Digital Receiver for Four-Phase-Modulated Carrier

My present invention relates to a digital receiver designed to demodulate an incoming carrier wave subjected at a transmitting end to four-phase modulation.

The term "four-phase modulation" is used in the telecommunication art to describe a technique by which a sinusoidal carrier wave is selectively shifted into any of four possible phase positions relative to a reference wave of the same or a subharmonically related frequency. These phase positions represent respective pairs of consecutive bits of a data word, known as dibits, which can be decoded at a receiver by coherent phase demodulation involving a comparison with a reference wave in step with that used at the transmitter. The phase-modulated carrier can be regarded as the sum of a cosine wave and a sine wave of like frequency and of amplitudes of the same unit value .+-.1; by combining a cosine wave in phase with the reference wave and a sine wave in quadrature therewith, the resulting carrier will be phase-shifted by selective inversions of the sign of either or both constituent waves. Thus, the logical values of the first and second bits of a dibit may be respectively represented by the signs of the cosine component and the sine component, for example; conventionally, phase positions in the first, second, third and fourth quadrants (45.degree., 135.degree., 225.degree. and 315.degree.) may denote the bit pairs 11, 10, 00 and 01, respectively. The actual synthesis of the carrier at the transmitting end may be performed digitally with the aid of predetermined filter coefficients by which instantaneous samples of the cosine and sine components are logically multiplied before being added together; a circuit arrangement performing these operations is disclosed in commonly owned U.S. application Ser. No. 100,908, filed Dec. 6, 1979 in the names of Francesco Gandini et al as a continuation of an earlier application, Ser. No. 902,164, filed May 2, 1978.

The use of four-phase modulation is particularly advantageous in a telephone system in which it is desired to transmit data at high speed among participating subscribers, e.g. digital speech at 64 Kbit/sec., signalization and synchronization messages at 16 Kbit/sec., or slow video signals. Full-duplex data transmission over a two-way signal path is possible with the aid of two carriers of different frequencies. The need may occasionally arise for interchanging the transmitting and receiving frequencies since the high-frequency channel is subjected to greater attenuation whereas the lower frequency channel may be affected by a more pronounced amplitude dissymmetry. Also, a given terminal may receive carriers of higher and lower frequency from different signal links.

An important object of my present invention is to provide means in a receiver of four-phase-modulated carrier waves for insuring their efficient coherent demodulation without the need for a separate transmission of the reference wave used at the transmitting end.

Another object is to provide a demodulator for this purpose which can be used practically unchanged for several harmonically related carrier frequencies available for two-way transmission.

A receiver according to my present invention comprises a squarer which converts an incoming carrier wave into a phase-modulated square wave, a generator of a reference square wave having a frequency harmonically related to that of the incoming carrier wave, phase-modulation means with inputs connected to the squarer and to the reference-wave generator for emitting a binary sequence of a configuration determined by the relative phasing of the two squarer waves, and a source of timing pulses controlling the operation of that generator, these timing pulses having a cadence equal to a multiple of the frequency of the reference square wave. A shift register connected to the phase-demodulation means receives samples of the binary sequence issuing therefrom, these samples being taken upon the occurrence of the timing pulses and being loaded into respective stages of the shift register which is stepped by the same pulses; the number of register stages is sufficient to accommodate samples representing at least one full cycle and preferably two cycles of the reference square wave. Logic circuitry connected to the source of timing pulses and to at least some of the stage outputs of the shift register generates a reading command upon detecting certain characteristic bit groupings in the corresponding register stages which indicate optimum instants for reading the contents of the register; these stage outputs are further connected to decision means emitting a decoded dibit in response to the reading command.

According to another feature of my invention, certain stage outputs of the shift register are connected to phase-locking means controlled by the reading command for recognizing predetermined bit groupings other than these referred to above which are indicative of spurious phase shifts, the phase-locking means delivering in that instance corrective signals to the timing-pulse source for correspondingly readjusting the operation of the reference-wave generator.

