
 
Queue Type on Mikrotik
Queue Type on MikrotikQueue Type on Mikrotik
Queue Type on MikrotikGLC Networks
 
Zabbix for Monitoring
Zabbix for MonitoringZabbix for Monitoring
Zabbix for MonitoringGLC Networks
 
Monitoring with Ganglia
Monitoring with GangliaMonitoring with Ganglia
Monitoring with GangliaFastly
 
Torch the light - Implementing Observability for Microservice Architectures
Torch the light - Implementing Observability for Microservice ArchitecturesTorch the light - Implementing Observability for Microservice Architectures
Torch the light - Implementing Observability for Microservice ArchitecturesSven Bernhardt
 
Stable OSPF: choosing network type.pdf
Stable OSPF: choosing network type.pdfStable OSPF: choosing network type.pdf
Stable OSPF: choosing network type.pdfGLC Networks
 
Mikrotik IP Settings For Performance and Security
Mikrotik IP Settings For Performance and SecurityMikrotik IP Settings For Performance and Security
Mikrotik IP Settings For Performance and SecurityGLC Networks
 
DOC ROI Presentation 2pm NZ3 - Duane Wilkins
DOC ROI Presentation 2pm NZ3 - Duane WilkinsDOC ROI Presentation 2pm NZ3 - Duane Wilkins
DOC ROI Presentation 2pm NZ3 - Duane WilkinsDuane Wilkins
 
Troubleshooting load balancing
Troubleshooting load balancingTroubleshooting load balancing
Troubleshooting load balancingAchmad Mardiansyah
 
MongoDB World 2019: Packing Up Your Data and Moving to MongoDB Atlas
MongoDB World 2019: Packing Up Your Data and Moving to MongoDB AtlasMongoDB World 2019: Packing Up Your Data and Moving to MongoDB Atlas
MongoDB World 2019: Packing Up Your Data and Moving to MongoDB AtlasMongoDB
 
CCNA : Intro to Cisco IOS - Part 1
CCNA :  Intro to Cisco IOS - Part 1CCNA :  Intro to Cisco IOS - Part 1
CCNA : Intro to Cisco IOS - Part 1GLC Networks
 
Building Local-loop Services for Customers
Building Local-loop Services for CustomersBuilding Local-loop Services for Customers
Building Local-loop Services for CustomersGLC Networks
 
Best Current Practice (BCP) 38 Ingress Filtering for Security
Best Current Practice (BCP) 38 Ingress Filtering for SecurityBest Current Practice (BCP) 38 Ingress Filtering for Security
Best Current Practice (BCP) 38 Ingress Filtering for SecurityGLC Networks
 
GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...
GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...
GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...Otávio Carvalho
 

Similar to Service Composition for Mobile Ad Hoc Networks using Distributed Matching (20)

Memory Efficient Graph Convolutional Network based Distributed Link Prediction
Memory Efficient Graph Convolutional Network based Distributed Link PredictionMemory Efficient Graph Convolutional Network based Distributed Link Prediction
Memory Efficient Graph Convolutional Network based Distributed Link Prediction
 
Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)
Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)
Networking in Telecommunication (signalling, tcp, ucp, ss7, sctp, sigtran)
 
Empowering Real-Time Decision Making with Data Streaming
Empowering Real-Time Decision Making with Data StreamingEmpowering Real-Time Decision Making with Data Streaming
Empowering Real-Time Decision Making with Data Streaming
 
Network LACP/Bonding/Teaming with Mikrotik
Network LACP/Bonding/Teaming with MikrotikNetwork LACP/Bonding/Teaming with Mikrotik
Network LACP/Bonding/Teaming with Mikrotik
 
Presentation_final.pdf
Presentation_final.pdfPresentation_final.pdf
Presentation_final.pdf
 
Odessa Enabling Interactive Perception Applications on Mobile Devices
Odessa Enabling Interactive Perception Applications on Mobile DevicesOdessa Enabling Interactive Perception Applications on Mobile Devices
Odessa Enabling Interactive Perception Applications on Mobile Devices
 
Queue Type on Mikrotik
Queue Type on MikrotikQueue Type on Mikrotik
Queue Type on Mikrotik
 
Zabbix for Monitoring
Zabbix for MonitoringZabbix for Monitoring
Zabbix for Monitoring
 
Monitoring with Ganglia
Monitoring with GangliaMonitoring with Ganglia
Monitoring with Ganglia
 
CQRS: Theory
CQRS: Theory CQRS: Theory
CQRS: Theory
 
Torch the light - Implementing Observability for Microservice Architectures
Torch the light - Implementing Observability for Microservice ArchitecturesTorch the light - Implementing Observability for Microservice Architectures
Torch the light - Implementing Observability for Microservice Architectures
 
Stable OSPF: choosing network type.pdf
Stable OSPF: choosing network type.pdfStable OSPF: choosing network type.pdf
Stable OSPF: choosing network type.pdf
 
Mikrotik IP Settings For Performance and Security
Mikrotik IP Settings For Performance and SecurityMikrotik IP Settings For Performance and Security
Mikrotik IP Settings For Performance and Security
 
