FR2662889A1 - Method and device for resolving the phase ambiguity in digital transmission with coherent demodulation - Google Patents

Abstract

Method and device for resolving the phase ambiguity, without using differential coding, in digital transmission with coherent demodulation. On reception, a time base (25) is provided able, by simultaneous searching for the MVT on two of the trains generated as well as on their complements, to supply the binary number (C1, C2) of the carrier. This binary number is then applied to a logic correction stage (22) in order to obtain the correct trains received.

Description

METHOD AND DEVICE FOR LIFTING AMBIGUITY
OF PHASE IN A DIGITAL TRANSMISSION A
CONSISTENT DEMODULATION
The present invention relates to a method and a device for removing phase ambiguity in a digital transmission with coherent demodulation.

 FIGS. 1A and 1B attached diagrammatically represent a known installation for transmitting a digital multiplex on a channel using a coherent demodulation method, FIG. 1A representing the transmission part, and FIG. 1B representing the reception part.

 Referring to this figure lA, a multiplexer 1 constructs a frame on 2n digital binary trains, referenced ao, ai, ai, ... an-l, and bo, bi, ... bi, bn-l, to from incoming information from various tributaries 2 to be sent.

 This operation is synchronized by a time base providing different synchronization signals.

A frame constitutes the longest periodic pattern with respect to these synchronization signals. This frame is divided into P sectors each containing Q bits for each of the 2n bit streams leaving multiplexer unit 1.

The length of the frame is then P.Q.

On these 2n trains, one of them, ao for example, is chosen so that the instants kQ are reserved for the insertion of bits intended for synchronization. The value of ao at these times kQ is equal to Vk modulo P, where vo, vi, ... vp-1 are the binary elements of the "frame alignment word" or "MVT". This
MVT is, remember, intended to regain the synchronization of the time base on reception.

 The aforementioned 2n trains are introduced into a differential encoder 4, generally followed by absolute Gray encoders 5, 6, which converts them respectively into 2n other trains xo, xi,. . . Xit. . . Xn-l and yo, yl, ... yì, ... yn-l, which depend on the 2n incoming trains and the 2n previous trains.

 After passing through two respective digital / analog converters 7 and 8, these signals are applied to the modulator 9. The latter uses these 2n trains to modulate, according to the chosen modulation, according to 22n phase states and possible amplitude of the carrier wave at radio frequency. A modulated carrier is then obtained at 11 to be applied to the radio power transmitter itself.

The output 11 of the modulator 9 can be represented by a complex number: the phase and amplitude states of this modulator are determined from x0, x1, ... xn-1 and yo, yl, ... yn- l by the complex number
C = (xo20 + xl 21 +... + Xn-i 2n-1) + j (yo20 + yl2 +... + Yn-l2n-l)
Ci is then understood to mean the sequence of the different symbols emitted successively.

On reception (FIG. 1B), the reverse processing is carried out, the demodulator restoring the symbolic sequence C'i. This reception part therefore successively comprises a demodulator 12 to which the signal is applied at 13
received two analog / digital converters 14, 15 of which
2n trains x'i and y'i leave two Gray 16, 17 decoders followed by a decoder
differential 18 to obtain the 2n trains a'i and b'i
and
a demultiplexer 19, coupled to the time base
reception 20, which ultimately provides the
tributaries received 21.

 It is impossible, for reasons of symmetry, to reconstruct a reference to the reception having a fully known phase relation with respect to the reference of the transmission. The phase ambiguity between transmission and reception is k.X / 2 with k = 0, 1, 2 or 3, and k a priori unknown.

 We say that we have four possible carriers.

We have then
C'i = Ci ejklt / 2 with k = 0, 1, 2 or 3.

 The differential coding and decoding are designed in such a way that the restored trains a'o, ... a'n-l, b'o, ... .b'n-i are nevertheless identical to the trains transmitted ao,. . an-i, bo,. . bn-i in the absence of errors on the transmission.

 By way of example, it is assumed that this is a four-phase modulation for which we have: n = l. In this case, there is no Gray coding because it is useless, and a possible differential coding is then that which codes the phase variations of a + bj. Its inputs are a and b and its outputs are x and y. In Figure 2 attached, the states (a, b) are represented by circles and the transitions between states are labeled with the corresponding values of (x, y).

 In the case of a more complex modulation (n> l) a possible differential coding consists in carrying out a differential coding of the four phase type on the most significant bits (an-l and bn-i), and in performing an absolute coding on the remaining trains which, however, depends on the quadrant in which we are located (ie Xn-l and yn-l).

The Gray coding in this case is an absolute coding carried out separately on the two channels in quadrature X and
Y and intended to reduce the number of bit errors following a symbol error. Such coding is designed so that two symbols which are neighbors on a constellation differ in a single bit. An example of a 16-state constellation coded with a Gray code is given in Figure 3 attached.

 Gray coding is designed so that on an error corresponding to a received symbol close to the transmitted symbol, only one bit error appears. However, its effectiveness is degraded by the presence of differential coding. Indeed, a symbol error inside the differential encoder results in two erroneous transitions. There is therefore a multiplicative ratio on the bit error rate which is due to the differential coding. In other words, this differential coding has the disadvantage of having a multiplier effect on the errors.

