Patent Application
20070115826
This invention relates to multiple access data networks and more particularly to systems and methods for avoiding collisions while increasing throughput.
Ethernet is a common multiple access control (MAC) protocol used to handle data flow in local area networks. Because data from multiple points flows over a common transmission media, collisions in the form of multiple simultaneous contentions for the same medium among data from different points can occur. To handle such contentions, networks are sized on a probability basis to minimize contentions and tokens are used to mediate among multiple simultaneous accesses. Sizing of the network to reduce contention issues results in lowered capacity as measured by throughput.
One known possibility to reduce contentions is to join every station together via switches, routers or bridges. Such a solution is costly. Another costly solution is to increase the number of switches in the local network, thereby isolating the stations while reducing contentions. This approach may not allow the data to flow faster, but it does have the advantage of collision reduction and thus reduces the number of retries. One advantage of using more switches is that some local traffic, such as traffic to a printer, need not travel the entire network but rather can be switched to the printer branch. The cost of each additional switch is high, roughly six or seven times the cost of a hub, and thus not an effective solution.
One example of a multiple access system implemented for satellite communication is the Aloha system between Hawaii and the U.S. mainland. The multiple access protocol is based on the partitioned interval in time. All the stations have equal opportunity of accessing the medium at the beginning of each interval only. If two or more stations attempt to transmit at the same interval a collision would occur and the transmissions would be lost. The “lost” transmissions would be retransmitted a short time later. The Aloha protocol has a throughput expression with the maximum of 36.8%. Thus, the collision rate of 63.2% means that almost ⅔ of all transmissions must be repeated. Since 63% of the retransmission also collide, the actual throughput is very low.
One conclusion that can be drawn from this type of system is that the allowance of collisions, or medium contentions, has a large negative effect on throughput. Another conclusion is that traffic control is important and, in the case of the Aloha system, a level around 10% would be ideal.
Ethernet uses an enhanced multiple access protocol called Carrier Sense Multiple Access with Collision Detection (CSMA/CD). The scheme provides for each station wishing to transmit data to sample the signal level on the shared transmission medium. If the transmission medium is idle for a fixed duration, then the station can transmit data at that time. If a collision occurs between multiple stations (all of whom begin transmission at roughly the same time), all transmitting stations are required to stop transmitting. The data that otherwise would have been transmitted (but for the collision) is retransmitted after a random delay. The delay being random is thus different for every station.
It is difficult to make throughput calculations for an Ethernet protocol since throughput model for the Ethernet protocol has too many variables. In some situations, there are conditions for a high throughput and other conditions lead to very poor throughput. Under high traffic conditions there has been no demonstration of lowered collision rate. Finally, while a station might have an “advertised” throughput, the actual throughput could be much lower depending upon factors outside the control of the station and thus not calculable or manageable by the station. Accordingly, traffic management is difficult.
The concepts discussed herein are directed to providing a multiple-access contention-free environment for a local area network without using centralized control and without using information contained in the data. Systems and methods are disclosed for reducing potential data collisions by buffering data from connected devices when the bus (or other common transportation media) is not available for immediate use. In one embodiment, the buffering is controlled by hubs that can accept information or hold it up for a period of time until the buffer clears. By using the hub approach, data packets can be buffered and then when the bus is available, multiplexed onto the bus. The buffering can occur several times if necessary. In some embodiments, buffer fullness is used as a measure as to which buffer to draw from first. When buffers are full, signals are sent to the stations to reduce their access to the network on a temporary basis. In this manner, collisions are avoided without requiring the network to look into a packet to obtain header information. The concepts discussed herein can be used in either a bus or tree configuration.
The foregoing has outlined rather broadly the features and technical advantages of the present invention in order that the detailed description of the invention that follows may be better understood. Additional features and advantages of the invention will be described hereinafter which form the subject of the claims of the invention. It should be appreciated by those skilled in the art that the conception and specific embodiment disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present invention. It should also be realized by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the invention as set forth in the appended claims. The novel features which are believed to be characteristic of the invention, both as to its organization and method of operation, together with further objects and advantages will be better understood from the following description when considered in connection with the accompanying figures. It is to be expressly understood, however, that each of the figures is provided for the purpose of illustration and description only and is not intended as a definition of the limits of the present invention.
For a more complete understanding of the present invention, reference is now made to the following descriptions taken in conjunction with the accompanying drawing, in which:
Prior to beginning a detailed discussion of some illustrative embodiments of the invention, it might be helpful to review the basic architecture of Ethernet protocol implementations. In terms of Ethernet local area network (LAN) configurations, there are bus and tree configurations. The bus configuration allows direct access by every station. Ethernet CSMA/CD was designed for this purpose, i.e., to a share medium for all stations.
