GB2290931A - Computer network operating method - Google Patents

Abstract

A computer network has a number of nodes, each having a computer 40 and an adaptor plugged into the computer's parallel port. A modified ethernet protocol is used to communicate between nodes in the network. The modification involves the ethernet packet being divided into sub-packets of a predetermined size. Data is transferred between the nodes one sub-packet at a time. This has the result that the requirement for memory on the adaptor and also for clock encoding and decoding means on the adaptor is removed. <IMAGE>

Description

A NETWORK DEVICE The present invention relates to a computer network, a method of operating a computer network and an adaptor for use in a computer network. It is envisaged that the invention will have particular application to local area networks (LANs) which provide for communication between personal computers distributed around an office, building or industrial facility.

The benefits of LANs are well-known. They improve communication between computer users and allow the members of a workgroup to work consecutively on a computer file without requiring them to leave their PCs.

Accordingly, there is an increasing demand for the provision of LANs. However, although operating systems and software applications have developed rapidly to provide networking features, there has been comparatively little development in the hardware used to provide a LAN between PCs. Typically, this hardware is expensive and may additionally be difficult to install.

For example, present peer-to-peer networking systems rely on the same basic Ethernet cards that have been on the market for the past ten years, and although the price of these cards has fallen, they are still complex to set up and require a desktop PC user to open their PC. Users of portable computers ( which do not provide for the installation of such cards ) must purchase a parallel port adaptor or PCMCIA card. Both these options are expensive.

In addition to this, implementation of a LAN is further complicated by the wiring of the network. Either the user must wire-up with coaxial cable, which is relatively expensive and bulky, or purchase further hardware in the form of a hub if they wish to take advantage of the telephone type cable of lOBase-T. Clearly, the use of such telephone cable is a cheaper option and leaves open the possibility of utilising telephone lines already installed in a building between voice, fax and LAN traffic.

There are two principal reasons why conventional network adaptors are expensive.

Firstly, the nature of the Ethernet communications between different PCs in a network is that all the information present on the network is passed to the adaptor. The adaptor is then responsible for storing this data, and identifying whether or not it is required by the host computer. This is done by inspecting the start of each data packet, which typically contains either a destination address or a command for all stations to receive and acknowledge it.

Thus, conventional designs of network adaptor cards for use with personal computers rely heavily on the asynchronous interface between the adaptor card and the personal computer. Every implementation has assumed that the PC may not be able to respond to an incoming packet of data from the network, and that the adaptor card must be capable of accepting and interpreting the full data packet before alerting the PC of its presence. To accomplish this, a number of designs have been developed, all of which require memory management control within the hardware adaptor in order to store the data as it is received. This may require memory to be present on the adaptor itself, or may use physical memory within the host computer; however this memory is under the direct control of memory management circuitry upon the adaptor card itself.

Secondly, conventional implementations of the industry standard ETHERNET networking system, as defined by the IEEE802 specification, have found it necessary to precondition the raw digital data that is output on the network cabling. This requirement has arisen for a number of reasons. Ethernet packets can be of variable size, and conventional implementations typically need to encode the raw data such that a clock signal is preserved which can be extracted from the data stream. This ensures that long sequences of logic "O"s or "l"s can be correctly interpreted.It has also been necessary to maintain continuous excitation of the network bus during data transmission for wiring configurations where data transmission has been by current drive into a resistively terminated cable, so that other transceivers on the cable can detect the presence of data and refrain from transmission.

In traditional Ethernet circuits, this pre-coding of the data has always been by the use of Manchester encoders and decoders in the Serial Network Interface section of a network adaptor circuit. These have also needed to recover the clock signal inserted by the transmitter to correctly interpret the data for the receiver.

In addition to the above reasons for the expense of conventional adaptors, the Ethernet standard imposes further requirements on the cabling that is used in the network. The international IEEE802 standard which defines the industry accepted system for networking computers, covers all layers of the network, and defines the manner in which the packets must be assembled for transmission on the network. The two major linchpins of this specification are the variable size of the packets to be transmitted and the need to encode the data packets so that the transmission clock can be extracted from the data packet on the network. Both of these requirements place a heavy burden on the control circuitry and cabling requirements. Hence, it will be seen how these two requirements have prevented the adoption of a simple wiring system and kept network adaptor prices high.