 The invention aims to remedy this drawback.

To this end, it relates to a method and to a device for removing the phase ambiguity in a digital transmission with coherent demodulation, this method consisting in at least taking the following measures: no differential coding of the train is carried out.
transmitted a frame locking word is chosen which is not
not the complement of one of its permutations; it is carried out, unless it is unnecessary, taking into account
the simplicity of the constellation due to the modulation
adopted, a Gray coding of the transmitted train; at on reception, the frame lock word is
searched simultaneously on track a'o where it should
geographically find themselves in the absence of ambiguity of
phase, on its complementary path a'o, on its path b'o
which corresponds to its other complex component, and on
the complementary path b'o of the latter as soon as it is found on one of these four paths, we
deduces the number k of the carrier and therefore the
equations that affect the signal received due to
phase ambiguity, and we proceed accordingly,
in accordance with these equations, correcting this
signal to restore the correct trains.

In any case, the invention will be well understood, and its advantages and other characteristics will emerge during the following description of a nonlimiting exemplary embodiment, with reference to the appended schematic drawing in which
Figures 4A and 4B are general block diagrams
transmission and reception parts of an installation of
digital transmission implementing the invention
Figures 5A and 5B are the transcripts of the diagrams
according to FIGS. 4A and 4B in the simplest case of a
four-phase modulation
Figure 6 is a diagram of the simple device of
carrier correction for the installation according to
Figures 5B; and Figure 7 shows the word search device of
frame alignment and carrier determination
which equips the time base of an installation according to
Figure 5B.

 The transmission part shown in FIG. 4A is practically identical to that of FIG. 1A, with the essential difference that it does not include the differential coding device 4.

 In this example, it is for example assumed that the MVT is registered in the path ai. Of course, it could be carried by any other route.

The invention is based, in principle, on the following remark:
The phase ambiguity is due to the symmetry, along the two orthogonal axes of the complex plane, of the constellation which represents the modulation considered and to the central symmetry (the aforementioned FIG. 3 represents for example a constellation with 16 phase states in the complex plan
OXY). The latter explains the phase ambiguity of kX / 2, with k between 0 and 4.

If it is assumed that the frame alignment word has been introduced in a and that the connection is error-free, then, having regard to the linear character of the Gray coding, we have the following relationships
If a'o = ao then k = O
If b'o = ao then k = l (I)
If a'o = ao then k = 2
If b'o = ao then k = 3 searching for the frame locking word in one of the 4 auto trains, b'o, viola, b'o, therefore makes it possible to determine the value of k.

If we can determine k, i.e. the phase rotation, we can then easily find, from the signals received a'i, b'i (i = O, 1, ..., n-1 ), the signals emitted a "i, b" i by application of the following equations, deduced from the above, and valid, in the absence of prohibitive errors, for all the index values (a "o, b" o to a "ni, b" ni)
if k = O: a "i = a'i and b" i = b'i
if k = l: a11 = b'i and b "i = a'i (II)
if k = 2: a "i-ati and b" i = bti
if k = 3: a "i = b'i and b" i-ali with i = 0.1, ..., nl
According to the invention, the MVT is searched for on reception simultaneously in the four channels received a'o, b'o, a'o, and b'o, where this MVT can be found, which gives, when 'it was found in one of these four routes, the carrier number k by virtue of the preceding relations (I). We then find, knowing this number k, the trains received correct by applying the above equations (II).

As a consequence of the above, the reception part according to FIG. 4B comprises
. elements 12, 14, 15, 16, 17, 19 identical to those of
Figure 1B;
. no differential decoder, but a logical body 22
carrier correction according to equations
(II) above, which is inserted between the decoders of
Gray 16, 17 and the demultiplexer 21 and which receives, on
two control inputs 23, 24, two binary symbols
Co, C1 which come from the reception time base 25
and which represent, in binary digit, the number k of
the carrier
. this reception time base 25, which is connected to the
demultiplexer 19 to receive in particular the
values a "i and b" i, and which deduces the values Co and Ci
above when, after simultaneous search for MVT on
the four tracks a "o, b" o, a'' omit b "o, this MVT was
found, except too high error rate, on one
of these four paths.

 Reference will now be made to FIGS. 5A, 5B and 6, which illustrate an example of simple carrier correction, since it is applied to a modulation with four phase states.

 In such a case (FIGS. 5A and 5B), the Gray coders and decoders and the digital / analog and then analog / digital conversions are obviously not necessary, since the coding with 4 phase states corresponds by itself to a coding of Gray.

 The carrier corrector is represented in FIG. 6. I1 comprises two “four-in-one” multiplexers 26 and 27 which each receive on the one hand the number k of the carrier, coded in binary according to the levels co and ci coming from the reception time base, and on the other hand (the MVT is assumed here to be on transmission on channel a) the levels a ', b' coming from the demodulator 12 and their complementary a'and b 'obtained at using logic inverters 28 and 29. It is then easily verified, by the above equations (II), that the signals a "and b" actually obtained at the output of these multiplexers 26 and 27 are indeed the correct signals.

