Fast service restoration for lost IGMP leave requests

Abstract

A fast service restoration for a lost IGMP Leave request. The network node does a comparison to determine if Leave messages are lost in transmission from an end-user system to a supporting network node.

Description

BACKGROUND AND BRIEF DESCRIPTION OF THE INVENTION

[0001] As data communication systems continue to evolve, the bandwidth capabilities of such communication systems continue to increase. As such, applications that require large amounts of bandwidth become increasingly feasible. One such application is the distribution of digital media services. Digital media services can provide video information such as television programs or movies, audio programs, and text based information streams.

[0002] Typically, the various types of digital media that may be provided are sourced from one or more information sources. These information sources are intended to provide the multimedia programming to a large number of users that have access to the information via one or more communication networks. Destination routers coupled to the communication network provide interfaces to end users, or subscribers.

[0003] Typically, multiple subscribers are coupled to a single communication link that is coupled to the router. This communication link is utilized for distribution of the program data streams, or channels, to the end users. The Internet Group Management Protocol (IGMP) has been developed by the Internet Engineering Task Force (IETF) as a standard that relates to the communication between the router and the subscriber, which is often referred to as a host. The communication between the router and the set of hosts coupled to a particular communication link is accomplished using point-to-multipoint multicast transmissions. IETF specifications RFC1112 “Host Extensions for IP Multicasting” and RFC2236 “Internet Group Management Protocol, version 2” describe the use of the current IGMP standard in detail. (A table of acronyms used herein can be found at the end of this specification.)

[0004] The IGMP protocol as currently defined by the IETF dictates how the multicast transmissions between the router and host are managed. Each of the hosts coupled to a router determines which of the potential multicast transmission groups, or channels, that it receives. When the communication link between the router and a plurality of hosts is shared by the plurality of hosts, bandwidth limitations on the communication link can require intelligent management of the multicast transmission provided via the communication link. For example, if a number of users reside on a single communication link and the bandwidth limitations only allow for a limited number of multicast transmissions to be supported over the communication link, usage of the particular multicast transmissions by the plurality of hosts must be monitored to ensure that bandwidth is not wasted on multicast transmissions which none of the hosts are actively receiving or using.

[0005] The present invention relates to the problem of efficiently addressing IGMP Group Leave messages that are lost in transmission from an end-user system to its supporting network node (e.g. an ATM DSLAM). For example, traffic overload controls that discard packets could cause such messages to be lost

[0006] In conventional IGMP implementations, a lost Group Leave message would be detected by inspecting IGMP membership reports from end-user systems. These reports are sent in response to a general membership query (GMQ) message sent by the network node, typically every 125 seconds. In the case where other multicast operations (e.g. Group Coin requests) are dependent on the successful execution of the Leave request, it is not acceptable to wait 125 seconds for the loss of the Leave message to be detected before subsequent action can be taken to execute the Leave request. A problem with this solution, i.e. of waiting for and analyzing group membership reports, is that waiting 125 seconds to detect a lost message is too long when execution of the Leave message is gating another operation, such as a Group Join request.

[0007] U.S. application Ser. No. 09/469,034 filed Dec. 21, 1999 entitled “Method and Apparatus for an Improved Internet Group Management Protocol” (owned by the assignee hereof and incorporated herein by reference) addresses, inter alia, the problem of lost Leave messages. The application proposes solving the problem by sending out another Leave message if a host that initiated the lost Leave message does not receive a GSQ (group specific query) message within a predetermined time interval from when the lost Leave message was sent. It is also undesirable to wait for a time interval to determine whether a Leave message was lost, especially when the Leave request is gating another operation such as a Group Join request. Also, since this solution resides at the end-user system, it would be difficult to ensure that it is widely implemented in the network.

