a qos aware approach to service oriented communication in
play

A QoS Aware Approach to Service-Oriented Communication in Future - PowerPoint PPT Presentation

A QoS Aware Approach to Service-Oriented Communication in Future Automotive Networks Mehmet C akir , Timo H ackel, Sandra Reider, Philipp Meyer, Franz Korf and Thomas C. Schmidt 4 December 6 December 2019, Los Angeles, California 2019


  1. A QoS Aware Approach to Service-Oriented Communication in Future Automotive Networks Mehmet C ¸akir , Timo H¨ ackel, Sandra Reider, Philipp Meyer, Franz Korf and Thomas C. Schmidt 4 December – 6 December 2019, Los Angeles, California 2019 IEEE Vehicular Networking Conference (VNC) Dept. Computer Science , Hamburg University of Applied Sciences , Germany { mehmet.cakir, timo.haeckel, sandra.reider, philipp.meyer, franz.korf, t.schmidt } @haw-hamburg.de

  2. Outline 1. Introduction to In-Vehicle Networks 2. Automotive Service Classification 3. Middleware for QoS Aware Communication 4. Performance Evaluation 5. Conclusion & Outlook 1

  3. In-Vehicle Networks - State of the Art • Scenarios such as Autonomous driving and V2X pose new challenges on in-vehicle networks • Automotive services have heterogeneous communication requirements • Ethernet as high-bandwidth communication medium replaces legacy bus systems • SOME/IP introduces Service-Oriented Architecture (SOA) and promises flexibility • Time-Sensitive Networking (TSN) provides Quality-of-Service (QoS) with hard deadlines 2

  4. In-Vehicle Networks - State of the Art • Scenarios such as Autonomous driving and V2X pose new challenges on in-vehicle networks • Automotive services have heterogeneous communication requirements • Ethernet as high-bandwidth communication medium replaces legacy bus systems • SOME/IP introduces Service-Oriented Architecture (SOA) and promises flexibility • Time-Sensitive Networking (TSN) provides Quality-of-Service (QoS) with hard deadlines A mechanism is missing that merges the concepts of SOA and QoS-enhanced communication for dynamically changing communication relations. 2

  5. Our Contributions • We derived four QoS classes based on automotive service requirements • We developed an automotive specific multi-protocol stack • We designed a protocol for dynamic QoS agreements • We evaluated the performance of our middleware in simulation 3

  6. Classification of Automotive Services Class Description Examples 4

  7. Classification of Automotive Services Class Description Examples Web-based Services Globaly accessible high- Infotainment, (WS) level services Smart City 4

  8. Classification of Automotive Services Class Description Examples Web-based Services Globaly accessible high- Infotainment, (WS) level services Smart City Non time-critical Temperature, IP-based Services car control Windows Regulator (IPS) 4

  9. Classification of Automotive Services Class Description Examples Web-based Services Globaly accessible high- Infotainment, (WS) level services Smart City Non time-critical Temperature, IP-based Services car control Windows Regulator (IPS) Real-Time Services Time-critical Electronic Stability Control, (RTS) car control Rear Camera 4

  10. Classification of Automotive Services Class Description Examples Web-based Services Globaly accessible high- Infotainment, (WS) level services Smart City Dynamic Non time-critical Temperature, IP-based Services car control Windows Regulator Middleware (IPS) Services Real-Time Services Time-critical Electronic Stability Control, (RTS) car control Rear Camera 4

  11. Classification of Automotive Services Class Description Examples Web-based Services Globaly accessible high- Infotainment, (WS) level services Smart City Dynamic Non time-critical Temperature, IP-based Services car control Windows Regulator Middleware (IPS) Services Real-Time Services Time-critical Electronic Stability Control, (RTS) car control Rear Camera Static Real-Time Safety- & time-critical car Services Airbag, Brakes control (SRTS) 4

  12. Classification of Automotive Services Class Description Examples Web-based Services Globaly accessible high- Infotainment, (WS) level services Smart City Dynamic Non time-critical Temperature, IP-based Services Middleware car control Windows Regulator (IPS) Services Real-Time Services Time-critical Electronic Stability Control, (RTS) car control Rear Camera Static Static Real-Time Safety- & time-critical car Non-Middleware Services Airbag, Brakes control Services (SRTS) An in − depth explanation can be found in the paper. 4

  13. Multiprotocol Approach Services 7 Application QoS Middleware 6 Presentation 5 Session S/RTS WS IPS SOME/IP SOME/IP HTTP HTTP (optional) (optional) 4 Transport TCP UDP 3 Network IP IP 2 Data Link Time-Sensitive Networking enabled Ethernet Time-Sensitive Networking enabled Ethernet 5

  14. QoS-Negotiation Protocol Subscriber Subscriber Publisher Publisher Side Side Side Side Stage 1: Handshake Stage 1: Handshake Publisher Application register register Middleware Stage 2: Connection Stage 2: Connection Middleware Subscriber Application 6

  15. QoS-Negotiation Protocol Subscriber Subscriber Publisher Publisher Side Side Side Side start Stage 1: Handshake Stage 1: Handshake Publisher Application register register Middleware Stage 2: Connection Stage 2: Connection Middleware Subscriber Application register register 6

