System and Method for Communication Between Nodes of a Closed Loop Local Communication Path

The present invention relates to a system and method for communication between nodes of a closed loop local communication path.

Many efforts have been made to develop a ring type communications network because of the potential low cost and particularly because such a system can be implemented using fiber optic paths. Fiber optic data transmission is particularly attractive for a local communication system because of its immunity to electromagnetically induced noise.

The general objective of a ring distribution of processors is recognized, for example, in the text Computer Networks and Distributed Processing, by James Martin, Prentice-Hall, Inc., 1981, page 38, FIG. 3.2, the second illustration.

One of the traditional objections to such a ring type processing network relates to such a system wherein a central controller is required with a store and forward function with respect to each of the other nodes. In such a system, communication is dependent on the speed of operation of the central controller, and is less flexible because of the requirements that the central controller poll the recipient node prior to actual forwarding of the data. It would be desirable to avoid the use of a central controller not only for greater speed and flexibility of communication between the nodes of a closed loop communication path, but also to provide such a communication system where failure of a computer at one node would not prevent continued operation, as is the case where a central controller is present.

Other types of ring networks essentially involve a ring of data processing devices each of which must check an incoming message before forwarding it to the next node. In many check and forward systems communication is structured in relation to specified multiplex time slots or in relation to a circulating control message e.g. a so called token or "GO" control signal. In any event, complexity and added expense result from the necessity for recognizing the destination address of an incoming message within a processing cycle of brief time duration, and for selectively removing a message packet from the loop where the same originated at the local node and has traversed the loop. In one such proposed check and forward system, information was to be sent in individual packets eighteen bits in length including a total destination label size of eight bits, a data field of eight bits and start/stop bits.

Accordingly, an important object of the present invention is to provide a closed loop communication system including a series of nodes each of which may communicate with the others on a flexible basis without the requirement of a permanent network controller and without the use of the check and forward principle.

In accordance with an important feature of the invention, a particular method for effecting communication in such a closed loop communication path has been developed wherein data messages are propagated about the loop path without processing delays at the respective nodes and without restriction on the length of a given data message.

A further object of the invention is to provide a closed loop communication system and method which is extremely simple and economical and highly flexible in its adaptability to the needs of relatively small enterprises, for example individual restaurants, and which system can very readily be expanded as needed with minimal cost and downtime.

In accordance with a method aspect of the present invention the nodes of a closed loop communication path initiate communication with other nodes by monitoring the closed loop communication path for a gap in communication along the closed loop communication path of a defined time span. Upon detection of such a communication gap, a node desiring to initiate communication, termed a bidding node, interrupts the loop by means of a transmission link switch at the bidding node, and transmits a bid message about the loop. The nonbidding nodes upon receipt of the bid message respond by insuring that their transmission link switches are closed. The bidding node monitors the closed loop communication path for the bid message (which will traverse the loop and arrive at the bidding node if the transmission link switches of the other nodes along the loop are in the closed condition). Upon a successful bid at a bidding node, a data message may be transmitted via the closed loop communication path while the transmission link switch at the bidding node is in the loop interrupting condition. At a recipient node upon receipt of the data message, the transmission link switch is placed in the loop interrupting mode, and a response is transmitted via the closed loop communication path. At the bidding node the transmission link switch is switched to the closed transmission condition to assure against a communication gap and then is placed in the loop interrupting mode. With such a method, decentralized control can be effected without requiring a common timing reference for the nodes of a closed loop communication path.

In accordance with an aspect of the apparatus invention, a series of nodes of a closed loop data processing network have communication links for enabling data communication about the loop path which is undelayed rather than requiring a processing step at each node. Means is associated with each data processing unit for placing an associated transmission link switch in a loop interrupting or cutoff mode during transmission of a data message from the data processing unit. Said means controls the transmission link switch to avoid introducing a communications gap of a predetermined time duration during reception of a response to the transmitted data message. In this way the response can circulate about the closed loop and prevent an undesired communications gap from being perceived at other nodes which may desire to transmit messages. This feature makes possible a decentralized closed loop communications network for data communication wherein the data processing units at the nodes operate autonomously without requiring a supervisory control such as a common timing reference.

A highly significant concept leading to successful operation of a closed loop data communication system as described herein was that of a control system embedded at the respective decentralized nodes such that one of the transmission link switches about the loop is maintained in cutoff mode at all times to insure that data messages do not repeatedly traverse the loop communications path. In a specific implementation of this concept, a bidding node seeking to initiate a data message and a recipient node transmitting a response coordinate the condition of their transmission link switches such that the bidding node prevents repeated circulation of a bid message or a data message originating therefrom while a recipient node prevents repeated circulation of its response message.

Other objects, features and advantages will be apparent from this detailed disclosure and from the appended claims.

FIG. 1 is a diagrammatic view illustrating a ring type local computer network implemented in accordance with the system and method of the present invention;

FIG. 2 is an electric circuit diagram illustrating a specific implementation of a transmission link switch for the nodes of a ring type computer network such as shown in FIG. 1 and indicating pulse amplification inserted in series with the transmission link switch, and node circuits for receiving and transmitting data and for controlling the transmission link switch; and

FIGS. 3 and 4 comprise flow diagrams for explaining the system and method in accordance with the present invention.

FIG. 1 illustrates a ring type computer network comprising a series of nodes or stations 1 through 5. A loop communication path is provided by successive transmission links 6 through 10 and successive transmission link switches 11 through 15. The switches 11 through 15, in the illustrated embodiment, have a first mode wherein the associated incoming and outgoing transmission links are directly connected for the transmission of data without appreciable time delay, and have a second mode (indicated diagrammatically for switch 14) wherein the transmission link switch prevents a direct transmission of signal pulses from the incoming transmission link to the outgoing transmission link.