DOC ROI Presentation 2pm NZ3 - Duane Wilkins
DOC ROI Presentation 2pm NZ3 - Duane WilkinsDOC ROI Presentation 2pm NZ3 - Duane Wilkins
DOC ROI Presentation 2pm NZ3 - Duane Wilkins
 
Troubleshooting load balancing
Troubleshooting load balancingTroubleshooting load balancing
Troubleshooting load balancing
 
MongoDB World 2019: Packing Up Your Data and Moving to MongoDB Atlas
MongoDB World 2019: Packing Up Your Data and Moving to MongoDB AtlasMongoDB World 2019: Packing Up Your Data and Moving to MongoDB Atlas
MongoDB World 2019: Packing Up Your Data and Moving to MongoDB Atlas
 
CCNA : Intro to Cisco IOS - Part 1
CCNA :  Intro to Cisco IOS - Part 1CCNA :  Intro to Cisco IOS - Part 1
CCNA : Intro to Cisco IOS - Part 1
 
Building Local-loop Services for Customers
Building Local-loop Services for CustomersBuilding Local-loop Services for Customers
Building Local-loop Services for Customers
 
Best Current Practice (BCP) 38 Ingress Filtering for Security
Best Current Practice (BCP) 38 Ingress Filtering for SecurityBest Current Practice (BCP) 38 Ingress Filtering for Security
Best Current Practice (BCP) 38 Ingress Filtering for Security
 
GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...
GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...
GaruaGeo: Global Scale Data Aggregation in Hybrid Edge and Cloud Computing En...
 

Service Composition for Mobile Ad Hoc Networks using Distributed Matching

  • 1. Service Composition for Mobile Ad Hoc Networks using Distributed Matching UCAmI 2012 Unai Aguilera and Diego López-de-Ipiña DeustoTech - Deusto Institute of Technology, University of Deusto http://www.morelab.deusto.es 3 December 2012 The degeneration of rationalism 1/6
  • 2. Problem definition ►Automatic service composition on MANET ● User or app. → specification of required service ● Description of service inputs and outputs ● Distributed process → Composition / Graph UCAmI 2012 2/18
  • 3. Architecture ►Layered architecture → Peer-to-peer nature UCAmI 2012 3/18
  • 4. Dissemination ►Service discovery → not central directories ►Nodes propagate service information ● Service input and outputs ►Parameter table ● Table Message → node to node ● Maximum distance for dissemination ● Triggered by ● Neighbour dis/appearance ● Reception of new information UCAmI 2012 4/18
  • 5. Dissemination ►Disseminated information is grouped ● To reduce transmitted data ● According to a taxonomy of parameter types ● Equality, subsumption, no-relation ● Only the most generic type is propagated UCAmI 2012 5/18
  • 6. Service Discovery ►Uses previously disseminated information ►Searching node propagates an special message ● Required inputs and outputs UCAmI 2012 6/18
  • 7. Route Management ►Combined with search process ● Search message → route to start node ● Response message → route to responding node ►Routing table update ● Message processing ● Neighbour dis/appears ►Supported communication ● Unicast ● Multicast UCAmI 2012 7/18
  • 8. Distributed Matching ►Detect relations among compatible services ►No centralized solution → distributed information ● No easy way to perform service matching ► Proposed solution ● Create and maintain during topology change ● Distributed Service Graph (DSG) ● Service matching → intermediate nodes ● Detect collisions among compatible parameters ● UCAmI 2012 8/18
  • 9. Distributed Service Graph ►Graph distributed across the network → DSG UCAmI 2012 9/18
  • 10. Collision Detection 1.Intermediate nodes check propagated information 2.Match compatible I/O parameters → collision occurs UCAmI 2012 10/18
  • 11. Collision Detection 3.Search for compatible services → New routes 4.Responses from services → New routes 5.DSG connection update UCAmI 2012 11/18
  • 12. Composition Search ►DSG used during composition search ● Obtain a compatible work-flow ● Pre-computed paths ►Searchprocess: 1.Create composition search special services 2.Register special services 3.Connection if compatibility exists 4.Forward search UCAmI 2012 12/18
  • 14. Evaluation ►Network simulator ns-2 ►Extended with AgentJ → Java code ►Simulation parameters ● 30 nodes over a 300 x 300 meters area ● 100 m transmission range ● Random Waypoint → 0-5 m/s uniform dist. ● 54 Mb/s transmission rate ● Max. packet size 1500 bytes ● 30% of the nodes → 1 search/second UCAmI 2012 14/18
  • 15. Evaluation Results ►Composition length: 3, 5, 7 services ● 10 executions → average values ● Distributed Service Matching (DSM) ● Compared with Flooding Based Solution (FBS) UCAmI 2012 15/18
  • 16. State of the Art ►Service Graph has been previously applied ● Centralized solutions ● Networks with infrastructure ►Semantic descriptions of services already used ● Service described as a whole ● Not their input and output parameters ● More adequate for functional composition UCAmI 2012 16/18
  • 17. Conclusions ►Proposal ● Layered architecture: Discovery, Composition ● Response to MANET topology changes ● Avoids centralization ►Composition search time reduction ● Pre-computed DSG ►Increases search failures ● Graph maintenance / Invalid routes ►Future work ● Evaluate with larger networks ● Test with real devices UCAmI 2012 17/18