IP-based mobility management mechanisms

In document Interoperabilidade e mobilidade na internet do futuro (Page 59-63)

2.4 Mobility management

2.4.1 IP-based mobility management mechanisms

Whenever a node connects to a network, it acquires an IP address belonging to the network of the access router. If we consider the mobility of that given node, its IP address will change as it moves across different access network. Such change has consequences regarding IP reachability, requiring the reestablishment of ongoing connections by the MN. For example, a TCP connection is identified by a 4-tuple (source and destination IP address and source and destination port) and, therefore, a change on the IP address of the MN (i.e., the source IP address) will break ongoing TCP connections. To overcome this issue, IP-based mobility management mechanisms have been proposed in order to maintain the mobile node’s reachability over the same IP address, allowing IP session continuity while on the move. Most of the presented proposals tackle IPv6 instead of IPv4, due to the capability of the hosts to be configured with several IPv6 addresses in the same network interface.

2.4.1.1 Centralized mobility management

Mobile IP (MIP) (both the MIPv4 [85] and MIPv6 [60]) is one of the first mechanisms developed by the Internet Engineering Task Force (IETF) to support node mobility. It defines that each MN is identified by an IP address that is independent of its location, called home address, and whenever away from its home network is associated with a second IP address, called Care-of Address (CoA), that identifies the current location of the MN. This enables the MN to change its PoA and still continue to be identified by its home address allowing seamless connection. MIP is also independent of the access technology of the current and the new PoA, i.e., it is suitable for mobility across homogeneous and heterogeneous networks. Two new functional entities are introduced in the network architecture: the Home Agent (HA) and the Foreign Agent (FA). While the HA is a router on the home network of the MN that maintains the current location of the MN and tunnels the messages towards the MN when it is away from its home network, the FA is a router on the visited network that routes the messages to the MN while registered. Specifically to Mobile IP for IPv6, there is no need to deploy the FA,

which is able to operate in any location without support required from the local router [60].

When the MN is at the home network, packets to and from it are delivered by using standard IP routing. However, whenever the MN roams to a foreign network, the MN registers its new CoA with its HA to inform about its new location (i.e., binding procedure), establishing a tunnel between the HA and the CoA of the MN. In doing so, messages destined to the MN (i.e., addressed to its home address) are sent towards its HA which are then tunneled towards the its CoA and delivered to the MN. For messages sent by the MN there is no need to implement new routing mechanisms since the route of the messages to the destination is made according to conventional IP mechanisms. This routing approach, as depicted in Figure 2.9, is commonly named triangular routing.

Figure 2.9: Triangular Routing in Mobile IP

Extensions to the base behavior of the protocol have been proposed to optimize the performance of MIP and, consequently, of the handover procedure. Fast Handovers in Mobile IPv6 (FMIPv6) [67] aims at reducing the handover latency, by enabling the MN to quickly detect that it has moved to a new subnet, providing the new access point and the associated subnet prefix information when the MN is still connected to its current subnet. Hierarchical Mobile IPv6 (HMIPv6) [17] aims at reducing the signaling overhead between the MN and its HA (and correspondent nodes) resulting from the handover procedure by introducing a new entity, called Mobility Anchor Point, which is located on the visited network and acts like a local HA.

Network-based mobility management solutions do not require the MN to be involved in the signaling exchanged with the home agent, with the mobility procedure being handled by a proxy mobility agent on the behalf of the MN. Proxy Mobile IPv6 (PMIPv6) [24], the network-based counterpart of MIP, proposes a mobility management solution within localized domains, providing mobility support for MNs without their direct involvement in the related signaling. Mobility entities in the

2.4. Mobility management 37 network are responsible for tracking the movement of the MN, initiating the mobility signaling and setting up the required routing states. For that, it introduces two new network functional entities: the Mobile Access Gateway (MAG) and the Localized Mobility Anchor (LMA). While the MAG, deployed on the access link where MNs connects to, is responsible for performing the mobility management on behalf of the MN, the LMA is responsible for managing and maintaining the reachability towards the MN by establishing tunnels with the MAG.

When the MN connects to an access network, the MAG on that link detects the attachment of the MN and it triggers the binding update procedure on the behalf of the MN in order to register it with the LMA. Upon successful registration, a bidirectional tunnel between the MAG and the LMA is established, used for routing the MN’s traffic between the MAG and the LMA (i.e., used for routing packets from and to the MN). Packets destined to the MN are delivered to the LMA, which in turn forwards them towards the serving MAG via the established tunnel for the MN, to be finally sent towards the MN by the serving MAG. Whenever the MN switches its point of attachment from one MAG to another, a new tunnel is established between the new MAG and the LMA for the MN and the previous tunnel is teared down. Nevertheless, since the new MAG advertises the same home network prefix as the previous MAG, the MN is able to continue to use the same address that was obtained from the previous MAG.

An illustration of the previously described mobility management solutions is presented in Figure 2.10.

2.4.1.2 Distributed mobility management

The approaches of addressing mobility described above have in common the fact that they are centralized, with traffic requiring to go through a centralized anchor point in the network. This introduces several limitations regarding to non-optimal routing, low scalability and reliability aspects [23]. To mitigate the previous limitations, Distributed Mobility Management (DMM) [115] solutions started to be explored, where mobility anchors are deployed closer to the MNs (i.e., are relocated to the network edge). In doing so, MNs become capable of moving between mobility anchors, avoiding traffic to transverse a single mobility anchor far from the optimal route. DMM solutions can be categorized as partially distributed, where only the data plane is distributed, or fully distributed, where both the control and the data plane are distributed. Moreover, existing DMM solutions can be divided into two categories: (i) client-based; and (ii) network based solutions.

(a) MIPv6 (b) Fast Handovers in Mobile IPv6

(c) Hierarchical Mobile IPv6 (d) Proxy Mobile IPv6

Figure 2.10: Centralized mobility management solutions

approaches to work in a distributed way. Regarding client-based [51, 11, 12] and network-based [52, 49, 106, 18] DMM solutions, most of these proposals focus on distributing the operation of MIP and PMIP protocols, respectively. Other proposals, namely routing-based DMM solutions, leverage its operation on IP routing protocols to support mobility instead of tunnel establishments, being categorized as network-based solutions.

In both client-based and network-based DMM solutions, multiple mobility anchors are deployed in the network edge and, therefore, anchoring is distributed across multiple access networks. Instead of using a single IP address anchored at a central anchor entity, the MN configures additional IP addresses at each visited access network. Such approach allows the MN to use the locally-anchored address for new communications, while active communications can still use previous assigned IP addresses through bi-directional tunnels established between the MN and the corresponding mobility anchor in MIP-based DMM solutions or between the current and the corresponding mobility anchors in PMIP-based DMM solutions. In partially distributed approaches, a dedicated server (e.g., central mobility database), which tracks the MN and maintains mobility sessions of MNs, is responsible for managing the mobility of MNs.

2.4. Mobility management 39

In document Interoperabilidade e mobilidade na internet do futuro (Page 59-63)