Each transmission link switch in FIG. 1 is diagrammatically indicated as having an incoming signal branching means for directing incoming signal pulses both to the transmission link switch path and also to an incoming signal receiving path such as 17. Similarly an outgoing signal junction is indicated for transmitting signal pulses to the outgoing transmission link either from the transmission link switch (when the transmission link switch is in its first transmitting mode) or from an outgoing signal sending path 18.

Considering a given station such as that at 1 in FIG. 1, if one of the other nodes (e.g. node 4) transmits a signal pulse, the pulse travels along the loop communication path via transmission links such as 9 and 10. If the transmission link switch 11 at node 1 is in the first transmitting mode, the signal pulse will be directly coupled to the outgoing transmission link 6 and for example may be again received at the sending node (e.g. node 4) after a time interval essentially dependent on the transmission velocity of the transmission links 6 through 10 and of the switches 11 through 15. As illustrated in FIG. 1, a sending node such as node 4 is to have its transmission link switch 14 in the second nontransmitting mode, so that the signal pulse traverses the loop path only once.

In a system in accordance with the present invention, each node may be provided with pulse amplification. Such pulse amplification may be provided at each transmission link switch path. Where the transmission links such as 6 through 10 are implemented as fiber optic transmission lines, it is convenient in the present state of the art to convert incoming light pulses into electronic form for amplification. In this case the transmission link switches such as 11 through 15 may be in the form of electronic switches interposed in series with electronic pulse amplification circuitry.

For such a system it is advantageous to provide a standardized fiber optic interface module such as indicated at 20, FIG. 2, having a standard fitting 21 for receiving an incoming fiber optic transmission line and having a standard fitting 22 for connection with an outgoing fiber optic transmission line. Further standard electric cable fittings are indicated at 23 and 24 for providing an electric signal output and an electric signal input for the fiber optic interface module. The interface 20 may comprise an electronic receiver circuit 30 associated with an optical detector 31 and an electronic transmitter circuit 32 for driving an optical emitter 33.

The electronic receiver circuit 30 may be of a standardized design suitable for driving any of the various types of node circuitry such as indicated at 40. Further, the circuits 30 and 32 together may provide an optimum degree of amplification such that when cable fittings 23 and 24 are directly connected by means of a jumper cable, interface module 20 may serve as a simple repeater. In the event that module 20 does not include its own power supply, a further connector means may be provided as indicated at 39 which may be coupled to cooperating means 41 of node circuitry 40 so that power may be supplied to the interface module 20 from node circuitry 40. If node circuitry 40 is to be disconnected from module 20, e.g. because of a malfunction, a separate power supply module (not shown) may be connected at 39 to provide for operation of the interface module 20 independently of the node circuitry 40.

In general, a ring network such as shown in FIG. 1 may have fiber optic links such as 6 through 10 of a maximum length compatible with the amplification provided by components 30 and 32 of a standard interface module such as 20. Thus if node circuits 1 through 5, FIG. 1, comprise actual working computer circuits such as microprocessor units, electronic display units (e.g. cathode ray tube displays), keyboards, printers, and the like), the fiber optic links may each consist of a single continuous fiber optic strand without any amplification. A passive jumper cable between fittings 23 and 24 would then be used only to bypass the working node circuitry such as 40 in the event of a malfunction or the like.

A basic aspect of the system of FIG. 1 resides in the provision of a decentralized network where in principle any of the working node circuits 1 through 5 may be bypassed (e.g. at 23, 24, FIG. 2) without detriment to communication among the remaining nodes. Furthermore the transmission of data via an interface module is not subject to a processing (clock cycle) delay when the associated transmission link switch is in its first transmission mode (or is bypassed because of a malfunction). In particular, where electronic node circuitry, such as 40, FIG. 2, requires a given clock cycle time for executing a processing step, the signals being sent along the loop circuit such as shown in FIG. 1 traverse the node circuitry at an electronic transmission speed which is independent of the duration of such a clock cycle or processing interval and is dependent only on the actual response times of the circuits disposed in the transmission link switch path. Referring to FIG. 2, the switch path may include optical detector 31, receiver circuit 30, a cable 42 between fittings 23 and 43, pulse amplifier 44, NOR circuits 45, 46, pulse amplifier 47, a cable 48 between fittings 49 and 24, transmitter circuit 32 and optical emitter 33. In the particular exemplary embodiment of FIG. 2, which is based on commercially available components at the time of development, each of the transmission link switches such as 11 through 15 in FIG. 1 is implemented by means of a TTL (transistor transistor logic) integrated circuit module, and in particular by the NOR gate 45. An input line 51 of gate 45 is held at a logical zero potential to maintain switch 45 in its first pulse transmitting mode. To shift the switch 45 to its second nontransmitting mode, line 51 is shifted to a logical one "Transmit Cutoff" condition.

In FIG. 2, an electronic branching point 56 is provided at the output of electronic amplifier 44 so that incoming signal pulses can be supplied via a line 52 to a serial adapter component 53 as well as being supplied to the transmission link switch 45 via its second input 54.

When the node circuitry 40 has placed switch 45 in its second transmission cutoff mode, the serial adapter component 53 may effect a data transmission along the loop transmission path (FIG. 1) by supplying electronic pulses to an input line 55 of the NOR gate 46. Gate 46 thus provides a junction for coupling locally generated signal pulses onto the outgoing path leading from the transmission link switch (NOR gate 45) to the outgoing transmission link. Thus the incoming signal branching means which is indicated diagrammatically in FIG. 1 may be comprised of an electronic junction point as indicated at 56, FIG. 2. The "Receive Data" path 52 in FIG. 2 corresponds with the incoming signal branch path 17 indicated in FIG. 1. Similarly the "Transmit Data" line 55 in FIG. 2 corresponds to the outgoing signal sending path 18 indicated in FIG. 1.

