NEtwork MObility (NEMO) Speaker: Yi-Ting Mai Contact info. :wkb@wkb.idv.tw Date: 2010/05/11

Outline MIPv6 (Mobile IP version 6) NEMO (NEtwork MObility ) NEMO research article “On the Design of Micro-mobility for Mobile Network” Conclusion

Mobile IPv6 Mobile IPv6 While a mobile node is away from home subnet Allow IPv6 nodes to move from one IP subnet to another While a mobile node is away from home subnet It sends information about its current location to the home agent The home agent intercepts packets addressed to the mobile node and forwards them to mobile node’s current location Mobile IPv6 是解決 IPv6 節點移動到不同IPv6 subnet時所產生的位址改變，使得通訊中斷的問題。 主要的解決方法概念是在IPv6行動節點本來所處的子網路下設置home agent (HA)，當節點離開時，把新的位址向HA註冊，之後當有指向此節點位置的封包到達時，就會由HA接受，並轉送到目前節點所處的位址。

A Mobile IPv6 scenario Mobile Node move Link B Home Link Link A Internet Home Agent Correspondent Node Mobile Node Router Home Link Link A Link B Link C move 從此圖中我們可以看到，當Mobile Node從Link A移到Link B時，會產生位址的改變，使得原本與Mobile Node (MN) 通訊的Correspondent Node (CN)找不到它的通訊對象，這就是Mobile IPv6所欲解決的問題。

Correspondent Node communicating with the mobile node Mobile IPv6 routing Home Network Internet HA Home Agent with which node’s home address is associated Correspondent Node communicating with the mobile node Mobile IPv6 解決上述問題的方式可由此圖表示，HA 成為 MN 在它 home link 上的代理者，HA維護著MN在home link上的位址Home address與MN目前所在位址(care-of address, CoA)對應關係，當MN移動時，會向HA註冊目前的CoA，對於與MN通訊的CN而言，所有要送給MN的封包都以MN的home address為目的位址，這些封包都會由HA接收，並轉送到目前MN的CoA，對於CN來說，MN的位址就好像沒有改變一樣，透過此方法或許會造成通訊的delay但是不會使得通訊中斷。 Visited Network Mobile Node with a care-of address

Mobile IPv6 handover IPv6 router with HA function move MN sends a Binding Update to an HA on previous network HA returns a Binding Acknowledgement HA tunnels packets to MN MN sends a Binding Update to CN (1) (3) (4) (2) CN MN 我們之前提過CALM希望通訊最好能夠不間斷，也就是使handover所產生的data loss降到最低，Mobile IPv6中的smooth handover可以達到這點，也就是採用所謂的先接後斷的handover方式，從圖可以看到，MN在(1)中傳達要移動的通知，在經過(2)HA的確認之後，HA把目前MN與CN的通訊轉移到它現在的位址(3)後，MN再透過目前所在地的router通知CN他現在的位置，並透過現在所處的router建立新通訊(4)，之後就可以結束透過HA的連線。

NEtwork MObility (NEMO) Manage the mobility of an entire network Mobile network includes one or more mobile routers (MRs), which connect it to the global Internet Mobile Network Nodes (MNNs) Nodes behind the MR(s) can be either fixed or mobile. Mobile network example Networks of sensors and computers deployed in vehicles Vehicles are increasingly equipped with a number of processing units for safety and ease of driving reasons, as advocated by ITS applications 前面所說的Mobile IPv6是針對host mobility的，也就是單一節點(主機)的行動性的支援，然而在ITS的應用情境下，有可能整部車本身就是個網路，整個網路視為一個可移動的單位，構成了所謂的mobile network, 整個網路都會移動，移動後，連接到網際網路接點就改變了，也因此也改變了此網路在整個網際網路拓撲(topology)中的可達性(researchability)。 NEMO就是用來解決因此產生的問題用的。

NEMO network scenario TTS Server ADSL/WiFi WiMax/3.5G Workstation V-I communication V-I communication DSRC/802.11 DSRC/802.11 V-V communication V-V communication

NEMO terminology (1/2) Mobile network A IP subnet is composed of a set of nodes A mobile network as a single unit Connected to the Internet through one or more mobile routers (MR) via AR (access router) Internet AR AR MR MR MR MR 我們先來介紹NEMO下的幾個專用名詞。 從圖中可以看到mobile network是以整個網路為單位來作移動的，它可能經由車上的mobile router透過不同地方的access router來連接到網際網路。 而一個行動網路內部也有可能包幾種不同的節點，我們在下面幾張投影片介紹。

