WO1985004540A1

WO1985004540A1 - Two-wire duplex data transmission system

Info

Publication number: WO1985004540A1
Application number: PCT/US1985/000413
Authority: WO
Inventors: Alfredo Anuff
Original assignee: American Telephone & Telegraph Company
Priority date: 1984-04-02
Filing date: 1985-03-13
Publication date: 1985-10-10
Also published as: JPS61501741A; CA1226918A; EP0175756A1

Abstract

A system (10) for transmitting data between two units (12, 14) in a full duplex mode over two lines (16, 18) transmits data in one direction in the form of a duobinary dipulse line code format and in the other direction in either a duobinary 100 % duty cycle or a bipolar 100 % duty cycle line code format. Separation of the signals in each unit (12, 14) is by both filtering and canceling. Detection of the duobinary dipulse signal in the data unit (12, 14) receiving it includes demodulation to the baud frequency for improved accuracy.

Description

TWO-WIRE DUPLEX DATA TRANSMISSION SYSTEM

Technical Field

The invention relates generally to the transmission of data and more particularly to simultaneous transmission of data in both directions over two wires. Background of the Invention

Data is typically transmitted electronically between two data units in the form of a digital line code. Where simultaneous two-way transmission capability is desired, separation of these signals can be achieved by the use of two wire pairs. One wire pair is identified with transmission, and the other wire pair is identified with reception. A simultaneous two-way signal capability is generally known as "full duplex" mode. The full duplex mode is advantageous in that one unit need not wait for the transmission of the other unit before transmitting to it. Therefore, total transmission time is saved.

There is existing equipment, such as the subscriber loop connecting a customer to a telephone office facility, which has only a single wire pair available for transmission. It would be advantageous to have the capability of transmitting data in full duplex mode over this single wire pair. This requires each data unit to have a means of differentiating between the transmitted and received signals.

One way of differentiating between the transmitted and the received signals is to use the so called "burst" or "ping-pong" mode, which amounts to time compression multiplexing (TCM). The data is compressed in each unit to occupy less than half the time interval of its original form. It is then transmitted in non-overlapping bursts with a guard band to provide adequate separation. This approach, however, requires a higher frequency band due to the higher data rate, and therefore results in increased loss, especially where the wire pair contains one or more bridge taps. Another approach, which does not require a higher data rate, is the so called "hybrid" mode. For this mode the transmitted and received signal are similar, but each data unit is provided with an echo canceler for differentiation. However, in order to adequately serve this function within the existing loop plant, the canceler must have a minimum of sixty dB (decibel) attenuation for the minimum 64kb (kilobit) rate needed to accommodate at least one digital voice channel. An echo canceler of this type is relatively complex and is not presently available commercially.

A third approach to differentiating between the transmitted and received signals is by Frequency Division Multiplexing (FDM) , in which the signals occupy different frequencies separated by a guard band. Such a system is described, for example, in U. S. Patent 4,236,244. High pass and low pass filters are used to provide separation of the signals. There is a tradeoff with this arrangement. If the guard band is made narrow, the filters required for the signal separation must be relatively complex. On the other hand, if the guard band is made larger to permit more effective filtering, then the higher frequency signal will begin to be outside the frequency range for which the system is designed and as a result will be more susceptible to transmission imperfections, such as crosstalk, noise, and attenuation. Summary of the Invention

In the novel system in accordance with the present invention, the transmitted and received signals have the same nominal frequency, but have different symbol, or pulse shapes, which are chosen to enhance the concentration of power within different frequency regions. The line code formats are chosen so that the primary lobes of the energy spectra of the codes overlap, but have distinctly separate peak regions. Separation of the nonoverlapping portions of the lobes is performed by filters. Separation of the overlapping portions of the lobes is performed by cancelers. In this manner the cancelers and filters uniquely complement each other in that each is required to perform separation of only that portion of the respective lobes for which it is particularly suited. Consequently, neither the filters nor the cancelers need be very complex. Brief Description of the Drawing

FIG. 1 is a block diagram of a data transmission system in accordance with one example of the present invention including central office unit and a customer unit.

FIG. 2 is an exaggerated schematic diagram of the energy spectrum of a duobinary dipulse line code format of the signal transmitted by the office unit of the data transmission system of FIG. 1.

