(CS6204) Advanced topics in mobile computing NOV.10.2009

1. Terminal Mobility Support ProtocolIEEE Transaction on mobile computing Vol. 8, No.6, June 2009Teck Meng Lim, Chai Kiat Yeo, Francis Bu Sung Lee, and Quang Vinh Le (CS6204) Advanced topics in mobile computing NOV.10.2009

2. Table of Contents • MOTIVATIN OF Terminal Mobility Support Protocol (TMSP) • BACKGROUND OF TMSP (BRIEF MOBILE IP) • TMSP FEATURES • COST COMPARISON WITH MOBILE IP, SIP-OVER-MIPV6 • PERFORMANCE ANALYSIS • CONCLUSION

3. Terminal Mobility Support Protocol (TMSP) • TMSP is a mobility method that resolves IP mobility for the mobile node(MN) like 802.11 or 802.16. • TMSP uses an innovative IP-to-IP address mapping method to provide IP address transparency for applications. • TMSP locate a single user using SIP Uniform Resource Identifier(URI) • TMSP is much more efficient than Mobile IP in terms of the number of hops as well as overhead.

4. Motivation • TMSP is proposed to overcome the drawbacks of Mobile IP like Triangular routing, packet, redirecting, increased in IP header size, and the need for new infrastructure support. • Unlike MIPv4, TMSP uses an IP-to-IP address mapping method instead of Permanent IP (Home Address) to acquired current IP (Care-of-Address)binding in Mobile IP. • Unlike MIPv6, TMSP ensures transparency in IP address with providing Uniform Resource Indicator(URI)-to-IP address mappingto locate a user and module installed on MN and CN.

5. Background of Mobile IP • [1-2]A CN ‘s packets are routed to a MN via the HA of MN [HA intercepts these packets] • [3] HA makes a tunnel [encapsulates the original packets inside a new IP packet] and forward the packets. • [4] After taking off the outer header, FA forwards the packets directly to the MN. • [5-7] The MN sends packets back to the CN. Source of the figure : NCC 2009, January 16-18, IIT Guwahati

6. Background of Mobile IP • Packets are routed via triangular shape (Triangular routing) • Required dedicated infrastructure like a Home Agent(HA) • Packets can not be sent to directly from a CN to MN (redirecting) • In a tunnel from HA to FA or MN, the packet is encapsulated in a new outer header (doubles IP packet size from encapsulation)

7. Background (Mobile IP) • In enhanced Mobile IPv6, most of drawbacks such as triangular routing and packet redirecting can be removed by the route optimization mode. So, Packets from CN can be routed directly to the Care-of-Address of MN and vice versa. • Examples of Route optimization in MIPv6 Next slide

8. Background (The route optimization in MIPv6)

9. Background of Mobile IPv4, IPv6IP-in-IP tunnel • The IP-in-IP encapsulation in tunneling increases the packet sizes. • This leads to inefficiency for a small IP packet like a voice packet. Added outer header while a packet is passing through a tunnel from HA to FA Source of the figure : http://www.cse.wustl.edu/~jain/cse574-06/ftp/mobile_ip.pdf

10. Background (IPSec on IP Mobility) • IPSec protocol adds security to IP by enabling the sending and receiving of cryptographically protected packets [Packet level security] • A security association(SA) between CN and MN requires following entities in the Header: Authentication Header[AH], Encapsulated Security Payload Extension[ESP], and Security Parameter Index[SPI]. • The SA is only valid at the time of negotiation between CN and MN using the three entities. • Once MN enters another network and changes its IP address, the SA will be invalid. It means a new Internet Key Exchange(IKE) negotiation that gives rise to computation overhead and latency must be performed. TMSP has a solution that avoids IKE re-running

11. Six major features of TMSP • Does not assign a permanent IP address to each MN • Uses a single IP address at any time for each network interface • Does not introduce new network servers or infrastructure support to enable IP mobility • Does not need IP packets redirection • Does not incur extra IP header extension • Allows IPSec without rerunning Internet Key Exchange (IKE)

12. Entities of TMSP 5.CN’s/MN’s SIP Redirection Server (SIP-RS) 1. Keeps states for each data connection in MN and its CN (three linked state-tables) 1.1. Mapping Table (M-table) - Network ports, SRC IP, DEST IP 1.2. URI table (U-table) - information for CN - URI and Care-of-Address 1.3. SID table (S-table) - a mapping between each network port and a SID for each network port at MN 2. User agent (UA) • Application that initiates requests or returns a response on behalf of the user 2.UserAgent (UA) 1.1 M Table 1.2 U Table 1.3 S Table Header Restore Module 3.(HRM) Address mapping Module 4.(AMM) Correspondent Node Or Mobile Node

