The Proximity-Aware Transparent Handoff mobility scheme presentation in INTERCON 2011 in Lima, Peru

1. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú A Proximity-Aware Transparent Handoff MobilityScheme for VoIP Communication overInfrastructure Mesh Networks Juan Carlos Basurto, Patricia Chávez, Hernán Córdova ESPOL - Ecuador

2. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú Key Concepts 1. Wireless Mesh Network 2. QoS & VoIP 3. Mobility & Handoff 31/07/2011 2

3. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú Motivation Mobilityisessential in VoWLANimplementations. WirelessMesh Networks offerscalability, versatility and ease of implementation and access. GoodQoSlevels + FastHandoffsshould be amust. 31/07/2011 3

4. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú RetaledWork SMesh DSN Labs @ Johns Hopkins University, USA LCMIM NICTA, University of Queensland, AU 31/07/2011 4

5. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú RelatedWork - SMesh Tricksclientsintobelievingthere’soneonmipresentnode. Handoff: gARPinjectionbasedonclient’s RSSI. Custom DHCP server totrackclientmovement. Client and Routergroupsconcepts (Spines). Customaddressingscheme. Customroutingprotocol: RedundantMultipathbased. 31/07/2011 5

6. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú RelatedWork - SMesh Pros Robust FastHandoffsachieved Unmodifiedclientswitheasysetup Cons DHCP leasing couldtakeseveralseconds Excessivefloodduringhandoffs(overhead) CannotimplementotherRoutingProtocols 31/07/2011 6

7. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú RelatedWork - LCMIM PeriodicgARPbroadcasting Handoff and Mobility: Round RobinbasedgARPbroadcasting Allnodeshavetwo interfaces: Forbackhaul (different) and Forclient (unique) communication Clients’ IPsconfiguredstatically Allclientshavethesame Default GW Clientliststomanage active clients and PINGsto inactive Modification of ADOV routingprotocol Oranyother reactive routingprotocol 31/07/2011 7

8. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú RelatedWork - LCMIM Pros Highlyindependentnodes = lessoverhead FastHandoffsachieved Easytodeploy Cons ExcessivemediumfloodcausedbygARPbroadcasts Possibleclientoscillation Tiedto reactive routingprotocolsmodifications 31/07/2011 8

9. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú Proximity-AwareTransparentHandoff (PATH) Challenges: Fasthandoffs High grade of node Independence Easytoinstall and deploy Notboundtoanyroutingprotocol UnmodifiedClients 31/07/2011 9

10. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH - Overview Handoff and Mobility gARPinjectionbasedonClient’s RSSI Static IP addressing Allnodeshavetwo Interfaces: backhaul(different) and client (unique) Allclientshavethesame Default GW Clientlisttomanageconnectedclients and PING inactive Neighborliststomanagemobility No specificroutingprotocolneeded Highlydependenton WMN coverageimplementation 31/07/2011 10

11. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH – ClientMonitoring One interface is set in monitor mode Packet capture with RSSI analysis If RSSI exceeds a certainthresholdvalue: A gARPissenttothe new client TheclientisaddedtoClient’sList 31/07/2011 11

12. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH 31/07/2011 12

13. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH – InactivityCheck Keeptrack of thelast time a Clientcommunicatedwiththerouter IftheClientexceeds a fixedInactivityPeriod in value, a PING isbroadcastedtotheclient. Twopossibilities: Clientrespondstotherouter UpdateClient’sList Clientrespondsto a new router New routerwillsendgARP Ifnothingisheardafter a certainadditionalperiod of time, theclientwill be removed fromtherouter’sClient’sList 31/07/2011 13

14. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH – Sinergy Client’sMonitoring Add new clientstoClient’sList InactivityCheck PING idle clients Removeclients in Client’sList 31/07/2011 14

15. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH – Implementation WRT54G + OpenWRT 802.11 channel 11 -65 dBm of nodeoverlapping Variables set: Threshold set to -65 dBm Inactivitycheck set to 5 and 7 seconds Routing: OLSR protocol Testsran @ ESPOL Campus 6 othernetworkstransmitting at thesame time Averagenoisefloor of -90 dBm 31/07/2011 15

16. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH – Testbed 31/07/2011 16

17. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH – Tests VoIPCalls Asterisk PBX 2 Softphones 1 Laptop 1 PC PacketLoss UDP packetinjection, capture and analysis UDP Throughput iPerf 10 Mbps UDP maxtransmission 31/07/2011 17

18. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results – Delay 31/07/2011 18

19. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results - Jitter 31/07/2011 19

20. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results – DelayStationaryClient 31/07/2011 20

21. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results – DelayMovingClient 31/07/2011 21

22. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results – Handoff Forward Direction Handoff from MeshRouter1 to MeshRouter2 Handoff from MeshRouter2 to MeshRouter3 31/07/2011 22

23. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results – HandoffReversedDirection Handoff from MeshRouter1 to MeshRouter2 Handoff from MeshRouter2 to MeshRouter3 31/07/2011 23

24. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results - Throughput 31/07/2011 24

25. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú PATH Results – PacketLoss 7,171 packetssent 4 Lost Nonelostduring a handoff Packetloss = 0,05% 31/07/2011 25

26. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú Conclusions PATH achievedveryfasthandoffs High grade of nodeindependence MinimumOverhead Causedby a smallnumber of PINGs and gARPs Most of theoverheadis set bytheroutingprotocol and notbythescheme UnmodifiedClients Notboundto a specificroutingprotocol. Anyroutingprotocolmay be implemented Easytoinstall No additional DHCP leasedelays 31/07/2011 26

27. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú WaystoImprove and FutureWork Determine theoptimumInactivityPeriod and pollingintervaltowhichoverheadisfurtherminized and clientconnectivityeffectivenessismaximized. Test PATH withotherroutingprotocols and determine whichonesuitsbest. Developoptimizedversions of PATH tocovereachroutingprotocolpossibleflaws. Test with dual band routerstodifferentiateclientwithbackhaulcommunicationoptimizingcapacity and throughput. Addsecuritytotests. 31/07/2011 27

28. Universidad Nacional de Ingeniería | IEEE INTERCON 2011, Lima, Perú Thankyouforyourattention 31/07/2011 28