US 4,086,566 AGrant
Error Detector for Modified Duobinary Signals
Issue Date:1978-04-25
•8 Claims
•7 Drawing Sheets
Abstract
A modified duobinary signal is separated into two (odd and even) pulse trains, each of which is representative of a bipolar pulse train. Each representative bipolar pulse train has a bit rate one-half that of the modified duobinary signal. Violations of the modified duobinary coding rules appear as violation of the bipolar coding rules. The bipolar violations which occur are detected separately for the odd and even pulse trains. The detected errors are read into an error combiner and are read out serially at the original modified duobinary bit rate.
Metadata
Assignee
- GTE Automatic Electric Laboratories Incorporated
Inventor
- Adam Lender
Application Information
Application Number:US 7421680
Filing Date:1976-11-15
Priority Date:1976-11-15
Art Unit:236
Classifications
IPC:
G06F 1100
Field of Search:
340325146.1 AB41;42
Patent Drawings (7 sheets)
Description
Background of the Invention
1. Field of the Invention
This invention relates to error detection in digital transmission wherein correlative level coding is employed, and more particularly relates to error detection for a modified duobinary signal.
2. Background of the Invention
Duobinary systems are explained, for example, in an article appearing in IEEE Transactions on Communications and Electronics, vol. 82, May 1963, pp. 214-218, as well as a variety of generally available publications such as IEEE Spectrum article February 1966. In particular, the duobinary system is disclosed and claimed in U.S. Pat. No. 3,238,299 entitled, "High-Speed Data Transmission System" by the present inventor. Subsequent publications by this inventor have disclosed a number of variations in the basic duobinary concepts. The duobinary technique, as originally developed, contemplated a baseband three-level, correlative level-coded signal in which the frequency spectrum of the wave increased with decreasing frequency, reaching a maximum value at a frequency which was essentially zero. Important among the additional publications of interest is the article entitled, "Correlative Digital Communication Techniques," appearing in IEEE Transactions on Communications Technology, vol. 13, June 1965, pp. 203-208. In addition, a number of correlative level-coded techniques are described in U.S. Pat. No. 3,388,330 entitled, "Partial Response Multilevel Data System." More particularly, a technique for converting a binary signal into a modified duobinary signal and the particularities of the characteristics of the modified duobinary signal are described in considerable detail in U.S. Pat. No. 3,457,510 entitled, "Modified Duobinary Data Transmission" by the present inventor. The subject U.S. Pat. No. 3,457,510 is incorporated herein by reference. A digital technique for generating a modified duobinary signal is contained in the copending application, Ser. No. 623,292, filed Oct. 17, 1975, and the present inventor is a coinventor of that application.
A technique for the detection of errors in a modified duobinary signal is disclosed in U.S. Pat. No. 3,461,426. In the subject error detection patent, it is to be noted that the transmitted modified duobinary waveform and the binary waveform reconstituted therefrom by decoding are monitored and coincidences detected between extreme level signals of the former and correlated portions of the latter to indicate errors. In contrast, the error detector of the present invention does not require the use of the binary output signal obtained by decoding in the receiving circuitry. Further, the prior error detection technique does not teach the concept of dividing the modified duobinary into two (odd and even) pulse trains each of which is representative of a bipolar pulse train.
Summary of the Invention
The error detector of the subject invention contemplates dividing the modified duobinary signal, having a predetermined bit rate, into first and second pulse trains, each of which is representative of a bipolar pulse train, and each having a bit rate equal to one-half said predetermined bit rate. The pulse trains are delayed one from the other by one bit interval. Errors which occur in each of the separate pulse trains are separately detected and serially combined to obtain the error output signal for the modified duobinary pulse train.
Brief Description of the Drawings
FIG. 1 is a waveform diagram showing transmitted and received modified duobinary pulse trains and the odd and even bipolar pulse trains that may be derived from the received modified duobinary pulse train.
FIG. 2 is a block diagram of the error detector of the invention.
FIG. 3 is a more detailed block diagram of the pulse train divider.
FIG. 4 is a more detailed block diagram of the clock divider.
