Multiplex communications – Data flow congestion prevention or control – Control of data admission to the network
Reexamination Certificate
1999-09-14
2004-10-05
Nguyen, Chau (Department: 2663)
Multiplex communications
Data flow congestion prevention or control
Control of data admission to the network
C370S235000, C370S395400
Reexamination Certificate
active
06801501
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates in general to packet networks, and more particularly to a method and apparatus for performing measurement-based admission control using peak rate envelopes.
2. Description of Related Art
Packet networks are fundamental element of the Internet and other IP-based WANs. The Internet is increasingly becoming a medium for converging traffic as diverse as data, video, audio, and voice over IP (VOIP). Multimedia network applications involve real-time transmission of data, voice, and video over networks. Some existing applications are video conferencing, multimedia virtual presentation, video-on-demand, and network games. These applications present new challenges to the current networking technology.
For example, multimedia applications have much higher data volume than email and file transfer. The current Internet does not support bandwidth reservation, therefore there is no guarantee that such amount of bandwidth is available when needed. Further, most multimedia data, especially video, are compressed and encoded to decrease the redundancy of data. This creates Variable Bit Rate (VBR) data, which presents another challenge to the networks because the bandwidth need of an application varies over time. Accordingly, the amount of the reserved bandwidth either has to be the peak bandwidth, i.e., the peak-rate reservation, or has to be changed over time to reflect the variability of traffic. Reserving for the peak bandwidth results in waste of bandwidth because the reserved peak bandwidth is not utilized all the time. However, changing the bandwidth reservation over time is much more difficult to accomplish. In addition, multimedia applications are delay-sensitive and loss-sensitive by nature. Most multimedia applications require an explicit delay bound on their transmitted packets because of delay-sensitivity. For example, the delay bound of a live video transmission is related to the receiver's latest playback time: if the packets arrive beyond that point, they may be useless for the receiver. The delay bound also varies from one application to another due to the differences in their delay-sensitivities and playback environment setups.
While, ATM was designed with different types of quality of service (QoS) categories in mind, the Internet has historically offered a single level of service, which is often referred to as “best effort.” Best effort service merely ensures that all data packets are treated with equity in the network. However, because of the network dynamics, the Internet does not offer a guaranteed level of service quality, but rather some Internet connections exhibit high levels of congestion that result in poor quality, while other Internet connections exhibit consistent levels of high quality service. Thus, IP, as a connectionless protocol, is not a technology that was initially designed for quality of service. New standards such as RSVP, 802.1q and MPOA are attempting to address this shortcoming, either in various transport environments or native to IP.
To satisfy the QoS requirements of real-time multimedia applications, networks must employ resource reservation and admission control techniques. To develop an admission control method, the necessary prerequisites for being able to promise any QoS in packet networks and what the QoS requirements of different user applications must be defined. Note that the term packet network, as used herein, also covers the term cell network, since a cell is a packet of fixed size.
The QoS in packet networks can be defined with delay, delay variation and packet loss ratio experienced by the packet flow of the user. Any packet network providing explicit QoS promises must somehow take care that the total amount of traffic sharing the same resource, like bandwidth of a link, does not exceed a certain level which would cause buffer overflow or does not require queues that are too long thereby resulting in either violation of packet loss or delay promises. Obviously, conforming to QoS promises is a matter of queue behavior estimation. In order to estimate the behavior of packet queues, the network needs to know something about the statistics of traffic sent through connections. Therefore the user and the network must make some kind of traffic contract, including a description of the traffic source, at least. In the simplest case, the network provides the same QoS for all connections and the traffic contract contains only the source traffic description. In a more sophisticated case, the user specifies the QoS level needed; giving requested upper bounds for the delay, the delay variation and the packet loss. In ATM, the sophisticated contract is being used.
Still, there is a possibility of a violation of QoS promises. The traffic contract itself does not prevent a malicious user from sending more traffic than contracted. If the connection is charged according to the contracted traffic rate, then the violation of the contract is tempting. In the worst case, one malicious user can cause violation of the promised QoS of every connection. To avoid this, the network may perform packet level traffic control.
The main purpose of the connection admission control procedure is to protect the user and the network so that the agreed QoS is achieved and the usage of network is optimized. Thus, the connection admission control is a set of actions taken by the network during the call set-up phase in order to determine whether a connection request can be accepted or rejected. Admission control within the network should effectively control the admitted applications so that the existing applications' QoS are not violated.
To increase bandwidth usage where multiple flows are competing for network resources, statistical multiplexing is often used. Statistical multiplexing is a bandwidth management feature that makes full use of network's capacity by dynamically allocating bandwidth to applications only when they actually have traffic to send. This type of multiplexing of traffic streams differs a lot from time division multiplexing currently used in telephone networks. Due to the asynchronous nature of the network, packets from more than one incoming link may arrive concurrently and in that case, the most important packets must be carried and the others must be queued. Basically, networks consist of a number of switches and each switch of a number of queues, and every queue in the network is dependent of several other queues. The role of admission control in the network is to maximize the performance of a switch according to chosen overall policy.
As suggested above, two QoS parameters are fundamental to admission control: maximum delay and the probability of violating the maximum delay. Clearly packets must not be queued longer than the QoS parameters allow. From the queue service rate and the maximum delay allowed, the maximum queue length allowed may be calculated as follows:
maximum
queue
length
=
maximum
delay
service
rate
of
the
queue
Some buffer space for queue is always needed when the aggregate arrival rate of the connections exceeds the service rate of the queue. The admission control must ensure that the peaks of the aggregate arrival rate results in only a small probability of exceeding the maximum queue length allowed. This probability must be equal to or smaller than the probability of violating the maximum delay. Secondly, the packet loss ration requirements of the connections must be fulfilled.
The admission control ensures that the maximum delay or the packet loss ratio does not exceed given bounds using either preventive admission control or measurement-based admission control. Preventive admission control is a more traditional admission control method that relies on the parameters of the source traffic descriptor and calculates the queue length or cell loss ratio for the case the new connection is accepted.
In contrast, measurement-ba
Knightly Edward
Qiu Jingyu
Hyun Soon-Dong
Nguyen Chau
Nokia Corporation
Squire Sanders & Dempsey L.L.P.
LandOfFree
Method and apparatus for performing measurement-based... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for performing measurement-based..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for performing measurement-based... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3297842