Patent Application
20050259631
The present invention relates to a routing method and system for routing data packets from a source terminal to a destination terminal via at least one communication network.
Recent developments in communication technology lead to communication networks operating based on the Internet Protocol (IP). In connection with mobile communication, Mobile IP is thus becoming more and more important. In communication networks, or network systems comprising plural individual networks interconnected with each other, data are forwarded in units of so-called data packets from a source terminal to a destination terminal. A mobile source terminal is referred to as mobile node MN, while a destination terminal (which may be a fixed or mobile terminal) is referred to as a correspondent node CN. Each of the interconnected networks comprises at least one mobility agent entity for each of said terminals. A mobility agent is any network entity implementing functionalities supporting mobility of the terminal within the network/network system while assuring that communication remains possible. For example, the expression “mobility agents” as used in the present text comprises
As is generally known, Mobile IP includes a method of routing packets through a Home Agent (HA) to provide mobility transparency to the Correspondent Nodes (CN) and the Transport and Application layers in the Mobile Node itself. Routing packets through the HA results in long routes, especially when the MN is roaming in a network topologically/geographically distant from the home network. This is also known as the triangular routing problem.
Routing can be optimized through the usage of a dynamically assigned home agent from the visited network, or using other locally assigned addresses for communication with the Correspondent Nodes. This involves letting the correspondent node CN know the binding between the Home Address and the Care_of_Address. Binding in this connection denotes a triplet of numbers that contains the mobile node's (MN) home address (permanent address, e.g. IP address), its temporary address, i.e. CoA, and the registration lifetime (i.e. how long the mobility agents may use the binding).
GPRS networks as an example for mobile packet data networks manage mobility in conjunction with the link layer connectivity in the cellular access network. The currently defined methods assign an address to the MN from the address pool of either the local GGSN or a GGSN in the home network (GGSN=Gateway GPRS Support Node, GPRS=General Packet Radio Service).
It is to be noted that GPRS represents an example of a network only to which the present invention is applicable. The invention as subsequently described is, however, applicable to any routing of data packets from a source terminal to a destination terminal via at least one communication network, said at least one communication network comprising at least one mobility agent entity for each of said terminals. Also, the protocol used in such packet data networks is not limited to any specific protocol type. For example, Mobile IP version 4 (IPv4) or version 6 (IPv6), or GPRS specific protocols can be adopted.
An earlier patent application of Applicants which was filed in September 2001 describes a method of managing a Mobile IP Binding Cache outside of the Correspondent Nodes in their access network routers.
Thus, the problem resides in providing an IP routing between a Mobile Node MN and a Correspondent Node CN (often the correspondent node is itself a Mobile Node as well) where routing loops, or unnecessarily long routes in general are avoided, and at the same time the location privacy of both communicating nodes is protected. Also, it is essential to the system responsiveness and scalability that no connection state as such is required before the packets can be routed to the mobile node. Furthermore, signaling and other overhead over the air interface should be avoided.
The key in location privacy is that the address(es) used in the communication with the correspondent node reveal no information about the mobile's current location, or point of attachment in the packet network topology. Such information is included and/or can be deduced from the Care Of Address, for example.
There are at least two ways to provide addressing meeting this criterion:
The second option above has the added benefit that the mobile's use of the network will be harder to profile over time, if different addresses are used at different times by the same mobile device.
When the address conveys no information about the current point of attachment, the network must be able to map the address to an address in the access network where the mobile node is actually attached. The point in location privacy is that this mapping is hidden from the correspondent node.
Furthermore, to ensure the scalability of the network, the address used in the communication must be routable—packets sent with the address must reach a point in the network where the current location of the mobile node is known, so that the packet may be further forwarded to the mobile node in a timely fashion. The alternative, where a location look-up over the network is required before the packet sent to the mobile can be forwarded at all has the problem that the packets need to be queued at the origin access network while the location request is being served. This will cause initial delay, additional burstiness, and possible packet loss due to buffer overruns. Also, this will require location lookup for every small session of communication, while it would have been more preferable to take a hit in the routing efficiency (if any) for the benefit of less location signaling and state maintenance. Such a look-up concept involving so-called location privacy agents is disclosed in Applicants former patent application filed with the US PTO under Ser. No. 09/986,602 on Nov. 9, 2001.
