KR20040040448A - Method and arrangement in an ip network - Google Patents

Abstract

The present invention relates to a method and arrangement of an IP network. It is an object of the present invention to provide a scalable solution for pre-allocating resources to obtain predictable end-to-end QoS in heterogeneous IP networks. This object is achieved by classifying the destination domain into domain property labels that indicate the resource availability of the destination domain and how to obtain QoS as an endpoint in a particular domain category.

Description

METHOD AND ARRANGEMENT IN AN IP NETWORK

The Internet is based on the Internet Protocol (IP) standardized by the IETF. Some initial objectives with the IP protocol are to interconnect different kinds of physical networks into one large virtual network and to provide a uniform platform for supporting a wide range of applications. Some technical reasons for significant success in achieving this goal are:

Stateless packet forwarding: IP datagram forwarding is stateless for the application data stream. Forwarding is performed according to the prefix table of the called party address.

Dynamic and scalable routing: Routes are distributed to distributed dynamic intra- and inter-domain routing protocols such as Open Shortest Path First (OSPF) and Border Gateway Protocol (BGP). Set up. This routing protocol automatically detects network errors and sets up new routes to avoid them. Inter-domain routing is well scaled to the destination rooted sink tree because of the aggregation of network address prefixes.

IP is designed for use in networks where different traffic flows share network resources evenly. This means that the received QoS depends on the current load on the network.

At present, the Internet has been developed in terms of link technology from fiber optics to wireless, in terms of application service demand from real-time bidirectional to asynchronous bulk data transmission, and also from user-critical business applications to unstructured home entertainment. In terms of diversification. This development leads to the need for service differentiation of the network. A requirement in the QoS mechanism is that it must be developed according to the basic principles of stateless packet forwarding and scalable aggregation described above.

The technical state of QoS in IP network is as follows.

Integrated Service (IntServ) / Resource ReSerVation Protocol (RSVP)

The IntServ structure and RSVP are signaled structures for guaranteeing end-to-end QoS of individual application data streams. This solution guarantees particulate service at the price per flow state compound packet classification within the router along with the route.

In RSVP, it has been proposed to set up an aggregated tunnel between an aggregator and a de-aggregator. This is a more hierarchical but still a model where aggregated tunnels must be established between pairs of edge routers. This edge router suffers at least the same complexity as the standard IntServ / RSVP. For network policy management, RSVP relies on policy servers.

Differentiated Service (DiffServ)

DiffServ standardizes router support for class-based forwarding. DiffServ forwarding at the core router is stateless for the application data stream. Traffic conditioners at domain boundaries are used to protect domains from overloads.

The problem with DiffServ is that it must meet the QoS demands for a wide range of applications. Resources (bandwidth) for various traffic classes may be calculated and provided semi-statically according to expected service characteristics and estimated usage statistics. However, to provide predictable service levels only through provisioning, resources must be over-ready. This may be possible in homogeneous networks with homogeneous applications and user demands. In a real network, there are many applications / users whose links with very different characteristics are connected to each other (e.g., fiber optic connections and wireless connections) and which have demands beyond those prepared on all hops.

To provide predictable services in heterogeneous environments, DiffServ must rely on Dynamic Network Resource Management (NRM) to control the quality and use of prepared resources. To meet scalability needs, resource management must support the aggregation of resource needs.

Multi-Protocol Label Switching (MPLS)

MPLS is a method of extending a typical IP network layer routing and control protocol with label-switched forwarding. MPLS provides connection-oriented switching in IP networks. The label is associated with a specific data stream (known as Forwarding Equivalence Classes (FEC)). The labels and their FEC combinations are distributed through a network that is an MPLS domain to establish a label-switched route. Upon entering the domain, a packet is assigned one or more labels (label stacks). As the domain passes, the packet is forwarded based on the label. MPLS can be used to provide QoS by allocating resources to specific label-switched paths. MPLS only works within individual label-switched domains. Inter-domain resource reservation is not currently supported.

All of the above methods require additional support for inter-domain resource preparation. This can be provided by a server-based architecture. For RSVP, the architecture of the policy server has been proposed. For DiffServ, QoS agents and bandwidth brokers have been proposed. For MPLS, a QoS agent that understands the meaning of MPLS can be used.