NEMO terminology (2/2) Mobile Network Nodes (MNNs) -- different kind of nodes in a mobile network Mobile Router (MR) Border router of the mobile network Local Fixed Nodes (LFNs) It permanently located in mobile network Local Mobile Nodes (LMNs) Nodes belong to a mobile network Visiting Mobile Nodes (VMNs) Do not belong to the mobile network CN AR LFN MR VMN MN LMN Internet 任何在行動網路內的節點都統稱為MNN (Mobile Network Node)，它有可能用來連接網際網路用的MR，或者是有固定連接點的LFN、可以在內部網路移動的無線內部節點LMN，也有可能是從外部來的暫時移動節點VMN，而CN如同前面介紹過的，是與這些MNNs做通訊的對應節點。 CNs Nodes communicating with MR, LFNs and LMNs, VMNs

ITS as a mobile network Several IPv6 nodes in a single mobile entity Multiple IP subnets Each node has an IPv6 address Every IPv6 address contains a common IPv6 prefix Different communication media (3G, M5, MM, WiFi, WiMax...) available Mobile Router 如前所述，在ITS中，一部車輛，就可能是一個行動網路。 對於CALM來說，行動網路中可能有數個IPv6節點，根據內部不同的網路拓撲，也可能具備幾個不同的IP子網路 (例如多個MRs的情況下)，每個節點在內部網路中，都有他們唯一的IPv6位址，但對外(行動網路之外)他們有一個共同的IPv6 prefix，並可能透過不同的通訊medium連接到網際網路。

IETF NEMO support (1/2) Mobile router manages mobility of the entire network Current location (care-of address) is associated with HA Nodes inside the mobile network don't need to participate in mobility management Mobile Network Home Agent 簡單來說，NEMO解決行動網路移動性的方法，就是由MR來處理整個行動網路的移動性，這時MR的角色跟Mobile IPv6中的MN的角色類似，也是透過HA來解決移動所產生的問題，只是他代表的不僅是它自己，而是整個行動網路。 在行動網路內的節點，不需要參與行動性的管理，或甚至沒有察覺到網路正在移動的事實。 Correspondent Node

IETF NEMO support (2/2) Bi-directional tunnel between MR and HA to maintain sessions Address registration Care-of address (CoA) is bound to mobile network prefix (MNP), not home address (HoA) Association in HA MNP (RFC 4885) MR-CoA instead of MR-HoA  MR-CoA HA records a network-specific route rather than a host-specific one MR-CoA is the next hop to the mobile network determined by MNP 在行動網路移動時，內部的節點透過HA與MR之間的雙向tunnel (IPv6 in IPv6 Encapsulation) 來使application session不至於中斷，要達到此目的，同樣需要MR向HA註冊目前所在位址，只是不同於Mobile IPv6，此時向HA註冊的不僅是是CoA，而是可以在網際網路中唯一辨識出此行動網路的Mobile Network Prefix (MNP)，HA紀錄的不只是MR的CoA了，而是MNP配合MR-CoA，記錄在HA的routing table中，要達到MNP這個prefix指定的網路的next hop就是MR的CoA。 Mobile Network Prefix (MNP): ㄧ個可以在網際網路中唯一辨別整個行動網路的IPv6j位址，所有在由MNP所決定的行動網路中的節點之位址必定包含此prefix. (defined in RFC4885) 假設CN要傳送封包給在一個行動網路下的LFN，這個封包會由HA所接收，然後HA根據他所記錄的位址對應，把此封包封裝在另一個IP封包中，此時封包的來源位址變成HA的位址，而目的位址是MR的CoA，當封包送達後，經過MR的拆解後，此封包如從CN送出時一樣，原封不動地到達LFN。 目前更有效率的NEMO解決方案還持續在發展中，此方法在效率上並不是最好的解決方案，但目前這個方法是最直覺，最方便，直接由MIPv6的概念推衍出來的解決方案。

Benefit of network mobility Only the mobile router manages mobility and maintains Internet access through one or several access media Nodes behind MR can be standard IPv6 nodes without mobility support mechanisms NEMO basic support Mobile IPv6 已經有了Mobile IPv6, 為什麼還要繼續發展NEMO呢？ 這是因為它可以降低整個行動網路在管理上的複雜度，並且就算節點本身不支援Mobile IPv6所支援的host mobility，仍然可以透過NEMO的network mobility連接到網際網路。

On the Design of Micro-mobility for Mobile Network Junn-Yen Hu, Cheng-fu Chou, Min-Shi Sha, Ing-Chau Chang and Chung-Yi Lai, pp. 401-412, LNCS 4809, Springer, Dec. 2007.

