Multiplex communications – Diagnostic testing – Determination of communication parameters
Reexamination Certificate
1997-03-24
2001-08-21
Zimmerman, Brian (Department: 2635)
Multiplex communications
Diagnostic testing
Determination of communication parameters
C370S232000, C370S238000, C370S230000
Reexamination Certificate
active
06278693
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to communications systems, and in particular to the quality of service provided by such a communications system.
Communication systems are conventionally divided into two main forms. In the first of these, referred to as a circuit switched system, a dedicated line is provided for a communication between two end terminals (note that it may be possible to multiplex a plurality of such lines onto a given physical communications link). The effect of this is to assign a certain chunk of the network bandwidth to the communication. The most prominent example of a circuit switched system is the conventional telephone network. The drawback with this approach is that it does not maximise bandwidth usage of the network. For example, a telephone call still has bandwidth assigned to it even during periods when the parties at each end are both silent, and so no useful data is being transferred.
The second main type of communications system is known as a packet-switched system. In such a network, information to be transmitted is formed into blocks of data known as packets, which are then presented to the communications system for delivery to their destination. In general no bandwidth is pre-assigned to any particular user, rather the packets compete for available bandwidth. The implication of this is that a network can transmit additional packets whenever network bandwidth is not being fully utilised. The result of this is that bandwidth usage is optimised compared to a circuit switched network, which runs out of capacity when all its bandwidth is assigned (as opposed to being actually utilised).
A packet switched system in some ways is similar to a road network, with the packets being equivalent to the cars on the roads. In both types of network, problems arise when network traffic starts to approach or exceed network capacity, resulting in a backlog of traffic—ie queues. It is therefore important when designing a packet switched system to try to ensure that network capacity will not be exceeded.
Computer communications often have extremely variable bandwidth requirements; eg two computers may exchange a large file, requiring significant network capacity, followed by a period of no data exchange. Such properties normally render circuit switched systems unsuitable for computer communications, since assigned bandwidth would be idle for a relatively high proportion of time. Therefore computer communications have usually been implemented over packet-switched networks.
The designers of such packet switched networks have generally planned the capacity of such networks based on a statistical understanding of the average level of communications. The implication of this is that during peak levels of communications, the network may have insufficient capacity to transmit all packets, resulting in queue formation as described above. Of course, the network could in principle be provided with sufficient capacity to handle the peaks, but then a large chunk of bandwidth would be idle when communications were only at an average level (the overall network efficiency would then be poor, as for circuit-switched networks).
Traditional computer communications have not been too severely impacted by delays in data transmission caused by peaks in network traffic; the precise arrival time of a data file transmitted from one computer to another is not normally too important. However, more recently there has been considerable interest in using computer networks for the transmission of multimedia data, in particular sound and video. A particular example of this is the desire to be able to transmit sound/video information across the Internet/Worldwide Web. Contrary to the traditional forms of computer communications, such multimedia data is very sensitive to network delays, especially the sort of variable delays (often referred to as “jitter”) which are common on a packet-switched network. This is because multimedia systems output data in real-time at a constant rate (they are “isochronous”). Therefore, if a packet containing the next piece of sound or video is delayed too much, then the multimedia system may exhaust its supply of data, producing an interruption in the sound or video. Such an interruption can seriously degrade the quality of the multimedia output; too many interruptions can render the sound/video output totally useless or unintelligible.
There has been considerable work on how to overcome the above problems in order to allow effective multimedia communications over packet switched networks. Some of this development has concentrated on optimising processing at the receiving terminal (see eg U.S. Pat. No. 5,404,446). Another very important strategy has been to introduce some form of bandwidth allocation, conceptually equivalent to simulating a circuit-switched connection on top of a packet-based network (see eg EPA 625838).
It can be seen therefore that the bandwidth available on the network and the jitter suffered by packets traversing the network represent important parameters describing network performance. These parameters are of great significance to a user or application wishing to use the network for data transmission.
There are other parameters relating to the network which are also very important. One of these is latency, which represents the mean end-to-end delay for data transmissions across the network. Latency is likely to be high when a data packet is processed by lots of intermediate nodes in a network on its way to its destination. High latency is particularly deleterious in real-time communications, eg sound or video conferencing, where a long delay can prevent natural interaction. Note that this difficulty also exists in circuit switched networks where the transmission path is excessively long, as is perhaps the case with trans-atlantic telephone calls routed by satellite.
Another parameter reflecting network performance is error rate, which represents the number of errors in the received data stream compared to the original. Potential errors include an incorrectly received bit, perhaps due to noise on the transmission line, or a lost packet, perhaps due to a network failure, or to excessive network congestion resulting in queue overflow and packet loss at an intermediate node in the network. U.S. Pat. No. 4,697,281 teaches reducing packet size when the error rate is high, since if the error rate is too great, the probability of a large packet being correctly received in its entirety becomes very small.
Parameters such as bandwidth, latency, error rate etc may collectively be referred to as quality of service (QoS) parameters. Many networks provide users or applications with a mechanism to indicate their desired quality of service. For example, under the Internet Protocol, datagrams contain three bits, denominated D, T, and R, which if set request low delay, high throughput, and high reliability respectively (see “Internetworking with TCP/IP; Principles, Protocols, and Architecture” by D Comer, Prentice Hall, 1988, section 7.7.4). The Internet does not guarantee to satisfy these requests, but may be influenced by them in its route selection.
Many networks provide users and applications with a rather greater degree of control of quality of service. For example, the bandwidth reservation procedure for the above-mentioned EPA 625838 can be regarded as specification of quality of service. In this case a numerical bandwidth is requested: if the network is prepared to guarantee the availability of the requested bandwidth, then the request is granted; otherwise the request is refused.
A more comprehensive approach to quality of service is described in “Integrated Packet Networks with Quality of Service Constraints” by W Lee and P Kamat, p223-227, IEEE Globecom '91. This paper describes a mechanism whereby users or applications can request quality of service for a particular parameter by specifying two values. The higher of these values represents the desired or requested value, whilst the smaller represents the lowest acceptable va
Aldred Barry Keith
Lambert Howard Shelton
Mitchell Harry David
Renshaw David Seager
Herndon Jerry W.
International Business Machines Corp.
Zimmerman Brian
LandOfFree
Communications systems with quality of service parameters does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Communications systems with quality of service parameters, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Communications systems with quality of service parameters will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2511373