FIG. 5 is a block diagram illustrating the two bipolar error detectors and the input paths including delay circuits 10 and 12.
FIG. 6 is a block diagram of the error combiner; and
FIG. 7 is a waveform diagram showing the detection of an error in the even bipolar error detector.
Table I illustrates the positive and negative representations of the time slots for the waveform polarity and further illustrates the odd and even representations that will be obtained from the waveform T-characteristic.
Detailed Description of the Invention
In general, the modified duobinary waveform operated upon by the present invention comprises a three-level signal in which the extreme levels represent one binary state, such as MARK, and the center or intermediate level signal represents the other binary state, such as SPACE. This convention can, of course, be reversed. This signal follows a predetermined set of rules. These rules may be readily understood by grouping all of the successive MARKS in pairs and assigning the pair number of each MARK as illustrated in FIG. 1 for waveform T. Successive MARKS are indicated by the numerals 1 and 2, with a repetition of this numbering for the next pair of MARKS. A MARK bearing number 1 in a pair of two successive MARKS will be seen to always have the opposite polarity relative to the previous MARK which, of course, carries the number 2. The polarity of the MARK, identified by number 2, relative to the previous MARK bearing number 1, is governed by a set of odd and even rules, as in the straight duobinary system and method. Specifically, if the number of intervening SPACES between a pair of MARKS numbered 1 and 2 is odd, then the polarities of these two MARKS are opposite; and if the number of intervening SPACES between a pair of MARKS numbered 1 and 2 is even, then the polarities of these two marks are the same.
By referring to waveform R, FIG. 1, it is seen that an error has been introduced at time slot 6. It is apparent that at this point an error detector would not recognize the negative pulse in time slot 6 as an error, since the number of intervening SPACES from the last MARK is even, and since the previous MARK was negative, the appearance of a negative in time slot 6 would not be indicative of an error. Following the rules as set forth hereinabove, it is apparent that the presence of a negative MARK in time slot 8 is indicative of the fact that an error has occurred either by the presence of this MARK in time slot 8 or by the presence of the previous MARK in time slot 6. Thus, an error may be detected by the violation of the modified duobinary coding rules.
Waveforms A and B in FIG. 1 show, respectively, the odd and even bipolar waveforms which may be derived from the modified duobinary pulse train. The waveforms as shown may be considered to be return-to-zero or 50% duty cycle bipolar signals at the bit rate equal to half of the modified duobinary waveform. The coding rule for the bipolar signal is that successive MARKS will have opposite polarities. Referring now to the waveform B, wherein the error occurs because time slot 6 is associated with this waveform, it is to be noted that here again the violation of the modified duobinary coding rules also appears as a violation of the bipolar coding rules. Thus, detection of the bipolar violations of the odd and/or even wave forms will result in the detection of the errors which have occurred during transmission of the modified duobinary pulse train. Thus, the essence of this error detection technique is to regard the modified duobinary signal as an algebraic sum of two independent bipolar pulse trains, delayed with respect to each other by T seconds, where T is the bit interval duration of the modified duobinary, and 1/T is the bit rate in bits per second of the modified duobinary pulse train. This representation is valid as shown in a waveform diagram of FIG. 1, as it conforms with the rules and patterns of the modified duobinary signal. The pulse train R is modified duobinary, following legitimate duobinary rules. Waveform A represents the odd pulses only (1, 3, 5, etc.) and waveform B represents only the even pulses (2, 4, 6, etc.). An algebraic sum of waveforms A and B is, indeed, R. Except for the error which occurs at time slot 6, waveforms A and B follow the bipolar pulse train rules in that successive MARKS alternate in polarity. Thus, the keypoint in the implementation is to separate the pulse train R in FIG. 1 into bipolar pulse trains. A and B and to check these bipolar waveforms separately for bipolar violations. In order to effectuate this representation using a binary logic, it is necessary to approach the required result in a somewhat different manner.
