Patent Grant
8189512
The invention relates to the field of Mobile IP, and in particular to routing traffic in a Proxy Mobile IP network.
Mobile IP (MIP), which is described in IETF RFC 3344, allows users of mobile communications devices to move from one network to another whilst maintaining a permanent IP address, regardless of which network they are in. This allows the user to maintain connections whilst on the move. For example, if a user were participating in a Voice Over IP (VoIP) session and, during the session the user moved from one network to another, without MIP support the user's IP address may change This would lead to problems with the VoIP session.
A Mobile Node (MN) is allocated two IP addresses: a permanent home address and a care-of address (CoA). The CoA is associated with a node in the network that the user is currently visiting. To communicate with the MN, packets are sent to the MN home address. These packets are intercepted by a Home Agent in the home network, which has knowledge of the current CoA. The Home Agent then tunnels the packets to the CoA of the MN with a new IP header, whilst preserving the original IP header. When the packets are received by the MN, it removes the new IP header and obtains the original IP header. The MN sends packets directly to another node via a foreign agent in the visited network. The foreign agent maintains information about visiting MNs, including the CoA of each visiting MN.
Proxy Mobile IP v6 (PMIPv6), IETF draft-sgundave-mip6-proxymip6-01, describes a Proxy Mobile Agent (PMA) function. This function emulates home link properties in order to make a MN behave as though it is on its home network and allows support for mobility on networks that would not otherwise support MIPv6.
A PMA is usually implemented at the access router. The PMA sends and receives mobility related signalling on behalf of a MN. When a MN connects to an access router having a PMA, the MN presents its identity in the form of a Network Access Identifier (NM) as part of an access authentication procedure. Once the MN has been authenticated, the PMA obtains the user's profile from a policy store. The PMA, having knowledge of the user profile and the NAI, can now emulate the MN's home network. The MN subsequently obtains its home address from the PMA. The PMA also informs the MN's Home Agent of the current location of the MN using a Binding Update message. The Binding Update message uses the NAI of the MN. Upon receipt of the Binding Update message, the Home Agent sets up a tunnel to the PMA and sends a binding acknowledgement to the PMA. On receipt of the Binding Acknowledgement, the PMA sets up a tunnel to the Home Agent. All traffic from the MN is routed to the Home Agent via the tunnel.
The Home Agent receives any packet that is sent to the MN, and forwards the received packet to the PMA through the tunnel. On receipt of the packet, the PMA removes the tunnel header and sends the packet to the MN. The PMA acts as a default router on the access link. Any packets sent from the MN are sent via the PMA to the Home Agent, which then sends the packet on to its ultimate destination.
Mobility Support in IPv6 (IETF RFC 3775 June 2004) describes route optimization initiated by the MN for messages sent from and to the MN. However, the Proxy MIPv6 specification, which is a variant of MIPv6, doesn't assume any mobility management protocol in the MN. The techniques for route optimization specified in MIPv6 cannot be applied to PMIPv6 without modification.
As PMIPv6 does not assume mobility management protocol in the MN, the PMA processes the signals defined in MIPv6 on behalf of the MN. The PMA is therefore well placed to process route optimization signalling on behalf of the MN, although to date there has been no suggestion or description of how the PMA should perform route optimization. A trivial solution would be to apply the principles of route optimization specified in MIPv6. As illustrated in
As illustrated in
Using MIPv6, the MN is aware of the handover, and can initiate a new route optimization. Using PMIPv6, if the PMA is not aware of the handover of the MN, then the PMA cannot initiate a new route optimization. Provided the MN moves within the PMIPv6 domain, the MN's IP address remains unchanged, and so even if the MN is a MIPv6 client, the MN will not send a signal for a new request of route optimization or for cancellation of the current route optimization to the CN.
Another problem with handover of the MN from PMA1 to PMA2 is packet loss. Packet loss during a handover is a problem for all mobile protocols, and sending packets over an optimized route increases the risk of packet loss because the anchor point (the HA) does not have control of the optimized route taken by packets.
According to a first aspect of the invention, there is provided a method of routing communications traffic in a Proxy Mobile IP communications network, in which a Mobile Node is attached to a first Proxy Mobile Agent, and an optimized route is established between the Mobile Node and a Corresponding Node via the first Proxy Mobile Agent, and the Mobile Node subsequently detaches from the first Proxy Mobile Agent and attaches to a second Proxy Mobile Agent, the method comprising:
It is preferred that the step of determining when the Mobile Node is no longer attached to the first Proxy Mobile Agent comprises determining that a Binding Cache entry for the Mobile Node at the Mobile Node's Home Agent has changed.