The present invention overcomes some or all of the above-mentioned problems.

According to the present invention there is provided a method of operating a computer network, said network comprising two or more nodes and data carrying means, each node comprising an adaptor and a computer, said method comprising the steps of: assembling the data to be transmitted into an ethernet data packet; dividing said packet so as to create smaller subpackets of a predetermined size; placing a first sub-packet onto the data carrying means; and identifying at other nodes whether the data is intended for that node, and if the data is so intended waiting for the computer of that node to accept that subpacket; and, on said acceptance placing the subsequent sub-packet of said ethernet data packet onto the carrying means; repeating the above step until the entire ethernet data packet has been transmitted.

Preferably, each sub-packet is placed onto the data carrying means consecutively, and the predetermined length of the sub-packet is sufficiently small to obviate the need for incorporating a clock signal into the subpacket, whereby the requirement that the adaptor of the transmitting mode has a data encoding means is removed.

Advantageously, the method further comprises the step of providing an acknowledgement signal on the data carrying means indicating that the sub-packet has been accepted, access to the data carrying means being denied to nodes other than the transmitting and receiving nodes unless said acknowledgement signal has been provided.

Preferably, the method will further comprise the step of reassembling the ethernet data-packet from said subpackets at the receiving node.

Preferably, said reassembling step will involve a mere concatenation of the data portion of the sub-packets.

The invention will now be described further with reference to and as illustrated in the accompanying drawings in which Figure 1 is a block diagram showing the various components of a conventional network adaptor.

Figure 2 is a block diagram showing the various components of a conventional printer sharing module.

Figure 3 is a block diagram of the network adaptor of a specific embodiment of the present invention.

Figure 1 shows a conventional network adaptor which both transmits and receives data-packets conforming to the Ethernet Protocol according to IEEE802. These datapackets are transmitted over transmission lines between nodes in a local area network.

Such a conventional network adaptor may either be provided as a card to be placed in one of the PCbus slots in a desktop computer or alternatively may be provided as a module to be plugged into the parallel port of either a portable computer or a desktop computer.

Typically, when a request is made by the user to send information to another node on the network, data representing this information is passed to the interface logic 1 and then subsequently to the memory 2 on the adaptor card. Thereafter, the packet assembler 4 adds data concerning the recipient and sender of the message and perhaps start and stop data. Other data required in the Ethernet packet ( e.g. a cyclic redundancy checksum), may be inserted by the packet assembler 4 but is more typically inserted by software running on the PC. The Ethernet packet is then converted to a serial stream of bits by the transmit serializer 5, this serial stream of bits subsequently having a clock signal incorporated therein by the Manchester encoder 6.This encoding is necessary to ensure that the Ethernet packet ( whose length typically varies from a few bytes to many thousands of bytes ) is transmitted to the receiving node without a loss in the synchronisation in that transmission. It will be realised that the transmission will then be made in accordance with the typical "carrier-sense multiple access/collision detection" sending protocol. In order to implement this protocol, the adaptor is provided with a collision detector 7.

On receiving a data-packet from the network, the Manchester decoder 8 first extracts from the encoded data stream a clock signal which is used to correctly interpret the incoming data. The coded signal is then passed to a de-serializer 9, which converts the serial stream of data to parallel data and passes this parallel data to a first-in/first out (FIFO) device 10. The first few bytes of the data stream are stored in the FIFO 10, and compared to the contents of hardware address comparator 11 to establish whether the data present on the network is destined for the PC to which the network adaptor is connected. However, whatever the result of this comparison the remainder of the data is passed through the FIFO to the memory 2.If the data is destined for the PC to which the adaptor is attached, the contents of the memory 2 are subsequently passed in a known manner across the interface into the PC.

Hereafter, software in the PC establishes whether the data-packet has been transmitted without error and may transmit a further packet of data requesting the original packet of data to be re-sent if an error is found.

Figure 2 shows a conventional design for a printer sharer which when connected to a number of PCs in a network enables those PCs to share the facilities offered by the printer. As in the case of the conventional network adaptor illustrated in Figure 1, the printer sharer may either be configured on a card to be inserted onto the PCbus or as a module to be plugged into the parallel port of the PC.