For the particular node circuitry shown in FIG. 2, data is received as a time sequence of pulses on a single input channel 52. The pulses may represent successive hexadecimal characters and may be accumulated in a serial to parallel input register of component 53. For example, a given station of FIG. 1 may transmit a two character bid code followed by an address code representing the transmitting or bidding station or device. If the input register of component 53, FIG. 2, has received such a two character bid code, it can be transferred in parallel to a component 60 for processing, via a multiconductor data path indicated at 61. Such a transfer can be initiated by applying a momentary logical one signal to a read line 62. Similarly data from component 60 to be transmitted may be transferred in parallel via data path 61 to an output register of component 53 in response to a momentary logical one signal to a write line 63. Such data can then be output serially at a clock rate of an associated clock circuit 64 via transmit data line 55. Line 65 may connect a clock rate signal and the inverse of the clock rate signal with components 60 and 70. Component 70 may be a printer control circuit, for example. Data may be transmitted from component 60 to component 70 via a multiconductor path (not shown) and logical control signals may be supplied from component 60 to component 70 for effecting a desired output operation, with suitable acknowledgement and status signals being supplied from component 70 to component 60 in a conventional manner.

Component 60 may be implemented, for example, as a type 8748 integrated circuit, commercial sources for this microcomputer with erasable programmable read only memory being given in IC MASTER for 1981, page 555, lines 100 et seq.

A further understanding of the system of FIGS. 1 and 2 and its operation may be had by reference to the flow diagrams of FIGS. 3 and 4. Reference numerals referring to these figures have an initial number corresponding to the figure number followed by a hyphen and a further digit or digits.

FIG. 3 is a simplified flow diagram for explaining operation at a node where a device is ready to transmit a data message. Such a node may have circuitry as shown in FIG. 2. FIG. 2 will be referred to in describing both sending and receiving nodes since it will be apparent when similar parts of two different nodes are intended. The sending node monitors its serial adapter 53 for the presence of data on the loop communication path. For example, component 60 may respond to each "Data Ready" signal supplied via line DR. In this case, the absence of a logical one signal at line DR for a predetermined time interval would indicate the existence of a communications gap. For example if the input register has eight stages, and if data is supplied to the input register at 125 kilobits per second, a data ready signal could occur after 64 microseconds. For the case of a printer station ready to transmit data, the microprocessor of component 60 can simply check for a Data Ready logical one condition on line DR at intervals of ten microseconds. The program can be such that if the Data Ready line DR remains at a logical zero condition during each of about ninety-eight monitor intervals, corresponding to a communications gap of about one millisecond, an affirmative decision is reached at the decision block 3-1. Decision block 3-1 can be implemented as a register which is initially set to a count value of ninety-eight and which is decremented each time the Data Ready line is found to be at the logical zero level and inspected for a zero count value representing the affirmative decision. If the Data Ready line DR is found to be at a logical one level, the register is reset to ninety-eight and a negative decision is reached as represented by the return path leading from the right hand corner of block 3-1.

Where the communications gap is detected, the microcomputer of component 60 proceeds as represented by step 3-2. In this event, a logical one is placed on the associated Transmit Cutoff line 51 to shift the transmission link switch 45 to its second transmission cutoff mode. At this point, having opened the direct loop path of FIG. 1, e.g. at transmission link switch 14, the microprocessor of component 60 checks for another communications gap of about one millisecond (e.g. a count of ninety) as an allowance for mistiming. If this second gap is also found to be present, then a bid message is loaded into an output register of the serial adapter 53 and transmitted to the Transmit Data line 55 at the transmission rate desired for the loop communication path of FIG. 1. The bid message may include a first two hexadecimal characters, e.g. 55 (hexadecimal), followed by two hexadecimal characters representing the address of the bidding (originating) station (e.g. a number between one hundred and twenty-eight and two hundred and fifty-five). Where data channel 61 and/or the output register of component 53 has a capacity less than that required for a given message length, it will be understood that suitable portions (e.g. successive eight bit bytes) of the complete message may be transferred via an output buffer register to the output register in successive operating cycles. Thus if eight bits at a time are loaded via path 61, a first eight bits may be transmitted in a burst with a duration of sixty-four microseconds after which a second eight bit portion of the message is loaded into the output register and sent. Alternatively, the component 60 may be operated at a higher clock rate, e.g. six megahertz, and may load a complete message into a sixteen bit output register, eight bits at a time, whereupon the complete sixteen bit message is sent over a desired transmission interval.

For a specific serial adapter 53 (e.g. a type IM 6402-1 Universal Asynchronous Receiver-Transmitter), each eight bit byte of data may be sent as part of a character group comprising: one start bit, eight data bits, one parity bit and one stop bit, a total of eleven bits. With the system and method herein disclosed, a message may be of any desired length. For example with an eight-bit HASH character representing the sum of bytes starting with a one-byte local address, and a one-byte status, the data part of the message could contain over 250 bytes.

Once the bid message has been transmitted, the bidding station monitors its serial adapter 53 to determine if its bid message has completely traversed the loop communication path. If this is the case, the bid message will be received at the input register of the serial adapter 53 of the bidding station (e.g. node 4) signifying a successful bid as represented by block 3-3, FIG. 3. For the case where two hexadecimal characters at a time are transferred via the data link 61 from an input buffer register associated with the input register of serial adapter 53, the program may provide that upon receipt of the characters 55 (hexadecimal), the program awaits arrival of the next two characters and compares the received address with the address of the bidding station. If the address is equal to that of the bidding station, the program proceeds to block 3-4 as represented in FIG. 3. It will be understood that block 3-4 is executed without allowing the occurrence of a communications gap (of about one millisecond) at any of the other nodes, and that the other nodes are so programmed that the node link transmission switches will all remain in the first transmitting mode in the absence of the necessary communications gap. It will be understood that each block of data as transmitted will be received at each of the other nodes and should also traverse the loop and arrive at the input register of the sending station.

