 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a method of measuring echo delay, used in an
echo-cancelling arrangement incorporated in a transceiver arrangement to
cancel an echo signal occurring in the receive path in response to a
signal supplied to the transmit path and consisting of a substantially
undelayed near echo and a delayed distant echo, this echo cancelling
arrangement comprising a near echo canceller receiving a signal from the
transmit path and a distant echo canceller receiving said signal from the
transmit path with a delay substantially equal to the measured distant
echo delay.
2. Description of the Prior Art
It is known that echo cancellers are adaptive devices which are
incorporated in, for example, data transmission modems connected to a
two-way transmission circuit in order to cancel automatically undesirable
echoes occurring in the one-way receive path of the modem in response to
the signal supplied to the one-way transmit path of the modem.
Conventional echo cancellers are generally designed to cancel echo signals
which are not delayed or relatively little delayed in relation to the
transmitted signal and which may occur during a time interval of 40-50 ms
following the instant of originating the transmitted signal. In practice,
these characteristics are sufficient to cancel the echoes occurring on all
national and international terrestrial circuits.
However, international communications are being conducted more and more via
communication satellites. In a circuit of this kind, including a satellite
link between two radio-relay stations, it is possible that there will
occur in the receive path of a modem a echo which is not or little delayed
and is generated in the part of the circuit preceding the satellite link,
as well as a distant echo which is generated in the part of the circuit
after the satellite link and which is therefore subject to a considerable
delay .tau., depending particularly on the wave-propagation time in the
satellite link. Depending on whether the satellite is geostationary or not
and on the variation in the terrestrial link, it can be estimated that in
the international switched network the delay .tau. of the distant echo may
vary between approximately 220 and 630 ms.
To cancel simultaneously the near echo and the distant echo, which each
have a relatively short duration of the order of 10 ms or several tens of
ms but which are separated by a large time interval of the order of the
delay .tau., it is an advantage to use the configuration described above
and known from the article by Stephen B. Weinstein, entitled "A Passband
Data Driven Echo Canceller for Full-Duplex Transmission on Two-Wire
Circuits" and published in the journal IEEE Transactions, Vol. COM-25, No.
7, July 1977, pp. 654-666. This configuration comprises a section for
cancelling the near echo, a delay line simulating the delay .tau. and
connected thereto a section for cancelling the distant echo. This
configuration makes it possible to use two conventional echo cancellers of
reasonable complexity, but necessitates measurement of the delay .tau. of
the distant echo in order to introduce this information into the delay
line. In the above-mentioned article it is proposed, in order to measure
this delay, to apply to the transmit path a pulse formed by a 1000 Hz
sinusoidal signal lasting several ms and to determine the resulting
distant echo delay with a moving-window power detector. With this method,
the signal used to measure the echo delay is therefore only transmitted
for a short time, which is a disadvantage in satellite communications for
which the transmission of energy has to be permanent in a communication
path in order to maintain that path.
SUMMARY OF THE INVENTION
For measurement of the delay .tau. of the distant echo, the present
invention provides an entirely different method in which, in order to
generate the echo, use is made of a particular data sequence which can be
transmitted with the same type of modulation as that subsequently used to
transmit the useful signal; at the receiving end this method can be easily
implemented digitally and yields the delay with a perfectly defined
precision which can be reduced to the right value desired.
According to the invention, this method of measuring the delay .tau. of the
distant echo comprises the following steps:
transmission, by modulation of a carrier, of successive data sequence each
having a periodic autocorrelation function with sidelobes of negative or
zero value, and having a duration at least equal to the maximum value of
the delay to be measured and a number of data elements determined by the
precision desired for measurement of the delay;
after each transmitted data sequence, calculation of the correlation
function between the data of the next transmitted sequence and the data
recovered by demodulating the received echo signal resulting from the
preceding sequence and storing successive samples representative of said
correlation function in a memory;
after calculation of said correlation function, processing of the stored
samples consisting, after elimination of samples which may depend on the
near echo, in determining the sample corresponding to the maximum value of
said correlation function and calculating the delay of the distant echo
from the rank of such sample among all samples of the correlation
function.