The above means that a rendezvous point like the Mobile IP Home Agent (HA) is essential to any connectionless packet network providing mobility with location privacy. But routing everything via the home agent HA will cause unnecessary routing loops, especially when the mobile node MN is roaming in networks topologically far away from the home network, and is communicating with correspondent nodes outside of the home network. At an extreme this could for example mean to route the packets from U.S. to Finland and then back from Finland to U.S., if the mobile node's MN home agent HA is located in Finland, but the MN is roaming in the U.S. and communicating with an IP host in U.S.
Thus, from the foregoing it becomes clear that currently both GPRS and Mobile IP solutions suffer from the lack of optimized mobility and location privacy at the same time. They can provide for either optimal routing, or location privacy, but not both simultaneously.
Consequently, it is an object of the present invention to provide an improved routing method and system for routing data packets from a source terminal to a destination terminal via at least one communication network, which method is free from the above mentioned drawbacks.
According to the present invention, the above object is for example achieved by a routing method for routing data packets from a source terminal to a destination terminal via at least one communication network, said at least one communication network comprising at least one mobility agent entity for each of said terminals, the method comprising the steps of: establishing a route from the source via at least one first mobility agent associated to said source, at least two consecutively arranged second mobility agents associated to said destination, to said destination, deciding that said route is to be optimized, and upon said decision, rerouting said route from one of said at least one first mobility agents directly to one of the at least two consecutively arranged second mobility agents such that at least one intermediate mobility agent in said route is bypassed in the resulting rerouted route.
In this connection, it is to be noted that rerouting can happen before a single packet has been transmitted via the (initially) established non-optimized route. That is, the route may still be in the process of being established or may have already been established. In each of the cases, an appropriate indication in a signaling such as resource reservation signaling may trigger rerouting even before the establishment of the (initial, non-optimized) route is actually completed.
According to favorable further developments
Furthermore, according to the present invention the above object is for example achieved by a routing system for routing data packets from a source terminal to a destination terminal via at least one communication network, said at least one communication network comprising at least one mobility agent entity for each of said terminals, the system comprising: route establishment means adapted for establishing a route from the source via at least one first mobility agent associated to said source, at least two consecutively arranged second mobility agents associated to said destination, to said destination, decision means adapted for deciding that said route is to be optimized, and, rerouting means, adapted to perform in response to said decision a rerouting of said route from one of said at least one first mobility agents directly to one of the at least two consecutively arranged second mobility agents such that at least one intermediate mobility agent in said route is bypassed in the resulting rerouted route.
According to favorable refinements of the present invention,
By virtue of the present invention an advantageous routing method which simultaneously provides for optimal routing and for location privacy is realized, i.e. the location information of the destination is not leaked to the source or other non-trusted entities in the network
In the following, the present invention will be described in greater detail with reference to the accompanying drawings, in which
The present invention will now be described in detail with reference to the drawings.
Furthermore, MN1, MN2 denote mobile nodes acting as a source as well as a destination, respectively. The mobile nodes are identified and addressable by their home addresses H1, H2, respectively. Each of the mobile nodes access the network NW1 via an access router AR1, AR2, respectively, as a mobility agent for a respective one of said terminals. In addition, the network is provided with a respective Home agent HA1, HA2, respectively, for said mobile node MN1 and/or MN2. Even though it may not be excluded that HA1 and HA2 are locally close to each other or located at the same site, for the further explanations it is assumed that they are physically different entities. The same applies to the access routers as it is assumed that the source and destination terminals are geographically distant from each other so that they have to rely on the use of different access routers for accessing the network.