Referring now to FIG. 2, which shows a general block diagram of the preferred embodiment of the invention, the received modified duobinary signal as shown at R, FIG. 1, is applied to pulse train divider 4 via path 2. Pulse train divider 4 separates the waveform R into waveforms C and D, as shown in FIG. 7. By referring to FIG. 7, it may be seen that waveform C is a unipolar representation of the positive pulses of waveform R and that waveform D is a unipolar representation of the negative pulses of waveform R. With respect to the latter, the pulses are shown as positive in waveform D. Thus, as will become apparent later, the bipolar signals are representative only because they are derived from a unipolar representation. A clock signal at the pulse repetition rate of the modified duobinary signal in FIG. 2 is applied from clock 34 via path 36 to pulse train divider 4. The clock 34 may be a local clock, but most often is derived from the incoming data, as is well known in the art. Techniques for deriving a local clock are well known and these will not be described in this specification. It is also seen that the output from clock 34 is applied to clock divider 40, and clock divider 40 divides the clock rate in half and, in addition, provides 42 and 44 at one-half of the clock 34 rate, but in opposite phases one from the other. Reverting now to the output of pulse train divider 4, it is to be seen that two similar circuit paths are employed. The first path includes delay 10 and bipolar error detector 22, and the second includes delay 12 and bipolar error detector 24. Since these operate similarly, only the upper one will be considered in the following discussion. Outputs 6 and 8, i.e., waveforms C and D, are applied to inputs of delay 10 along with the input 42 from clock divider 40. Delay 10 provides, on output path 14, a delayed representation of the input signal on path 6 and, at output path 16, the delayed output representation on path 8. It should be noted that because of the timing from clock divider 40, the representations on paths 14 and 16 are applicable only to the odd time slots of the original waveform. In addition to the representations on paths 14 and 16, the original waveforms on paths 6 and 8 are applied to bipolar error detector 22 as well as the timing signal along path 42 from clock divider 40. The bipolar error detector detects pipolar error violations which may occur by comparing the polarities of previous and present MARKS, and applies these via path 26 to error combiner 30, which combines the errors from both paths and applies these to output 32. Pertinent portions of the error detector shown in FIG. 2 will be explained in more detail hereinbelow.
A technique for generating waveforms C and D is shown in FIG. 7 is illustrated in FIG. 3. The modified duobinary waveform, which is received from the transmission facility, is applied to center-tap transformer 46 via path 2 and provides the positive and negative outputs, respectively, at paths 48 and 50. The positive output is applied to top slicer 52, which identifies the pressence of a positive pulse, such as is shown in time slots 1 and 2 of FIG. 7. The bottom slicer 54 identifies the negative- going pulses such as is shown in time slots 3 and 8 of waveform R, FIG. 7. These pulses are amplified in amplifiers 56 and 58, which are then applied, respectively, to flip-flops 60 and 62 from which the output waveforms C and D are obtained from the Q-outputs of the respective flip-flops. The timing information for flip-flops 60 and 62 is applied via path 36 from clock 34 to the clock inputs of the respective flip flops. It should be noted that in certain applications, the waveforms C and D, i.e., the positive representation in a unipolar format of the positive and negative pulses of a modified duobinary waveform, may already be available from equipment associated with the error detector, and thus the elements as shown in FIG. 3 would not be required.
In order to obtain the odd and even output as disclosed in the conceptual discussion above and illustrated in FIG. 1, it is necessary to derive the clock pulse signals, each having a clock rate of one-half the bit of the modified duobinary signal and each being of opposite phase one from the other. This is accomplished by clock divider 40, which is shown in FIG. 4. The clock frequency in the form of a square wave appears on path 38 and is applied to inverter 64. Inverter 64 may include a plurality of inverters as is necessary to obtain the time delay required to maintain timing integrity. The output of inverter 64 is applied to a D-type flip-flop 66 which has the Q output and the D input strapped as shown. When the positive-going edge of the square wave appears at the clock pulse input, the Q-output becomes Q, i.e., the same as the D-input. Equivalently, Q always changes state. Further, it is apparent that Q and Q have opposite phases. Thus, the outputs Q and Q are square waves at a frequency which is half that of the modified duobinary bit rate, and they have opposite phases. Essentially this is the same as having the clock square waves delayed by T seconds, where T is equal to the reciprocal of the modified duobinary bit rate. NAND-gates 70 and 72 operate as inverters to clean up the waveforms. These are followed by an RC differentiator for shaping and further clean-up by the NAND-gates 90 and 92, which provide the required divided clock pulses which have opposite phases. One phase of the clock output, i.e., that which is provided on path 44, is shown in FIG. 7 as waveform 44.