Features of the invention will be more fully appreciated from the following
description of an exemplary embodiment when considered in conjunction with
the accompanying.
DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a data transceiver or modem which includes an echo cancelling
circuit to which the method according to the invention may be applied;
FIG. 2 shows diagrams illustrating the time sequence of the steps of the
method according to the invention;
FIG. 3 is a diagram of correlation function calculating circuit 18 in FIG.
1 for implementing the second step of the method according to the
invention;
FIG. 4 is a diagram of the elementary correlator circuit 31 in the
correlating function calculating circuit shown in FIG. 3;
FIG. 5 shows time-sequence diagrams which illustrate the processing
effected during the delay calculating step of the method of the invention;
FIG. 6 is a diagram of a circuit for calculating the echo signal delay from
a correlation function modulus sample signal V.sub.2 as shown in FIG. 5(b)
.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The modem provided with an echo-cancelling arrangement as shown in FIG. 1
comprises a transmit path 1 and a receive path 2 coupled to a two-way
transmission path 3 via a (hybrid) coupling circuit 4.
The transmit path 1 comprises an encoder 5 which converts the binary data
to be transmitted, originating from a terminal which is not shown, into a
baseband signal B which is generally complex and whose phase and amplitude
may change in value, as a function of the binary data to be transmitted,
at a clock frequency H supplied by a clock generator 7. The complex signal
B is applied to a circuit 50 in which its phase is incremented at each
period 1/H by the phase variation (during this period 1/H) of the carrier
used for transmitting. During the transmission of useful data this phase
variation of value .DELTA..rho..sub.1 is applied to circuit 50 via a
switch 51 set to position r. The complex signal D supplied by circuit 50
and thus subjected to the phase variations of the carrier is applied,
during the transmission of useful data, to a band-pass filter 6 for
complex signals via a switch 17 also set to position r. The passband of
filter 6 is centered on the frequency of the carrier used for the
transmission of useful data and this filter 6 thus supplies the analog
modulated carrier signal which is applied to the transmit access of hybrid
coupling circuit 4. The modulation rate of the carrier is determined by
the clock frequency H. In the case of, for example, a standardized modem
using eight-phase modulation, the modulation rate is 1600 Baud and the
frequency of the carrier is 1800 Hz.
At the receive access of hybrid coupling circuit 4 there should only appear
a carrier signal received from transmission path 3 and modulated with data
originating from a remote modem (not shown). This receive access is
connected via a low-pass filter 8 to a demodulator 9 which recovers the
baseband data signal generated in the transmitter of the remote modem.
Demodulator 9 is connected to a decoder 10 which effects the inverse
operation of that effected in the remote modem by an encoder corresponding
to encoder 5, thereby restoring the data bits of the baseband signal, and
provides them to a terminal which is not shown.
In fact, when a useful signal is transmitted towards a remote modem (not
shown) via transmit path 1, undesirable echo signal, caused particularly
by imperfections in the two-wire/four-wire coupling circuits of the
transmission path 3, may appear at the receive access of the coupling
circuit 4 and interfere seriously with the restoration of the data in
receive path 2. As has been explained, when the transmission path 3
includes a satellite link, the undesired echo signal may simultaneously
incorporate a non-delayed near echo generated between the local modem and
the satellite link and a distant echo generated between the satellite link
and the remote modem. These two types of echo are of substantially the
same length, with a maximum of several tens of ms; but the distant echo
has, in relation to the near echo, a delay .tau. which may vary, for
example, between 220 and 630 ms.
To effect economically the cancellation of an echo signal formed by a near
echo and a distant echo it is possible to use the configuration shown in
FIG. 1. This configuration includes a near echo canceller 11, comprising
in particular a memory 11-1 having the function of a delay line, a memory
12 having the function of a delay line, and a distant echo canceller 13,
comprising in particular a memory 13-1 having the function of a delay
line. The three delay lines 11-1, 12 and 13-1 are connected in cascade and
they receive the transmitted data signal subjected to the phase variations
of the carrier by circuit 50. It may be assumed for the sake of
simplification that the sampling frequency of signal D applied to
configuration 11, 12 and 13 is equal to the clock frequency H fixing the
modulation rate. In fact, in the event that the sample frequency is a
multiple of frequency H, it is known that the signal D may be considered
as being distributed in time over several identical configurations each
operating in the same manner with the sample frequency H.