  16. QoS-Negotiation Protocol Subscriber Subscriber Publisher Publisher Side Side Side Side start Stage 1: Handshake Stage 1: Handshake [Service exists] [Service exists] QoSRequest Check Publisher Application register register Request QoSResponse Middleware negotiate negotiate Stage 2: Connection Stage 2: Connection Middleware Subscriber Application register register 6

  17. QoS-Negotiation Protocol Subscriber Subscriber Publisher Publisher Side Side Side Side start Stage 1: Handshake Stage 1: Handshake [Service exists] [Service exists] QoSRequest Check Publisher Application deliver deliver register register Request QoSResponse Middleware Endpoint negotiate negotiate Stage 2: Connection Stage 2: Connection [Stage1 Success] [Stage1 Success] ConnectionRequest Create Endpoint Middleware Subscriber Application register register 6

  18. QoS-Negotiation Protocol Subscriber Subscriber Publisher Publisher Side Side Side Side start Stage 1: Handshake Stage 1: Handshake [Service exists] [Service exists] QoSRequest Check Publisher Application deliver deliver register register Request QoSResponse Middleware Endpoint negotiate negotiate Stage 2: Connection Stage 2: Connection [Stage1 Success] [Stage1 Success] ConnectionRequest Create Endpoint Middleware Endpoint ConnectionResponse Subscriber Application register register deliver deliver Create Endpoint 6

  19. QoS-Negotiation Protocol Subscriber Subscriber Publisher Publisher Side Side Side Side start Stage 1: Handshake Stage 1: Handshake [Service exists] [Service exists] QoSRequest Check Publisher Application deliver deliver register register Request QoSResponse Middleware Endpoint negotiate negotiate publish publish Stage 2: Connection Stage 2: Connection [Stage1 Success] [Stage1 Success] ConnectionRequest Create Endpoint Middleware Endpoint ConnectionResponse Subscriber Application register register deliver deliver Create Endpoint Finished 6

  20. Performance Evaluation • Impact of cross-traffic on the latency of different QoS classes • Scaling of setup time in relation to the number of services • Setup time in a realistic automotive network with cross-traffic 7

  21. Latency Behaviour of Mixing Different QoS Classes CT Node 1 CT Node 2 Publisher Node Cross Traffic ≈ 950Mbit/s Critical Link Switch 1 Switch 2 RTSSubscriber Linkspeed: 1Gbit/s IPSSubscriber 8

  22. Latency Behaviour of Mixing Different QoS Classes CT Node 1 CT Node 2 50 Publisher Node Cross Traffic ≈ 950Mbit/s L End 2 End [ µ s] Critical Link 40 Switch 1 Switch 2 RTSSubscriber Linkspeed: 1Gbit/s IPSSubscriber 30 20 0 . 5 0 . 55 0 . 6 0 . 65 0 . 7 0 . 75 0 . 8 Simulation time [s] 9

  23. Latency Behaviour of Mixing Different QoS Classes CT Node 1 CT Node 2 IPS 50 Publisher Node Cross Traffic ≈ 950Mbit/s L End 2 End [ µ s] Critical Link 40 Switch 1 Switch 2 RTSSubscriber Linkspeed: 1Gbit/s IPSSubscriber 30 20 0 . 5 0 . 55 0 . 6 0 . 65 0 . 7 0 . 75 0 . 8 Simulation time [s] 9

  24. Latency Behaviour of Mixing Different QoS Classes CT Node 1 CT Node 2 IPS 50 Publisher Node Cross Traffic ≈ 950Mbit/s L AVB MAX L End 2 End [ µ s] Critical Link 40 Switch 1 Switch 2 RTSSubscriber Linkspeed: 1Gbit/s IPSSubscriber 30 L AVB max = t MTU + 3 · t AVBFrame + 2 · t Switchdelay + IPG 20 + 2 · t Nodedelay 0 . 5 0 . 55 0 . 6 0 . 65 0 . 7 0 . 75 0 . 8 Simulation time [s] 9

  25. Latency Behaviour of Mixing Different QoS Classes CT Node 1 CT Node 2 IPS 50 RTS Publisher Node Cross Traffic ≈ 950Mbit/s L AVB MAX L End 2 End [ µ s] Critical Link 40 Switch 1 Switch 2 RTSSubscriber Linkspeed: 1Gbit/s IPSSubscriber 30 L AVB max = t MTU + 3 · t AVBFrame + 2 · t Switchdelay + IPG 20 + 2 · t Nodedelay 0 . 5 0 . 55 0 . 6 0 . 65 0 . 7 0 . 75 0 . 8 Simulation time [s] 9

  26. Latency Behaviour of Mixing Different QoS Classes CT Node 1 CT Node 2 IPS 50 RTS Publisher Node Cross Traffic ≈ 950Mbit/s L AVB MAX L End 2 End [ µ s] Critical Link 40 Switch 1 Switch 2 RTSSubscriber Linkspeed: 1Gbit/s IPSSubscriber 30 L AVB max = t MTU + 3 · t AVBFrame + 2 · t Switchdelay + IPG 20 + 2 · t Nodedelay 0 . 5 0 . 55 0 . 6 0 . 65 0 . 7 0 . 75 0 . 8 Simulation time [s] Result: QoS can be guaranteed for heterogeneous client requirements 9

Recommend


More recommend