Multiplex communications – Data flow congestion prevention or control – Control of data admission to the network
Reexamination Certificate
1998-08-28
2001-02-27
Nguyen, Chau (Department: 2739)
Multiplex communications
Data flow congestion prevention or control
Control of data admission to the network
C370S468000
Reexamination Certificate
active
06195332
ABSTRACT:
TECHNICAL FIELD
The present claimed invention relates to the field of computer network communications. More specifically, the present claimed invention relates to a method and apparatus for transmitting packet-based data via ethernet hardware over a ring topology using rate-based controls.
BACKGROUND ART
Data communication is the process by which two or more data terminals or peripherals send information to each other. In its simplest form, a data communication system, or network, is comprised of a transmitter, a transmission path, and a receiver.
Networks are classified according to different characteristics. One such characteristic is its size. A local area networks (LAN) typically connects devices within a building or a school campus while a wide-area network (WAN) interconnects devices within one city or between different cities.
Network topologies in data communication are greatly diversified. The topology refers to a map, or physical location, of the cables linking the devices in the network. Network topologies typically fall into one of the following types: bus, ring, star or mesh network. The ring topology connects all nodes by a serial connection while a star network connects each device to a central hub. Each of the topologies has advantages and disadvantages depending on its application.
Finally, the method of transferring data within the network is a very important characteristic. The communication options include choosing parallel or serial transmission, synchronous or asynchronous transmission, and single or multiple signal transmission capabilities.
The hardware and software rules and procedures, called protocol, dictate how and when data is transferred. Most networks utilize serial asynchronous transmission protocols. Asynchronous transfer, used in most LANs, uses continuous bursts of fixed-length packets, or cells, to transmit data. But because the transfer is not synchronous, signal information indicating the start and stop point of the signal must be included with the substantive data being sent. As a result, asynchronous transfer loses transmission efficiency by having to send procedural data.
While many systems utilize serial asynchronous transmission, they differ on the protocol used in the system. A protocol is a set of rules that determines how and when the users get to send information across the network. Protocol options include Ethernet carrier sense multiple access/collision detection (CSMA/CD), token-passing, or multiplexing. Token-passing protocol only allows one device, the one with the token, to transmit on the network at any one time. Conversely, the ethernet carrier sense multiple access/collision detection (CSMA/CD) protocol is essentially a random access, first in time rule. The first device to transmit on the network without encountering a collision wins the right to finish its transmission. When a collision does occur, both devices to cease transmission, reset, and wait a random period of time before retransmitting. Lastly, multiplexing spaces out the time or wavelength of the transmission between all the inputs, so they each take turns sending information.
A typical network communication involves computers and peripherals connected in a LAN by a mesh or ring topology. Different LAN systems are then connected in a larger network, like a WAN, using a ring topology. Often, the internetwork link is called a backbone, a very high data-rate link.
While Ethernet protocol and hardware is ubiquitous for LAN systems, it has not been a viable option for WAN operation. The ethernet collision detection protocol limited the overall length of the network. For a data packet to successfully transmit on ethernet protocol, the time for the packet to transmit with no collision detection being returned to the sender is theoretically equivalent to the round trip of the furthest path in the network. For a WAN, this distance can be very high. Using ethernet collision detection protocol on a WAN would increase the frequency of collisions, and increase the response time to a point where the system may be perpetually incapacitated. Hence, a need existed for a different protocol that would successfully transmit an ethernet packet on ethernet hardware over a long distance while assuring equity to transmit on the network. In this way, communications between LANs could be seamless, efficient, reliable, and cost-effective.
As networks become larger and devices transfer more data, a need arises for the links between networks to operate at a very high transmission rate with an associated high-speed protocol. Also, because of the high-cost of the physical layer of a network, a need exists for expansion technology to utilize existing equipment and interfaces while allowing increases in operating performance. As such, it would be prohibitively expensive to tear out existing ring topology WAN networks and replace them with a new type of topology.
In an effort to meet increased bandwidth transmission demands, fiber optic systems have been favored for many backbone connections in a WAN network due to their greater traffic-carrying capacity. In response, the industry developed a standard for fiber optic communication called synchronized Optical Network (SONET) to allow transmission systems to be interoperable. With transmission rates in the order of gigabits per second, the prior art has the capability of meeting the high transmission-rate needs. However, the prior art has many limitations that detract from its usefulness.
SONET requires the use of SONET protocol for transmission over the fiber optic link. Thus, all LAN data must be translated twice. For example, if data is to be sent from LAN ‘X’ to LAN ‘Y’ over a SONET link, then the LAN frame configuration protocol (i.e. ethernet) must be translated into the SONET frame configuration at the data enters the SONET Link. Likewise, the SONET frame configuration must be translated to the LAN frame configuration when data leaves SONET and enters the LAN. This step requires complex and computationally-intensive translations. The translations incur extra costs, overhead, and latencies when ethernet packets running over LANs are converted to frames in a SONET or an equivalent protocol. Consequently, the conversion/reconversion of data frames from one protocol to another is computationally expensive, time-consuming, and inefficient.
Furthermore, the SONET frame is inflexible. It cannot be modified to a specific network link or network protocol. Thus even if the network connection is point-to-point, SONET still requires source and destination addresses in the frame overhead. Hence, the overhead can be redundant and wasteful. Furthermore, the circuit-switch based modulation/demodulation is an unnecessary complication for packet switching. Thus a need exists for a network communication to communicate with LAN's without timely and complex translations and without redundant overhead.
Another limitation of the prior art is its high cost. As its name implies, SONET requires the use of fiber optic cable. SONET does not have the flexibility to operate on different types of a physical layer. Hence, if a user has a metal cable network (i.e., on a WAN) and wants to use the SONET standard, the system must undergo a costly transformation from cable to fiber optics. In light of this limitation, a need exists for a high transmission rate network that has the flexibility to run on either a cable physical layer with its standard equipment or on the more costly fiber optic physical layer.
The prior art protocol uses time division multiplexing to divide the transmission resources between multiple nodes. A guaranteed minimum bandwidth for each ensures access to the network. Similarly, a preassigned maximum bandwidth from each node prevents overloading the network. Unfortunately, this protocol does not take advantage of local traffic conditions between nodes in a WAN. Thus, for example, one node may need to transmit data at a rate equal to the capability of the system to an adjacent node. However, even if that data is the only traffic on that local li
3Com Corporation
Lee Chiho Andrew
Nguyen Chau
Wagner , Murabito & Hao LLP
LandOfFree
Rate-based flow control protocol on an ethernet-over-ring... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Rate-based flow control protocol on an ethernet-over-ring..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rate-based flow control protocol on an ethernet-over-ring... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2587104