FIG. 3 is an exaggerated schematic diagram of the energy spectrum of a bipolar, 100% duty cycle line code format of the signal transmitted by the customer unit of the data transmission system of FIG. 1. FIG. 4 is an exaggerated schematic diagram of the main power spectrum lobes of the diagrams of FIGS. 2 and 3 in superposition.

FIG. 5 is a schematic diagram in block form of the circuit a customer unit of the data transmission system of FIG. 1.

FIG. 6 is a schematic circuit diagram in block form of the office unit of the data transmission system of FIG. 1.

Detailed Description The data transmission system 10 of FIG. 1 is one example of the present invention. It includes a customer unit 12 and an office unit 14 which are interconnected by a pair of wires 16, 18. The wires 16, 18 carry both a dipulse duobinary line code format digital signal transmitted by the office unit 14 to the customer unit 12 and a 100% duty cycle bipolar signal transmitted from the customer unit 12 to the office unit 14. The signals are at the same baud rate. The nature of duobinary line codes is discussed in detail in, for example, A. Lender, "The Duobinary Technique for High-speed Data Transmission," IEEE Transactions on Communications and Electronic, 82 (May 1963), pp.214-218. The manner and equipment for generating duobinary line codes of a duobinary dipulse form or of a 100% duty cycle bipolar form are known to those in the art. It is also known that these line code formats have different power spectra. The curves 20 and 22 of FIGS. 2 and 3 illustrate the corresponding power spectra, with the ordinate representing increasing power and the abscissa representing increasing frequency in terms of 1/T, where T is the period of the pulses.

The basic impulse duobinary code contains spectrum nulls at odd multiples of the Nyquist frequency

1/2T. The nulls of the 100% duty cycle bipolar signal code occur at even multiples of 1/T. The maximum power of the duobinary signal occurs at d.c, while that of the 100% duty cycle bipolar code occurs at 1/2T. If the form of the pulse used with the duobinary code is changed to that of a dipulse, however, then d.c. is eliminated, and the power spectrum becomes that of FIG. 2, with maximum power centered over the frequency 1/T.

The main power spectrum lobes of the curves 20 and 22 of FIGS. 2 and 3 are shown in superposition in the diagram of FIG. 4. Here again, the ordinate is increasing power, while the abscissa is increasing frequency. It is noted that there is an overlapping portion of the lobes which is shown as the shaded area 24. One form of the customer unit 12 is shown in

FIG. 5 of the drawings. It includes a transmitted data section made up of a low-pass filter 30, a binary to ternary conversion unit 32 and an in-data selecting switch 34. A binary data stream enters through an input select switch labeled the in-data 34 and passes to the binary-to-ternary converter 32, where it is converted by means of a derived clock signal to a 100% duty cycle bipolar line code format transmitted signal. Next, it passes through the low-pass filter 30 to reduce energy in the opposite band and through an impedance matching resistor R to a hybrid isolation transformer 35, through which it is coupled to the wires 16, 18 for transmission to the office unit 14.

A receiving section of the customer unit 12 includes an adaptive equalizer 36, a high-pass filter 38, a demodulator 40, a low-pass filter 42, a timing extraction unit 44, a phase unit 46, an analog-to-digital (A/D) converter 48 and an out-data selecting switch 50.

There is associated with the receiving section an echo canceler section which includes a high-pass summer 52 and an adaptive transversal filter 54 with up to four parameter coefficients, and a resistor array 56. A control section, which includes a detector 58, a control unit 60, and a resistor array 62 is associated with all three of the other sections.