One way of determining when the Mobile Node is no longer attached to the first Proxy Mobile Agent comprises:
The forwarding of packets to both the Mobile Node and the Home Agent is termed ‘bi-casting’. An advantage of this method is that the packets sent to the Home Agent are addressed to the Mobile Node, and so the Home Agent will automatically return then to the Proxy Mobile Agent. This is the normal behaviour of the Home Agent, and so no modification of the Home Agent or the Mobile Node is required.
The method may comprise, prior to establishing the optimized route between the Mobile Node and the Corresponding Node:
An alternative method for determining when the Mobile Node is no longer attached to the first Proxy Mobile Agent comprises receiving at the Proxy Mobile Agent a signal informing from the Home Agent informing the first Proxy Mobile Agent that the Mobile Node is no longer attached to the first Proxy Mobile Agent. The sending of this signal from the Home Agent may be triggered by a change in the Binding Cache entry for the Mobile Node in the Home Agent. An advantage of this method over the ‘bi-casting’ method is that traffic on the network is reduced.
Another way of determining when the Mobile Node is no longer attached to the first Proxy Mobile Agent is when an indication is received at the Proxy Mobile Agent from link-layer protocols that the Mobile Node is unreachable from the first Proxy Mobile Agent. Again, this reduces the amount of traffic on the network compared to the ‘bi-casting’ method.
The method may further comprise, prior to determining when the Mobile Node is no longer attached to the first Proxy Mobile Agent, forwarding packets destined for the Mobile node to the Mobile Node, a Home Agent, and at least one neighbouring Proxy Mobile Agent. This ensures that packets are sent to the likely Proxy Mobile Agents that the Mobile Node may subsequently attach to.
Preferably, the Mobile Node connects to the first Proxy Mobile Agent via a mobility-aware access network. This allows the first Proxy Mobile Agent to become aware when the Mobile Node subsequently attaches to a second Proxy Mobile Agent.
Once it has been determined that the Mobile Node is no longer attached to the first Proxy Mobile Agent, it is preferable to cancel the forwarding of packets from the first Proxy Mobile Agent to the Home Agent.
The request to cancel the route optimization may be a request to delete at the Corresponding Node a Binding Cache entry for the Mobile Node. Once the Binding Cache Entry has been deleted, the Corresponding Node will revert to its default behaviour and send all packets to the Mobile Node's Home Agent.
According to a second aspect of the invention, there is provided a Proxy Mobile Agent comprising:
The Proxy Mobile Agent may further comprise:
By ‘bi-casting’ the packets to both the Mobile Node and the Home Agent, the invention takes advantage of the behaviour of the Home Agent, which returns to the Proxy Mobile Agent all packets addressed to the Mobile Node. Using this embodiment, no modification is required to the Home Agent in order for the Proxy Mobile Agent to become aware that the Mobile Node is no longer attached to it.
The Proxy Mobile Agent preferably comprises:
According to a second aspect of the invention, there is provided a Home Agent for use in a Proxy Mobile IP communications network, the Home Agent comprising:
Proxy Mobile Agent PMA1, as shown in
PMA1 detects that the downlink packets are not returned from the HA, and so becomes aware that MN is no longer attached to PMA1. This indicates that the MN is now attached to another PMA. This detection is a trigger for the PMA1 to perform handover related procedures such as cancellation of the route optimization with the CN that has previously been performed. PMA1 does not detect a lack of packets from the MN in an uplink to detect that the MN is no longer attached to PMA1, because in the case where unidirectional traffic is being sent from the CN to the MN via PMA1, there would be no uplink packets to detect. Furthermore, even where communication between the CN and the MN is bi-direction, a handover of the MN to PMA2 during transmission of packets from the CN to the MN would not be detected until PMA1 next expects uplink packets from the MN.
A further advantage of bi-casting is that the risk of packet loss during handover from PMA1 to PMA 2 is reduced, because the downlink packets sent to PMA1 are sent to the HA and forwarded to PMA2 immediately after the HA performs a change of the binding cache entry for the MN when the MN attaches to PMA2.
Referring now to
When the MN moves to a new access network, it performs a handover from PMA1 to PMA2, as illustrated in
The packets sent from PMA1 to the HA are not returned to PMA1 from the HA, and so PMA1 becomes aware that the MN is no longer attached to PMA1. This is a trigger for PMA1 to initiate cancellation of the route optimization, as illustrated in
The procedure to cancel the route optimization that the PMA1 performs after the handover of the MN from PMA1 to PMA2 is illustrated schematically in
PMA1 bi-casts packets received from MN1 (source=MN1, destination=MN2) to both MN2 and HA2, and, as described above, PMA1 discards packets returned from HA2 because the same packets have been already sent to MN2. Similarly, PMA1 bi-casts packets received from MN2 (source=MN2, destination=MN1) to both MN1 and HAL As before, the packets returned from the HA1 are discarded by the PMA1 because same packets have been already sent to MN1.