The sampling of the binary sequence to be loaded into the shift register advantageously takes place at the squarer and at the reference-wave generator, thus upstream of the demodulation means which may be constituted by an Exclusive-OR gate.

The above and other features of my invention will now be described in detail with reference to the accompanying drawing in which:

FIG. 1 is a block diagram of a receiver according to my invention designed to demodulate four-phase-modulated carrier waves;

FIGS. 2, 3 and 4 show details of certain blocks depicted in FIG. 1;

FIG. 5 is a diagram showing the output connections of a shift register included in the receiver of FIG. 1; and

FIG. 6 is a set of graphs relating to the operation of the receiver.

FIG. 1 shows a transmit/receive modem at a terminal of a telecommunication system connected to a remote station via a two-way signal path 1 such as a line loop of a telephone system. This signal path terminates at a coupling network FO, including a hybrid coil and the usual band-pass filters, which forwards outgoing carrier waves from an output load 3 of a transmitting section TR and feeds incoming carrier waves by way of a lead 2 to a squarer SQ forming part of a receiving section. The transmitting section TR, which may be of conventional design, is assumed to be identical with a similar section at the remote station in which the incoming carrier is phase modulated in the aforedescribed manner to convey a succession of dibits forming part of 10-bit data words, for example. This corresponds to a recurrence rate or Baud frequency of 40 kHz for the dibits if each data word is allotted a time slot of 125 .mu.sec. Carriers modulated with these dibits by the four-phase technique, transmitted in two directions over signal path 1, may have respective frequencies of 80 and 240 kHz.

The squarer SQ converts the incoming sinusoidal carrier into a binary square wave in which amplitude peaks of opposite polarities are respectively translated into a high and low voltage. The resulting square wave is sampled by timing pulses on an output lead 5 of a clock circuit comprising a crystal-controlled local high-frequency oscillator OS working into a programmable frequency divider DI whose step-down ratio can be adjusted up or down by a corrective signal delivered to a control input thereof by an output lead 6 of a phase-locking circuit PL. The square-wave samples are fed on a lead 4 to one input of a phase demodulator MO, advantageously constituted by an Exclusive-OR (EX-OR) gate, to which a reference square wave is delivered by an output lead 7 of a local generator PO also controlled by the timing pulses on lead 5. The same timing pulses, assumed to have a cadence of 16 times the Baud frequency (i.e. 640 kHz), are supplied to a time base TE and to a stepping input of a 16-stage shift register SR whose loading input is connected to an output lead 9 of phase demodulator MO. An output lead 8 of time base TE extends to generator PO for changing its phase under conditions and in a manner described hereinafter. Another output lead 10 of time base TE terminates at phase-locking circuit PL as well as at a decision circuit DE which, in response to a reading command from that time base, detects the dibits conveyed by the incoming carrier wave and sequentially emits the bits of each pair on a lead 14. Shift register SR is provided for this purpose with a number of stage outputs connected to two four-lead multiples 11 and 12 which extend on the one hand to decision circuit DE and on the other hand to time base TE; a three-lead multiple 13 connects respective stages of register SR to circuit PL. A control lead 15, carrying a binary discriminating signal, extends to time base TE and may also have a branch connected to phase-locking circuit PL as indicated in phantom lines; the discriminating signal, designed to inform the receiver whether a low-frequency carrier of 80 kHz or a high-frequency carrier of 240 kHz is to be demodulated, may be manually generated by a nonillustrated selector switch.

In FIG. 5 I have shown the shift register SR in greater detail, along with the logical values of 16 samples stored at a particular instant in its stages SR.sub.1 -SR.sub.16. The four leads 11a, 11b, 11c, 11d of multiple 11 are shown connected to a first group of nonadjacent stages SR.sub.1, SR.sub.5, SR.sub.9 and SR.sub.13 whereas the four leads 12a, 12b, 12c and 12d of multiple 12 are connected to a second group of nonadjacent stages SR.sub.3, SR.sub.7, SR.sub.11 and SR.sub.15 interleaved with the first group. Moreover, stages SR.sub.7, SR.sub.8 and SR.sub.9 are connected to respective leads 13a, 13b and 13c of multiple 13.