13. Entities of TMSP 3. IP-to-IP Address Mapping Module (AMM) – intercepts outgoing and incoming IP and IPSec packets at the network layer to perform an IP-to-IP address mapping function 4. IP Header Restoring Module (HRM) – restores IP address in the IP header of an IPSec packet to the IP addresses used at the instant when the connection was first established 5.SIP-Redirection server – provides a URI to Care-of-Address resolution of each client

14. Bootstrapping • Mobility for an MN begins when its UA completes a SIP registration procedure with MN’s SIP redirect server, specified in the server part of the SIP URI as provided by the user. • MN and CN use SIP register message to register themselves with their respective SIP-RS. • It is important to highlight that CN and MN do not need to be registered with the same SIP redirect server. A SIP URI is an email-like address that contains two parts: user and server parts, separated by an “@,” e.g., “user@server” The user part contains the user name to identify the user and the server part contains the IP address or domain name of the SIP server.

15. Connection establishment • A CN sends a SIP invite request to MN’s the SIP-RS by using MN’s Care-of-Address. • The SIP-RS replies with a SIP moved temporarily (Msg. No 302) message containing the Care-of-Address of MN. • The CN sends ACK to the SIP-RS • By using MN’s new Care-of-Address, the CN sends a new invite message to the MN directly. • The MN replies to CN with a OK message. • The CN sends ACK to MN.

16. Connection establishment Kernel space Kernel space

17. Connection establishment Functions called in TMSP entities • GETHOSTBYNAME (): is called to map SIP URI to the Care-of-Address • ‘connect (): usual data connection • ‘kernel-space-connect ():is called to establish connection with MN. • ‘accept (): data connection packet is captured • TMSP-connect-hook (): incorporating function which generates a unique SID for data connection if IPSec is the requested protocol and to call its UA to perform • TMSP-accept (): After accept function for TCP, it registers the network port number used for the connection and update

18. General IP packets’ Pathway • Assume that • The CN has acquired a new Care-of-Address, C, from Access point M when this packet is generated, while the MN has acquired a new Care-of-Address, D, from Access point N. • Both the MN and CN have also informed each other about their new Care-of-Address using a SIP invite request procedure

19. IPSec packets’ Pathway IPSec module encrypts the packet as a payload and encapsulate it with an IPSec header Application do not need to be aware of the Care-of-Address changes on the MN IP Address transparency

20. IPSec packets’ Pathway • The HRM will replace the source IP address in the IP packets’ header with RSA(Original IP) IP Header used to generate security checks for IPSec. Without this IP header restoration, the IP Packet will fail the IPSec module’s security check. • This method allows IPSec module to use the share key established at the instant when the connection is established. • The AMM replaces2 the destination IP address of the IP or IPSec packet by the Care-of-Address of the MN based on its entry in the M-Table • Finally, for IPSec packet, it pastes the MN’s SID into the reserved bits of the AH • This packet is then directly routed to the MN. At the network layer of MN, the AMM replaces the source and destination IP Addresses of the IP or IPSec packet by the DSA and RSA respectively

21. Update procedure • After MN enters a foreign network, the SIP-Redirection Server(RS) and CN update the MN’s new Care-of address • MN uses the SIP registration procedure to inform the change of location to its SIP-RS. • MN re-invites its CN using the SIP invite request message to report the newly acquired Care-of-Address. • Upon receiving this re-invitation from MN, CN will update the MN’s Care-of-Address in its U-Table. • The MN’s Care-of-Address is used by the AMM to deliver packets to the current IP address of MN.

22. Performance Comparison Terminology • When an MN moves, a series of message exchanges begin to keep the MN’s movement up to date. • C movement : the total number of hops that messages traversed as MN moves in a new network (Cost for a new movement connection establishment with a new Agent) • C maintain : the total number of hops that messages traversed to maintain existing sessions between CNs and MN (Cost of maintaining the seamless connection as MN moves) • C tx : counts the number of hops that a message needs to traverse from CN to MN (Cost as CN sends a message to MN) • TMSP versus MIPv6 basic(=MIPv4) versus application-layer protocol using SIP-MIPv6

23. Comparison in terms of Signaling cost,Transmission cost Signaling Cost can be broken into Movement cost and Session maintenance cost • C signaling = C movement + C maintain • Each cost can be estimated by DA-B which means the number of hops from A to B • C protocoltx= Transmission Cost of each protocol • For simplicity of comparison, we assume that the DHCP server is one hop away from MN and there are five hops between two network entities.