In Schelen, O. Quality of Service Agents in the Internet, Doctoral Thesis, Department of Computer Science and Electrical Engineering, Division of Computer Communication, Lulea University of Technology, Lulea, 1998, Network Resource Manager; NRM ) Was introduced. The NRM may provide inter-domain resource preparation and call admission control separately from, or in conjunction with, the mechanism described above. Among them, the combination of differential forwarding and NRM operates along the basic lines of inter-domain set and stateless forwarding described above. NRMs can be identified by path-dependent admission control, resource scheduling over time, resource demand management for immediate and future use, resource signaling between resource manager entities (ie inter-domain communication), and incoming prefixes identified by address prefixes. Has an aggregation of resource requests towards the side domain. NRM knows the topology and characteristics of the network and can thus keep track of the resources that exist within the routing domain based on the topology. For each domain in the network, the NRM is responsible for admission control. The entity of the NRM may perform admission control in its own domain and may allocate resources to neighboring NRMs for other called parties. Therefore, NRM can provide predictable QoS.

The funnel concept is also introduced in Schelen. The Parnell concept is a hierarchical model for the harmonization of resource needs. The panel concept uses NRM, which requests resources from other NRMs. Reservations made to the same called party from different sources are aggregated where they meet along the path, so each NRM has at most one reservation per destination domain with neighboring agents. The NRM in charge of the domain in which the destination point is located may generalize the received reservation request for that point to cover all of the endpoints of that domain.

1 shows how resource requests are aggregated towards the called party domain. 1 is a network 100 comprising four domains (A, B, C and D). Each domain has NRM (a, b, c, d). Dx, Dy and Dz may be sub-networks or base station controllers in a radio access network. NRM (a) and NRM (b) require resources in domain D; NRM (a) is connected to Dy and NRM (b) is connected to Dx. Since the NRM knows the network topology, they know that the packet must be sent over domain C. In this example, NRM (a) sends 20 resource unit requests to NRM (c) (109), and NRM (b) sends 10 resource unit requests to NRM (c) (111). NRM (c) needs 10 resource units in domain D for its domain, and therefore sends a request for 40 resource units to NRM (d). Then, NRM (d) sends a confirmation to NRM (c) that 40 resources have been pre-allocated to domain D (114), and NRM (c) sends one more confirmation to NRM (a) (110). ) Also transmits one more acknowledgment to NRM (b) (112). Packets using reservations are marked by the application or edge router and identified and / or remarked by the policy point. This confirms that packets with only assigned QoS-classes utilize pre-allocated paths.

In the panel concept, it is assumed that the destination domain is well prepared and other mechanisms are used to ensure QoS in the destination domain. In large networks, it is not desirable to use all of the aforementioned panel concepts as endpoints because they are not sufficiently hierarchical. Instead, the panel is used to reach the appropriately sized destination domain (eg, subnet). No resources are reserved for the final part of the path in the called party's domain. Therefore, the panel concept cannot provide QoS on its own end-to-end, ie from source endpoint to destination endpoint, if the destination domain is insufficiently reserved. However, there is a called party that is not linked to a well-prepared called party domain. One example of such a domain is a radio access network, where the last hop, i.e., the hop between the base station and the terminal is the bottleneck link. Another problem occurs when the host is a mobile terminal. The QoS mechanism must allow for quick local recomputation when handing over between base stations in a radio access network.

The present invention relates to a method and arrangement in an IP network. More specifically, the present invention relates to reserve resources for obtaining a predetermined end-to-end quality of service (QoS) for an IP package of a particular stream.

1 shows an example of the prior art in which resource requests are aggregated toward the called party domain using a funnel concept.

2 illustrates a network comprising two domains in accordance with the present invention.

3 shows an example of an inter-domain resource reservation in accordance with the present invention.

4 is a flowchart illustrating resource reservation according to the present invention.

The goal is to provide a hierarchical solution for preallocating resources to obtain predictable end-to-end QoS in heterogeneous IP networks.

This problem is solved by an IP network having the features of claim 1 and also by a network resource manager (NRM) having the features of claim 16. This problem is also solved by a method having the features of claim 27 and by a computer program product having the features of claims 42 and 44.

The method implemented in the IP network provided by the present invention,

The second NRM informing the first NRM of the called party's domain attribute label; And

-Respectively carrying out an appropriate operation for transmitting an IP packet with a predetermined QoS between the source terminal and the called terminal according to the domain attribute label in which the first NRM and the second NRM are known;

This makes it possible to allocate resources in order to obtain predictable end-to-end QoS in heterogeneous IP networks.