According to FIG. 2, which shows details of time base TE, the two multiples 11 and 12 terminate at respective coincidence circuits C.sub.1 and C.sub.2 designed as respective EX-OR gates with negated outputs 16 and 17 extending to an AND gate AN. The output 18 of the latter is connected to a loading output of an incrementer IN which is clocked by timing pulses on lead 5 and is connected in a loop with a 16-cell memory ME via a feed-forward lead 20 and a feedback lead 19. Memory ME is addressed by a circuit CL also stepped by the timing pulses on lead 5, these pulses being further applied to a stepping input of a four-bit counter CB which measures dibit cycle of 16 timing pulses; Baud counter CB periodically energizes the lead 10 terminating at components TE and DE of FIG. 1. The discriminating signal on lead 15 reaches a presetting circuit PR whose output is the lead 8 which extends to reference-wave generator PO as also shown in FIG. 1.

The phase-locking circuit PL is seen in FIG. 3 to comprise two multipliers EX.sub.1 and EX.sub.2, also designed as EX-OR gates. Two inputs of multiplier EX.sub.1 are connected to leads 13b and 13c of multiple 13 whose third lead 13a is connected to multiplier EX.sub.2 along with a branch of lead 13c. The output 23 of multiplier EX.sub.1 is tied to a control input 24 of an up/down counter CM which has a stepping input connected to the output 22 of multiplier EX.sub.2 and an enabling input connected to lead 10; the output of this counter is the lead 6 terminating in FIG. 1 at the control input of frequency divider DI. An inverter DV, controlled by the discriminating signal on lead 15, may be inserted between the multiplier output 23 and the counter input 24 as indicated in phantom lines.

FIG. 4 shows the decision circuit DE as comprising two majority-logic networks MA.sub.1 and MA.sub.2 each having four inputs connected to respective leads of multiplier 11 or 12, these networks further having enabling inputs connected to lead 10. Their outputs 25 and 26 extend to a decoder DD which detects the dibit extracted from shift register SR (FIG. 1) and feeds the two bits thereof to a serializer PS emitting them on output lead 14.

I shall now describe the operation of the receiver of FIG. 1 with reference to FIG. 6 whose first nine graphs (a)-(i) relate to the demodulation of a four-phase-modulated low-frequency carrier of 80 kHz while its remaining graphs (a')-(i') have the same significance in conjunction with a high-frequency carrier of 240 kHz.

Graph (a) represents the reference wave emitted by generator PO which graphs (b), (d), (f) and (h) show a square wave derived from the incoming carrier with leading phases of 45.degree., 135.degree., 225.degree. and 315.degree., respectively; the remaining graphs (c), (e), (g) and (i) of this group show the voltage differences between the reference wave and the incoming carrier. Since, however, both the carrier and the reference wave are sampled in the rhythm of the timing pulses on lead 5, only intermittent segments of the waveforms shown in these graphs will actually come into existence. These waveform segments, corresponding to samples loaded into the stages of shift register SR, lie between pairs of vertical lines marked with the corresponding stage designations SR.sub.1 -SR.sub.16.