The near echo canceller 11 comprises a calculation circuit 11-2 which forms
the weighted sum of the samples of signal D, stored in the delay line
11-1, with weighting coeffcients determined in a control circuit 11-3.
Delay line 11-1 produces a delay .tau..sub.1 which is substantially equal
to the duaration of the near echo. Delay line 12 produces a delay which is
substantially equal to .tau.-.tau..sub.1 so that the samples of signal D
arrive at the input of the delay line 13-1 with a delay which is
substantially equal to the delay .tau. of the distant echo. The distant
echo canceller 13 comprises a calculation circuit 13-2 which forms the
weighted sum of the delayed samples of signal D, stored in delay line
13-1, with weighting coefficients determined in a control circuit 13-3.
Delay line 13-1 produces a delay .tau..sub.2 which is substantially equal
to the maximum duration of the distant echo and which is of the same order
of magnitude as .tau..sub.1.
Memories 11-1, 11-2 and 13-1, 13-2 serve as transversal filters whose echo
correcting output signals .epsilon..sub.p (for the near echo) and
.epsilon..sub..rho. (for the distant echo) are applied to an adder 14. The
signal .epsilon..sub.p +.epsilon..sub..rho. leaving adder 14 is applied to
the (-) input of a difference circuit 15. This difference circuit 15 is
inserted via its input (+) and its output in receive path 2, between
hybrid coupling circuit 4 and filter 8. The output signal e of difference
circuit 15 is also applied to the two control circuits 11-3 and 13-3 in
order to adjust the coefficients of the transversal filters of the near
echo canceller 11 and the distant echo canceller 13. When these
coefficients are suitably adjusted, the correcting signal .epsilon..sub.p
supplied by the filter of near echo canceller 11 is practically equal to
the near echo signal .epsilon..sub.p which appears on the receive path 2
and the correcting signal .epsilon..sub..rho. supplied by the filter of
distant echo canceller 13 is practically equal to the distant echo signal
.epsilon..sub..rho. which likewise appears on the receive path 2. The
effect of this is that, in the output signal e of difference circuit 15,
the received signal .epsilon..sub.p +.epsilon..sub..rho. resulting from
the near and distant echoes is practically cancelled. The adjustment of
the filter coefficients of echo cancellers 11 and 13 may be effected, for
instance by successive iterations in accordance with a gradient algorithm,
so as to minimize the mean square value of the difference signal e.
However, for this known configuration to operate correctly in this fashion,
it is necessary for the distant echo canceller 13 to act on a transmitted
signal D which has undergone the same delay .tau. as the received distant
echo. The present invention provides a simple and efficient method for
measuring this delay, which varies with the path for the distant echo.
This method comprises steps which occur in a time sequence in accordance
with the diagrams in FIG. 2. Diagram 2a refers to the operations carried
out in transmit path 1 using steps I and II; diagram 2b refers to
operations carried out on the basis of signal D in transmit path 1 and
signal R in receive path 2 during step II; diagram 2c refers to processing
effected during step III.
Step I, which extends from instant t=0 to instant t.sub.1 (see diagram 2a),
consists in transmitting via transmit path 1, by modulation of a carrier,
a data sequence with a duration T having a periodic autocorrelation
function with sidelobes whose value is negative or zero. For the
transmission of this sequence the same carrier can be used as that used in
the useful-data transmission mode, the switches 51 and 17 then occupying
position r. However, as will be explained below, it may be an advantage in
the transmission of this sequence to use another carrier with a different
frequency. In that case the switches 51 and 17 are set to position t. In
this way, the phase of the baseband signal B is incremented in circuit 50
by the phase variation .DELTA..rho..sub.2 of this other carrier during a
period 1/H, while the signal D thus obtained is applied to band-pass
filter 16 for complex signals whose passband is centred on the frequency
of this other carrier. The duration T of the transmitted data sequence is
at least equal to the maximum value of the delay .tau. to be measured. The
number of elements in this sequence is determined by the precision desired
for the measurement of the delay. As will be explained clearly below, the
precision of the measurement of the delay is determined by the duration of
each element in the sequence.
