Multiplex communications – Data flow congestion prevention or control – Flow control of data transmission through a network
Reexamination Certificate
1998-01-21
2002-06-11
Marcelo, Melvin (Department: 2663)
Multiplex communications
Data flow congestion prevention or control
Flow control of data transmission through a network
C370S465000
Reexamination Certificate
active
06404738
ABSTRACT:
I. BACKGROUND OF THE INVENTION
IA. Field of the Invention
This invention relates to quality of service (QoS) for network communication. Specifically a new form of QoS, called soft-QoS, is introduced to bridge the gap currently existing between the efficient provision of network-level QoS and the requirements of multimedia applications. This invention is embodied in a network system that uses soft-QoS, in a method for utilizing soft-QoS and a computer program product that enables a network to use soft-QoS.
IB. Related Work
Presently, communication over a network has become widely prevalent. Connections that use such a network require a guarantee of a certain quality of service. Quality of service (QoS) offered by a network connection is measured in terms of QoS parameters. QoS parameters, in conventional systems is expressed in statistical terms such as cell loss rate, delay jitter and cell transfer delay.
In order to provide guarantee of a certain QoS, a network needs to make long-term assumptions on the statistical characteristics of its traffic. But then, network traffic is comprised of various connections with their own respective characteristics. Additionally, assumptions made by a network regarding a specific connection must be valid for its entire duration. Above mentioned, statistical characteristics are obtained in conventional systems using methods including stochastic modeling or statistical analysis of empirical data.
But then, traffic profiles of an interactive multimedia connections is generally unknown when it is setup. Therefore, above mentioned long-term statistical representations are unsuitable for such interactive multimedia connections. Specifically, parameters used in long-term statistical representations do not represent the specific QoS requirements of multimedia applications. It is generally accepted that distributed multimedia applications require a network service that can match the dynamic and heterogeneous bandwidth requirements that are typical of such applications.
FIG. 1
shows an example of how video bit-rate significantly changes over a session. On the x-axis frames corresponding to the video data are represented. Size of the data in terms of number of bits are represented in the y-axis.
FIG. 2
shows frame rate from a source and a compounded multimedia traffic on a network.
Moreover, a network has to rely on conservative traffic assumptions to guarantee a certain QoS. Such reliance on conservative assumptions lead to reservation of network capacity far in excess of what is required. Consequently, there is poor network utilization.
It should be noted that problem of poor network utilization due to capacity over-reservation has been addressed by practitioners in this field using VBR
+
. See D. Reininger, G. Ramamurthy and D. Raychaudhuri, “VBR MPEG Video Coding with Dynamic Bandwidth Renegotiation,” Proceedings of ICC (June 1995). VBR
+
is a network service class, using which a connection can renegotiate its bandwidth while the connection is in progress.
Using this methodology overall utilization of a network is improved since bandwidth can be dynamically allocated and reallocated during a connection. See D. Reininger, W. Luo, “Statistical Multiplexing of VBR
+
video,”
Proceedings of SPIE International Symposium on Voice, Video & Data Communications
, Dallas, Tex. (November 1997).
But then, formulae used in connection bandwidth calculation involve QoS parameters like equivalent bandwidth. Consequently reallocating bandwidth changes the QoS of a connection. In such a case, procedures used for bandwidth reallocation, as in VBR
+
, have to ensure that new QoS parameters satisfy the original QoS requirements of a connection. In order to realize this objective, the authors use a concept of “QoS satisfaction”.
Methods known conventionally are based on representing QoS satisfaction as a “hard” binary (“yes
o”) parameter. If, for example, the delivered (or estimated) cell loss ratio is larger than a required cell loss ratio, the connection is deemed unsatisfied. This result is notwithstanding however small the discrepancy between the required and delivered cell loss ratio is.
Taking the above into consideration, it is thus clear that using binary parameters lead to an overgeneralization of QoS satisfaction. This has been partially rectified in conventional systems by using two sets of QoS parameters at the setup phase, they being desired QoS and acceptable QoS.
A user connection and a network engage in a sequence of QoS negotiations using the above mentioned two parameters. In such a case, the user connection is assigned acceptable QoS parameters which may be different from the desired QoS originally requested by it. Although this methodology presents an improvement over the earlier mentioned binary characterization of QoS satisfaction, providing a desired and acceptable QoS does not lead to an adequate mechanism to handle QoS that fall between the two. It is amply clear that the region between the QoS boundaries, desired and acceptable, is not used effectively.
Also, conventional systems do not have a QoS satisfaction scheme that can be employed while a connection is in progress. Besides, it is not possible for a network to achieve high utilization while maintaining hard-bound long-term QoS guarantees under the requirements imposed by distributed multimedia-applications.
Therefore, at least the following problems exist in conventional network systems.
“hard” QoS is inadequate for multimedia traffic.
“hard” QoS leads to unsatisfactory utilization of networks.
Conventional QoS satisfaction schemes can not be modified while a connection is in process.
Conventional QoS satisfaction schemes do not provide quality-fair resource allocation since they use equal bandwidth allocation.
II. SUMMARY OF THE INVENTION
It is an objective of this invention to solve the above identified problems in network systems. Specifically, it is an objective of this invention to provide a soft-QoS for balancing network utilization and application-level QoS in distributed multimedia systems.
It is another objective of this invention to use during congestion a soft-QoS for utilizing the tolerance of an application to network bandwidth shortage. It is an objective of the present invention to represent soft-QoS by a satisfaction index and a softness profile.
In case of video transmission, satisfaction index is an indication of perceptual quality. For different multimedia applications, softness profile correlates the satisfaction index to the resulting bandwidth allocation during network congestion. For example, video applications exhibit a non-linear quality response to bit-rate changes. Scaling a video source reduces its bit-rate, but impacts the perceptual quality. While “soft” applications (such as teleconferencing or multimedia-on-demand browsing) can tolerate relatively large reductions in bit-rate, “hard” applications (such as video on-demand or medical applications) cannot tolerate bit-rate scaling without significantly degrading the application-level QoS. Softer applications are more robust to network congestion since they exhibit a slower decay in satisfaction index with increasing bandwidth shortage.
It is yet another objective of the present invention to use softness profiles of various applications in a network to balance bandwidth utilization and user satisfactions. It is yet another objective of this invention to enable softer applications to receive a favorably priced service for their added flexibility to the bandwidth allocation allowing networks to operate at more aggressive utilization regimes.
It is another objective of this invention to address the issues relating to QoS and expand the notion of QoS satisfaction beyond its current “hard” binary status.
Specifically satisfaction index is gradually scaled from “unsatisfied” to “satisfied. Satisfaction index represents the discrepancy, between the requested equivalent bandwidth and the allocated bandwidth during a network congestion. The exact relationship between the satisfaction index and
Izmailov Rauf
Reininger Daniel
Marcelo Melvin
NEC USA Inc.
Nguyen Phoungchau Ba
Sughrue & Mion, PLLC
LandOfFree
Dynamic network bandwidth allocation for multimedia... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Dynamic network bandwidth allocation for multimedia..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Dynamic network bandwidth allocation for multimedia... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2961057