Mobile Communications Exercise: Adhoc Networks and Transport Layer Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 1/34
Exercise: Adhoc Networks 1 N 1 a) Why is routing in multi-hop ad hoc networks complicated, what are the special challenges? Why should existing routing protocols from classical wired networks not be reused? b) Recall the distance vector and link state routing algorithms for fixed networks. Why are both difficult to use in multi-hop ad hoc networks? Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 2/34
Solution: Adhoc Networks 1.a a) Why is routing in multi-hop ad hoc networks complicated, what are the special challenges? Why should existing routing protocols from classical wired networks not be reused? Very frequent topology changes compared to wired networks Varying propagation characteristics Lack of a central instance Reuse of routing protocols may technically be possible, but very inefficient and error prone Specialized protocols can greatly increase efficiency Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 3/34
Solution: Adhoc Networks 1.b b) Recall the distance vector and link state routing algorithms for fixed networks. Why are both difficult to use in multi-hop ad hoc networks? Both algorithms assume a (more or less) stable networks (seldom topology changes compared to routing information exchange) BUT: topology of ad-hoc networks may change often Both algorithms build routing tables independent of demand High communication burden for each topology change Routing information may be outdated, when communication happens right after topology change Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 4/34
Exercise: Adhoc Networks 2 N 2 a) Please outline the categories or classes into which MANET routing protocols can be characterized. Please also give a brief explanation of the differences of the classes and name at least one example protocol per class. b) What are the benefits of location information for routing in ad hoc networks, which problems do arise? Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 5/34
Exercise: Adhoc Networks 2.a a) Please outline the categories or classes into which MANET routing protocols can be characterized. Please also give a brief explanation of the differences of the classes and name at least one example protocol per class. Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 6/34
Solution: Adhoc Networks 2.a Table-driven / Proactive routing q Maintain routes to all other nodes permanently q Constant, high signalling overhead Hybrid routing On-demand-driven / Reactive routing q Routes are discovered if needed q Delayed packet forwarding since route must be established first q Signalling overhead depends on traffic patterns Cluster-based/Hierarchical q Nodes have different roles, cluster-heads determine routes q Dependant on scenario and application Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 7/34
Solution: Adhoc Networks 2.a Unicast routing protocols for MANETs (topologie-based) Table-driven/ pro-active Hybrid On-Demand -driven/reactive Cluster-based/ hierarchical Distance- Vector DSDV... Link- State OLSR TBRPF FSR STAR... ZRP... DSR AODV TORA... not covered: position-based routing protocols LANMAR CEDAR... Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 8/34
Solution: Adhoc Networks 2.b b) What are the benefits of location information for routing in ad hoc networks, which problems do arise? Benefits Outgoing route can be selected based on direction of destination Problems Privacy Local Minimum E" A" Z" B" Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 9/34
Exercise: Adhoc Networks 3 N 3 a) What makes DSR different from Distance-Vector and Link-State Routing? b) Please outline the steps of route discovery between nodes A and Z in the following network. Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 10/34
Solution: Adhoc Networks 3.a a) What makes DSR different from Distance-Vector and Link-State Routing? Distance-Vector / Link State Routes are discovered pro-active Routers keep routing tables to route packets Dynamic Source Routing Routes are discovered on-demand Source node stores (and determines) route, no routing tables on Routers Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 11/34
Solution: Adhoc Networks 3.b b) Please outline the steps of route discovery between nodes A and Z in the following network. Step 1 [Z, A, 42] A E B D G C F Z Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 12/34
Solution: Adhoc Networks 3.b b) Please outline the steps of route discovery between nodes A and Z in the following network. Step 2 A [Z, A/E, 42] E [Z, A/B, 42] B D G C F Z Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 13/34
Solution: Adhoc Networks 3.b b) Please outline the steps of route discovery between nodes A and Z in the following network. Step 3 A E B D G [Z, A/E/G, 42] C [Z, A/B/D, 42] [Z, A/E/D, 42] F Z [Z, A/B/C, 42] Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 14/34
Solution: Adhoc Networks 3.b b) Please outline the steps of route discovery between nodes A and Z in the following network. Step 4 A E B D G C F Z Path: A E G Z [Z, A/B/D/F, 42] Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 15/34
Exercise: Transport Layer 4 N 4 a) What is the reaction of standard TCP in case of packet loss? In what situation does this reaction make sense and why is it quite often problematic in the case of wireless networks and mobility? b) Can the problems using TCP be solved by replacing TCP with UDP? Where could this be useful and why may this be dangerous for network stability? c) Please name and very briefly describe the mechanisms discussed in the lecture to mitigate problems of TCP in wireless networks. Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 16/34
Solution: Transport Layer 4.a a) What is the reaction of standard TCP in case of packet loss? In what situation does this reaction make sense and why is it quite often problematic in the case of wireless networks and mobility? Problem Statement TCP interprets packet loss as (temporary) overload situation TCP reacts with the slow-start mechanism In wired networks: Packet losses usually indicate overload. In mobile networks, packet losses may occur due to Transmission errors Roaming due to mobility Actual congestion in the network Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 17/34
Solution: Transport Layer 4.b b) Can the problems using TCP be solved by replacing TCP with UDP? Where could this be useful and why may this be dangerous for network stability? UDP + Higher throughput for (some) users - No congestion control - no reliable data transmission Fast overload of networks Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 18/34
Solution: Transport Layer 4.c c) Please name and very briefly describe the mechanisms discussed in the lecture to mitigate problems of TCP in wireless networks. Indirect TCP Split of TCP connection into 2: MN <-> AP, AP <-> CN Snooping TCP AP snoops into packet stream and retransmits packets locally Mobile TCP Set sender window to 0 upon network interruption, freeze transfer Fast retransmit / Fast recovery After roaming, MN sends multiple ACKs, CN avoids slow-start Transmission/time-out freezing MN + CN freeze connection when wireless network becomes unavailable Selective retransmission Receiver ACKs single packets not sequences, avoids retransmissions Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 19/34
Solution: Transport Layer 4.c Approach Mechanism Advantages Disadvantages Indirect TCP splits TCP connection isolation of wireless loss of TCP semantics, into two connections link, simple higher latency at Snooping TCP M-TCP Fast retransmit/ fast recovery Transmission/ time-out freezing Selective retransmission snoops data and acknowledgements, local retransmission splits TCP connection, chokes sender via window size transparent for end-toend connection, MAC integration possible Maintains end-to-end semantics, handles long term and frequent disconnections simple and efficient handover problematic with encryption, bad isolation of wireless link Bad isolation of wireless link, processing overhead due to bandwidth management avoids slow-start after roaming mixed layers, not transparent freezes TCP state at independent of content changes in TCP disconnect, resumes or encryption, works for required, MAC after reconnection longer interrupts dependent retransmit only lost data very efficient slightly more complex receiver software, more buffer needed Wolf-Bastian Pöttner, IBR, TU Braunschweig, www.ibr.cs.tu-bs.de 20/31