Step II is performed in a circuit 18 of FIG. 1 which receives the baseband
signal B provided by encoder 5 and the signal R received in receive path
2. During this step II a succeeding data sequence identical to the prior
data sequence transmitted during step I is transmitted in the same manner
from instant t.sub.1 to instant t.sub.2 (see diagram 2a). At the same
time, during the duration T of this succeeding transmitted sequence (see
diagram 2b), the correlation function between the baseband data in the
succeeding transmitted sequence and the received data formed by
demodulating the echo signal received in receive path 2 from the
transmitted prior data sequence is calculated. Successive samples of this
correlation function are stored in a memory. As shown schematically in
diagram 2b, there may be found among such stored samples two particular
samples with a greater amplitude than the others, namely a sample EC.sub.p
due to the near echo and near instant t.sub.1 and a sample EC.sub..rho.
due to the distant echo and occurring at an instant which is shifted by
the delay .tau. to be measured in relation to instant t.sub.1.
Step III may be performed by a circuit such as 19 of FIG. 1 immediately
after step II, from instant t.sub.2 to instant t.sub.3 (see diagram 2c).
It consists in a processing of the successive samples of the correlation
function stored in the memory during step II, in order to determine, after
elimination of samples which may depend on the near echo, the sample
corresponding to the maximum of the correlation function; the rank of this
sample, i.e. in practical terms its location in the memory, then provides
information characterizing the delay .tau. of the distant echo. It will be
readily appreciated that the precision of measurement of the delay .tau.
depends on the duration of each correlation-function element, i.e. in fact
on the duration of each element in the transmitted sequence.
The information characterizing the delay .tau. is used in delay line 12 of
the echo cancelling arrangement to delay accordingly the data signal D
applied to the distant echo canceller 13. To define the length of delay
line 13-1 of this echo canceller the precision of measurement of the delay
.tau. has to be taken into account.
It will now be shown how the various steps of the method according to the
invention can be performed. To help clarify the picture, it will be
assumed by way of example that in the modem associated with the echo
cancelling arrangement the useful data are normally transmitted using
eight-phase modulation, with an 1800 Hz carrier and a modulation rate of
1600 Baud.
For the sequence to be transmitted during steps I and II a first
possibility consists in using sequences known as maximum-length sequences.
The possible lengths in terms of binary elements are N=2.sup.N -1, n being
an integer. These sequences consisting of +1 and -1 bits have a periodic
autocorrelation function whose main lobe has the value N and all the
sidelobes have the value -1. With a binary sequence of this kind applied
to encoder 5, the baseband signal b is real and has a phase of 0.degree.
or 180.degree. according as the bits in the sequence are of the value +1
or -1. If use is made of the 1800 Hz carrier for the useful-data
transmission mode, switches 51 and 17 will be set to position r so that
the data signal D delivered by circuit 50 will be subject to the phase
variation .DELTA..tau..sub.1 of the carrier during the modulation period
of 1/1600 s and this signal D will be applied to filter 6. However, as
will be seen below, it is more advantageous to use a carrier frequency
which is equal to the modulation rate, namely 1600 Hz. In that case
switches 51 and 17 are set to position t. For this carrier frequency the
phase variation .DELTA..rho..sub.2 to be used in circuit 50 is 360.degree.
and its action on the baseband signal B may be omitted; the data signal D
is applied to filter 16.
Another possibility with regard to the sequence to be transmitted consists
in using multiphase data sequences whose periodic autocorrelation function
has sidelobes which are all zero. These sequences are referred to in, for
example, U.S. Pat. No. 3,099,796 and in an article by D.C. CHU entitled
"Polyphase code with good periodic correlation properties", published in
the journal IEEE Transactions on Information Theory for July 1972. A
sequence of this kind is formed by complex data and constitutes the
baseband signal B which, as has been explained, is applied to circuit 50
to be subjected to the phase variation of the carrier during a modulation
period. It has been possible to verify, for example that with eight-phase
modulation, sequences with a length N=64 having a periodic autocorrelation
function with zero sidelobes can be obtained; with four- or two-phase
modulation, sequences of this kind with lengths N=16 or 4 can be obtained.