A response to the message sent by the successful bidder (e.g. node 4) is required. If this is an initial transmission from the bidder, then the response can be data (e.g. to be printed). It can also be a Message Accepted signal (ACK), Message in Error, Retry (NAK), Message Accepted and Acted Upon (DEND), or Message Cannot Be Accepted by Addressed Device At This Time (BUSY). As indicated at block 3-5, the bidder waits for a response for a suitable time interval, e.g. about one millisecond. If the first part of a response is received before the interval elapses, then the transmission link switch at the bidding station (e.g. switch 14, FIG. 1) is placed in its first transmitting mode as represented by block 3-6. This prevents a communication gap from being observed at stations beyond the original bidding station, e.g. stations no. 5 and no. 1, if station no. 2 was being addressed and is responding. The original bidding station can now receive and process any remaining part of the response (e.g. from station no. 2).

The original bidding station must again shift its transmission link switch (e.g. 14) to the second transmission cutoff mode as represented by block 3-7.

FIG. 4 indicates in a simplified flow diagram the operation of a receiving station in responding to the transmission of a communications message (sent as outlined in FIG. 3). When a receiving station receives a bid message, the first part of the message is symbolized as CLOSEL (close link) and is the number 55 (hexadecimal). The second part of the message is the address of the bidding node. When a bid is recognized as represented by block 4-1, the first step at the receiving station is to ensure that the transmission link switch (e.g. switch 12 of node no. 2) is in the first transmitting mode as shown by step 4-2.

The message transmitted by the bidding station may have the following component parts: a Start Character, (STCHR, e.g. 5E, hexadecimal); an Address of the Device to Receive Data, (TOADDR): the bidding device address, (ADDR); a Command (COM) to the Device Receiving the Message or the Status (STATUS) of the Bidding Device Which is Responding to a Command; Optional Data Intended for the Receiving Device (DATA); a Longitudinal Redundancy Check (LRC); and a Byte Sum of All Bytes Starting with ADDR through DATA, (HASH).

The message may include the TOADDR part twice so that if there is a discrepancy between the two receiving values for TOADDR, the message is ignored.

The receiving nodes each compare the TOADDR value with the address assigned to its device. If the received message is addressed to its device, as determined at block 4-3, then the DATA, LRC and HASH parts are read as represented at block 4-4. If the DATA part conforms with the LRC and HASH parts, the DATA is considered to be acceptable. In any event the addressed device must respond in less than one millisecond. A response message may take the format given above, or it may include the following parts: Start Character, STCHR (5E hexadecimal), and a response character, RCHR, sent twice. The response character may be ACK (78 hexadecimal), NAK (19 hexadecimal), DEND (same code as CLOSEL), or BUSY (2C hexadecimal).

In general the station receiving a message will first place its transmission link switch (e.g. switch 12 for node no. 2) in the second cutoff mode. Then the one of the response messages is sent as represented at block 4-5. Having sent the response, the responding station executes a time delay so as to allow for transmission propagation about the loop (so that the response only can traverse the loop once), and having blocked a second traverse of the loop, the transmission link switch (e.g. switch 12 of node 2) is placed in its first transmitting mode.

In an actual system which has been successfully operated, provision is made for resolving conflicts where several nodes issue bid messages simultaneously. A more detailed outline of programming under these circumstances is as follows, with most of the steps of the outline being assigned sequential numbers for convenience of cross reference.

GAPREC--step one

This is an entry point at a one-millisecond gap for checking for input data. The sequence of steps following this entry point is entered after a communications gap is recognized pursuant to decision block 3-1 of FIG. 3.

SNDREQ?--step two

If the device has data to send (Send Request is set), then the bidding/sending routine is executed beginning at step eleven of this outline, and which corresponds to the sequence beginning at block 3-2 in FIG. 3.

FIRST CHAR.=CLOSEL?--step three

If the device is to receive data, the first character received must be the close link request character CLOSEL. See the discussion of block 4-1, FIG. 4, and compare step thirteen et seq. of this outline. A bidding station which is unsuccessful is branched to this step three from step fifteen of this outline.

If the first received character is not CLOSEL, then the program may return to a MAIN CODE routine not relevant to the present disclosure.

CLOSE LINK--step four

Make sure the link switch is in transmitting (closed) mode at this device. If the link switch was in cutoff (open) mode at this device, the bidding device will retry thus getting through.

READ NEXT CHAR--step five

START CHAR?--step six

Wait for bidding to finish, returning from step six to step five until a Start Character STCHR is received.

MESSAGE TO THIS DEVICE?--step seven

The "to" address in the message indicates whether the data is directed to the local device. The "to" address is sent twice and both address values must be the same before acceptance.

If the message is not to this device, the program may return to a MAIN CODE routine not relevant to the present disclosure.

RECEIVE DATA--step eight

See the discussion concerning block 4-4, FIG. 4.

DATA OK?--step nine, part A

OPEN LINK, NAK, DELAY (later inserted), CLOSE LINK--step nine, part B

In part A of step nine, the processor of component 60 may compute the LRC and HASH values for the received data, and compare the computed values with the received values for LRC and HASH. If the computed and received values do not match, then a NAK response is transmitted to indicate that the received data is defective. As explained hereafter in reference to step seventeen of this outline, a master device may count NAK responses to a message, and reset its Send Request register after a predetermined number of failures. For minor devices, attempts to send the data may continue indefinitely. Where the sending device is to retry sending the message, another communications gap will occur before the message may again be received. Thus if the received data is defective, the program branches from part A to part B of step nine, and then after execution of part B returns to a MAIN CODE routine not relevant to the present disclosure.

