1. Ding-Chau Wang, Weiping He, Ing-Ray Chen Virginia Tech Presented by Weisheng Zhong and Xuchao Zhang CS 5214 (Fall 2015)

2. 2  Background  DMAPwSR Scheme  Performance Model  Experiment

3.  Mobile IPv6 (MIPv6) is a network protocol for enabling mobility in IPv6 networks.  It allows mobile nodes (MN) to move within IP-based networks while maintaing on-going connections.  Two major sources of traffic in MIPv6 systems are due to mobility management and service management. 3

4.  MNs are expected very active with significant mobility.  The mobility rate with which subnets are crossed by MNs can be high, causing a high signaling cost for MNs to infrom the MN's home agent (HA) and coresponding nodes (CNs) of the address change.  Hierarchical IPv6 uses a Gateway Foreign Agent (GFA) to keep track of the MN's current care-of- address (CoA) when MN moves within a region.  And when MN moves to a new region, it registers with a new GFA whose address is updated to the HA as current regional CoA. 4

5.  In addition to a CoA, a regional CoA ( RCoA) is also allocated to a MN whenever it enters a new DMAP domain.  The HA and CNs only know the MN's RCoA, therefore:  When MN moves across a MAP domain and triggers a RCoA address change, new RCoA address needs to be propagated to the HA and CNs.  When MN moves from one subnet to another but is still within a region coverd a MAP domain, the CoA change is only propagated to the MAP, thus saving the signaling cost for mobility management.  The number of subnets coverd by a MAP domain is static in HMIPv6.  Without considering service management 5

6.  Propose a cross-layer intergrated mobility and service management scheme call DMAPwSR-- dynamic mobility anchor points with smart routers.  Identify the best DMAP domain size that can minimize the network traffic via model-based evaluation with simulation validation.  The goal is to minimize the overall mobility and service management cost.  The basic idea is that each mobile node (MN) can choose smart routers to be its DMAPs to balance the cost associated with mobility services versus packet delivery services. 6

7.  MN enters AR 1 in DMAP service area 1, it selects AR1 as the DMAP.  MN acquires RCoA 1 and CoA 1 from AR 1,and an entry (RCoA 1, CoA 1 ) is recorded.  The HA and CNs are informed of the MN's RCoA address-- RCoA 1 7

8.  MN moves across AR 2 but still within DMAP service area 1, MN only informs the DMAP of the new CoA address--CoA 2  And an entry (RCoA 1, CoA 2 ) is updated. 8

9.  MN moves to AR 4 in DMAP service area 1, it crosses the DMAP domain area, so it will acquire a new RCoA (RCoA 2 ) and a CoA (CoA 4 ) from AR 4.  AR 4 becomes the new DMAP whose reouting table records an entry (RCoA 2, CoA 4 ).  The HA and CNs are informed of the MN's RCoA address-- RCoA 4 9

10.  In the scheme, MN appoints a new DMAP only when it crosses a DMAP service area whose size is determined based on knowledge regarding the MN mobility and service characteristics in the new DMAP service area.  1.DMAP not often change (large size): high service delivery cost  Path: CN-DMAP-MN  2.DMAP often change (small size): low service delivery cost, but raise the cost of informing the HA and CNs of the RCoA address change. 10

11.  DMAPwSR is movement-based, which means the DMAP service area size is detemined by the number of subnet crossings, say K, the MN moves away from the DMAP.  An area covering K moves from the last DMAP.  MN can easily keep track of the number of subnets it has crossed.  When the number of subnet crossings is equal to K, the AR it just moves into will become its new DMAP and update the HA and CNs with its RCoA.  The optimal size Kopt is depend on MN's mobility and service behaviors characterized by MN's service to mobility ratio (SMR).  A MN measures its SMR periodically and apply Kopt dynamically. 11

12. 12

13. 13  Transitions (i.e.Move)  Places (i.e. Intra)  Tokens and arcs  Place Xs holds the number of subnet crossings since the last DMAP registration. Initailly there is no token.  By inspecting the number, we will know if the next subnet crossing is an intra-domain move or an inter-domain move.