Now, a routing is described as illustrated in
Thus, after step 4, the initial packet to be routed from the source to the destination has been delivered from MN1 to MN2. Assuming that MN2 answers the message received from MN1, the packet flow is as denoted in
Apparently, there is established a route 1, 2, 3, 4 (from MN1 to MN2) and/or 5, 6, 7, 8 (from MN2 to MN1) from the respective source MN1/MN2 via a first mobility agent AR1/AR2 associated to said source, at least two consecutively arranged second mobility agents HA2, AR2/HA1, AR1 associated to said destination, to said destination MN2/MN1.
Every subsequent packet from MN1 to MN2 and vice versa will take the same routing through the network and will involve such a triangular routing of e.g. AR1->HA2->AR2. This may represent a rather long distance causing undesirable delays.
The present invention, when adhering to the example of
In this connection, it has to be noted again that for explanatory purposes the drawings are simplified. Thus, it should be kept in mind that basically it is not required for the invention that the *first* one does the rerouting to the *last* one. The only condition is be that the path is significantly shortened by the route optimization. For example, it may be that a 2nd mobility agent reroutes the packets to a 5th one, thus by-passing 3rd and 4th ones. More generally, there is established a route from the source via at least one first mobility agent associated to said source, at least two consecutively arranged second mobility agents associated to said destination, to said destination, decided that said route is to be optimized, and upon said decision, a rerouting of said route from one of said at least one first mobility agents directly to one of the at least two consecutively arranged second mobility agents is performed such that at least one intermediate mobility agent in said route is bypassed in the resulting rerouted route. Simultaneously, this is done so that the location information of the destination is not leaked to the source or other non-trusted entities in the network. This will be explained in more detail in the following. A comparison of the packet flow between
Upon such rerouting a subsequent packet from MN1 at step 6 to AR1, AR1 will route the packet in step 7 directly to AR2, as denoted by 7:AR1>AR2(H1>H2), thereby bypassing HA2 for this and the subsequent packets. Stated in other words, the rerouting comprises the step “5” of informing said first mobility agent AR1 of a current care_of_address AR2 of the destination, wherein said informing comprises the steps of sending (5) a message from the first one HA2 of said consecutively arranged second mobility agents to said first, AR1, mobility agent including the current care_of_address of the destination.
Thus, routing distance is shortened, delivery of packets becomes faster and delay sensitive applications may benefit from such a routing. Simultaneously, the CoA of MN2, is not revealed to the source MN1. Rather, the CoA of MN2 is only informed to AR1 and kept within the network. Hence, location privacy of MN2 is maintained while routing is optimized. Of course, in case MN2 moves so that another AR becomes “responsible”, the binding will be updated.
As regards the decision for initiating route optimization, said decision is taken at one of said at least two second mobility agents associated to said destination, i.e. at HA2 or AR2. Since HA2 is the first of the mobility agents associated to MN2, it could be preferred to let HA2 decide on whether to perform route optimization or not.
The decision is for example based on an indication by the source to optimize the route. In this case, a packet sent from MN1 to MN2 includes, e.g. in its header, a corresponding indication such as a specific bit set to a predetermined value indicating that the routing is to be optimized.
Also, said decision can be based on a service type of the traffic between the source and the destination. In such a case, the service type (of the application to which the packet belongs) is indicated in the data packet or a signaling message. The mobility agent checks whether the service type of the packet matches a predetermined service type for which route optimization is to be performed, and if so, performs optimization as described above. Examples for such a service type may be a service type indicating a service imposing delay requirements, such as indicating real-time traffic.
Additionally or alternatively to the above, said decision can be based on an estimated distance between said source and said terminal, and in case said estimated distance exceeds a predetermined threshold distance value, it is decided by said mobility agent to reroute said established route, as described above. Stated in other words, the mobility agent (e.g. HA2 in the above example) evaluates/estimates the length of the packet route and dependent on the estimation decides to reroute packets. Thus, any packet may be rerouted dependent on the route length or only packets of a specific service type are rerouted dependent on the route length estimation.