[0008] U.S. Ser. No. 10/059,697 filed Jan. 29, 2002 entitled “Facilitating Improved Reliability of Internet Group management Protocol Through the Use of Acknowledge Messages” (owned by the assignee hereof and incorporated herein by reference) also addresses, inter alia, the problem of lost Leave messages. The patent application discloses methods of performing Group Leave and Group Join operations aided by the use of new negative and affirmative acknowledgement messages, By using these messages, the reliability of the IGMP protocol is improved since an end-user system that has issued Leave or Join requests is informed of the result of the requests and can respond appropriately In conventional IGMP implementations, when a Group Leave or Group Join request is not successful, such as would be the case in a lost Leave message, the end-user system that issued the request is not notified. In this case, the end-user systems and the network nodes (DSLAMs) must be provided with functionality to support the new acknowledgement messages. While upgrading network nodes to provide this functionality may be relatively easy to accomplish, it would be much more difficult to ensure that all end-user systems were upgraded to detect and correctly respond to the new acknowledgement messages.

[0009] In view of the problems and disadvantages discussed above, a means at a network node of quickly detecting and responding to the loss of Group Leave messages is needed.

[0010] It is an object of the invention to provide rapid or fast service restoration for lost IGMP Leave requests.

[0011] Another object of the invention is to provide, at the network node, a method and apparatus for quickly detecting and responding to the loss of Group Leave Requests.

[0012] Another object of the invention is to free up the no-longer required bandwidth and accept new Join messages more quickly than prior art solutions and with functionality that is implemented at the DSLAM only, which makes ensuring a network-wide implementation easier.

[0013] Another object of the invention is to improve the robustness and response it controls, such as in the case of IGMP supported services.

[0014] According to the invention, whenever a DSLAM (digital subscriber line access multiplexer) or a network node receives a Group Join message, it compares the MAC (media access control) address of the originating STB (set-top boxes) to that of previous Join messages with their existing connections. If a match is found, then the DSLAM will initiate a GSQ for the group of the matching previous Join request. If there is not enough resources to satisfy the requirements of the Group Join message (or subsequent Group Join messages), then it may either be ignored, or queued for later processing, once the normal GSQ sequence is complete and resources are available. If the multicast group is no longer needed, the connection to the group is then terminated, as per normal GSQ processing. This frees up bandwidth for any new Join messages. The normal GSQ processing which takes normally around two seconds, clears the trouble caused by the loss of Leave Message.

BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and other objects, advantages and features of the invention will become more apparent when considered with the following specification and accompanying drawings wherein:

[0016] FIG. 1 shows a typical network architecture for multicasting video or other high-speed services in an IP (internet protocol) over ATM (IPoATM) environment, and

[0017] FIG. 2 represents an IGMP messaging center scenario incorporating the invention.

DETAILED DESCRIPTION OF THE INVENTION

[0018] FIG. 1 shows a typical network architecture for multicasting video, or other high-speed services, in an internet protocol (IP) over an asynchronous transfer mode (ATM) (IPoATM) environment. Referring to FIG. 1, a video head end provides statically configured channels to a DSL access multiplexer (DSLAM) via an ATM network. The video channels are dynamically cross-connected to customer premise equipment (CPE) units as IP multicast video connections over ATM. Each CPE unit performs IPoATM to IP interfacing to provide IP multicasted video channels to one or more set-top boxes. Each STB sends IGMP messages to the DSLAM to effect certain actions, such as joining or leaving a multicast group. If any of the messages required for such actions are lost, e.g. a Group Leave message, then the action will be delayed until the loss is detected and corrective measures are taken.

[0019] FIG. 2 represents the IGMP messaging scenario between a STB and a DSLAM. Referring to FIG. 2, the STB 1 sends the DSLAM a Group Join message requesting to join the multicast group for channel 1. Subsequent to a connection to channel 1 being established for STS 1 at the DSLAM, STB 1 sends a Group Leave message requesting a leave from the multicast group for channel 1. However, the channel 1 Leave message is lost before it is processed by the DSLAM. The STB 1 next sends a Group Join message to the DSLAM requesting to join the multicast group for channel 2 If the link between the CPE and DSLAM is currently bandwidth limited, the join request for channel 2 cannot be executed until more bandwidth becomes available, which would have been the case had the Leave request for channel 1 not been lost. Even if the link is not bandwidth limited, the loss of the Leave request for channel 1 will result in the bandwidth consumed by this connection being unavailable until the connection is cleared as the result of normal GMQ processing.