24. Signaling cost(Movement cost) ofTMSP, MIPv6(basic),and SIP-MIP v6 • When MN moves into a foreign network, it acquires a new IP address via DHCP. • In TMSP, it starts a registration procedure using a message pair • (Register and Okay messages) with its SIP-RS. • While in MIPv6, it starts a binding update procedure using a message pair (binding update and acknowledgment messages) with its HA. • In SIP-MIPv6, it triggers both the registration and the binding update procedures. • http://en.wikipedia.org/wiki/Dynamic_Host_Configuration_Protocol

25. Signaling cost(Maintain cost) ofTMSP, MIPv6(basic),and SIP-MIP v6 • When MN moves into a new foreign network, it informs its CNs of its new Care-of-Address to continue existing sessions. • In TMSP and SIP-MIPv6, MN informs its CNs using the SIP invite request messages that consist of a message pair (SIP invite request and okay messages). • In MIPv6, MN does not inform its CNs of its new IP address. Thus, it does not incur any signaling cost. • Eta = the number of CN

26. Transmission cost of TMSP, MIP v6(basic),and SIP-MIP v6 • When TMSP is used, a message transmitted from CN to MN follows the conventional IP routing for both TCP and UDP packets. • When MIPv6 is used, a message transmitted from CN to MN is first sent to the MN’s HA where the message is tunneled to MN. Likewise, a message from MN to CN is reverse tunneled back to the MN’s HA before it is sent to the CN. • When SIP-MIPv6 is used, the message route path is • similar to TMSP for UDP packet and is similar to MIPv6 for TCP packet.

27. Data Transmission Gain The hop count is measured from the point a packet is sent from MN to CN after MN has moved into a new foreign network. It includes the cost of one movement cost function

28. Average Signaling cost By using C movement and C maintain, a *Handover cost can be derived Where alpha is the possibility of active handover and 1-alpha is idle handover probability. * The cost of a handover refers to the duration from the point a host enters a new network until it has successfully attained a new IP address

29. Data Traffic overhead The CBR source generates a total of T = R * (8L) * t bits for a duration of ts, where R is the rate (in packets per second) of the traffic source and L is the size of an IP packet. For MIPv6, the amount of traffic generated onto the network from CN to MN is given by where LIP-in-IP is the size of the outer IP header used to encapsulate the packet, HMIPv6CN-HAis the number of hops from CN to the MN’s HA, and HMIPv6HA-MNis the number of hops between MN and its HA. For TMSP, the amount of traffic generated onto the network from CN to MN is given by where HTMSP is the number of hops between CN to MN.

30. The result of traffic overhead This figure shows the gain in terms of traffic generated when TMSP is used over MIP V6 The X-axis indicates the difference in hops between the triangular route and the direct Route CN-MN Gain = TMIPv6 – TTMSP / TMIPV6 For example, if Gain = 0.6, then TTMSP = 0.4 * TMIPV6

31. The result of traffic overhead This figure shows the gain of using TMSP when the number of hops for the direct Route is fixed at 10, 15, 20, and 25 and the number of the triangular route is fixed at 30 But distributed in proportion (HMIPv6CN-HA / HMIPv6HA-MN)

32. Effect of TMSP on the throughput of Wireless Lan • Experiments are run to compare the saturation throughput of an MN using an IEEE 802.11g WLAN interface. MN floods the air channel using a CBR traffic of 30 Mbps. • The reported throughputs are approximately equal regardless of whether TMSP is installed on MN. Thus, this shows that TMSP does not affect the achievable throughput of the WLAN in this scenario.

33. Performance of the IP-to-IP Address mapping • The additional delay of 0.56 micro seconds (12%) introduced by TMSP is therefore negligible as it is much smaller than the transmission delay of the minimum frame size of Ethernet. Thus, TMSP will not affect the transmission rate of a host. • The transmission delay for the minimum frame size (64 bytes) of Ethernet on a 100-Mbps LAN is approximately 5.12 micro seconds and that for the minimum frame size (512 bytes) of gigabit Ethernet on a 1-Gbps LAN is approximately 4.10 micro seconds. Compared to the transmission delay of both, 0.56 microseconds of delay is reasonable

34. Applicability of TMSP • TMSP provides a perfect solution for seamless communications • TMSP works for point to point communication software such as VoIP and Video conferencing • TMSP does not affect the legacy applications running on a non-TMSP-enabled server • One biggest limitation of TMSP is that when MN and CN move and change their IP address at the same time, the IP addresses of both are not updated correctly • Due to the reuse of IP addresses, there exists very slight possibility of IP address mapping confusion for connectionless service hosted by UDP server (many clients & single port receive)

35. Conclusion • TMSP enables IP mobility for MNs. It creates an association of IP addresses and port numbers for a connection in MN and its CNs and uses this association to hide the changes in IP address when MN moves from one network to another. • Owing to the manipulation of IP addresses in the IP header of each packet, TMSP ensures that packets are sent directly between CN and MN without packet redirection. • TMSP taps on the pervasiveness of SIP server as an IP directory service and uses SIP as a signaling mechanism to maintain location information. • Each MN is identified by a SIP URI. Thus, TMSP does not require permanent IP address for each MN, does not require new network infrastructure support, and does not enforce packet redirecting during routing.

36. Appendix-Mobile IP terminology

37. Appendix-Mobile IP terminology

38. Question Thanks