The second NRM includes means for informing the first NRM of the domain attribute label of the called party domain,

Wherein the IP network comprises means for performing appropriate operations for transmitting IP packets with a predetermined QoS between the source terminal and the called terminal according to the domain attribute label in which the first NRM and the second NRM are known,

In heterogeneous IP networks, it is possible to reserve resources to obtain predictable end-to-end QoS.

An advantage of the present invention is that the NRM path vector acts as a tool to identify the NRM of the requested destination domain and the NRM along the path.

Another advantage of the present invention is that the NRM path vector provides a tool for detecting rejections and errors along the path towards the endpoint.

As another advantage, the present invention provides a tool for distinguishing destination domains having different characteristics.

As another advantage, the present invention may utilize the scalable properties of a funnel model in a network having a provisioned called party domain.

2 shows an IP network 200 according to a first embodiment of the present invention. This network 200 comprises a first domain E and a second domain F. The domain is the logical part of the IP network and is divided for management reasons. A domain refers to a routing domain in the present invention.

The domain E includes a router 201, a network resource manager (NRM) (e), a subnetwork 208 having a server 210 and a terminal 207. In the embodiment shown in FIG. 2, domain E may be a source domain. Alternatively, the source domain may be a third domain that transmits a packet through domain F to reach the called party domain F. The domain in which the transmitting terminal is located is referred to as a source domain.

The called party domain F includes an endpoint within one of the server 211, the router 202, the NRM (f), the subnetwork 203, and the subnetwork 203. The domain where the endpoint is located is referred to as the called party domain.

Each subnetwork 203, 208 further includes one or more terminals 204, 207. Each terminal 204 is assigned a dynamic or static IP address by the subnetwork 203,208. The terminal 204 to which the packet is to be sent is referred to as an endpoint. Sub-networks 203 and 208 are illustratively LANs having one or more gateways, one or more servers and one or more terminals, or one or more Radio Network Controllers (RNCs), one or more base stations (BSs) and It may be a wireless network including one or more wireless terminals. Terminals 204 and 207 are preferably PCs or IP phones in a wired network, or may be wireless phones or laptops in a wireless network.

Routers 201 and 202 interconnect each other with different networks 203 and 208, eg different LANs with terminals. The NRM (e, f) may comprise, for example, a computer program for resource preallocation and may be executed, for example, on the servers 210, 211 or alternatively within the routers 201, 202. A server is actually a device that stores and calculates data while the router is primarily routing IP packets.

NRM uses “fun backgrounds”, such as admission control, inter-domain communications 205 and 210, and aggregation of resource requests, for example by using both funnel concepts in the NRM of the called party's domain. It has the same characteristics as described in ”. The NRM is further responsible for aggregation of called party address prefixes by indicating the appropriate called party address prefix, and labels this called party with a domain property label in accordance with the present invention. By categorizing each domain with domain attribute labels, it is possible to separate between domains with different characteristics such as availability of resources, for example bandwidth. The domain attribute label contains information about the method used in this domain to obtain QoS to endpoints in the domain. The panel concept is useful for pre-allocating all resources in a scalable manner to the NRMs in the destination domain, but the path from the NRMs to endpoints in the destination domain still remains. Therefore, this attribute, namely the domain attribute label of the called party domain, is of particular interest.

The NRM (f) in the receiving domain (F) that receives the resource request sends an acknowledgment message (if the request is approved) to the NRM (e) and in some cases other units involved in the request. The acknowledgment message indicates that a certain amount of resources are pre-allocated so that the requested QoS can be met at the called party NRM (F). The domain attribute label may be added to the confirmation message or transmitted in a separate message. By reading the domain attribute label, the NRM and, in some cases, other units involved in the request, know whether or not there is a need to reserve resources. If the destination domain is insufficiently prepared and resources need to be pre-allocated, the domain attribute label tells how resource reservations should be handled.

Domain Property Label

Domain attribute labels are defined in the domain attribute label field. The label field may include, for example, 16 bits and may be part of data transmitted between NRMs. The label field makes it possible to define a large number of domain attribute labels. The NRM communicates with an application protocol for the Transmission Control Protocol (TCP), which defines a domain attribute label field. The information is routed in a general manner and there may be resources pre-allocated for transmission of domain attribute labels. Attribute labels have four types of definitions:

The domain attribute label “ Privisioned” provides information that the domain is well-prepared for the resource and that no resource reservation is required to provide QoS to endpoints in the domain. For example, a well-prepared LAN (Local Area Network) No action is required by the requesting unit, e.g., terminal 207 or NRM (e).