The DELAY procedure of part B of step nine was later inserted as explained in connection with step nineteen hereafter.

RESPOND--step 10

If the received data is good, an appropriate response is sent. See block 4-5, FIG. 4.

SEND (Bidding/Sending Routine)

As stated under step two above, if the device has data to send, then a branching takes place after step two to a bidding/sending routine. The entry point for this routine is designated "SEND". The steps of the SEND routine are designated eleven through 18, and are outlined as follows:

OPEN LINK--step eleven

Since the device needs to originate a message, it must cut off the direct loop path (otherwise oscillation will occur in the ring network because of the progressive amplification at the successive nodes during each traverse of the loop path).

GAPSND?--step twelve

In this step, the prospective sending station monitors the communication loop for another millisecond communication gap before transmission is allowed.

If data is received during the gap timing interval, the program branches to step three above.

SEND BID--step thirteen

In this step a bid for link control is made, the step thus corresponding to a portion of block 3-2, FIG. 3. The bid message comprises the close link request character (CLOSEL) and the sending device address (ADDR). The close link request character requests the previous link controller (e.g. the previous successful bidder prior to the communications gap) to shift its link switch to the transmitting (closed) condition. The address which forms part of the bid message enables the resolution of simultaneous bids by allowing the bidding device with the lowest address to rebid, while forcing devices with higher addresses to drop out of the bidding.

CLOSEL CHAR RECEIVED?--step fourteen

If the bidding device awaits for a predetermined time interval without receiving a communication, or if a close link request character is not received, the program branches to the MAIN CODE routine, and the bid is retried after detection of a further communications gap.

ADDR: DEVICE'S ADDR?--step fifteen, part A

WAIT 100 MICROSECONDS--step fifteen, part B

If a bid message has been received, the address part ADDR as received is compared with the local device address. If the address values are the same, there is no bidding conflict and the local device proceeds to send its data. See block 3-4, FIG. 3, and steps twenty-four et seq of this Outline.

If the received address part is lower than the local address, the local device has lost the bidding. Thus the program branches to step three above.

If the received address is higher than the address of the local device, the local device executes part B of step fifteen so as to allow time for losing bidders to follow steps three and four above. This prevents an unsuccessful bidding station from cutting off a first portion of a new bid message.

If the local device has the lower address, it rebids after part B of step fifteen, by returning to step thirteen above. In this way, contention can continue until all link switches except that of the bidder with the lowest address have been placed in transmitting (closed) condition. With a situation such as shown in FIG. 1 with only one link switch cutoff, a bid message from the associated device, e.g. at node 4, will be transmitted about the loop (via link switches 15, 11, 12, and 13) and be received at the bidding device. Thus the address in the received bid message will correspond to the local address. See block 3-3, FIG. 3.

SEND DATA--step sixteen

Send data and await a response. The details of this SEND DATA routine begin at step twenty-four of this outline.

SEND FAIL?--part A of step seventeen p0 >255 RETRIES?--part B of step seventeen

If the data transmission has failed (see part B of step twenty-seven), the program branches from part A to part B of step seventeen. The number of retries is incremented after each failure, and if the number of retries is not greater than 255, the program returns to the MAIN CODE routine without resetting the Send Request register (SNDREQ). It is then necessary to wait for a communications gap before executing a retry. If transmission has failed more than 255 times, the Send Request is reset before returning to MAIN CODE. See step eighteen which follows part B in this case.

RESET SNDREQ--step eighteen

If transmission is successful as determined in part A of step seventeen then the resetting of SNDREQ signals completion of data transmission, and the program returns to the MAIN CODE routine.

RESPOND

When a receiving station determines that received data is good (e.g. as a result of part A of step nine above), then a response is to be sent as represented at block 4-5, FIG. 4. This routine is outlined in the following which includes steps nineteen through twenty-three.

IMMEDIATE EXEC?--part A of step nineteen

OPEN LINE, ACK, DELAY (later inserted), CLOSE LINK--part B of step nineteen

When data has been received, the receiving station may determine that an acknowledgement (ACK) is all the response that is required; in this case, the program branches from part A to part B of step nineteen. Also where no response can be made, the local device will send the ACK message as indicated by part B. On the other hand, if the response can be made immediately, the program progresses to step twenty.

The DELAY procedure of part B of step nineteen, and the corresponding delays in part B of step nine and in part B of step twenty were later inserted, and may have a duration of 200 microseconds. These inserted time delays insure that the sending station will have received the respective response and shifted its link switch to cutoff mode. See step 3-6, FIG. 3, and the "OPEN LINK" step following step twenty-eight in this Outline.

DATA TO SEND--part A of step twenty

OPEN LINK, DEND, DELAY (later inserted), CLOSE LINK--part B of step twenty

In part A of this step, the processor determines whether the response requires the transmission of data. If not, the program branches to part B of step twenty and transmits an indication of command acceptance (DEND). The DEND transmission indicates an end of communication.

SEND DATA--step twenty-one

This routine begins at step twenty-four of this outline and corresponds to block 4-5, FIG. 4.