In the printer sharer, the information to be transmitted to the printer is transmitted byte-by-byte across the transmission means to the printer. On acceptance of a byte, the printer outputs an acknowledgement signal which is received at the transmitting node. On receipt of this signal the transmitting node transmits the next byte of data and this process is repeated until the entire data-packet is transmitted. Because of the byte-by-byte nature of this transmission it is not necessary to provide a memory capability or a Manchester encoding capability on the module. The function of each element shown in Figure 2 is similar to the function of the corresponding element shown in Figure 1.

Turning now to Figure 3, a specific embodiment of the network adaptor of the present invention shown therein is adapted to be connected to the parallel port of the PC. It will be appreciated by those skilled in the art, that the specific embodiment could also be configured as a card to connected to the PCbus. However, the module configuration allows the specific embodiment to be used with a portable computer.

When it is desired to send a data-packet over the network, a means for assembling an ethernet data-packet assembles the information to be transmitted across the network together with data signifying the length of the data-packet, a cyclic redundancy checksum and other data that will be known to those skilled in the art. This data is then passed to the means for dividing the datapacket into sub-packets each of which is one byte long.

A first of these bytes is then presented via the interface logic 43 together with data indicating the receiver and sender of the data-packet and start and stop data provided by the packet assembler 44 to the transmit serializer 45. The start and stop data is added if needed by the receiver to determine the position of the sub-packet within the data-packet.

In addition, the basic access system of Ethernet for sending data over a connecting bus is modified. The same CSMA method is used to initially place the first of the data sub-packets onto the network, but the transmission protocol varies in that the receipt of each sub-packet by an appropriate receiver must be acknowledged at the transmitter before transmission of each subsequent subpacket. During this phase of data-transmission, access to the data bus is restricted to the relevant transmitter and receiver and other nodes are unable to access the bus for transmission. If receiver fails to acknowledge receipt of any sub-packet in a defined time, the protocol allows the sub-packet to be re-transmitted, or the transmission to be aborted and the bus freed for other transmitters.

The data carrying means is provided by a RS485 bus.

The data rate for transmission along this bus is set at 1000kBits/sec to provide a compromise between the speed of the network and electromagnetic compatibility (EMC) considerations, thereby allowing operation on unscreened cable. In addition, the module further contains a replica of the PC's printer socket. When the network is not running this directs data directly to the printer.

When the network is running any user can access any printer using the network software features.

When operating as a receiving node, the first few bytes of the incoming data-packet are first de-serialised by the de-serializer 47 and then passed to the FIFO which allows the temporary storage of those bytes. The hardware address comparator 49 then outputs a signal indicating whether the data-packet on the network is intended for the PC 40 and if the data is so intended, the adaptor requests that the network suspends the transmission of the packet until it has passed the first element of the data to the PC 40. A reference clock (not shown) is based on an oscillator having a frequency at least five times greater than the transmission bit rate to provide unambiguous sampling and retrieval of the data. Once the first sub-packet has been transmitted to the PC, data transmission is allowed to resume and the next sub-packet is provided on the network.This procedure is repeated until the entire data-packet is transferred to the PC 40.

It will be seen that, on receipt of a data-packet, the data-packet is only passed to the PC if the address comparator establishes that the data-packet is destined for that PC. In this way, the need for memory on the adaptor is removed, as is the requirement for the adaptor to perform memory management functions across the host computer bus. A corollary of this is that it is possible to construct a network adaptor on the host computer's parallel port which does not contain memory. This has not been possible before, as it is not possible to perform DMA across a parallel port. Previously, this has meant that the full data-packet has had to be placed in memory on the external adaptor prior to transmission to the host.

It will also be seen how the division of the datapacket into a number of sub-packets allows network information to be sent over a low-cost analogue bus.

Further, this allows the information to be presented to the bus without the need for special encoding. By defining a method to perform this, expensive additional circuitry can be dispensed with, whilst retaining data security on the network.

It will further be seen that the need for pre-coding and clock recovery is removed by limiting the size of the Ethernet packet to a maximum defined length.

The sending of the data-packet in discrete data chunks may be achieved either by software or by hardware on the adaptor. The size of each chunk may be controlled so that there is no loss of synchronisation over the transmission time, so that the data can be unambiguously recovered without recourse to a reconstituted clock frequency by using a reference clock of sufficiently high frequency.

It will also be seen how the receiving station is able to analyze the first incoming segments of each full Ethernet packet and determine whether the data is to be accepted prior to informing the host computer of its presence.