Note that although
In
The same notation regarding the signals/steps as in
As shown in
Thus, within network NW1, packets routed to the mobile node MN1 (from the source in the external network NW2) are passing via the rather long route from ERn to HA1 to AR1 and then to MN1.
According to the present invention, when applied to this scenario, such drawback is prevented, as will become apparent from
The present invention, when adhering to the example of
This will be explained in more detail in the following. A comparison of the packet flow between
Upon such rerouting a subsequent packet from ENx at step 9 to MN1, ERn will route the packet in step 10 directly to AR1, as denoted by 10:ERn>AR1(Ex>H1), thereby bypassing HA1 for this and the subsequent packets. Stated in other words, the rerouting comprises the step “8” of informing said first mobility agent ERn of a current care_of_address AR1 of the destination, wherein said informing comprises the steps of sending, 8, a message from the first one HA1 of said consecutively arranged second mobility agents to said first, ERn, mobility agent including the current care_of_address of the destination.
Thus, routing distance is shortened, delivery of packets becomes faster and delay sensitive applications may benefit from such a routing. Simultaneously, the CoA of MN1, i.e. AR1 is not revealed to the source ENx. Rather, the CoA of MN1 is only informed to ERn and kept within the network NW1. Hence, location privacy of MN1 is maintained while routing is optimized. Of course, in case MN1 moves so that another AR becomes “responsible”, the binding will be updated.
It is to be noted that the present invention can be implemented at nearly any time by taking the decision to reroute the established route. Therefore, the explanation referring to an initial “first” packet and subsequent “second” packet has been chosen as a mere example for enhancing understanding of the invention. As regards the decision to be taken, e.g. by HA1 in the case of
Stated in other words, as mentioned above, it is very likely that most of the time routing via the HA will not cause any significant routing inefficiency, for example, when the subscriber is located close to his home, and the Home Agent situated topologically “close by” is used, no significant saving can be attained by route optimization. In general, it should therefore be decided on a case-by-case basis, whether route optimization will result in saving of delay, or network resources offsetting the cost of the route optimization related signaling and state maintenance. Taking such a decision can be based on criteria as explained above, while of course additional criteria may also be applied.
Taking the above into consideration it is the Home Agent itself, who has all the information needed to make the decision for the route optimization. Home Agent sees the address of the correspondent node CN, as well as the current care-of address CoA of the mobile node. The home agent can also profile and/or monitor the traffic between the two and decide if and when to initiate route optimization. Optionally, the AR or even the MN itself could be utilized in triggering the route optimization and thus reducing the burden on the Home Agent related to following the traffic patterns being forwarded. An example of this could be some resource reservation signal originating from the MN for requesting certain Quality of Service for a forthcoming traffic stream (e.g. real.time traffic). If low delay is requested, the related Home Agents could be asked to proactively arrange route optimization to decrease the end-to-end transmission delay.
Route optimization itself cannot be performed by the Mobile Node, or the Correspondent Node, since doing that would reveal the care_of_address of the other party to the other, and thus having no location privacy. Due to this the care-of addresses of the communicating entities need to be kept inside the network (Mobile Network). Mobility Agents at the edges of the network will take care of the route optimization, as signaled by the home agents.
The Access Routers providing network access for the Mobile Nodes will take care of the route optimization and are trusted not to reveal the care-of address of the correspondent node to the mobile node they are serving. Edge routers interfacing the other networks will terminate all mobility and route optimization related signaling to guard the location revealing information from leaking to non-trusted networks/entities.
In an ideal case, the whole global network (e.g. the Internet) would be mobile (the Mobile Internet) and utilizing this invention (
The edge router will then tunnel the packet sent to the mobile node's home address to the MN's Home Agent. The tunneling method used is immaterial, but it is essential that the edge router's address will be carried or otherwise made known to the Home Agent. The Home Agent will further forward the packet to the access router serving the Mobile Node.