The domain attribute label "Catered" provides information that the domain deals with locally set up QoS via NRM, for example called by an endpoint, proxy or Radio Network Controller (RNC). When an endpoint is located in a radio access network, if the resource is handled by a Radio Network Controller (RNC) in cooperation with the local IP resource manager, the RNC negotiates with the local NRM for the resource. End-point) and knows when the terminal requests a service that requires end-to-end QoS.

The domain attribute label “Requested” may be called by a requesting unit such as, for example, the sending terminal 207, NRM (e) or proxy, in order for the domain to extend QoS from NRM to a specific endpoint. It provides information that QoS is handled through NRM The address of the NRM is known through the NRM path vector, which is further described in “NRM path vector”.

The domain attribute label "Signalled" provides information that QoS in the domain is handled by signaling. The sending entity is responsible for "resource" in the data to cause receiving terminal 204 to request QoS in the called party's domain. Resource ReSerVation Protocol (RSVP) route "message is transmitted, and the receiving object transmits an" RSVP resv "message. This implies that the sending object (also the receiving object) should be located along the path of traffic as the terminal 207. However, the NRM (e) cannot send this RSVP message and must inform the router 201 that the message should be sent to the called party through the proxy.

The four domain attribute labels described above are given in order to be able to distinguish recipient domains having different characteristics. Nevertheless, other domain attribute labels can be defined and used in connection with the methods described below.

NRM path vector

A Network Resource Manager (NRM) path vector is introduced in the present invention to identify the network resource manager along the path to the called party. For each funnel for a given called party, the NRM path vector tells you the sequence of NRMs that granted the resource. The NRM path vector is a tool for identifying the NRM of the requested called party domain. Rejections and errors can also be detected by the NRM path vector, for example if the request is denied, the path vector shows where the rejection occurred and if the NRM is inaccessible, the path vector shows where the problem is. NRM path vector is used for the requested label. However, the NRM path vector can be used for labels that are in signaling state, ready state, and ready state to identify the NRM.

The IP network 300 according to the second embodiment of the present invention is shown in FIG. IP network 300 includes five routing domains (G, H, I, J, K), one domain (G) is the source domain and one domain (I) is the called party domain. The source domain G includes an end point constituting the NRM (g) and the terminal 301, and the called side domain I includes an NRM (i), a called side unit 311 and an end point 302. Moreover, intermediate domain H comprises NRM (h), endpoint 312 and device 313, domain J comprises NRM (j) and domain K comprises NRM (k). do. Each NRM can communicate with other NRMs and also endpoints in other domains.

Referring to FIG. 3, the terminal 301 wants to transmit an IP packet requesting a predetermined QoS to the terminal 302. According to the network topology of this embodiment, the IP packet requests to pass through domain H to reach domain I. In order to satisfy the requested QoS, the resource, which is ten units in this embodiment, is pre-allocated from the terminal 310 to the endpoint (terminal 302) in the called party domain (I). (More resources than may be required to meet the requested QoS may also be requested.) The amount of resources may be measured by the amount and / or bandwidth required in delay and / or jitter. The following steps are performed.

The terminal 301 first requests ten units from the NRM (g),

NRM (g) requests ten units from NRM (h) to endpoint 302 (303). This second request is aggregated with other requests from other domains, for example domain J sends a request 307 for five units to an endpoint located at domain K, where the request is also a domain. It has data to transmit that must pass through (H) and its called party is in domain I or beyond, for example domain K. Each NRM has only one or several reservations per destination domain. For example, QoS is divided into different classes by delay, bitrate, and the like. Thus, one reservation per destination domain and also per QoS-class is possible.

Subsequently, NRM (g) requests a resource from the adjacent NRM (h) (303), and two different methods for resource pre-allocation to NRM (i) can be used. "Alt 1" is used when NRM (h) pre-reserves resources in domain (I), and "Alt 2" is used when NRM (h) is pre-prepared for domain (I). Used when no resources are available.

Alt 1 : In most operations, resources can be pre-reserved according to the heuristics of demand over time and target over-allocation, such as time of day and day of week. May be accepted during the path from the first NRM to the destination domain. The request will therefore be acknowledged by the first NRM (h) for resources in the path to the NRM (i) of the called party's domain.