DELAY, CLOSE LINK--(later developed steps)

These steps follow step twenty-one and correspond to block 4-6, FIG. 4. A delay of 200 microseconds provides time for data to be transmitted through the buffers of the serial adapter such as 53 of the responding station, and time for the bidding station to shift its link switch to cutoff mode. The time delay also accommodates any time lags in the repeater circuitry intervening between the responding device and the bidding device. Thus, before the responding station has shifted its link switch (e.g. switch 12 at node 2) to its transmitting (closed) mode, the bidding station will have received the complete data message transmitted by the responding station (block 3-6, FIG. 3), and will have again shifted its link switch to cutoff (open) mode (block 3-7, FIG. 3). The objective is to maintain one of the link switches such as 11 through 15, FIG. 1, in cutoff mode at all times (while avoiding a communications gap at stations beyond the bidding station during the transmission of response data). Such concept of maintaining one link switch in cutoff mode at all times was considered crucial to successful operation of the actual system of the present Outline, this system being designated the A Line System. (The lack of a step number for these steps of the Outline indicates their later insertion to effect successful operation. Delays were also later inserted before shifting the link switch of the responding station to transmitting mode at part B of steps nine, nineteen and twenty.)

SEND FAIL?--step twenty-two

If the failure indication of part B of step twenty-seven has not been actuated, the response has been successfully transmitted, and the processor returns to the MAIN CODE. If the response failed to get through, processing progresses to step twenty-three which follows.

SET SNDREQ--step twenty-three

By setting the Send Request, SNDREQ, the processor is set up to retry sending the response upon detection of a communications gap of one millisecond.

After setting up to retry sending the response when a new communications gap is found, the program returns to the MAIN CODE routine.

SEND DATA

This routine is entered at steps sixteen and twenty-one of this outline, and is summarized in the following which includes steps twenty-four through twenty-eight.

OPEN LINK, SEND START CHAR, TOADDR, TOADDR, and DEVICE ADDR--step twenty-four

When a bidding station determines that the loop is complete at step fifteen, part A, the bidding station should have its link switch (e.g. 14, for node 4) in the cutoff (open) mode. To insure this, however, an Open Link command is transmitted by the processor of component 60 for making certain that line 51, FIG. 2, is at the logical one transmit cutoff potential. Transmission of the Start Character, STCHR, terminates bidding. The further parts of the message have been discussed in relation to block 3-4, FIG. 3. The address of the intended recipient, TOADDR, is sent twice as previously mentioned. Then the local device address, ADDR, is sent.

SEND REST OF DATA--step twenty-five

The bidding station, having won the bid, may send a command (COM) to the device addressed or may send the status (STATUS) of the local device in responding to a command from the device addressed.

SEND LRC, HASH--step twenty-six

LRC is defined as a longitudinal redundancy check, namely the exclusive OR bytewise of all bytes starting with ADDR, and including any command, COM, or STATUS response, and any optional data (DATA) for the receiving device.

HASH is defined as the byte sum of all bytes starting with ADDR through DATA.

This step has been modified to omit a final CLOSE LINK command, so as to insure that a responding station will have shifted its link switch to cutoff mode before the bidding station shifts its link switch (e.g. 14, FIG. 1) to the transmitting (closed) mode.

RESPONSE?--part A of step twenty-seven

INDICATE FAILURE--part B of step twenty-seven

A response to the message of steps twenty-four, twenty-five and twenty-six is required, at least an acknowledgement, (ACK), or a Message in Error, Retry (NAK). If the message was an original transmission from the bidder, then the response can be a transmission of data. It can also be ACK, NAK, Device End (DEND), or BUSY. See the previous discussion in relation to block 3-5, FIG. 3, and in relation to block 4-4, FIG. 4. The DEND response indicates that the communication has been accepted and the command performed. No status report (STATUS) can be returned by the responding device. The BUSY response indicates that the responding device cannot handle the request at this time. A master station will automatically retry 255 times when a BUSY response is received. Minor devices will discontinue trying to send the message.

If there is no response within one millisecond, then communication failure is indicated, and the program proceeds from part A to part B of step twenty-seven. Having registered the failure, the processor proceeds to the OPEN LINK step following step twenty-eight.

CLOSE LINK (later inserted step)

Before the link switch of the sending station is placed in the transmitting (closed) mode, the sending station is to ensure that a receiving station has received the transmitted data and placed its link switch in cutoff (open) mode to transmit the response. Thus the sending station waits for the response as shown by step twenty-seven, part A, and only shifts the link switch to transmitting mode if a response is actually received.

If no response is received in part A of step twenty-seven within one millisecond, the link switch is not closed, the program branching from part B of step twenty-seven to the step following step twenty-eight of this outline.

NAK?--step twenty-eight

After closing the link switch, if the sending station determines that the response received was a NAK (message in Error, Retry), then the program returns to part B of step twenty-seven to indicate a communication failure, and then goes on to the OPEN LINK step which follows herebelow.

OPEN LINK (later inserted step)

If a response is received which is not a NAK, then when the complete response has been received, the sending station promptly shifts its link switch to the cutoff (open) mode before the responding station shifts its link switch to the transmitting (closed) mode (following a time delay). See block 3-7, FIG. 3.

If a failure in the transmission occurred resulting in a NAK, the link switch of the sending station will have been shifted to transmitting (closed) mode; accordingly the program branches from part B of step twenty-seven to the present "OPEN LINK" step.

Thus in any event at the end of the SEND DATA routine, the link switch of the sending station will be in the cutoff (open) mode.

If a responding station has data to send (part A of step twenty), it follows the routine of steps twenty-four et seq above, but then shifts its link switch to transmitting (closed) mode after a time delay of 200 microseconds, so that the sending station (the bidder) has time to shift its link to cutoff (open) mode after receiving the complete message.

In the system which has been successfully operated, exemplary components had the following implementation:

Optical detector OD1 (31, FIG. 2)--type SD 4478.

