1. Hierarchical Mobility Management in IP-based Wireless/Mobile Networks 8 November 2004 Sangheon Pack Seoul National University

2. Outline Introduction Hierarchical Mobile IPv6 (HMIPv6) Analytical Modeling Fault-Tolerance Scalability Adaptability Summary and Future Work References

3. Recent trend in IETF… New working groups MIP4: Mobility for IPv4 MIP6: Mobility for IPv6 MIPSHOP: MIPv6 Signaling and Handoff Optimization IP Mobility Optimizations (Mob Opts) in IRTF Analysis of Mobile IP Route Optimization considering such parameters as traffic pattern, link conditions, topology etc Alternative mechanisms for discovering a Mobility Anchor Point (MAP) in Hierarchical Mobile IP (HMIP) Evaluation of existing and new mechanisms for discovering, and selecting a target base station and/or router for handover Background (1/2)

4. Related Protocols Mobile IPv4 Low latency handoffs –draft-ietf-mobileip-lowlatency-handoffs-v4-09.txt, June 2004. Regional registration –draft-ietf-mobileip-reg-tunnel-06.txt, March 2002. Mobile IPv6 Fast Handover –draft-ietf-mipshop-fast-mipv6-03.txt, October 2003. Hierarchical Mobile IPv6 –draft-ietf-mipshop-hmipv6-02.txt, June 2004. Background (2/2)

5. Introduction Hierarchical Mobile IPv6 (HMIPv6) Mobile IPv6 De facto standard for mobility support in all IP networks High signaling overhead and long handoff latency HMIPv6 Mobility anchor point (MAP) –Local home agent (HA) Reduction of signaling overhead and handoff latency Location Privacy Benefit to network maintenance –Authentication, Authorization, Accounting etc

6. HMIPv6 Operation (1/3) MAP HA CN Internet MAP old AR new AR MAP domain MN Local BU (Home address, RCoA) (RCoA, LCoA) Home BU

7. HMIPv6 Operation (2/3) MAP HA CN Internet MAP old AR new AR MAP domain MN Local BU (Home address, RCoA) (RCoA, LCoA’)

8. HMIPv6 Operation (3/3) MAP HA CN Internet MAP old AR new AR MAP domain Local BU (Home address, RCoA’) (RCoA’, LCoA’) MN Home BU

9. Performance Study Hierarchical Mobile IPv6 Multi-level Hierarchical Mobile IPv6

10. Analytical Modeling HMIPv6 Random walk model Pedestrian movement Fluid flow model a high mobility, but static speed/moving direction Multi-level HMIPv6 (MH-MIPv6) Fluid flow model Rectangular MAP domain configuration Cost Functions Binding update cost + Packet delivery cost

11. Random walk model Discrete Time Markov Chain (DTMC) 01R-1RR-22 Random Walk Model

12. Fluid flow model 0 1 1 1 1 1 1 2 2 2 22 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 C L c /6 Fluid Flow Model

13. Binding update cost Random-walk model Fluid-flow model MAP Domain Cost functions (1/2)

14. Packet delivery cost Processing cost at the MAP Processing cost at the HA Transmission cost Total packet delivery cost CN HA MAP MN Cost functions (2/2)

15. Binding update cost vs. User mobility (Random) R=1 R=4 Numerical Result (1/6)

16. Binding update cost vs. User mobility (Fluid) R=1R=4 Numerical Result (2/6)

17. Packet delivery cost vs. User population Random walk model Fluid flow model Numerical Result (3/6)

18. Total cost vs. SMR Random walk model Fluid flow model Numerical Result (4/6)

19. Optimal domain size (Random) Static MN Dynamic MN Numerical Result (5/6)

20. Optimal domain size (Fluid) Static MN Dynamic MN Numerical Result (6/6)

21. Multi-level HMIPv6 (MH-MIPv6) Multi-level HMIPv6 Decrease of binding update cost Further localization of binding update traffic Increase of packet delivery cost More processing at the MAP and tunneling Trade-off relationship BU cost vs. PD cost What is the optimal level a given network environment? Effect of SMR (Session to mobility ratio)

22. MH-MIPv6 System Model RMAP: Level 0 IMAP: Level 1 IMAP: Level 2 LMAP: Level D ………….. Level 0 Level 1 Level 3 Level 4 Level 2

23. BU Cost Fluid-flow model

24. PD Cost Optimal Level

25. Numerical Result (1/3)

26. Numerical Result (2/3)

27. Numerical Result (3/3)

28. Fault-Tolerance in HMIPv6 Robust Hierarchical Mobile IPv6 (RH-MIPv6)

29. RH-MIPv6 Robust Hierarchical Mobile IPv6 Multiple Binding Update Primary MAP: Primary RCoA Secondary MAP : Secondary RCoA Pros Faster failure detection/failure recovery Higher throughput Cons (Minor) Modification to the end host

30. CN P-MAP HA MN S-MAP AR 1. Receive RAs with MAP option from all available MAPs 2. Select a P-MAP and a S-MAP 4. BACK 5. BU with P_RCoA 5. BU with P_RCoA (P=1) 3. Local BU with P_RCoA (P=1) RH-MIPv6 Operation (1/2)

31. CN P-MAP HA MN S-MAP AR 6. BU with S-RCoA (P=0) Data 7. BU with S-RCoA (P=0) RH-MIPv6 Operation (2/2)

32. CN P-MAP HA MN S-MAP AR 1. Failure detection 3. Data transmission through S-MAP 5. Packet Tunneling 7. BU to HA with S-RCoA 6. Failure detection and serving MAP replacement 4. Mapping table update 8. Binding table update (P-RCoA -> S-RCoA) 2. Binding cache update Failure detection/recovery (1/2)