14. 14  If Moves holds a token it means a subnet crossing event just happens.  A token is put in place Moves.  Transitions rate is σ.  1.If the current move is an intra-domain move, i.e., number of tokens in Xs is less than K-1, transition A will be triggered.  MN only inform the DMAP of the CoA change.  Moves--A--Intra--MN2DMAP--Xs

15. 15  2. If the current move is an inter-domain move, i.e., the number of tokens in Xs is equal to K-1,transition B will be triggered.  Moves--B--Xs  When the number of tokens in Xs is K, it means MN moved into a new DMAP service domain, NewDMAP will be triggered.  Xs (K) --NewDMAP

16. 16  The SPN model is a continuous-time Markov chain with state presentation of (a,b) where  a is the number of tokens in place Moves.  b is the number of tokens in place Xs.  P i is the steady state probability that the system is found to contain i tokens in place Xs.

17. 17  C i,service is the communication cost for the network to service a data packet given that MN has moved across i subnets since last DMAP registration.  It includes:  a delay between the DMAP and a CN in the fixed network (βτ)  a delay from DMAP to the AR of the MN's current subnet in the fixed network (iτ)  a delay in the wireless link from the AR to the MN (γτ)

18. 18  Let Cservice be the average communication cost to service a data packet weighted by the respective Pi probabilities, and is calculated as follow:

19. 19  Let C i,location be the network signaling overhead to service a location handoff operation given that the MN has moved across i subnets since the last DMAP registration.  If i<K, the MN only informs the DMAP of the CoA address change. So the cost includes a delay from MN to AR(γτ) and a delay from AR to DMAP(iτ).  Total cost: γτ+iτ

20. 20  If i=K, A new DMAP will inform the HA and N CNs of the RCoA address change. The cost includes a delay from MN to AR (γτ) and a delay from AR to HA and N CNs (ατ+N*βτ).  Total cost: γτ+ατ+N*βτ  C DMAPwSR deterimines the K opt  λ is the data packet rate between the MN and CNs, and σ is the MN's mobility rate

21. 21 Basic Concepts: DMAP service area Movement-based DMAP Distance-based DMAP - DMAP service area size is determined by the number of movements the MN moves away from the DMAP. - DMAP service size is determined by the distance between the current subnet and the DMAP. Purpose: evaluate whether simulation results are sensitive to the definition of DMAP service areas.

22. 22 Hexagonal-shape network coverage model Basic Concepts: Network Coverage Model(1) Assume MN moves in accordance with random walk by which a MN stays in a subnet for a while and then moves It moves from the current AR to one of the 6 neighbor ARs randomly with equal probability of 1/6. Applied to both distance-based DMAP and move-basded DMAP

23. 23 Basic Concepts: Network Coverage Model(2) Mesh network coverage model MN moves from the current AR to one of the 4 neighbor ARs randomly with equal probability of ¼. Distance-based DMAP service areas with distances 1 through 4 are marked.

24. 24 Access point locations at Dartmouth College campus Basic Concepts: Network Coverage Model(3) Based on real trace data. E.g. real world wireless network consisting of access points(APs) on the campus of Dartmouth College. 695 APs on campus. Consider two APs as neighbor Aps if they are separated in distance in the range from 100m to 200m. When a MN leaves an AP, it randomly selects one of its neighbor APs to move into.

25. Comparison with MIPv6 and HMIPv6 MIPv6 Communication Cost: HMIPv6 Communication Cost: DMAPwSR Communication Cost:

26. DMAPwSR dominates MIPv6 when SMR is low. Cost difference between MIPv6, HMIPv6 and DMAPwSR DMAPwSR performs better than HMIPv6 whenever SMR is low or high.

27. Cost ratio between MIPv6, HMIPv6 and DMAPwSR

28. Batch mean analysis  95% confidence level and 10% accuracy > 20,000 observations Simulation versus analytical result for total Cost Simulation versus analytical results: cost difference

29. Distance-based DMAP service DMAP service area size is determined by the distance between the current subnet and DMAP. Cost difference under movement-based versus distance-based service area simulation Movement-based DMAP service DMAP service area size is determined by the number of movements the MN moves away from the DMAP.

30. Cost difference under different residence time distribution Optimal K versus SMR under various time distributions

31. Cost difference under different network coverage model Consistent with different network coverage models!!

32. 32 Thank You Q & A