The duobinary dipulse received signal is coupled from the wires 16, 18 through the hybrid transformer 35 to the node 64. Both the transmitted and the received signals are present at the node 64, but the received signal is at a much lower signal level than the transmitted signal. The transmitted signal, due to its high signal level relative to the level of the received signal, readily passes into the receiving section high pass summer 52. In the high- pass summer 52 it is summed with a canceling signal generated by the adaptive transversal filter 54 in response to the transmitted signal. The canceled transmitted signal is thereby highly attenuated in that portion of its energy spectrum which overlaps the main energy lobe of the received signal as shown by the shaded area 24 in FIG. 4, although its total signal level still remains higher than that of the received signal. From the high-pass summer 52, the signals pass to the adaptive equalizer 36, in which the signal pulses are reshaped for the desired level and slope. The reshaped signals then pass through a high-pass filter 38 in which the main energy lobe of the transmitted signal echo, which does not overlap the received signal lobe, is removed. For convenience of decoding, the signal is demodulated to the baud rate by the demodulator 40 and passed through a low-pass filter 42 to remove the remains of the undesired transmitted signal. The filter 42 is low-pass because after demodulation of the received signal, the transmitted signal is left at a higher frequency than the received signal. The demodulation is carried out by means of a timing frequency for the signal which is derived by the timing extraction unit 44. This extracted frequency is also sent to a phase detector 46, which generates a derived clock signal connected to the binary-to-ternary conversion unit 32 in the transmitting section. From the low-pass filter 32, the signal passes to an analog-to-digital converter 48 and then out through the out-data switch 50.

The adaptive transversal filter 52 is made up of four delay units which each have a coefficient determined by resistors of the resistor array 56. Input to the transversal filter 54 is from the node 66, which is common to the resistor R and the low-pass filter 30 and which is signal ground for the received signal. With this input, the transversal filter 54 generates as precisely as it can a signal counterpart which will directly interfere with the transmitted signal in a destructive manner in the high-pass summer 52. With a maximum of four coefficients, the transversal filter 54 is relatively uncomplicated. Yet its precision is entirely adequate, since the signal will later pass through the high-pass filter 38 and the low-pass filter 42 for removing those portions of the remaining undesired transmitted signal which still remain after the high-pass summer.

The control unit 60 is a general purpose signal processor which is programmed to control the values of the resistors in the arrays 56 and 62 in order to set the coefficients for the transversal filter 54 and the adaptive equalizer 36. The resistor values for the resistor arrays 56, 62 could alternatively be provided by applying a least-means-square algorithm or a sign-update to the transmitted signal and coupling it to the resistor array 56 or 62. Commonly used elements and element algorithms suitable for this are readily available in the art. The adaptive equalizer 36 is of the two parameter type. It is not essential to the operation of the customer unit 12, but significantly improves the operation of the remaining portion of the receiving section by providing a reshaped signal.

The combination of the demodulator 40, the low- pass filter 42, the timing extraction unit 44, and the phase detector 46 make up a linear block which has the effect of reducing detection sensitivity for the detector 58 but which is not essential to the operation of the customer unit 12. The signal could pass directly from the high-pass filter 38 to the analog-to-digital converter 48. However, each baud of duobinary dipulse received signal includes two or more higher frequency pulses from which a decision must be made as to the presence, the absence, and the sign of the pulse at the baud rate. The demodulation and low-pass filtering eliminates these higher frequency pulses and thereby provides a signal "eye" at the node 68 which greatly facilitates detection by the detector 58. The demodulation by the demodulator 40 is readily accomplished by simply using the frequency of the extracted timing, that being the baud rate. Because of its frequency content, the duobinary dipulse signal can advantageously be treated as if it were a suppressed carrier double sideband amplitude modulated signal. After equalization and simple high pass filtering, the received signal is squared in a nonlinear device and a phase-locked loop recovers the squared 2/T "carrier". A divide-by-two circuit generates accurate 1/T clock, which is also the "carrier". The equalized received signal is demodulated by beating against the 1/T^* "carrier" to recover the base and duobinary signal. The 1/T clock also drives a slicer to recover the data. Since the 1/T clock pulses are derived by dividing the 2/T clock pulses, however, the phase is ambiguous. To resolve the ambiguity, the data samples are compared to a reference value and fed to an up/down counter. When the slicer phase is proper, the number of samples above and below the reference value remains equal. When the phase is improper, however, the count rapidly exceeds the capacity of the counter and the phase is slipped.

