Multiplex communications – Data flow congestion prevention or control – Control of data admission to the network
Reexamination Certificate
2000-06-07
2004-08-17
Cangialosi, Salvatore (Department: 2661)
Multiplex communications
Data flow congestion prevention or control
Control of data admission to the network
C370S238000
Reexamination Certificate
active
06778496
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to network communication systems. It finds particular application in conjunction with controlling admission of Voice-Over IP (VoIP) calls to a packet-based network, and efficiently selecting paths for the admitted calls, so as to balance the packet loads within the network. However, it is to be appreciated that the present invention is also amenable to any type of service requiring connection management, irrespective of traffic burstiness or other factors, and to other like applications such as TCP requiring admission control, video or data transport where a need exists to improve the quality of service (QoS) of data transfer over a network including network scenarios with multiple applications having distinct QoS constraints.
The internet, intranets, and other internet protocol (IP) networks are handling ever increasing volumes of data. Beyond the worldwide web and e-mail capabilities of such networks, applications and protocols are being developed to further add to the volume of traffic. Among these are voice related applications such as telephony or Voice-Over IP, and video applications, such as video telephony, video conferencing, and the like. Unfortunately, even at current usage levels packet loss due to congestion is becoming problematic degrading the performance of data transfer.
Streams of packets typically enter the network from packet switching edge devices or gateways which serve as portals to the interconnected web of routers comprising the network. Typically these gateways are indiscriminant in their treatment of packet streams in that they merely port the packet streams onto the network without regard for congestion levels or likelihood of the packets reaching their final destination. Moreover, the networks typically are unaware of any coherence or association among packet streams, and merely forward individual packets from router to router on a first-come-first-served basis, without regard to their relative priorities. These two limitations severely constrain the ability to provide quality-of-service guarantees for real-time services such as voice in IP-based networks.
Some attempts have been made to address portions of this problem. For example, packet prioritization schemes such as differentiated services or Diffserv distinguish packet streams among several classes. Protocols are also evolving which route higher priority packets more reliably, for example, by allocating certain bandwidth on links between routers for each class. Another partial solution that has been articulated is that of establishing explicit routing paths through the network between frequently traveled points. Multi-protocol label switching (MPLS) is a protocol which enables a label to be assigned to a packet stream which specifies a predetermined path through the network. This allows better monitoring and control of congestion over the paths taken by voice streams, for example. However, the problem introduced by edge-devices not being aware of the congestion levels within the interior of the network still remains. One strategy being pursued to tackle this limitation is to dedicate a certain amount of bandwidth for each MPLS path, on each network link that it traverses. This creates a set of voice trunks among all pairs of nodes, much akin to the telephone trunk routes currently employed between circuit switches, and hence abandon the inherent flexibility afforded by the IP network. In particular, this strategy does not lead to a scalable solution. The number of trunks grows as the square of the number of edge nodes, and the consequent bandwidth fragmentation among hundreds or thousands of MPLS paths can exhaust the link capacities rather quickly. Furthermore, the servicing and provisioning of the multitude of voice trunks across the network are both cumbersome and slow to accommodate new nodes within the network.
The present invention contemplates a new and improved method and apparatus for call management over IP networks which overcomes the above-referenced problems and others associated with the existing approaches.
BRIEF SUMMARY OF THE INVENTION
The above problems are alleviated and an advance is made over the prior art in accordance with the teachings of applicants' invention wherein, a method of regulating admission of packet streams to a network includes at selected times sending path status messages along a set of paths in the network. A cost metric in each path status message is updated at the intermediate nodes as the message progresses along its defined path. Based on the final cost metric values collected upon receipt of the status messages at the respective path edges, subsequent packet stream arrivals are selectively blocked or admitted to the network.
In accordance with another aspect of the present invention, the method further provides choosing an optimal path between a source and a destination gateway.
In accordance with another aspect of the present invention, the paths through the network include a plurality of interconnected links, each potentially operating at different capacities or utilizations. The “choosing an optimal path” step optionally includes determining a most utilized bottleneck link for each of the alternative path choices through the network (for the source and destination gateways in question), and then selecting the path whose bottleneck link is the least utilized among the set of bottleneck links corresponding to the available path choices.
In accordance with another aspect of the invention, the paths through the network include a plurality of interconnected links, each path potentially having different numbers of links. The “choosing an optimal path” step optionally includes selecting a path through the network based on the combination criteria of having smaller bottleneck link utilization and having fewer links compared to other paths.
In accordance with another aspect of the present invention, the path includes a plurality of links interconnecting routers. The “updating a cost metric” step includes measuring link usage at the routers.
In accordance with another aspect of the present invention, each path is comprised of links that are adapted to discriminate between different classes of packet streams. The “updating a cost metric” step includes measuring usage by individual class.
In accordance with another embodiment of the present invention, a system for controlling admission of a packet stream along a path includes a plurality of routers adapted to update a cost metric in a path status message upon receipt of the message at each router. The system further includes a plurality of links interconnecting the routers where selected links and routers together comprise the path through the network.
In accordance with another aspect of the present invention, the plurality of routers are further adapted to generate, and send, the path status message along the determined path at selected times.
In accordance with another aspect of the present invention, the system alternately includes a gateway adapted to generate, and send, the path status message along the predetermined path at selected times.
In accordance with another aspect of the present invention, each of the plurality of routers is further adapted to select an optimum path through the network in response to receipt of a packet stream admission request.
In accordance with another aspect of the present invention, the plurality of routers are further adapted to discriminate between path status messages, and other (payload) packets.
In accordance with another aspect of the present invention, the plurality of links interconnecting the routers are adapted to discriminate between classes of packet streams, e.g., via Weighted Fair Queuing or WFQ.
In accordance with another embodiment of the present invention, a method includes sending a path status message along a defined path and, as the path message progresses through a plurality of routers in the path, updating a cost metric in the path status message at the routers that it transits through
Houck David J.
Meempat Gopalakrishnan
Cangialosi Salvatore
Lucent Technologies - Inc.
LandOfFree
Distributed call admission and load balancing method and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Distributed call admission and load balancing method and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distributed call admission and load balancing method and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3319600