Various optimization schemes may be utilized to reduce the tunneling overhead within the network itself. It should be noted that this invention does not call for tunneling over the air interface (the interface between the MN and an Access Router).
In the reverse direction, the MN will send it's packets with it's Home Address as the source address. The AR will authorize the MN's use of the specific Home Address by communicating with the Home Agent either directly or via other mechanisms, such as AAA (e.g. via the Diameter protocol). The rest of the Mobile Network will trust on the access routers to block any unauthorized source addresses. The edge routers will enforce this by not forwarding packets out of the Mobile Network with source addresses out of the pool of Home Addresses valid in the Mobile Network.
The destination address used in the packets sent by an MN in the Mobile Network is the address of the correspondent node known to the MN. It can be a home address of another Mobile Node either in the same or different network, or a normal IP address of a fixed node in either the Mobile Network itself or an external network. Normal IP routing will cause an optimal (shortest) path to be taken to the destination address. If the destination address is a home address, it will reach the Home Agent of the destination, from where it is forwarded to the access router of the destination. If the destination's Home Agent will want to utilize route optimization, the HA will send a Binding Update towards the Access Router or an Edge Router who sent the packets to the HA. (this will initiate rerouting, as explained herein before.)
The initial network access registration is required to convey the MN's Home Address from the Home Network to the Access Router (AR). After this the AR will allow the MN to send packets with the Home Address as the source address in the IP packets. The AR's address can be used as the Care-of Address (CoA) for the MN, if IP-in-IP tunneling (or equivalent, such as GRE or GTP) is used for the transport of user's packets in the Mobile Network. Alternatively, the AR may allocate an unique CoA for the MN, allowing the AR to provide a one-to-one mapping between the CoA and the Home Address, enabling optimized encapsulation in most cases.
The address ownership management will be made easier by the fact that it is the HA owning the address that will be involved in the binding update process. Access and Edge Routers can be configured to accept binding updates from known Home Agents of the Mobile Network only. After the Binding exists the tunnel endpoints are responsible of refreshing the binding. The refreshing need not necessarily happen via the Home Agent.
Optionally, the access routers could initiate the binding updates without involving the home agents directly, but that requires the Home Agent to use the sending AR's address as the tunnel source address when forwarding the packet to the destination AR. This way the destination AR knows of the source AR, and will be able to decide whether to do an binding update or not.
Different versions of the Internet Protocol (IP) can be used for the service to the MNs and the transport inside the Mobile Network. For example, MN's could be provided with IPv6 service, even when the internal transport in the Mobile Network is utilizing IPv4.
The invention is applicable to the GPRS networks. Here the GGSNs are the Access Routers. Home Address of the MN could be fetched from the HLR/HSS. Subscriber's home operator would maintain Home Agents, where the current GGSN under which the Mobile is located is known. Edge routers would be managing Binding Caches for the mobiles of the GPRS network, enabling optimal routing. The tunneling method utilized could be the GPRS Tunneling Protocol (GTP).
Optimally, the GPRS network has Edge Routers situated on all the major geographical locations. This makes the mobile network span widest possible area, enabling route optimization. Packets from external networks would be routed to the GPRS network through the Edge Router closest to the traffic source, allowing the GPRS network to provide optimal routing without revealing any location information to the external network entities.
The established roaming agreements should be utilized to allow different GPRS networks to be combined into a federated network, inside of which the location information (current point of attachment) could be utilized to provide the best routes.
The present invention as outlined above proposes that resource reservation signaling should indicate that routing optimization should be performed. Also, for example, the Edge Routers maintain Binding Caches for the correspondent nodes in the other networks. No information about the MN's current location is sent to the other networks. The whole definition of the “Mobile Network” involves trust between the elements in the Mobile Network. For example, if an external node would try to use Edge Router's or Access Router's address as the source address, that would be spotted on one of the routers on the edge of the Mobile Network (ingress filtering). In addition, it is not unfeasible to have internal keying infrastructure covering the network elements of the mobile network. So the tunneling between ARs/ERs and the HAs would need to be covered by (possibly transitive) trust relationships between them. The invention proposes the HA to terminate the tunnel the ER (or AR) sends to it. The tunneling from the HA to the target MN would happen normally, assuming that the “AR” is provided to the HA as the care-of address.