It is a particularly advantageous feature of the demodulator 40 that while it demodulates the received signal, it simultaneously modulates to a higher frequency whatever transmitted signal is still present, so that this undesired transmitted signal can be readily removed substantially completely by the low-pass filter 42. In essence, the demodulator 40 cohverts the dipulse signal to 100% duty cycle pulses, which are then enhanced by low-pass filtering, in which the dipulses tend to be filtered out. One form of the central office unit 14 of the data transmission system 10 is shown in FIG. 6. It has a transmitting section which includes an in-data buffer unit 70, a binary to ternary conversion unit 72, and a high-pass filter 74. Data enters through the in-data selecting switch 70 as a binary bit stream which goes to the binary to ternary conversion unit 72. There it is converted to a duobinary dipulse line code format transmitted signal which passes through the high-pass filter 74 and then through an impedance-matching resistor R to the wires 16, 18 through a hybrid transformer 76. A 100% duty cycle bipolar signal receiving section of the office unit 14 is formed by a low-pass summer 78, an adaptive equalizer 80, a low-pass filter 82, an analog digital (A/D) converter 84, and an out-data buffer 86. The A/D converter 84 receives the timing signal from a timing extraction unit 88 which is coupled to the common node of the analog digital converter 84 and the low- pass filter 82. A four-coefficient adaptive transversal filter 90 and a resistor array 92 for its coefficients are associated with the low-pass summer 78 to perform canceling of the transmitted signal. A control unit 94, a detector 96 and a resistor array 98 are associated with both the transmitting section and the receiving section.

It can be seen that, for the most part, the office unit 14 operates in a manner very similar to the operation of the customer unit 12. However, there are significant differences in their operation which account for differences in the diagrams.

Detection is relatively straightforward. A phase-locked loop derives the clock frequency 1/T, and the eye is sampled directly with the clock signal to recover the data.

Referring to the transmitting section, it is seen that the binary to ternary conversion unit 72 is driven by an external master clock rather than by timing derived- from the received signal, as was the case for the customer unit 12.- Such an external clock signal would typically be readily available within the environment of a central office. The binary-to-ternary conversion unit 72 converts the binary signal coming to it by modulating it with the external clock signal to produce a modulated dual binary signal which proceeds to the high-pass filter 74 to reduce energy in the opposite band and is then transmitted.

Referring now to the receiving section, the low- pass summer 78 adds to the received signal and the transmitted signal a canceling signal from the adaptive transversal filter 90 which greatly attenuates the overlapping spectral energy lobe portion of the undersired transmitted signal. Thereafter, the signal is reshaped to its proper level and slope by the adaptive equalizer 80, which is of a two-parameter type. From there it passes to a low-pass filter 82, where the remaining undesired high frequency transmitted signal is removed. The remaining signal continues to the A/D converter ^'84 and from there to the out-data selecting switch 86. The A/D converter 84 obtains its timing from the timing extraction unit 88, which in turn extracts its timing from the signal itself. The central office unit 14 does not include the linear block of demodulator 40, low-pass filter 42, timing extraction unit 44, and phase detector unit 46 that was found in the customer unit 12, because it is not receiving a modulated duobinary dipulse signal which can be demodulated. It is dealing instead with a 100% duty cycle pulses. The functions of the control unit 94, the detector 96 and the resistor array 98, as well as the functions of the adaptive transversal filter 90, the resistor array 92, and the low-pass summer 78 are otherwise entirely analogous to the function of the corresponding units of the customer unit 12.

The higher frequency content of the modulated duobinary dipulse signal which is transmitted by the office unit 14 to the customer unit 12 requires more amplification for equalization and has a greater bandwidth. This signal therefore has a higher susceptibility to noise than does the other signal format. For this reason it is advantageous to receive the duobinary dipulse signal at the quieter location, which would generally be the customer location. In addition, the modulated dual binary signal makes it possible by means of the demodulator 40, the low- pass filter 42, and the phase detector 46 to simplify the detection process and to make it more accurate. The 100% duty cycle bipolar signal which is transmitted by the customer unit 12 to the office unit 14, on the other hand, is somewhat less susceptible to noise interference.

The in-data and out data selecting switches 34, 50 of the customer unit 12 and the corresponding switches 70, 86 of the office unit 14 are provided to permit the control unit 60, 94 to place the circuit in a training mode in order to measure the impulse response for appropriately setting the adaptive equalizer 36, 80. This requires sending a known signal through it, one direction at a time.

It should be noted that a duobinary 100% duty cycle line code format could also be used as the signal transmitted from the customer unit 12 to the office unit 14, with largely equal performance results for the system 10. Processing of this signal would require somewhat less accurate filtering than the 100% bipolar line code format, but would require duty cycle restoration circuitry. It is therefore contemplated that various other line code formats may be used together with the modulated duobinary line code format with satisfactory performance. However, depending on which line code format is used for the other signal, there may be required either more accurate canceling or more accurate filtering, depending on how much overlap there is in the main lobes of the spectral energy of the two line code formats.