The acknowledgment message is transmitted by NRM (h) to NRM (g) 304 and NRM (h) to NRM (j) 308 by an adjacent NRM after each resource-negotiation.

NRM (h) and NRM (i) add their identifier, such as their IP address, to the NRM path vector to update the vector.

NRM path vector is included in the acknowledgment message.

NRM (h) informs NRM (g) of the domain attribute label of domain (I). NRM (h) receives a domain attribute label of domain I when pre-reservation of resources for domain I is performed.

Terminal 301 receives a known domain attribute label from NRM (g) and obtains confirmation that the resource has been pre-allocated to domain (I).

Alt 2 : In some cases, when no pre-reservation was performed, request 303 would be a series of requests between adjacent NRMs to set up resources. The confirmation is then propagated back towards the origin. Acknowledgment means that the resource is available to the destination domain as indicated in the acknowledgment message.

NRM (h) requests five plus ten units from NRM (i) (305).

The NRM (i) receives the request and notifies the called party that it is located in its domain.

A confirmation message is sent 306 from NRM (i) to NRM (h), where NRM (i) responds that 15 units have been pre-allocated to NRM (h).

NRM (i) is added to the NRM path vector, and the vector is transmitted in acknowledgment message 306.

NRM (i) informs NRM (h) of the domain attribute label of domain (I).

A confirmation message is sent by NRM (h) to NRM (g), where NRM (h) responds that 10 units have been pre-allocated to NRM (g) d.

The NRM (h) adds its identifier to the NRM path vector, and the NRM path vector now contains the identifiers of NRM (i) and NRM (h). The vector is included in the confirmation message 304.

NRM (h) informs NRM (g) of the domain attribute label .

The terminal 301 receives the known domain attribute label from the NRM (g) and obtains confirmation that the resource has been pre-assigned to the domain (I).

When Alt 1 or Alt 2 is performed, the appropriate action is performed according to the known domain attribute label.

If the domain attribute label is provisioned:

No resources are preallocated in the called party's domain.

IP packets are routed according to conventional routing protocols to endpoints 302 on unallocated paths.

If the domain attribute label is in the prepared state:

The called party domain (I) deals with locally set up QoS via NRM (i).

The terminating unit 311, which may be an endpoint, proxy, or preferably Radio Network Controller (RNC) in the wireless network, calls NRM (i) in the terminating domain. (RNC controls radio resources of the called terminal.)

The called party unit 311 negotiates with the called party NRM for the requested resource. Each destination side unit 311 must recognize the nearest NRM. This can be done while configuring each destination side unit 311.

If the domain attribute label is requested:

The requesting unit, which may be the endpoint 301, proxy or NRM (g), which is the origin of the IP packet, calls NRM (i). QoS is then handled through NRM (i) to further extend QoS to specific endpoint 302. The address of NRM (i) is known by the requesting unit via the NRM path vector.

If the domain attribute label is signaled:

Sender 301 sends an “RSVP path” message to allow receiver 302 to request QoS from endpoint 302.

The receiver then sends an "RSVP resv" message to allocate resources to the destination domain (I).

4 is a flowchart showing the method according to the invention in a general mode. This method is performed by an IP network and is for transmitting an IP network from a source terminal located in a source domain to a called terminal located in a called domain, wherein the source domain and the called domain each include an NRM. The method includes the following steps.

401. A first NRM (e) located in the source domain E requests a resource from a second NRM (f) located in the called party domain F.

402. The NRM (f) adds its identifier to the NRM path vector to update the vector.

403. The NRM (f) informs the first NRM (e) of the domain attribute label of the called party domain (F).

404. The NRM (e) and NRM (f) perform appropriate operations for transmitting IP packets with some end-to-end QoS.

The method is implemented by a computer program product comprising software code means for performing the steps of the method. The computer program product runs on the processing means of a server or router. The computer program may be loaded directly or by a computer usable medium such as floppy disk, CD, Internet or the like.

The present invention is not limited to the above-described preferred embodiment. Various alternatives, modifications, and equivalents may be used. Therefore, the above-described embodiments should not be treated as limiting the scope of the invention as defined by the appended claims.