Outline Introduction Related Work Micro-mobility scheme for NEMO Performance Evaluation Conclusion The topics that I am going to show you, are Introduction Related work Our proposed scheme, Micro-NEMO Performance evaluation Finally, Conclusion.

Introduction Mobility Management (differentiated by distinct types of handoff procedure) Macro-mobility scheme Mobile IP Micro-mobility scheme Hierarchical Mobile IP, Cellular IP and HAWAII In general, mobility management schemes can be differentiated by distinct types of handoff procedure, Macro-mobility scheme and micro-mobility scheme. Macro-mobility means that a large scale of movement of a mobile across diverse administrative domains. The major scheme for macro-mobility is Mobile IP. This scheme represents a simple and scalable global mobility solution while a MH is moving. But it is not suitable for movement in a small scale domain or at a high speed movement because it might incur a lot of global registration procedures and this could lead to heavy signaling overheads and significant performance degradation. In order to improve the performance for mobile Internet users, the micro-mobility concept was purposed. Micro-mobility means that local movement of a mobile within an administrative domain. Several Protocols such as Hierarchical Mobile IP, Cellular IP, HAWAII have been proposed for micro-mobility.

Introduction (cont.) Mobility Management (differentiated by distinct number of moving mobile hosts) Host Mobility (mobile node) Mobile IP, Hierarchical MIP, Cellular IP and HAWAII. Network Mobility (NEMO, mobile network) NEMO Basic Support Protocol MR Nested NEMO Pinball routing problem In addition, mobility management still can be discussed from the view point of number of moving MHs. If the subject of discussion is about a single MH moving, it is called host mobility. Such as Mobile IP, Hierarchical MIP, Cellular IP and HWAII. If a set of users are moving together via a certain vehicle, it is called network mobility. Or NEMO. That vehicle can be regarded as a mobile network. The NEMO basic support protocol is a major and popular scheme for network mobility. In NEMO, a mobile router is able to manage the mobility of a set of MHs within the same mobile network. So that mobile hosts inside the moving network should not perceive that mobile router has changed its point of attachment such that the binding update storm can be avoided. In addition, a MR might allow other MRs to associate with itself. One mobile network could get on in another mobile network. This is called nested NEMO, which might suffer from the pinball routing problem.

Pinball routing problem

Introduction (cont.) State of The Art Micro-NEMO scheme is presented The current state of the art for mobility management is described in the Table. We can see that no scheme support micro-mobility for network mobility. Hence, we try to develop a new micro-mobility scheme, called Micro-NEMO to support network mobility.

Related Work Hierarchical MIP (HMIP) Ohnishi Scheme An extension of HMIP To solve the pinball routing problem in NEMO rather than to provide a micro-mobility scheme for NEMO Each MN on a vehicle still perform binding update by itself In previous work, HMIPv6 a well-known micro-mobility scheme for host mobility. In this scheme, a mobile host moving within a micro-domain performs local binding update with the mobile anchor point rather than doing home registration. The Ohnishi scheme is a extension of HMIPv6. It is aimed to solve the pinball routing problem in network mobility rather than to provide a micro-mobility scheme for a moving network. In addition, each mobile nodes within a vehicle still perform binding update by itself. Hence, we think that is not really micro-mobility for NEMO.

Micro-mobility scheme for NEMO Micro-NEMO Micro-mobility + NEMO = Micro-NEMO Low handoff latency, less signaling cost and transparent to all MHs It is built from HMIPv6 To compatible with NEMO is important. NEMO and HMIPv6 are extended by MIPv6 So we try to design a really micro-mobility scheme -----Micro-NEMO for network mobility. Micro-NEMO, means that Micro-mobility plus NEMO. We hope it can achieve low handoff latency, less signaling cost and transparent to all mobile hosts. We believe that new proposed scheme to be compatible with NEMO basic support protocol is important. And we find that NEMO and HMIPv6 are extended by MIPv6. Hence, the Micro-NEMO is built from HMIPv6

Micro-mobility scheme for NEMO (cont.) This slide show four movement scenarios. F means that local fixed node, M means that mobile nodes, MAP means that mobile anchor point and AR is access router. First, scenario 1, a mobile network initially enters a new micro-domain. This is similar to the case when a MH first enters a new micro-domain for host mobility. And scenario 2, a mobile node gets on a vehicle. In this scenario VMN should be unaware of the movement of the mobile network by performing some signalings. Scenario 3, the mobile network - vehicle roams within a micro-domain. Scenario 4, a mobile node gets off the transportation.