33. CN P-MAP HA MN S-MAP AR 1. Failure detection 3a. BU to CN with S-RCoA 3b. Data transmission through S-MAP 3a. BU to HA with S-RCoA 2. Serving MAP replacement 4a. Binding table update (P-RCoA -> S-RCoA) 4a. Binding cache update 4b. Mapping table update Failure detection/recovery (2/2)

34. Performance Evaluation Semi-Markov Chain MAP unavailability Blocking probability Ns-2 simulation TCP/UDP throughput Normal 0 Undetected 1 Detected 2 P 0,1 P 1,2 P 2,0 Packet Arrival MAP Advertisement Interval (T A ) Failure t0t0 t FA tPtP Failure t1t1 TITI t 0 +T A tFtF Packet Arrival tAtA

35. Numerical Results (1/4) [MAP unavailability]

36. Numerical Results (2/4) [MAP blocking probability: low load]

37. Numerical Results (3/4) [MAP blocking probability: High load]

38. Numerical Results (4/4) [TCP Throughput]

39. Scalability Mobility-based Load Control (MLC) Scheme

40. MLC Mobility-based Load Control (MLC) at MAP Threshold-based admission control algorithm To give a higher priority to ongoing MNs than new MNs SMR-based replacement algorithm Invoked when there is no capacity at the MAP Pros Lower ongoing MN dropping and new MN blocking prob. Lower binding update traffic, MAP processing latency Cons MAP maintains the SMR for each MN

41. MLC Operation New MN?C used <C MAP C used <K Yes No Receive a BU MAP waits a binding update message from an MN The MN is accepted C used ++ The MN is accepted C used ++ Is there an MN with a higher SMR than δ The MN is rejected The MN is accepted The chosen MN is redirected Yes No Yes

42. Performance Analysis Markov Chain New MN Blocking Probability Ongoing MN Dropping Probability Binding Update Cost MAP Processing Latency: M/G/1 0 12 C MAP -1 C MAP K-1 K K+1 …… 23 K K-1K+1 C MAP -1 K+2 C MAP

43. Numerical Results (1/4) [New MN blocking probability]

44. Numerical Results (2/4) [Ongoing MN dropping probability]

45. Numerical Results (3/4) [MAP Processing Latency]

46. Numerical Results (4/4) [new MN load] [Ongoing MN load]

47. Adaptability Adaptive (Local) Route Optimization Scheme

48. ALRO Adaptive (Local) Route Optimization: A(L)RO SMR-based binding update Binding update to CN CN with a high SMR: binding update with LCoA CN with a low SMR: binding update with RCoA Binding update to MAP/HA Binding update with RCoA (No change) Pros Reduction of binding update, packet delivery costs Cons SMR monitoring and estimation

49. Non Optimal Local Route Problem AR MAP Non-optimal route Optimal route

50. ALRO Operation CN1MAPMNMAP BU (LCoA M, RCoA C1 ) BACK (RCoA C1, LCoA M ) CN2 BU (RCoA M, RCoA C2 ) BACK (RCoA C2, RCoA M ) (a) (b)

51. Analytical Modeling No Route Optimization (NRO) Global Route Optimization (GRO) Local Route Optimization (NRO) Adaptive Local Route Optimization (ALRO)

52. Numerical Result (1/2) [Optimal SMR threshold. vs. packet size] [Total cost vs. SMR]

53. Numerical Result (2/2) [Session delivery time]

54. Summary and Future Works In this dissertation Analytical Model Fault-tolerance, Scalability, and Adaptability Future works Periodical check-pointing and forced binding update Pointer forwarding strategy Adaptive MAP selection scheme Location Privacy

55. References Analytical Modeling S. Pack and Y. Choi, "A Study on Performance of Hierarchical Mobile IPv6 in IP-based Cellular Networks,” IEICE Transactions on Communications, Vol. E87-B, No. 3, pp. 462-469, March 2004. S. Pack, M. Nam, and Y. Choi, "A Study on Optimal Hierarchy in Multi- Level Hierarchical Mobile IPv6 Networks,” ” in Proc. IEEE GLOBECOM 2004, November 2004. (Under journal review) Fault Tolerance S. Pack, T. You, and Y. Choi, "Performance Analysis of Robust Hierarchical Mobile IPv6 for Fault-Tolerant Mobile Services," IEICE Transactions on Communications, Vol. E87-B, No. 5, pp. 1158-1165, May 2004. T. You, S. Pack, and Y. Choi, "RH-MIPv6: An Enhancement for Survivability & Fault Tolerant Mobile IP Systems," in Proc. IEEE VTC 2003-Fall, October 2003. (Under journal review)

56. References Scalability S. Pack, T. Kwon, and Y. Choi, "A Mobility-based Load Control Scheme at Mobility Anchor Point in Hierarchical Mobile IPv6 Networks,” in Proc. IEEE GLOBECOM 2004, Dallas, USA, November 2004. (Under journal review) S. Pack, B. Lee, and Y. Choi, "Load Control Scheme at Local Mobility Agent in Mobile IPv6 Networks, ” in Proc. WWC 2004, San Francisco, USA, May 2004. Adaptability S. Pack, T. Kwon, and Y. Choi, "Adaptive Local Route Optimization in in Hierarchical Mobile IPv6 Networks,” Submitted. S. Pack, T. Kwon, and Y. Choi, "A Comparative Study of Mobility Anchor Point Selection Schemes in Hierarchical Mobile IPv6 Networks,” in Proc. ACM MobiWac 2004, September 2004.