Claims (45)

It includes a plurality of domains including a source domain (E) and a called party domain (F), wherein the source domain (E) is a source terminal 207 and a first network resource manager (NRM) (e) Wherein the called party domain (F) comprises a called party terminal 204 and a second NRM (f); The source terminal 207 in the source domain E includes means for transmitting an IP packet requesting a predetermined QoS to the called terminal 204; The first NRM (e) in the source domain E includes means for requesting a resource from the second NRM (f) such that the transmission of the IP packet is sufficient to meet the QoS. To The first NRM (f) comprises means for informing the first NRM (e) of a domain property label of the called party domain F, The first NRM (e) and the second NRM (f) are each configured to transmit the IP packet having the QoS between the source terminal 207 and the called terminal 204, respectively, to the known domain attribute level. IP network (200) comprising means for performing appropriate operations in accordance with the invention. The method of claim 1, The IP packet from the source domain (G) is transmitted to the called party domain (I) via one intermediate domain (H), The intermediate domain (H) comprises at least one NRM (h) suitable for inter-domain communication with an NRM (g; i) in an adjacent domain (G; I). The method according to claim 1 or 2, Means for using an NRM path vector to identify the NRM (f) in the called party domain (F). The method according to any one of claims 1 to 3, Means for using an NRM path vector to identify the NRM (e, f) along the path from the source terminal (207) to the called terminal (204). The method according to any one of claims 1 to 4, Means for using an NRM path vector to detect rejection and / or errors along the path from the source terminal (207) to the called terminal (204). The method according to any one of claims 1 to 5, The second NRM (f) comprises means for adding its identifier to the NRM path vector. The method according to any one of claims 1 to 6, Wherein said first NRM (f) comprises means for transmitting a message 210 to said adjacent NRM (e), said message 210 comprising said NRM path vector. . The method according to any one of claims 1 to 7, NRM (h) comprises means for aggregating the resource request (303) with other requests of other domains (J). The method according to any one of claims 1 to 8, The known attribute label is provisioned, and the means for performing the appropriate operation comprises means for sending an IP packet to an unallocated resource. The method according to any one of claims 1 to 8, The known attribute label is cartered and the means for performing the proper operation means for invoking an NRM (f) in the called domain to guarantee QoS to the called terminal in the called domain. IP network (200), characterized in that it comprises a device further comprising. The method of claim 10, The device is a Radio Network Controller (RNC), and the called terminal 204 is a mobile terminal. The method according to any one of claims 1 to 8, The known attribute label is requested and the means for performing the appropriate operation comprises a requesting device further comprising means for invoking another NRM, wherein the other NRM is a feature called terminal 204 from the local NRM. And means for extending the resource reservation. The method of claim 12, The requesting device is the original requesting device (207) or NRM (e). The method according to claim 12 or 13, The IP network (200) comprises means for using the NRM path to identify the address of the other NRM. The method according to any one of claims 1 to 8, The known attribute label is Signaled and the means for performing the appropriate operation further comprises means for transmitting a "Resource ReSerVation Protocol (RSVP) path" to the called terminal 204. An IP comprising a source terminal 207, and the called terminal 204 further comprising means for transmitting an “RSVP resv” message to pre-allocate resources within the called domain (F). Network 200. A Network Resource Manager (NRM) unit (h) in a domain (H) in the IP network (300) according to any one of claims 1 to 15, located in the second domain (I) and In the NRM unit (h) comprising means for receiving a resource request 306 from 2 NRM (i), the NRM (h), Means for informing the second NRM (i) of a domain property label of the domain (H); And Means for performing appropriate operation between the first endpoint 312 and the second endpoint 302 in accordance with the known domain attribute label to provide end-to-end QoS. NRM unit (h) characterized in that. The method of claim 16, And further comprising means for using the NRM path vector to identify a third NRM (g) in the third domain (G). The method according to claim 16 or 17, NRM unit (h) further comprising means for using the NRM path vector to detect rejection and / or errors along the path between the first endpoint and the third endpoint. The method according to any one of claims 16 to 18, NRM unit (h) further comprising means for adding its identifier to the NRM path vector. The method according to any one of claims 16 to 19, Means for sending a message (305) to said second NRM (m), said message (305) comprising said NRM path vector. The method according to any one of claims 16 to 20, NRM unit (h) further comprising means for aggregating the resource request (306) with another request from another domain (J). The method according to any one of claims 16 to 21, The known attribute label is in a ready state, and the means for