Electronic Receiver (30, FIG. 2)--cascade connection six line receivers A1-1, A1-2, A1-3, A2-1, A2-2, and A2-3-type MC 10116. Input circuit of A1-1: pin 10 of A1-1 connected to upper terminal of grounded C12 (680 pF, 100 V) and to upper terminal of R11 (20 kilohms, 1%); pin 9 connected to upper terminal of R10 (20 kilohms, 1%) and to cathode of OD1; opposite terminals of C12 and OD1 to ground; opposite terminals of R10 and R11 to pin 11 of A1-1 and to upper terminal of grounded C8 (0.1 microfarad, 100 V), lower terminal of C8 to ground. Pin 7 of A1-1 to pin 5 of A1-2 and to grounded resistor R12 (200 ohms, 1%); pin 6 of A1-1 to pin 4 of A1-2 and to grounded resistor R13 (226 ohms, 1%); pin 3 of A1-2 to grounded resistor R14 (200 ohms, 1%) and to left terminal of C10 (0.001 microfarad, 100 V); right terminal of C10 to upper terminal of R16 (1 kilohm, 1%) and to pin 13 of A1-3; pin 2 of A1-2 to grounded R15 (200 ohm, 1%); pin 12 of A1-3 via one kilohm resistor (1%) to lower terminal of R16 and to pin 11 of A1-1; pin 15 of A1-3 to grounded R18 (301 ohm, 1%) and to left end of R20 (1K, 1%); right end of R20 to pin 10 of A2-1 and via R7 (10 kilohm, 1%) to pin 7 of A2-1; pin 14 of A1-3 to grounded R19 (301 ohms, 1%); pin 7 of A2-1 to R7, to grounded R6 (1 kilohm, 1%) and to pin 5 of A2-2; pin 9 of A2-1 to pin 6 of A2-1 via R8 (1 kilohm, 1%) and to pin 11 of A2-2 via R9 (1 kilohm, 1%); pin 6 of A2-1 to R8, to grounded R5 (1 kilohm, 1%) and to pin 4 of A2-2; pin 11 of A2-2 to R9 and to grounded C9 (0.1 microfarad, 100 V); pin 3 of A2-2 to grounded R4 (1 kilohm, 1%), and to pin 13 of A2-3; pin 2 of A2-2 to grounded R3 (1 kilohm, 1%) and to pin 12 of A2-3. Output circuit of A2-3: pin 15 to grounded R2 (1 kilohm, 1%) and to pin 3 of connector J2 (RRCV HI); pin 14 of A2-3 to grounded R1 (1 kilohm, 1%) and to pin 4 of connector J2 (RRCV LO).

Connector (23, FIG. 2): J2 pin 2-ground, J2 pin 3-RRCV HI, J2 pin 4-RRCV LO, J2 pin 5-ground.

Connector (30, FIG. 2): J1 pin 1-plus five volts (+5 VT, transmit), J1 pin 2-ground, J1 pin 5-ground, J1 pin 6-plus five volts (+5 VR, receive).

Power supply circuit: J1 pin 6 to plus terminals of grounded C5 (220 microfarads, 6.3 V) and grounded C6 (0.47 microfarad, 50 V) and to left terminal of L1 (56 microhenry; ground terminal (GND) to J1 pins 2 and 5, and to J2 pins 2 and 5; right terminal of L1 to grounded C7 (100 microfarads, 6.3 V), grounded C1 (0.1 microfarad, 100 V) and grounded C2 (0.1 microfarad, 100) and to +5 V output (C1 and C2 are decoupling capacitors for the integrated circuits packs A1 and A2).

Electronic Transmitter Circuit (32, FIG. 2): J1 pin 1 (+5 VT) to plus terminal of grounded C11 (220 microfarads, 6.3 V), and to pins 8 (VCC) and 5 (BCV EN) of A3 (Type 75117), and to upper terminals of R21 (68 ohms, 1 watt, 5%), and to plus terminals of C3 (0.1 microfarad, 100 V) and of C4 (0.1 microfarad, 100 V) and to +5 V output (C3 and C4 are decoupling capacitors for integrated circuit packs); A3 (type 75117) and A4 (type 75452); ground terminal (GND) o J1 pins 2 and 5, and to J2 pins 2 and 5, and to pin 1 (1N), pin 4 (GND) and pin 7 (D ENABLE) of A3; input (RXMIT HI) to J1 pin 3 and to pin 3 (A) of A3; input (RXMIT LO) to J1 pin 4 and to pin 2 (B) of A3; pin 6 (OUT) of A3 to pins 1, 2, 6 and 7 of A4 (type 75); pins 3 and 5 of A4 to lower terminal of R21 and to anode of optical emitter OE1 type SE 4352 (33, FIG. 2).

Optical emitter OE1 (33, FIG. 2)-type SE 4352-anode of OE1 to pins 3 and 5 of A4 and to R21; cathode of OE1 to ground.

Connector (43, FIG. 2): J3 pin 3 (RRCV HI) to pin 5 of A42, type 75118 (44, FIG. 2): J3 pin 4 (RRCV LO) to pin 7 of A42 (44, FIG. 2).

Pulse amplifier (44, FIG.2)-A421 (type 75118): pin 5 to J3 pin 3; pin 7 to J3 pin 4; pin 8 to ground; pins 10 and 13 via R4 (10 kilohms) to +5 V; pins 11 and 12 (RECEIVE DATA) to pin 8 of A221-type 6502 (NOR gate 45, FIG. 2) and to pin 20 of A13-IM 6402-1 (Serial Adapter 53, FIG. 2); pin 16 to +5 V.

NOR gate A221-LS02 (45, FIG. 2): pin 8 to pins 11 and 12 of A 421 and to pin 20 of A13; pin 9 (TRANSMIT CUTOFF) to pin 38-P27 of A41-8748 EPROM (component 60, FIG. 2); pin 10 to pin 5 of A222-LS02 (46, FIG. 2).

NOR gate A222-6502 (46, FIG. 2): pin 5 to pin 10 of A221; pin 6 to pin 6 of A12-LS04 (Inverter 57, FIG. 2); pin 4 to pins 14 and 15 of A422 (pulse amplifier 47, FIG. 2).

