US 7397763 Admissions control in a connectionless communications network

ABSTRACT – A method of providing call admission control which does not require using MIDCOM protocol methods, Packetcable protocols or COPS-RSVP s approaches is described which is simple to implement, cost-effective and which is able to deal with particular situations such as conference calls. Each link in a communications network over which it is required to perform call admissions control is provided with a middlebox connected at each end of that link such that admissions control can be carried out at one end of the link. Call services are provided by Call Servers, each of which has access to a database containing pre-specified information about all middleboxes in that call server’s realm. The database also has information about maximum bandwidths for the link associated with each middlebox. The call servers are used to keep a running tally of the amount of VoIP call bandwidth associated with each middlebox on the edge of a low-bandwidth link, and to accept or refuse calls on the basis of the bandwidth information on a per-call basis.

FIELD OF THE INVENTION

The present invention relates to a method and apparatus for admissions control in a connectionless communications network.

BACKGROUND TO THE INVENTION

Admissions control is a significant problem in communications networks and especially in connectionless, packet-based, communications networks. For example, consider a particular link in a communications network. If that link becomes congested, the traffic is unable to flow through the link and packets are dropped. This results in deterioration in quality of service for all services provided over that link. In particular situations this has especially severe impact, for example, when the link provides the main access route from a communications network of a particular enterprise or residential customer to a core communications network.

These problems are particularly relevant for voice over internet protocol (VOIP) solutions. If a link is already carrying the maximum number of VOIP calls, or other non-voice traffic, adding additional calls seriously degrades the voice quality of existing calls using that link. The new call added to the link also has poor voice quality. Continuing to add calls to the link degrades the quality of all calls until none of those calls are recognisable.

The term “Voice over Internet Protocol call” is used herein to refer calls involving any suitable type of media over internet protocol. For example, speech calls, fax calls, modem calls or video calls.

FIG. 4 shows a voice over internet protocol (VOIP) communications network in which admissions control is required. A local area network 40 (LAN), or any suitable type as known in the art, is connected via an access link A to a core communications network 42. Any suitable type of access link can be used as is known in the art, for example, Gigabit Ethernet, Digital Subscriber, or leased line. However, link A is unable to support calls into the core from all endpoints in the LAN simultaneously. Those endpoints are said to be “concentrated” behind link A. Concentration can be implemented in network designs for example where many of the calls are anticipated to stay on the LAN behind an access link to a core network and/or where not all of the endpoints will need to make calls into the core network at the same time. In this way a customer’s LAN may be connected to a core network via a single access link that supports both voice and data at the same time. In such situations there is a need to detect when over utilisation of the access link is likely to occur in order that preventative measures can be taken. However there are currently no suitable methods for detecting link over-utilisation and communicating this to a call server or other management node in order that link over-utilisation can be prevented.

Another example is illustrated in FIG. 5. In this case an enterprise network 50 is connected via a fixed wireless link 51 to a core network 52. The fixed wireless link has limited bandwidth and is unable to support calls from all endpoints in the enterprise network at one time. In addition varying amounts of bandwidth are required for calls, depending on the type of call required (e.g. voice calls can use a multitude of different codecs, each of which have their own bandwidth characteristics, fax call, etc.) and this further increases the complexity of the admissions control problem.

One known form of admissions control for the “access” portion of a network is found in the Packetcable (Trade Mark) standards for dynamic quality of service (DQOS). Packet Cable is a set of protocols developed by Cable Television Laboratories, Inc. The protocols are designed to enable quality of service enhanced communications using packetised data transmission technology to a subscriber’s home over the cable network. A network superstructure that overlays the two-way data-ready digital cable television network is used. The Packetcable protocols are thus specifically designed for such cable television networks. Another disadvantage of these protocols from the point of view of admissions control is that all the internet protocol media endpoints (e.g. user terminals and other packet media endpoints) and devices on the edge of low-bandwidth links (i.e. in the Packetcable architecture, the CMTS) are required to fully support the reservation protocol (RSVP). In addition those endpoints and devices are required to support mechanisms to send and receive session identifier information from a call server. Also the call server and the devices at the edge of the low bandwidth link need to support the common open policy service (COPS) protocol. This is problematic because many existing communications networks are formed from equipment made by different manufacturers and where many of the nodes or endpoints do not support RSVP or COPS where needed. In addition, the Packetcable protocols require all layer-3 aware devices in a media path to support RSVP in order that call admission control can be effected. However, this is not the case for many communications networks. For example, FIG. 5 shows a low-bandwidth link 51where nodes at either end of that link are only layer-2 aware devices. Therefore Packetcable protocol type call admission control mechanisms would not be effective. Other disadvantages of the Packetcable approach to call admission control include that no support for layer 2 flows is provided and the fact that all devices in the network which support RSVP are required to have some policy awareness.