The bits indicated in graphs (c), (e), (g) and (i) represent the contents of every stage at an instant when 16 samples of a complete dibit are loaded into the shift register SR. It will be noted that at such instants, in any phase position of the carrier, the stages of each group contain identical bits, namely 0/1 in graph (c), 1/1 in graph (e), 1/0 in graph (g) and 0/0 in graph (i). These characteristic bit groupings, indicating an optimum readout time, are detected by the inverting EX-OR gates of coincidence circuits C.sub.1 and C.sub.2 in FIG. 2 whose simultaneous conduction energizes the output lead 18 of AND gate AN whereby the amount stored in the concurrently addressed call of memory ME is increased by the incrementer IN. If mutually identical bits again appear in the stages of each group at 16-pulse intervals, i.e. when the same memory cell is being addressed, the contents of that cell will eventually reach a predetermined threshold; incrementer IN then emits on its output lead 21 an alignment signal which clears the memory ME as well as the presetting circuit PR and the Baud counter CB. From that point on, counter CB sends a reading command once per dibit cycle on its output lead 10 to the counter CM of phase-locking circuit PL (FIG. 3) and to the logic networks MA.sub.1, MA.sub.2 of decision circuit DE (FIG. 4). These reading commands mark the optimum instants for the extraction of the contents of the odd-numbered stages of shift register SR. Logic networks MA.sub.1 and MA.sub.2 then emit bits on their outputs 25 and 26 which correspond to the contents of the majority of the register stages in the respctive group connected thereto; if these contents happen to be two 0's and two 1's, the bits from the centrally located stages of the respective group will be selected since these stages are least susceptible to noise or line jitter. The voltages on leads 25 and 26 are translated by decoder DD, according to the code employed, into a bit pair sequentially emitted by serializer PS.

If the carrier to be demodulated has the higher frequency of 240 kHz, the same bit combinations as in the preceding instances will be read out for like phases of that carrier provided that the reference wave is shifted as indicated in graph (a'). For this purpose, the presetting circuit PR of FIG. 2 responds to the discriminating signal on lead 15 by emitting a phase-shifting command to generator PO (FIG. 1) via lead 8.

When the phase shift of the carrier relative to the reference wave of graph (a) or (a') is such that the bits entered in the first group of register stages SR.sub.1, SR.sub.5 etc. are different from those entered in the second group of stages SR.sub.3, SR.sub.7 etc., the middle stage SR.sub.8 will lie between two stages of different bit values as seen in graphs (c), (c'), (g), (g'). In that case the sample entered in stage SR.sub.8 is taken in the vicinity of a zero crossing of the incoming carrier so that the bit stored in that stage does not necessarily have the values shown in these graphs but may alternate between 0 and 1 in response to minor relative phase shifts between the carrier and the reference wave. I utilize this phenomenon, in accordance with a feature of my invention, for modifying the step-down ratio of divider DI (FIG. 1) to the extent necessary in order to re-establish the correct relative phasing of the reference wave and the carrier. Thus, as shown in FIG. 3, the EX-OR gate of multiplier EX.sub.1 will energize its output 22 whenever the voltages of leads 13a and 13c are different from each other, thereby stepping the counter CM either up or down; the stepping direction is determined by the output 23 of multiplier EX.sub.1 which is low when the bits in stages SR.sub.8 and SR.sub.9 are identical--as actually illustrated in the graphs referred to--but is high when the voltages on leads 13b and 13c are unequal. In the first instance, the contents of counter CM are incremented in a sense tending to shift the reference wave of graph (a) or (a') to the right; in the second instance the contents of the counter are correspondingly decremented in the sense of a leftward shift. A command temporarily increasing or decreasing the step-down ratio of divider DI (FIG. 1) to this effect is emitted only when the count of component CM reaches a certain upper or lower limit; as long as the relative phase of the carrier in the positions of the aforementioned graphs does not vary significantly in one direction or the other, the operation of the divider is not modified. In the event of a spurious phase shift, however, corrective action is taken by the phase-locking circuit PL so that demodulation continues to be carried out in a virtually coherent manner.

The inverter DV indicated in phantom lines in FIG. 3 will be needed under circumstances when, contrary to what is shown in FIG. 6, a spurious phase shift of the high-frequency carrier requires a corrective action opposite that needed for the lower-frequency carrier. Such a situation could occur, for example, when the reference wave in graph (a') retained the position of graph (a) upon a changeover to the 240-kHz carrier.