Accordingly, as has been described herein above, the present invention concerns a routing method for routing data packets from a source terminal MN1, H1; Enx, Ex to a destination terminal MN2, H2; MN1, H1 via at least one communication network NW1; NW1, NW2, said at least one communication network comprising at least one mobility agent entity HA1, HA2, AR1, AR2, ERn, ERm for each of said terminals, the method comprising the steps of: establishing a route 1, 2, 3, 4; 4, 5, 6, 7 from the source MN1, H1; Ex, ENx via at least one first mobility agent AR1; ERn associated to said source, at least two consecutively arranged second mobility agents HA2, AR2; HA1, AR1 associated to said destination, to said destination MN2, H2; MN1, H1, deciding that said route is to be optimized, and upon said decision, rerouting said route from one of said at least one first mobility agents AR1; ERn directly to one of the at least two consecutively arranged second mobility agents AR2; AR1 such that at least one intermediate mobility agent HA2; HA1 in said route is bypassed in the resulting rerouted route. The present invention also concerns a corresponding system.
In detail, even though not expressly depicted in the drawings, the foregoing description of the present invention apparently also discloses a routing system for routing data packets from a source terminal (MN1, H1; Enx, Ex) to a destination terminal (MN2, H2; MN1, H1) via at least one communication network (NW1; NW1, NW2), said at least one communication network comprising at least one mobility agent entity (HA1, HA2, AR1, AR2, ERn, ERm)) for each of said terminals, the system comprising: route establishment means adapted for establishing a route (1, 2, 3, 4; 4, 5, 6, 7) from the source (MN1, H1; Ex, ENx) via at least one first mobility agent (AR1; ERn) associated to said source, at least two consecutively arranged second mobility agents (HA2, AR2; HA1, AR1) associated to said destination, to said destination (MN2, H2; MN1, H1), decision means adapted for deciding that said route is to be optimized, and, rerouting means, adapted to perform in response to said decision a rerouting of said route from one of said at least one first mobility agents (AR1; ERn) directly to one of the at least two consecutively arranged second mobility agents (AR2; AR1) such that at least one intermediate mobility agent (HA2; HA1) in said route is bypassed in the resulting rerouted route.
The decision means is located at one of said at least two second mobility agents (HA2, HA1) associated to said destination.
Said decision is based on an indication by the source or destination to optimize the route or to request for a specific quality of service for which route optimization is beneficial. Alternatively and/or additionally, said decision is based on a service type of the traffic between the source and the destination. For example, said decision to optimize the route is taken in case the service type indicates a service imposing delay requirements, e.g. said service type indicates real-time traffic.
Said decision is based on an estimated benefit from route optimization between said source and said terminal, and in case said estimated benefit exceeds a predetermined threshold value, it is decided to reroute said route. The benefit can be measured/expressed in a delay reduction (as compared to non-optimized routing, which in turn may correspond to a (shortened) distance between source and destination).
Said rerouting means comprises informing means adapted for informing one of said at least one first mobility agents of a current care_of_address of the destination, wherein said informing means comprises sending means adapted to send a message from one of said consecutively arranged second mobility agents to one of said first mobility agents including the current care_of_address of the destination.
Said indication triggering the decision means for deciding for route optimization is included in a resource reservation signaling.
While the invention has been described with reference to a preferred embodiment, the description is illustrative of the invention and is not to be construed as limiting the invention. Various modifications and applications may occur to those skilled in the art without departing from the true spirit and scope of the invention as defined by the appended claims.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/IB02/02842 | 7/19/2002 | WO | 1/18/2005 |