The reservation protocol is defined in the Internet engineering task forces’ request for comments (RFC) 2205 whilst COPS is defined in RFC 2748.

The known approach to call admissions control mentioned above which uses the Packetcable standards is now described in more detail. This approach involves the Common Open Policy Service (COPS) with RSVP. An example of a typical architecture for COPS and RSVP admissions control is given in FIG. 6 which shows two access networks connected to a core communications network via Policy Enforcement Points (PEPs). The core network comprises a policy decision point (PDP) and a call server. Using this approach, the originating and terminating parties make an admissions request to the call server using H.248, or any other suitable device control protocol such as media gateway control protocol (MGCP) or NCS where NCS is the Packetcable specific version of MGCP as indicated in FIG. 6. The call server then grants an appropriate service ticket to each of those parties. Next, the originating party or originating PEP spawns a network admission request through the network. A similar request is spawned by the destination party or destination PEP to request a call flow in the opposite direction and so provide a 2-way flow. The PDP receives the admission requests and forwards those to the call server.

The call server verifies the service tickets. The PDP decides whether to accept or refuse the request on the basis of available bandwidth on the low-bandwidth access link and if accepted, opens a reserved path for media for the new call. However, the COPS and RSVP approach is problematic because significant post dial delay occurs as a result of the admission process and also the other problems mentioned above with respect to the Packetcable approach apply; In addition, the means by which the call server and PDP communicate is not yet fully standardized.

More recently the Internet Engineering Task Force (IETF) have set up a working group to consider ways in which middleboxes can be controlled. The term “middlebox” is used herein to refer to an entity in a communications network which is associated with a low-bandwidth link and which is able to allow or disallow individual traffic flows over that link. For example, the middlebox may be a node connected to one end of a low-bandwidth link. Also, the middlebox may be part of a node which is not directly connected to one end of a low-bandwidth link but which is able to allow or disallow individual traffic flows over that link. The IETF working group is referred to as the MIDCOM (middlebox communications) working group. In the future it may be possible to use protocols developed by the MIDCOM working group to control such middleboxes in order that they themselves perform admissions control. However, these MIDCOM protocols are not yet developed and ratified. Indeed, we understand that the MIDCOM working group is currently working on the control of middleboxes for network address translation and firewall purposes, but not for admissions control purposes. It will be some time before this is the case and those protocols are deployed on all the required nodes in existing communications networks. In addition such MIDCOM methods would require a means by which a call server is automatically able to identify which middleboxes are relevant for a particular call. However, “middlebox discovery” mechanisms like this are not currently known.

Thus a means of providing call admission control which does not require using MIDCOM protocol methods, Packetcable protocols or COPS-RSVP approaches is required which is simple to implement, cost-effective and which is able to deal with particular situations such as conference calls, lawful intercept (known in North America as CALEA), and other potential call service situations is required.

The invention seeks to provide an improved method and apparatus for performing admissions control which solves or at least mitigates one or more of the problems mentioned above.

Further benefits and advantages of the invention will become apparent from a consideration of the following detailed description given with reference to the accompanying drawings, which specify and show preferred embodiments of the invention.