The delay and bipolar error detector circuit arrangements illustrated in block form in FIG. 2 are shown in more detail in FIG. 5. As noted hereinabove, the circuits for the odd and even time slots are identical, and therefore only the operation of one will be described in detail since the operation of the other is identical, except for the time slots for which they are applied. In this case, the discussion will focus on the even time slots, i.e., the lower circuit configuration, since the waveform R, as shown in FIG. 1, has an error in the even time slot no. 6. The only essential difference between the two is in the phase of the clock pulses, as described hereinabove. Each of these circuits detects errors. One detects the errors caused by bipolar violations of the odd bits, and the other detects errors cause by bipolar violation of the even bits. Thus, the two clocks, in effect, separate odd from even bits that appear at waveforms C and D along input paths 6 and 8. In the modified duobinary waveform, if one considers either the even only or odd only bits, then the successive pulses must alternate in polarity. Any departure from this rule constitutes a violation which indicates an error. In FIG. 5, the delay 10 is shown as J-K flip-flop having a timing input to select only the odd bits, while the delay 12 is shown as a J-K flip-flop with timing, such as to select only the even bits. In the following discussion, only the operation of the detection circuitry for detecting bipolar violations of the even bits will be considered.
The function of delay 12 is to "remember" the previous +1 represented in waveform C, applied on path 6, or to "remember" the previous -1 represented in waveform D, which is applied on path 8. This may be readily accomplished by using a J-K flip-flop as shown in FIG. 5. For example, a +1 represented by waveform 6 will result in the Q-output of flip-flop 12 being positive or high; however, should the input on path 8, i.e., waveform D, be positive, representing -1, Q will be high and Q will be low. This "reading in" into flip-flop 12 occurs at the instant of clock pulse on the input CP. Table 1 illustrates the positive and negative representations of the time slots for the waveform polarity, both prior to and subsequent to the representations of the waveforms C and D, and further illustrates the odd and even representations that will be obtained from the waveform T-characteristic, as illustrated in FIG. 1. Thus, no error occurrences appear in Table 1 as illustrated. It is interesting to note that the odd and even waveforms show a "bipolar" representation in that positive representations alternately occur in the C- and D- outputs for the odd time slots and also for the even time slots, as indicated. Reverting back to the waveform diagrams, FIG. 7, the representations at the Q and Q-outputs of delay 12 are shown at 18 and 20. When these, in conjunction with the inputs on paths 6 and 8, are combined in the logic gating arrangement consisting of NAND-gates 102, 104 and 106, it is noted that the output of NAND-gate 102 is always positive, i.e., a+1, during the intervals in question, whereas the output of NAND-gate 104 is positive, i.e., a+1, except for one time slot, thus, indicating that an error had occurred. Note that since this is detection only, the actual time of occurrence of the error is not obtained. The 1 and 0 inputs to NAND-gate 106 then cause an output pulse to appear as illustrated in waveform 106, which is then timed through flip-flop 108 to appear on output path 28. Stated in a more precise manner, when waveform C has a positive pulse and Q is positive, this indicates an error, because two positive successive pulses appear, which violates the bipolar pattern. Similarly for simultaneous occurrences of positive pulses in waveform C and a positive output at Q. The three NAND-gates which follow each of the respective J-K flip-flops constitute an AND-OR combination. This is well known and may be illustrated using the following Boolean equation:
thus, the output W of NAND-gate 106 is CQ + DQ. This means that simultaneous occurrence of either C and Q or D and Q will cause an output pulse indicating an error.