Although in the system 10 the one unit 12 is identified as a customer unit, while the other unit 14 is identified as an office unit because such an arrangement would typically be desirable due to differences in the environment of customer and offices, the roles of the two units 12 and 14 could equally well be reversed to suit a situation where the noise environment is just the opposite of that proposed above. For that matter the units 12, 14 could be assigned to a function regardless of the particular noise and environment and would still operate satisfactorily. However, it is thought that the optimum arrangement takes account of noise differences in making such assignments.

It is a particularly advantageous feature of the data transmission system 10 that different data rates can be accommodated by the system simply by changing the rate of the external clock at the central office unit 14. A data transmission system which relies entirely on cancellation for discriminating between the signals would generally need a cancellation attenuation of about 60 decibels. The canceling performed by the adaptive transversal filters 54, 90 in the units 12, 14 of the data transmission system in accordance with the present invention requires only about 30 decibels attenuation. The remaining attenuation is performed by filtering.

Claims

1. A data transmitting and receiving unit,

CHARACTERIZED BY receiver means adapted to receive a data signal in a first digital line code format; transmitter means adapted to transmit a data signal in a second digital line code format at the same baud rate as said first signal, and filtering and canceling means adapted to separate the received data signal from the transmitted data signal, the first and second line code formats selected permit some overlap in their main spectral energy lobes.

2. A unit as defined in claim 1 CHARACTERIZED IN THAT the first line code is in a duobinary dipulse format.

3. A system for simultaneously transmitting data between first and second stations coupled via wires, CHARACTERIZED BY transmitter means for transmitting data at the first station in a duobinary dipulse line code format; receiver means for receiving data at the first station in a 100% duty cycle bipolar line code format; transmitter means for transmitting data at the second station in a 100% duty cycle bipolar line code format, and receiver means for receiving data at the second station in a duobinary dipulse line code format.

4. The system as defined in claim 3 wherein the one of said units which receives the duobinary dipulse signal is

CHARACTERIZED BY means for demodulating the signal to the baud rate prior to phase detection.

5. A data transmission and receiving unit, the transmission section being. CHARACTERIZED BY means adapted to receive and convert a binary signal to a ternary signal; means for filtering out of the ternary signal frequencies above the baud rate, and transmission line impedance matching means, and means for coupling the signal to be transmitted to a pair of lines for transmission, and the receiving section being, CHARACTERIZED BY canceling means linked to said coupling means for canceling from the combined signal to be transmitted and signal received from said coupling means that portion of the signal to be transmitted which overlaps the main energy spectrum lobe of the received signal; means adapted to filter out from the combined signal after canceling the peak power spectrum portion of said signal to be transmitted; extraction means for obtaining timing information from the received signal and using the obtained timing information to demodulate the received signal to the baud rate; means for filtering from the demodulated signal frequencies higher than the baud rate, and means for converting the demodulated and filtered received signal to a binary signal.

6. The unit defined in claim 5 wherein the canceling means is

CHARACTERIZED BY a high pass summer connected to receive the combined signal and an adaptive transversal filter coupled to the signal to be transmitted, said transversal filter being connected to said high pass summer and to a control unit for providing the canceling signal to the said high pass summer.

7. A data transmission and receiving unit, the transmission section being CHARACTERIZED BY means adapted to receive and convert a binary signal to a ternary signal, means for filtering out of the ternary signal frequencies above the baud rate; transmission line impedance matching means, and means for coupling the signal to be transmitted to a pair of lines for transmission, and the receiving section being CHARACTERIZED BY canceling means linked to said coupling means for canceling from the combined signal to be transmitted and signal received from said coupling means that portion of the signal to be transmitted which overlaps the main energy spectrum lobe of the received signal; means adapted to filter out from the combined signal after canceling the peak power spectrum portion of said signal to be transmitted, and means for converting the demodulated and filtered received signal to a binary signal.

8. The unit defined in claim 7 wherein the canceling means is

CHARACTERIZED BY a high pass summer connected to receive the combined signal and an adaptive transversal filter coupled to the signal to be transmitted, to said high pass summer, and to a control unit for providing the canceling signal to the said high pass summer.