[0020] According to the invention, whenever a DSLAM receives a Group Join message, it compares the MAC address of the originating STB to that of previous Join messages for which there are existing connections. If a match is found, then the DSLAM will initiate a GSQ for the group of the matching previous Join request. If there is not enough resources to satisfy the requirements of the Group Join message (or subsequent Group Join messages), then it may either be ignored, or queued for later processing, once the normal GSQ sequence is complete and resources are available. If the multicast group is no longer needed, the connection to the group is then terminated, as per normal GSQ processing. This frees up bandwidth for any new Join messages. The normal GSQ processing, which takes normally around 2 seconds, clears the trouble caused by the lost Leave message.

[0021] Thus, the invention distinguishes from the prior art by the technique in which lost IGMP Group Leave messages are detected, and a corrective action is taken in response to that detection, specifically, the network node receives an IGMP Group Join message from an end-user, compares the MAC address obtained from a Join message of the end-user device to another MAC address obtained from a previous Join message for which there exists IP multicast connections to the network node. Then, the network node initiates a response to a match resulting from the comparison. A GSQ for the group of the matching previous Join request and if there is not a response as per normal GSQ processing, terminating the connection to the group.

ADVANTAGES OF THE INVENTION

[0022] The invention frees up the no-longer required bandwidth and accepts new Join messages more quickly than prior art solutions and with functionality that is implemented at the DSLAM only, which makes ensuring a network-wide implementation easier.

[0023] Response to controls is important to an end-user's satisfaction with any service. Therefore, improving the robustness and response of these controls, such as the present invention does in the case of IGMP supported services, is important to service providers and vendors of equipment that provides the services.

Acronyms Used Herein

[0024] Asynchronous Transfer Mode (ATM)

[0025] Customer Premise Equipment (CPE)

[0026] Digital Subscriber Line Access Multiplexer (DSLAMs)

[0027] General Membership Query (GMQ)

[0028] Group Specific Query (GSQ)

[0029] Internet Group Management Protocol (IGMP)

[0030] Internet Engineering Task Force (IETF)

[0031] Internet Protocol (IP)

[0032] Internet Protocol over ATM environment (IPoATM)

[0033] Media Access Control (MAC)

[0034] Set-top Boxes (STB)

[0035] While the invention has been described in relation to preferred embodiments of the invention, it will be appreciated that other embodiments, adaptations and modifications of the invention will be apparent to those skilled in the art.

Claims

1. A method of managing IP multicast connections in a network having a network node and a plurality of end users, said method comprising the steps of (a) receiving, at a network node an IGMP Group Join message from an end-user device serviced by the network; (b) comparing, at the network node, a MAC address, obtained from the Join message, of the end-user device to another MAC address obtained from a previous Join message for which there exists an IP multicast connection to the network node; (c) initiating, by the network node responsive to a match resulting from the comparison, a GSQ for the group of the matching previous Join request; and (d) terminating, responsive to a result of the GSQ processing indicating that a connection to the group is no longer needed, the connection to the group.

2. The method defined in claim 1 wherein if there is not enough resources to satisfy the requirements of the Group Join message received (or subsequent Group Join messages), then it may either be ignored, or queued for later processing, once the normal GSQ sequence is complete and resources are available.

3. The invention defined in claim 1 wherein multiple Join-Leave pairs are dealt within rapid succession.

4. In a data communications network having one or more network nodes, said one or more network nodes including digital subscriber line access multiplexers, said network implementing the internet group management protocol (IGMP), said one or more network nodes having means for receiving an IGMF Group Join message from an end-user serviced by said network and comparing the MAC address contained in said Join message of an end-user device to another MAC address obtained from a previous Join message, said network node initiating a group specific query (GSQ) with a group of the matching previous Join requests and said network node terminating the connection to the group in response to the GSQ indicating that a connection to the group is no longer needed.

5. The invention defined in claim 4 wherein said network node includes a digital subscriber line access multiplexer.

6. The system defined in claim 4 wherein said end-use device is a set-top box.