Next, the odd and even errors must be combined. This is shown by the error combiner block diagram in FIG. 6. Once again, the three NAND-gates constitute an AND-OR combination. The three NAND-gates are 110, 112, and 114, and their function is to combine odd and even output errors. Thus, the output on path 16 provides all errors, both odd and even. Gate 116 is merely an inverter to "clean" the error output waveform. The output from NAND-gate 114 occurs when one of the two possible events occurs; simultaneous occurrence of a positive output on path 26 and appearance of the clock pulse on path 42. This indicates an odd error. The second possible event is simultaneous occurrence of a positive output on path 28 in conjunction with a clock pulse on path 44. This indicates an even error.
While the invention has been particularly shown and described with reference to a preferred embodiment thereof, it will be understood by those skilled in the art that changes in form and detail may be made therein without departing from the spirit and scope of the invention.
Claims
What is claimed is:
1. An error detector for detecting errors in a modified duobinary pulse train, which comprises: means for converting the modified duobinary pulse train into a first unipolar pulse train representing the positive pulses at a first output, and a second unipolar pulse train representing the negative pulses at a second output; delay means for providing a first delayed representation of the first and second unipolar pulse trains for only the odd bit times, and a second delayed representation of the first and second unipolar pulse trains for only the even bit times; and logic means operatively connected to said delay means and to said converting means for detecting coincidences of occurrence of successive pulses in the first or second pulse trains only for the even or odd bit times.
2. An error detector in accordance with claim 1 wherein said logic means further comprises: a first gating means having a plurality of inputs operatively connected to receive the first unipolar pulse train and the delayed representations, said first gating means providing an output error signal whenever a bipolar violation occurs in said first bipolar pulse train; a first flip-flop connected to receive the output error signal from said first gating means, said first flip-flop responsive to error input signals to provide delayed output error signals; a second gating means having a plurality of inputs operatively connected to receive the second unipolar pulse train and the delayed representations, said second gating means providing an output error signal whenever a bipolar violation occurs in said second bipolar pulse train; a second flip-flop connected to receive the output error signal from said second gating means, said second flip-flop responsive to error input signals to provide delayed output signals; and means for combining the output error signals from said first and second flip-flops.
3. An error detector in accordance with claim 2 wherein said means for combining further comprises: timing means providing a first timing signal at one-half the bit rate and a second timing signal at the same one-half bit rate but delayed so as to be out of phase with said first timing signal; a third gating means having as one input said first timing signal and having as a second input the output of said first flip-flop, and having an output; a fourth gating means having as one input said second timing signal and having as a second input the output of said second flip-flop, and having an output; a fifth gating means having a one input connected to the output of said third gating means, a second input connected to the output of said fourth gating means, and having an output; an inverter having an input connected to the output of the fifth gating means and having an output.
4. An error detector for detecting errors in a modified duobinary pulse train, which comprises: timing means; pulse train dividing means having a first input connected to said timing means, a second input connected to receive said modified duobinary pulse train, said dividing means providing a unipolar pulse train at a first output which is representative of the positive pulses, and a second output which is representative of said negative pulses of the modified duobinary pulse train; a first delay means for combining the first and second divider outputs to obtain a binary representation of the odd bits of the modified duobinary signal, and the complements thereof, having one input connected to said pulse divider first output, a second input connected to said timing means, having a third input connected to said pulse divider second output, and having first and second outputs; a second delay means for combining the first and second pulse divider outputs to obtain a binary representation of the even bits, and the complements thereof, having one input connected to said pulse divider first output, a second input connected to said timing means, and a third input connected to said pulse divider second output, and having the first and second outputs; a first error detection means having four inputs and an output, the first being connected to said pulse divider first output, the second input being connected to said pulse divider second output, the third input connected to the first output of said first delay means, and a fourth input connected to the second output of said first delay means; a second error detection means having three inputs and an output, the first input being connected to said pulse divider first output, the second input being connected to said pulse divider second output, the third input being connected to the first output of said second delay means, and a fourth input being connected to the second output of said second delay means; and error combining means having one input connected to the output of said first error detection means, a second input connected to the output of said second error detection means, and a pair of inputs connected to said timing means, and having an output.