When MN2 moves to another access network, it attaches to PMA2 and is no longer attached to PMA1. This scenario is illustrated in
The case where two Mobile Nodes belong to the same Proxy MIP domain is known as domain-local communication, and is illustrated in
The request for route optimization is in a different format to that of a normal Mobile IPv6 Binding Update to ensure that the request is intercepted and understood by PMA2. The format can that of a Proxy Binding Update or any suitable equivalent. The request is addressed to MN2. Because the request is sent in a special format, it is intercepted by PMA2. If the request is a normal MIPv6 Binding Update, it is forwarded from PMA2 to MN2.
PMA2 responds to the request with an acknowledgement. When PMA1 receives the acknowledgement, it adds MN2 to the Binding Update list, and unlike a normal entry in a Binding Update list, the MN2 entry includes the source address of PMA2, as PMA2 is the target to which PMA1 sends the request of deleting the Binding Cache. The Binding Cache is therefore created on PMA2 and not MN2.
In a similar manner to that described above, PMA2 sends a request of route optimization to the MN1, as illustrated in
MN2 may subsequently handover from PMA2 to a new Proxy Mobile Agent, PMA3. This situation after handover is illustrated in
This embodiment requires no support from other nodes (MN, HA, CN, and other PMAs), and there is no requirement for Layer 2 functionalities. No modifications are required to Home Agents, Mobile Nodes or to the current Proxy MIPv6 specification. The invention provides low packet loss during a handover between PMAs. In some embodiments, no signalling is necessary to the previous Proxy Mobile Agent. Mobile Nodes start to receive packets from the previous Proxy Mobile Agent immediately after the handover is complete.
There are alternative ways to bi-casting by which a Proxy Mobile Agent can become aware that a Mobile Node has moved to another Proxy Mobile Agent. For example, a PMIPv6 HA knows the identity of the previous PMA to which the MN was attached before moving to the current PMA. The HA may send a signal to the previous PMA infomring it that the MN is now attached to a new PMA. This reduces the amount of traffic, as bi-casting is not required.
After handover of MN from PMA1 to PMA2, the HA informs PMA1 that MN is now attached to PMA2. In response to this information, PMA1 sends a request to the CN to cancel route optimization. The CN deletes the binding cache entry of MN, and as a result of this deletion sends all packets destined for MN to the HA, as illustrated in
The bi-casting described above may not be necessary if the PMA can detact that a Mobile Node has become unreachable immediately after the Mobile Node leaves the coverage of the PMA (e.g., the indication from the link-layer protocols).
When the MN attaches to a new Proxy Mobile Agent, PMA2, it becomes unreachable from PMA1. PMA1 detects the unreachability of the MN and so stops sending downlink packets directly to the MN. Instead, the PMA1 sends the downlink packets to the HA, which will be forwarded to PMA2 and then to the MN. PMA1 also sends a request to cancel the route optimization to the CN. The CN deletes the binding cache entry of the MN and sends packets destined for the MN to the HA, as illustrated in
In a mobility-aware access betwork, the access network can keep track of the PMA to which the mobile node is attached, as illustrated in
The case where the MN moves and becomes attached to PMA2 is illustrated in
PMA2 also detects that the MN has moved to PMA2, and so PMA2 starts to send downlink packets to the MN. Because the downlink packets (from the CN) are forwarded from PMA1, the PMA2 is aware that the downlink packets have been sent through the optimized route by the PMA1. This triggers PMA2 to a request for route optimization to the CN.
The embodiments described above that do not rely on bi-casting can be used in addition to the bi-casting embodiment. For example, the bi-casting embodiment can be used as a fall-back solution for cases where the movement of the Mobile Node has, for some reason, not been detected. Compared to the bi-casting embodiment, less traffic is sent between the PMA and the HA, and packet loss during handover can be reduced.
It will be appreciated by the person of skill in the art that various modifications may be made to the above described embodiments without departing from the scope of the present invention.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP2007/052829 | 3/23/2007 | WO | 00 | 3/26/2010 |
| Publishing Document | Publishing Date | Country | Kind |
|---|---|---|---|
| WO2008/116494 | 10/2/2008 | WO | A |
| Number | Name | Date | Kind |
|---|---|---|---|
| 20020039367 | Seppala et al. | Apr 2002 | A1 |
| 20060018291 | Patel et al. | Jan 2006 | A1 |
| 20070189219 | Navali et al. | Aug 2007 | A1 |
| 20070230410 | Thubert et al. | Oct 2007 | A1 |
| 20070254661 | Chowdhury et al. | Nov 2007 | A1 |
| 20080225807 | Patil | Sep 2008 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20100226256 A1 | Sep 2010 | US |