Micro-mobility scheme for NEMO (cont.) Scenario 1 For the first scenario, when a vehicle move into a micro-domain. The MR will register with MAP to establish a local binding and perform home registration with its home agent (MR_HA). After these initial signalings have been finished, that home network of the vehicle knows which micro-domain the vehicle is on. When a CN wants to transmit data packet to local fixed node, the data packets pass through HA of MR, MAP and reach local fixed node finally.

Micro-mobility scheme for NEMO (cont.) Scenario 2 In scenario 2, when a mobile node gets on a vehicle, it will obtain a care of address and perform home registration with its HA(VMN_HA). After that, its HA will know the mobile is in certain vehicle. Afterward, when a corresponding node (CN) wants to transmit data packet to mobile node, data packets will be tunneled by mobile node’s and mobile router’s home agent, pass through MAP, AR1, MR, and finally arrive at mobile node.

Micro-mobility scheme for NEMO (cont.) Scenario 3 There is no need for MR of a vehicle to perform home registration when it moves around within the same micro-domain. For the scenario 3, there is no need for MR to perform home registration when it moves around within the same micro-domain. Note that residing node, mobile nodes and local fixed nodes, do not perceive the vehicle has change the access router such that they do not perform home registration with home agent again. This achieves the network mobility concept.

Micro-mobility scheme for NEMO (cont.) Scenario 4 It is similar to that a VMN enters a new micro-domain. In scenario 4, when mobile node gets off the vehicle, it is similar to that a mobile node enter a new micro-domain first. It could perform HMIP by itself.

Micro-mobility scheme for NEMO (cont.) Advantage Reduce the number of global binding updates. Reduce the number of signaling of residing nodes Drawback Pinball Routing Solution: Enhanced Micro-NEMO In this slide, I will show you the advantage and drawback of the Micro-NEMO scheme. Because the mobile network does not perform home registration with HA unless it traverses to a new micro-domain, the number of global binding update will be reduced. Moreover, because the mobility of the vehicle is transparent to its residing nodes, the number of signalings of residing nodes can be reduced as well. The problem of the Micro-NEMO is pinball routing. It means that all the traffic of the mobile network passes through all home agents of the MRs inside the mobile network. This problem will be more critical if the number of levels of nested NEMO increases. In order to solve the problem, we designed an enhanced Micro-NEMO scheme further.

Micro-mobility scheme for NEMO (cont.) Enhanced Micro-NEMO Apply ROMIP concept This slide shows the enhanced micro-nemo scheme. We think that the pinball routing problem can be solved if the root mobile router’s home agent knows all the binding information of child mobile router and mobile node. Hence, the root mobile router’s home agent can perform binding update with the corresponding node by root mobile router’s care of address. Afterward, corresponding node could directly forward packets to VMN of the vehicle inside a micro-domain. In this figure, a people area network formed by a PMR and mobile node, M. PMR mean that the private mobile router, which is owned by some people. When the people get on the vehicle, his PMR will obtain a care-of-address from MR and perform home registration with it’s home agent. Besides, outer mobile router should perform additional binding with it’s home agent to inform it’s home agent of new visitor’s information. Once mobile router’s home agent receives packet sent from child node’s home agent, it could perform binding update with CN. Afterward CN is able to directly forward data packets through MAP to the VMN.

Performance Evaluation Simulation environment Simulation parameter We have described the Micro-NEMO and enhanced Micro-NEMO. Next, we will try to show some performance evaluation. Our simulation environment consists of 64 micro-domains. Each micro-domain administrated by a mobile anchor point. And they form a mesh grid as the figure shows. Each micro-domain consists of 16 access routers, which form a mesh grid as well. There are total 500 moving vehicles randomly scattered over all micro-domains. Each vehicle has a MR and 5 VMNs and the total number of VMN is 2500. In order to model the mobility of the vehicle, time is slotted and Movement Probability is used in the simulation. The movement probability is represents the probability of a vehicle leaves its current access router in the next time slot. And there are total 1000 time units to represent the simulation time. For the delay latency, we assume that Internet latency is 50 time units, local domain latency is 10 time units and backbone latency is 1 time unit.

Performance Evaluation (cont.) Performance metrics Total number of binding updates Average handoff latency End to end delay Packet overhead Four performance metrics are used to compare the proposed Micro-NEMO basic and enhanced scheme with other schemes. Total number of binding updates, Average handoff latency, End to end delay and Packet overhead. The average handoff latency is defined as the time to complete binding update after a handoff, the end to end delay is defined as the time interval for a data packet from sender to receiver, and the packet overhead is defined as the ratio of encapsulate packet headers size to total packet size.