Pulse amplifier A422-75118 (47, FIG. 2): pins 1 and 2 to lower terminal of R2-120 ohm (47a, FIG. 2) and to J4 pin 3 (RXMIT LO); pins 3 and 4 to upper terminal of R2 and to J4 pin 4 (RXMIT HI).

Connector (49, FIG. 2): J4 pin (RXMIT LO) to J1 pin 3; J4 pin 4 (RXMIT HI) to J1 pin 4.

Serial Adapter-A13, IM 6402-1, IOC, UART.alpha. Socket (component 53, FIG. 2): pin 1 (VCC) to +5 V; pin 3 (GND) to ground; pin 5 (RBR8) and pin 33 (TBR8) to pin 19 (DB7) of A41, etc. to provide eight conductors of the multiconductor path (61, FIG. 2); pin 23 (TBR1) to pin 10 (WR) of A41 via write line (63, FIG. 2); pin 4 (RRD) and pin 18 (DRR) to pin 8 (RD) of A41; pin 40-2 MHZ (TRC) and pin 17-2 MHZ (RRC) to 2 MHZ output of node clock circuits (64, FIG. 2); pin 21 (MR) to RESET; pin 19 (DR) to pin 1 of A41 via data ready line (DR, FIG. 2); pin 22 (TBRE) to pin 39 of A41; pin 20 (RBI) to RECEIVE DATA line (52, FIG. 2); pin 25 (TRO) to pin 5 of A12-LS04; pin 39 (P5), pin 35 (PI), pin 36 (SBS), and pin 16 (SFO) to ground; pin 38 (WLS1), pin 37 (WLS2) and pin 34 (CRC) to +5 V.

Inverter A12-LS04; pin 5 to pin 25 of A13 (53, FIG. 2); pin 6 to pin 6 of A222-LS02 (46, FIG. 2).

Processor and Protocol Control Circuits-A41-type 8748 EPROM (60, FIG. 2): pin 4 (RESET) to line RST; pin 2 (XTAL 1) to 6 MHZ output of the node clock circuits (64, FIG. 2); pin 3 (XTAL 2) to 6 MHZ of the node clock circuits (64, FIG. 2); pin 33 (P16) to line TST; pin 7 (EA) and pin 20 (VBS) to ground; pin 6 (INT) to J1 pin 39 (COVER OPEN); pin 32 (P15) to J1 pin 38 (LOW PAPER); pin 26 (VDD) and pin 40 (VCC) to +5 V; pin 39 (T1) to Transmit Buffer Empty line (TBRE, FIG. 2); pin 1 (T.0.) to Data Ready line (DR, FIG. 2); pin 38 (P27) to Transmit Cutoff line (51, FIG. 2); pin 37 (P26), pin 36 (P25), pin 35 (P24) to three high order bits of a node address selector switch (weights 64, 32 and 16); pin 24 (P23), pin 23 (P22), pin 22 (P21) and pin 21 (P20) to four low order bits of the node address selector switch (weights 8, 4, 2 and 1); pins 19, 18, 17, 16, 15, 14, 13 and 12 (DB7, DB6, DB5, DB4, DB3, DB2, DB1, DB.0.) to eight lines of data path to A13 (data path 61, FIG. 2); pin 10 (WR) to UART WR line (63, FIG. 2); pin 8 (RD) to UART RD line (62, FIG. 2); pin 29 (P10) to output circuits (70, FIG. 2) via ACT line; pin 28 (P11), via RTS line; pin 29 (P12), via TD line; pin 30 (P13), to DTR line; pin 31 (P14) to BTS line, lines RTS, TD, DTR and BTS also leading to the output circuits (70, FIG. 2).

In the foregoing, the abbreviation "pf" refers to picofarads, the abbreviation "V" refers to volts, and the abbreviation "K" stands for kilohm.

In an exemplary layout according to FIG. 1, a first fiber optic interface FOIA has its output (like 22, FIG. 2) connected via a first fiber optic line L1 to a second fiber optic interface FOIB. The second fiber optic interface FOIB was connected via a second fiber optic line L2 to a third fiber optic interface FOIC which in turn was connected to the first fiber optic interface FOIA via a third fiber optic line L3.

A printer control (such as 40, FIG. 2) had its electrical input port (such as 43) connected to the electrical output port (such as 23, FIG. 2), but had its electrical output port (such as 49, FIG. 2) connected to the electrical input port of a keyboard display control. The electrical output port of the keyboard display control in turn was connected to the electrical input port (such as 24) of the first fiber optic interface FOIA.

Similarly, the second fiber optic interface FOIB has an electric path between its electrical output and input ports (like 23, 24, FIG. 2) including two master computers, and the third fiber optic interface FOIC couples into the overall loop a printer control and a cathode ray tube display. Thus there are two local stations which are locally interconnected by inexpensive electrical cables associated with each fiber optic interface. Either of the local stations can be bypassed by rearranging the local electrical cables. The local stations are linked by fiber optic lines so as to assure freedom from electrical and magnetic interference over the extended paths between local sites. In a restaurant environment and printer and the keyboard and display may be located at a cashier's station while the remaining components may be in management offices or the like.

For the sake of background information, a copy of a computer printout is shown on the following pages, with titles as follows:

(I) ALINE MASTER I/O INTERFACE V2.6

(II) ALINE CONSOLE CONTROLLER V2.1

(III) ALINE PRINTER CONTROLLER V2.0

(IV) ALINE CRT CONTROLLER V1.1 ##SPC1## ##SPC2## ##SPC3## ##SPC4##

It will be apparent that many modifications and variations may be effected without departing from the scope of the novel concepts and teachings of the present invention.