performing the appropriate operation includes means for receiving a call from the device 313 in the domain H, and the call is made to the called terminal 312. NRM unit (h), characterized in that it further comprises a request for the NRM (h) to ensure QoS. The method of claim 22, The apparatus (313) is a Radio Network Controller (RNC), and the called terminal 312 is a mobile terminal, characterized in that the NRM unit (h). The method according to any one of claims 16 to 21, The known attribute label is requested, and the means for performing the appropriate operation comprises receiving means from the requesting device and extending resource reservation from the NRM unit h to a particular endpoint 312. NRM unit (h) characterized in that. The method of claim 24, NRM unit (h), characterized in that the requesting means is an endpoint (302) in domain (I) or NRM (i). The method of claim 24 or 25, NRM unit (h) further comprising means having its own address identified by the NRM path vector. 27. An IP network according to any one of claims 1 to 26, in order to obtain a predetermined QoS between the source terminal 207 in the source domain E and the called terminal 204 in the called domain (F). Method of pre-allocating resources within The resource reserve allocation method includes a first network resource manager (NRM) (e) located in the source domain (E) satisfying the QoS from a second NRM (f) located in the called party domain (F). Requesting a resource, for the purpose of sending an IP packet, which is necessary to make it possible, The resource reserve allocation method, Notifying, by the second NRM (f), a domain property label of the called party to the first NRM (e); And The first NRM (e) and the second NRM (f) for transmitting the IP packet having the QoS between the source terminal 207 and the called terminal 204 according to the known domain attribute label. And further performing each of the appropriate operations. The method of claim 27, The IP packet from the source domain (G) is transmitted to the called party domain (I) via a third domain (H), And said third domain (H) comprises at least one NRM (h) suitable for inter-domain communication with NRM (g; i) located in an adjacent domain (G; I). The method of claim 27 or 28, And using an NRM path vector to identify the NRM (f) in the called party domain (F). The method according to any one of claims 27 to 29, And using an NRM path vector to identify the NRM (e, f) along the path from the source terminal (207) to the called terminal (204). The method according to any one of claims 27 to 30, And using an NRM path vector to detect rejection and / or errors along the path from the source terminal (207) to the called terminal (204). The method according to any one of claims 27 to 31, And adding the identifier of the NRM (f) to an NRM path vector. The method of claim 32, Sending a message (210) to the adjacent NRM (e); And And including the NRM path vector in the message (210). The method according to any one of claims 27 to 33, And aggregating the resource request (303) with another request from another domain (J) in the IP network (300). The method according to any one of claims 27 to 34, wherein The known attribute label is provisioned. The appropriate operation includes transmitting an IP packet to an unallocated resource. The method according to any one of claims 27 to 34, wherein The known attribute label is in a ready state, The appropriate operation comprises calling an NRM (f) in the called domain (F) to guarantee QoS to the called terminal (204). The method of claim 36, A radio network controller (RNC) calls the NRM (f) in the called domain (F) to guarantee QoS to the called terminal, and the called terminal 204 is a radio terminal. Characteristic resource allocation method. The method according to any one of claims 27 to 34, wherein The known attribute label is requested. The appropriate action includes calling another NRM that extends a resource reservation from the other NRM to a particular called party terminal (204). The method of claim 38, And the original request device or NRM (e) calls the other NRM. The method of claim 38 or 39, And a dungeon using the NRM path vector to identify the address of the other NRM. The method according to any one of claims 27 to 34, wherein The known attribute label is signaling state, The proper operation is Transmitting, by the source terminal 207, a "Resource ReSerVation Protocol (RSVP) path" message to the called terminal 204 in the called domain (F); and And sending, by the called terminal (204), an “RSVP resv” message to pre-allocate the resource in the called domain (F). A computer program product directly loadable into a processing means in a router and / or server in an IP network, the computer program product comprising software code means for performing the steps of claims 27-41. . The method of claim 42, Wherein said processing means is also located in a wireless network controller, proxy or terminal, directly loadable into processing means in a router and / or server in an IP network. A computer program product stored on a computer usable medium, 42. A computer program product comprising a readable program for causing a processing means in an IP network to control the execution of the steps of any of claims 27-41. A computer program product stored on a computer usable medium, A readable program for causing the processing means in the server and / or the router present in the IP network to carry out according to claim 44, Said processing means is also located in a wireless network controller, proxy or terminal.