Further, it is important that the Ethernet data-packet can be reconstituted at a receiving node. For example, a hardware module could be provided which re-assembled the Ethernet packet at the receiving node thereby enabling the module of the present invention to be used in networks containing other Ethernet receivers. The packet could alternatively be re-assembled by software in the PC of the receiving node.

It will further be seen how the length of the subpackets is carefully controlled to remove the need to extract a clock signal.

In addition, the discrete sub-packets may be transmitted as serial, part-parallel or parallel data.

It will als6 be seen how the sub-packets are sequentially assembled from the raw Ethernet data-packet such that a concatenation of the data-part of each subpacket would in its simplest implementation reconstruct the Ethernet data-packet in its entirety.

Claims (19)

1. A method of operating a computer network, said network comprising two or more nodes and data carrying means, each node comprising an adaptor and a computer, said method comprising the steps of: assembling the data to be transmitted into an ethernet data packet; dividing said packet so as to create smaller subpackets of a predetermined size; placing a first sub-packet onto the data carrying means; and identifying at other nodes whether the data is intended for that node, and if the data is so intended waiting for the computer of that node to accept that subpacket; and, on said acceptance placing the subsequent sub-packet of said ethernet data packet onto the carrying means; repeating the above step until the entire ethernet data packet has been transmitted.

2. A method of operating a computer network according to claim 1, wherein said sub-packet is sufficiently small to obviate the need for incorporating a clock signal into the sub-packet.

3. A method of operating a computer network according to claim 2, wherein said sub-packet contains one byte of data.

4. A method of operating a computer network according to claim 1, further comprising the steps of: providing an acknowledgement signal on the data carrying means indicating that the sub-packet has been accepted; and denying access to the data carrying means to nodes other than the transmitting and receiving nodes unless said acknowledgement signal has been provided.

5. A method of operating a computer network according to claim 1, further comprising the step of re-assembling the ethernet data packet from said sub-packet at the receiving node.

6. A method of operating a computer network according to claim 5 wherein said re-assembling step involves a mere concatenation of the data portion of the sub-packet.

7. A network node for use in the method of claim 1 comprising: a computer; an adaptor; means for creating an ethernet packet containing data to be transmitted; means for subdividing said ethernet packet into subpackets of a predetermined size; said adaptor comprising: means for placing a first sub-packet onto the datacarrying means; means for identifying whether a first sub-packet from another node is intended for the node; means for providing a data interface between said computer and said adaptor; means for providing a data interface between said adaptor and a data carrying means; means for sequentially transmitting said sub-packets via said data interface to another node in the network; and means for sequentially receiving said sub-packets via said data interface from another node in the network.

8. A network node according to claim 7, wherein the means for subdividing said ethernet packet comprises components of the computer operating under the control of a program.

9. A network node according to claim 7, wherein said adaptor further comprises a hardware means for subdividing said ethernet packet into sub-packets, said means for subdividing the ethernet packet comprising said hardware means.

10. An adaptor for use in a method of claim 1, said adaptor comprising: means for providing a data interface between said computer and said adaptor; means for providing a data interface between said adaptor and a data carrying means; means for sequentially transmitting said sub-packet via said data interface to another node in a network; means for sequentially receiving said sub-packet via said data interface from another node in a network.

11. An adaptor according to claim 10 further comprising a hardware means for subdividing said ethernet packet into sub-packets of a predetermined size.

12. An adaptor according to claim 10, wherein: said means for providing a data interface between said computer and said adaptor comprises a connector for connection to the parallel port of said computer; and said adaptor further comprises a replica of said parallel port; the arrangement being such that the adaptor may operate as part of a printer sharing system at the same time as providing networking capability.

13. A network comprising: a plurality of nodes according to claim 10; and data carrying means.

14. A network according to claim 13 wherein said data carrying means comprises telephone type cabling.

15. A network according to claim 13 wherein said data carrying means comprises wireless communication means.

16. A method of operating a computer network substantially as hereinbefore described with reference to and as illustrated in the accompanying drawings.

17. A computer network substantially as hereinbefore described with reference to and as illustrated in the accompanying drawings.

18. A network node substantially as hereinbefore described with reference to and as illustrated in the accompanying drawings.

19. An adaptor substantially as hereinbefore described with reference to and as illustrated in the accompanying drawings.