The tree structure is a hierarchical structure where stations are located at the bottom of the hierarchical structure and the gateway to the external network is located at the top of the hierarchical structure. The number of layers in between the top to bottom layers is based on the number of stations and traffic volume. In the tree structure, each station has its own medium (cable, air link) to the next level in the hierarchy. Ideally, the second least level should be a router which would eliminate the shared medium issues. For economical reasons, a HUB is often used to convert the individual mediums into a shared medium by having a common bus for all inputs. The HUB has multiple ports; each port having (as shown in
The network configuration has the choice of duplex (two wires, one bi-directional medium) or full duplex (four wires, two uni-directional mediums) connectivity to stations. The embodiments discussed herein describe four wire operation, but the same principle is applicable to a two-wire medium. As will be seen, some of the advantages of the concepts discussed herein are the elimination of collisions by using buffers and the elimination of the traffic overload by sending a stopping transmission signal to sources which are causing the overload. It is anticipated that these concepts will increase throughput traffic at least by a factor of six because:
a. both media can transmit and receive at the same time versus uni-direction transmission by the CSMA/CD. Hence the throughput capacity is double;
b. Collisions are avoided thereby eliminating the need for retransmission. Hence, the throughput capacity increases by 2.5 times;
c. There is no medium idle time before transmission required. Hence, the throughput capacity would increase by 1.2 times; and
d. The quality of service is improved because retransmission is reduced thereby improving latency.
For ease of discussion herein, the term “Frame Multiplex” will refer to a HUB modified as per the discussion herein.
Advantage is taken of the property of the existing CSMA/CD protocol to stop transmission onto the media when the equipment receives a media busy signal. With backward compliable to Ethernet CSMA/CD, then the traffic from the station can stop by sending a carrier signal to the station. With enhancement to the Ethernet standard, a unique start and stop signal can be used to turn off the traffic flow from the station to the Super HUB. However, even when the station is not sending data to the network, data can still flow to the station. For example, if the station has transmitted too much signal to the super HUB, the super HUB would transmit a stop signal to the station. Upon receiving the stop signal, the station would continue to transmit its data to complete the packet and stop any new packet transmission until a resume signal is received. As discussed, at all times data can be sent to the station without interruption. This scheme provides bi-direction traffic which is twice as efficient as the standard CSMA/CD protocol.
With respect to
Let us assume the system is operated in normal condition. The normal condition means the data buffer is not overflowing. Under this condition, one or more stations are sending packets to FM. FM will store these received packets in TX buffer bank 32 and send buffered data as fast as possible to port A. The reverse path is the same, e.g., the packet received at port A will be stored in RX buffer bank 33 and will send the buffered data in a broadcasting mode to all ports from 1-N. This operation is contention free, i.e., no collision. Before discussing the situation of heavy traffic condition that could overflow the buffer, the three choices of buffer configurations will be discussed, namely totally sharing the buffer by all ports 1-N, 1-N individual port buffers or the combination of shared and individual port buffer.
The shared buffer would use all memories in the buffer to hold data from all ports. When the buffer is near to overflowing, the traffic flow control signal would apply to all ports. The advantage of this partitioning is more storage capacity for uneven traffic rate from ports 1-N, but less control on traffic rate guarantees per port. The individual port buffer partitioning has less storage for heavy traffic ports, but individual port traffic can be managed independent of other ports such that when any buffer is near to overflow, the data will stop, even if other buffers have capacity. The combination of shared and individual buffers is a reasonable combination of storage capacity and individual control of traffic flows.
In a “combination” buffer system, the data from any port will go to the shared buffer until it is full. Any new data would be stored in the individual port buffer which will transfer that data to the shared buffer as soon as space is available. When an individual buffer is near to capacity, new incoming data will be stopped. There are at least two ways to stop the incoming traffic to the port from the connected device (12-1 to 12-N). One way is to transmit a signal from the port to the connected device. This scheme is compatible CSMA/CD protocol. The signal could be the traffic received from port A in buffer 33. If buffer 33 has no real data, then some idle signal can be sent. In other words, buffer 33 would store real data from port A or an idle signal. Another scheme is to send stop and start signals to control the traffic flows from the connected device. The advantage is to make the two uni-direction transmissions into four wires that are independent to each other. This traffic flow control could be applied to port A as well (if the buffer has an idle signal, then the system does not need the off-line signal in
Although the present invention and its advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims. Moreover, the scope of the present application is not intended to be limited to the particular embodiments of the process, machine, manufacture, composition of matter, means, methods and steps described in the specification. As one of ordinary skill in the art will readily appreciate from the disclosure of the present invention, processes, machines, manufacture, compositions of matter, means, methods, or steps, presently existing or later to be developed that perform substantially the same function or achieve substantially the same result as the corresponding embodiments described herein may be utilized according to the present invention. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, compositions of matter, means, methods, or steps.
The present application claims priority to co-pending U.S. Provisional Patent Application Ser. No. 60/726,459, entitled “FRAME MULTIPLEXER FOR LOCAL AREA NETWORK,” filed Oct. 14, 2005, the disclosure of which is hereby incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 60726459 | Oct 2005 | US |