 Finally, reference will be made to FIG. 7 which represents the device, integrated into the reception time base 25 of FIGS. 4B or 5B, for seeking synchronization and determining the carrier number k represented therefore1.

This device receives
the values a ", a", b ", b",
. the clock-bit H provided by the time base.

 Q being the length of a frame sector, the pulses H are applied to four counters by Q, 30 to 33, from each of which comes out, normally once every Q clock-bit, i.e. the sector frequency, a pulse (at 34 to 37 respectively) for triggering an open shift register (38 to 41 respectively) which contains as many flip-flops as the MVT contains binary symbols 1 or O.

 These four shift registers receive on their respective inputs 42 to 45, the series of levels which correspond respectively to: a ", a", b ", and b".

The multiple outputs, respectively 46 to 49, of the flip-flops of these shift registers 38 to 41 are applied to a logic member, respectively 50 to 53, known per se, for comparison with the MVT, the output of which
(respectively 54 to 57) is applied to a logical organ
(respectively 58 to 61) for detecting loss and resumption of locking, this logic member also being of known type.

 In a very conventional manner, the phase of the clock clock applied, respectively at 34 to 37, to each shift register 38 to 41, is, for the search for synchronization or the detection of loss of synchronization, shifted each time by bit by application, respectively at 62 to 65, of an inhibition signal from the respective counter 30 to 33 during a bit-clock period, which is applied by the respective search circuit, 58 to 61.

 Note that these MVT search circuits are quite conventional, but here they are used simultaneously on four channels instead of one.

 When the MVT, in the absence of a prohibitive error rate on the link, has been found (a number d of successive times sufficient to be certain that it is indeed the MVT) on one of these four channels, the corresponding circuit 58 to 61 then sends, on a first output (66 to 69 respectively) a setting pulse to 1, or "set", of a flip-flop (respectively 70 to 73) initially at zero.

 These four flip-flops attack in parallel a "four-in-two" multiplexer 74, which provides on its two outputs 75, 76, the aforementioned binary number [co, ci] which gives the position of flip-flop 70, or 71, or 72 , or 73 which is in state 1, and therefore the number k sought.

 Of course, the search circuit, for example 59, which found the MVT then classically continues its investigation to possibly detect, after successive times, the possible loss of this MVT: in such a case, the corresponding flip-flop is reset to zero, by applying a "reset" signal to its reset terminal (respectively 77 to 80), and the MVT search process then starts again simultaneously on the 4 channels.

Finally, it should be noted that if the MVT is not changed, the time to
synchronization is not increased compared to
sector synchronization search process of
prior art. that it is important not to choose an MVT which is the
complementary to one of its permutations: in a
such case, we could not differentiate the carriers,
and the method of the invention could not be implemented
artwork.

 As it goes without saying, the invention is not limited to the embodiment which has just been described: it is capable of being implemented in many other equivalent forms, or even more improved.

Claims (4)

 1 - Method for removing the phase ambiguity in a digital transmission with coherent demodulation, using an amplitude modulation of two carriers in quadrature, characterized in that it consists in at least taking the following measures. there is no differential coding of the train  issued (11). a frame locking word is chosen which is not  it is not the complement of one of its permutations, unless it is useless taking into account the  simplicity of modulation, Gray coding of the train  issued ;; upon reception the MVT is searched simultaneously on  two trains generated and their complementary  as soon as it is found on one of these four routes, we  deduces the carrier number (k) and therefore the  equations (II) that affect the signal received due  phase ambiguity, and we proceed accordingly,  in accordance with these equations (II), with the correction of this  signal to restore the correct trains.  2 - Device for implementing the method according to claim 1, characterized in that its reception part comprises a time base (25) capable of searching for the word  frame alignment on these four channels, and at  provide, when found, on one of these routes,  the binary number (k represented by co and ci) of this  track; and a logic carrier correction member (22) which  receives this binary number (k represented by co and ci) and  which restores accordingly, according to the equations  (II) above, the trains received correct.  3 - Device according to claim 2, applied to a modulation with four phase states, characterized in that the carrier corrector (22) comprises two inverters (28, 29) to obtain the complementary (a'b '), channels (a ', b') to be corrected, and two four-in-one multiplexers (26, 27) which each receive these four values (a ', b', a'b '), as well as this binary number (Co C1 ) to extract the corresponding corrected values accordingly (a ", b").  4 - Device according to claim 2 or claim 3, characterized in that this time base (25) comprises. four circuits (30, 38, 50, 58; 31, 39, 51, 59; 32,  40, 52, 60; and 33, 41, 53, 61) as well as  detection of loss and recovery of the word  frame alignment, these four circuits operating  in parallel and each according to a mode known per se four flip-flops (70 to 73) which are  respectively associated with each of these circuits, and  which are respectively set to state 1 by the circuit  correspondent when he found the lock word  of frame, and respectively to the zero state when it does not  did not find it or when it declared it lost; and a four-in-two multiplexer (74) which receives the  outputs of these four flip-flops (70 to 73) and which provides  accordingly on its two outputs (75, 76) the number  binary (co ci) which corresponds to the rocker which was  state 1.