Performance Evaluation (cont.) Total number of binding updates This slide shows the total number of binding update under different schemes. To compare with basic NEMO protocol, the number of global binding update of Micro- NEMO is lower than basic NEMO protocol since it could effectively integrate the idea of the micro-mobility. In addition, the total number of binding updates in both Hierarchical MIP and Ohnishi scheme is higher than Micro-NEMO because they do not consider the concept of network mobility. In other words, when the MR changes its access point, all mobile hosts inside the moving network will not observe that change such that the binding update storm for all MHs can be avoided.

Performance Evaluation (cont.) Average handoff latency Next, this slide shows that comparison of the average handoff latency. Since the proposed scheme can provide the functionality of micro-mobility, it is not surprising that the latency is lower than NEMO. Simulation result shows latency of NEMO had been reduced nearly 65%. In addition, because both HMIP and Ohnishi are micro-mobility schemes as well, their latencies are identical with Micro-NEMO scheme.

Performance Evaluation (cont.) End-to-end delay This slide shows that results of the end-to-end delay for different schemes, which also consider the characteristic of the nested NEMO. Because the NEMO basic support protocol and the Micro-NEMO do not implement route optimization, both of them are identical in end-to-end latency. Moreover, the end-to-end delay will be increased with number of nested level. Since the proposed enhanced Micro-NEMO scheme is equipped with the process of route optimization without passing through any HA, the end-to-end delay has been significantly decreased. The figure also shows that result of end-to-end delays for HMIP and Ohnishi Scheme. Because their packet deliveries only pass through one HA regardless the nested level, their delays are higher than that of Enhanced Micro-NEMO and less than those of Micro-NEMO and Basic NEMO support protocols.

Performance Evaluation (cont.) Packet overhead Finally, This slide shows that comparison of packet overhead for different schemes. Due to some reason, route optimization functionality, it shows that enhanced Micro-NEMO is much better than other schemes in terms of packet overhead.

Conclusion Micro-NEMO Enhanced Micro-NEMO Improve the number of binding update, average handoff latency, end-to-end delay and packet overhead. A brief conclusion. In this paper, we proposed a micro-mobility scheme for mobile network, Micro-NEMO, which can provide local movement within an administrative domain for a moving network. We also proposed a enhanced micro-nemo to solve the pinball routing problem. The simulation result show that the proposed schemes are able to improve the number of binding update, average handoff latency, end-to-end delay, and packet overhead in comparison with other mobility schemes.

Reference (1/2) [1] R. Koodli, Ed., RFC4068 Fast Handovers for Mobile IPv6, July 2005,EXPERIMENTAL [2] D. Johnson, C. Perkins, J. Arkko, RFC3775 Mobility Support in IPv6, June 2004, PROPOSED STANDARD [3] T. Ernst, RFC4886 Network Mobility Support Goals and Requirements, July 2007, INFORMATIONAL [4] T. Ernst, H-Y. Lach, RFC4885 Network Mobility Support Terminology, July 2007, INFORMATIONAL [5] R. Wakikawa, A. Petrescu, P. Thubert, RFC3963 Network Mobility (NEMO) Basic Support Protocol,?V. Devarapalli, January 2005, PROPOSED STANDARD [6] Ed., M. Kojo, Ed., RFC3753 Mobility Related Terminology,?J. Manner, June 2004, INFORMATIONAL [7] IIS5711: Mobile Computing Mobile Computing and Broadband Networking Laboratory CIS, NCTU: Introduction to Mobile IPv6 [8] Thierry Ernst – ENABLE-DAIDALOS Workshop IST Mobile Summit - Budapest - July 2007: IPv6 Network Mobility (NEMO) Standardization & Usages, http://www.lara.prd.fr

Reference (2/2) [10] Elmic Systems - Your Partner in Internet Connectivity, Security and Mobility [11] Hesham Soliman, MONET: Network Mobility Support in IPv6, Thierry Ernst, 52th IETF - December 2001 [12] Thierry Ernst - Motorola Labs & INRIA (Planete), Network Mobility Support in IPv6, 51st IETF London, August 2001 [13] Nigel Wall, ITS comms – the CALM, 2005/09 [14] Keisuke UEHARA, InternetCAR: Internet Connected Automobile Researches [15] Thierry Ernst – French IPv6 Worldwide Summit - Cannes - November 2006: IPv6 Network Mobility (NEMO) and ITS Usages, http://www.lara.prd.fr [16] [17] ISO TC 204 WG16 CALM, http://www.calm.hu/ [18] ITS Application Overview, http://www.itsoverview.its.dot.gov/

Appendix Nested 2