5. Apparatus for detecting errors in a modified duobinary signal having top and bottom extreme levels, which comprises: clock means having an output and having a clock pulse rate equal to the bit rate of said modified duobinary signal; pulse train dividing means for forming first and second binary pulse trains from said modified duobinary signal, having a first input connected to the output of said clock signal and having a second input connected to receive said modified duobinary signal, providing at a first output said first binary pulse train which comprises all of the top extreme levels of said modified duobinary signal as one binary state and the absence of top extreme levels as the other state, and providing at a second output said second binary pulse train which comprises all of the bottom extreme levels of said modified duobinary signal as said one binary state and the absence of bottom extreme levels as the other said state; a clock divider having as one input the output clock pulses from said clock means, having at a first output a half-rate clock signal of a first phase and having at a second output a half-rate clock signal of a second phase which is 180.degree. from said first phase; a first delay means for combining the first and second binary pulse trains to obtain a binary representation of the odd bits, and the complement thereof, of the modified duobinary signal, said first delay means having a first input connected to receive said first binary pulse train, having a second input connected to receive said second binary pulse train, having a third input connected to the first output of said clock divider and having first and second outputs; a second delay means for combining the first and second binary pulse trains to obtain a binary representation of the even bits, and the complement thereof, of the modified duobinary signal, said second delay means having a first input connected to receive said first binary pulse train, having a second input connected to receive said second binary pulse train, having a third input connected to the second output of said clock divider, and having first and second outputs; a first error detection means for detecting errors occurring in the binary representation of the odd bits of the modified duobinary signal, said first error detection means having first and second inputs, respectively, connected to the first and second outputs of said first delay means, having third and fourth inputs, respectively, connected to the first and second outputs of said pulse train dividing means and having an output; a second error detection means for detecting errors occurring in the binary representation of the even bits of the modified duobinary signal, said second error detection means having first and second inputs, respectively, connected to the first and second outputs of said second delay means, having third and fourth inputs, respectively, connected to the first and second outputs of said pulse train dividing means, and having an output; and error combining means for combining the error information from said first and second error detection means, said error combining means having a first input connected to the output of said first error detection means, having a second input connected to the output of said second error detection means, having third and fourth inputs, respectively, connected to the first and second outputs of said clock dividing means and having an output.
6. Apparatus in accordance with claim 5 wherein said pulse train dividing means comprises: coupling means having an input connected to receive said modified duobinary signal, providing at a first output the top extreme levels as positive pulses and at a second output the bottom extreme levels as positive pulses; first slicing means having an input connected to the first output of said coupling means and providing an output; first pulse reshaping means having a first input connected to the output of said first slicing means, and having a second input connected to the output of said clock means and providing at an output said first binary pulse train; second slicing means having an input connected to the second output of said coupling means and having an output; and second pulse reshaping means having a first input connected to the output of said second slicing means, having a second input connected to the output of said coupling means and providing at an output said second binary pulse train.
7. Apparatus in accordance with claim 6 wherein said first error detection means further comprises: a first AND-OR logic means having a first input connected to the output of said first pulse reshaping means, having a second input connected to the output of said second pulse reshaping means, having a third input connected to the first output of said first delay means and having a fourth input connected to the second output of said first delay means and having an output; and pulse retiming means having one input connected to the output of said AND-OR logic means and having a second input connected to the first output of said clock divider and having an output.
8. Apparatus in accordance with claim 7 wherein said second error detection means further comprises: a second AND-OR logic means having a first input connected to the first output of said pulse reshaping means, having a second input connected to the output of said second pulse reshaping means, having a third input connected to the first output of said second delay means, having a fourth input connected to the second output of said second delay means and providing an output; and pulse retiming means having a first input connected to the output of said AND-OR logic means, having a second input connected to the second output of said clock divider and having an output.
Patent Citations (4)
| Patent | Date | Inventor | Cited By |
|---|---|---|---|
| US3337864 | 1967-08-01 | Lender | |
| US3343125 | 1967-09-01 | Lender | |
| US3418631 | 1968-12-01 | Sipress et al. | |
| US3461426 | 1969-08-01 | Lender |