SUMMARY OF THE INVENTION

A method of providing call admission control which does not require using MIDCOM protocol methods, Packetcable protocols or COPS-RSVP approaches is described which is simple to implement, cost-effective and which is able to deal with particular situations such as conference calls and/or lawful intercept. Each link in a communications network over which it is required to perform call admissions control is provided with a middlebox connected at each end of that link such that admissions control can be carried out at one end of the link. Call services are provided by Call Servers, each of which has access to a database containing pre-specified information about all middleboxes in that call server’s realm. The information in the database is manually configured for example although this is not essential. The database also has information about which media endpoints are behind what middle box, and maximum bandwidths for the link associated with each middlebox. The call servers are used to keep a running tally of the amount of VoIP call bandwidth associated with each middlebox on the edge of a low-bandwidth link, and to accept or refuse calls on the basis of the bandwidth information on a per-call basis

According to a first aspect of the present invention there is provided a call server for use in a connectionless, packet, communications network in order to provide admissions control, said communications network comprising a plurality of middleboxes, each middlebox being associated with a different link in the communications network and arranged to control packet flow over that link, said call server comprising:

    • an input arranged to receive a call admission request from an originating packet media endpoint, said call admission request comprising information about the originating packet media endpoint, and a destination packet media endpoint;
    • an input for accessing information about all first middleboxes associated with the originating node and all second middleboxes associated with the destination node, together with information about the amount of available bandwidth on the link associated with each of those middleboxes;
    • a processor for determining whether to accept the call admission request on the basis of the accessed information about available bandwidth;
    • an output arranged to output the results of the determination as to whether to accept the call admission request.

This provides the advantage that the call server effectively provides admission control capability on behalf of the middleboxes. Because the call server is able to access information about middleboxes and the available bandwidth on the low-bandwidth links associated with those middleboxes it is able to perform call admission control. This is achieved without the need to modify existing packet media endpoints such as media gateways and internet protocol endpoints. In addition, it is not necessary for those packet media endpoints to be fully RSVP enabled or for the middleboxes to be MIDCOM enabled with respect to call admissions control. Another advantage is that call admission control over the low-bandwidth links is achieved even where nodes at either or both ends of the low-bandwidth link are layer-2 but not layer-3 aware. This is because effectively the call server performs the admission control determinations.

The information about the destination packet media endpoint is preferably provided by a destination or called party number representing the packet media endpoint.

Preferably the processor is arranged to determine whether to accept the call admission request on the basis of the accessed information about available bandwidth together with information about the bandwidth requirements for the call. For example, a pre-specified value of the bandwidth requirements for any call can be used. Alternatively the call server can determine what bandwidth is needed as explained in more detail below.

Preferably the processor is further arranged to determine whether all the first middleboxes are the same as all the second middleboxes and to accept the call admission request in such cases. This provides the advantage that when the call path does not traverse a low-bandwidth link associated with a middlebox then the call is simply accepted.

Advantageously said processor is further arranged to identify which of the first middleboxes are not also second middleboxes and vice versa, and wherein said processor is arranged to determine whether to accept the call admission request on the basis of accessed information about available bandwidth only for links associated with those identified middleboxes. This ensures that when the call path or flow does traverse one or more low-bandwidth links associated with middleboxes, then call admission control is performed for each of those links.

According to another aspect of the present invention there is provided a method of performing admissions control in a connectionless, packet, communications network, said communications network comprising a plurality of middleboxes, each middlebox being associated with a different link in the communications network and arranged to control packet flow over that link, said method comprising the steps of, at a call server:

    • receiving a call admission request from an originating packet media endpoint, said call admission request comprising information about the originating packet media endpoint and a destination packet media endpoint;
    • accessing information about all first middleboxes associated with the originating packet media endpoint and all second middleboxes associated with the destination packet media endpoint, together with information about the amount of available bandwidth on the link associated with each of those middleboxes;
    • determining whether to accept the call admission request on the basis of the accessed information about available bandwidth; and
    • outputting the results of the determination as to whether to accept the call admission request.

The communications network is preferably an internet protocol communications network at layer 3. At layer 2, a variety of protocols could be used, such as ATM, Ethernet, PPP, etc. A variety of layer 1 physical layers can also be provided. The calls are preferably voice over internet protocol calls.

The middle boxes at both ends of the low-bandwidth link are arranged to use quality of service (QOS) mechanisms as known in the art to prioritise VoIP traffic from other non-VoIP traffic accessing the low-bandwidth link. This is accomplished using well known classification techniques such as packet marking, port based, VLAN based, etc., as well as traffic shaping, and traffic dropping as known in the art. The end-result is that the voice traffic always has priority over non-voice traffic. If there is ever voice traffic to be sent, it is always sent over the low-bandwidth link ahead of non-voice traffic. This results, essentially, in the voice traffic having complete access to all the available bandwidth on the access link. The term “voice traffic” is used here and in the document as a whole to refer to any suitable type of media in a voice over IP call, for example, speech, fax, video and modem.

Preferably said information about bandwidth requirements for the call comprises session description protocol (SDP) information received from both the originating and destination packet media endpoints. SDP is specified in the IETF’s RFC number 2327. This provides a simple and effective means by which bandwidth requirement information can be obtained by the call server on a per-call basis. The call server is also able to adjust the bandwidth requirements used for the call if these requirements change in the SDP information, for example, as codec requirements are negotiated between the packet media endpoints during call setup time, prior to answer. (A codec is a device for converting speech into signals suitable for transfer by a packet based protocol.) Even post-answer, it is possible to use SDP changes to change the codec being used, and the call server is preferably arranged to take this into account when doing admissions control. Suppose that the call server detects a change to a new codec that requires more bandwidth than previously required for a call. In that case, if there is not enough bandwidth on a link used in this call to change to the new codec, then the Call Server is preferably arranged to force the call to remain at the current, un-modified bandwidth.

In one embodiment said call is a voice call and said information about bandwidth requirements for the call comprises information about one or more codecs to be used in the call. Also, the call may be a conference call to be established using a conferencing service in the communications network. In addition the call may be subject to lawful intercept as explained in more detail below.

In another embodiment the communications network comprises two or more call servers, and wherein said method further comprises: receiving said call admission request at an origination call server associated with the origination packet media endpoint and determining whether a destination call server, associated with the destination packet media endpoint is the same as the origination call server. This provides the advantage that admissions control for calls which traverse realms of more than one call server can be carried out.

Advantageously, when said determination indicates that the destination call server and the origination call server are different, the method comprises allowing the origination and destination packet media endpoints to negotiate as to a codec to be used for the call and to send information about that codec to both the origination and destination call servers. This codec can be used to determine an indication of bandwidth requirements for the call based on the codec information. Preferably, the negotiation about which codec to use is accomplished using known methods such as via session initiation protocol (SIP) or SIP-T. These protocols are able to carry SDP information regarding the call between the origination and destination call servers as described below.

Preferably, for all middleboxes associated with the origination packet media endpoint, the origination call server accesses information about available bandwidth and for all middleboxes associated with the destination packet media endpoint, the destination call server accesses information about available bandwidth.

Furthermore, if the call request is refused, instructions are sent to the origination packet media endpoint to provide a refusal indication to a calling party terminal which initiated the call request.

Also, if the call request is accepted, a database of middlebox information is updated with information about the call and updated again when that call ends.

Advantageously, one or more of said middleboxes are arranged to perform call admission control themselves under MIDCOM protocol control. This means that the method of the present invention can be implemented in communications networks which contain a mixture of MIDCOM enabled and non-MIDCOM enabled equipment. However the present invention does not seek to provide a MIDCOM protocol based solution although some embodiments of the present invention are operable in networks which contain a mixture of MIDCOM enabled and non-MIDCOM enabled equipment.

The invention also encompasses a communications network comprising at least one call server as specified above. Preferably, every originating and every destination packet media endpoint connected to a particular middlebox is controlled by the same call server. By specifying requirements for network topology in this way it is possible to use a simplified method of call admissions control as described herein.

The invention also encompasses a computer program arranged to control a call server such that the method specified above is carried out. Any suitable programming language may be used for the computer program as is known in the art.

The preferred features may be combined as appropriate, as would be apparent to a skilled, person, and may be combined with any of the aspects of the invention.

 

Related Posts