The total duration T of the sequences to be transmitted (of either of the
two types given above) must be at least equal to the maximum value of the
delay .tau. of the distant echo, e.g. 630 ms. In measuring this delay
.tau. it may be possible to make do with an accuracy of e.g. +5 ms, which
means that each element in a sequence (of either type) must have a
duration of 10 ms. It will, for example, be possible to use a
maximum-length sequence of 63 elements or a multiphase sequence of 64
elements. These elements will then have to be transmitted at a rate of 100
Hz, i.e. a rate which is 16 times lower than the normal rate of 1600 Hz.
For this transmission at a rate of 100 Hz of the elements in sequences
used to measure the delay .tau., it is an advantage, in order to use the
same circuits in the transmit path as in normal operation, to have the
transmit path operate at the normal rate of 1600 Hz, repeating the same
value for each element in these sequences 16 times.
Circuit 18 which performs step II of the method according to the invention
may be embodied in accordance with the diagram of FIG. 3. Circuit 18
receives the signal R appearing at the receive access of coupling circuit
4 on the one hand and on the other the baseband signal B of the transmit
path which, during step II, is constituted by a sequence of one of the
types described above.
It is first necessary to demodulate the received signal R. Non coherent
demodulation can be effected with the aid of the in-phase transmission
carrier and the quadrature-phase transmission carrier. In the case that
the transmission carrier used for transmission of the delay-measurement
sequence has a frequency equal to the modulation rate (H=1600 Hz in the
example chosen), the demodulation of signal R can be effected in a
particularly simple manner as shown in FIG. 3. The received signal R is
applied to two sampling circuits 20 and 21, one of which, 20, is operated
by a sampling signal E having the same frequency (1600 Hz) and phase as
the transmission carrier and the other of which, 21, is operated by a
sampling signal E' having the same frequency as E, with a phase shift of
90.degree. corresponding to a time shift of 1/(4.1600) s. The sampling
effected in circuits 20 and 21 at the frequency of the carrier comes down
to carrying out a demodulation operation. Connected to the outputs of
sampling circuits 20 and 21 are low-pass filters 22 and 23, intended to
eliminate components with frequencies higher than the fundamental
frequency of the spectrum of the transmitted data sequence (1600 Hz in the
sample chosen). At the output of the two filters 22 and 23 there are thus
obtained, with a sample rate equal to the modulation rate H=1600 Hz, the
in-phase and quadrature-phase components of the received signal
transferred to the baseband. These components are applied to sampling
circuits 24 and 25 which are operated by two sampling signals S and S'
having a same frequency and being shifted in relation to each other, as
are the sampling signals E and E'. The sampling frequency determined by
the sampling signals S and S' should be at least twice the rate of the
data in the transmitted sequence. In the example chosen, in which this
data rate is 100 Hz, a sampling frequency of 200 Hz may be used.
At the output of the two sampling circuits 24 and 25 there are thus
obtained the two in-phase and quadrature-phase components R.sub.p and
R.sub.q, together forming a complex signal sampled with a frequency of 200
Hz. It is now necessary to calculate the correlation function between this
complex signal and the baseband signal B of the transmit path, constituted
by a maximum-length sequence or a multiphase data sequence. In the latter
case, signal B is composed of complex data and calculation of the
correlation function involves multiplications of complex terms whose real
and imaginary parts are different from zero. In the case of a
maximum-length sequence, signal B comprises real data with values of +1
and -1 and calculation of the correlation function only involves
multiplications of real terms. For simplicity's sake, only the case in
which a maximum-length sequence is used will be considered in the rest of
the description of FIG. 3. Signals R.sub.p and R.sub.q are then applied to
two correlators 26 and 27 which have the same configuration and which each
receive the baseband signal B. For the reasons explained above the
practical case will be considered in which this signal is a data sequence
formed of 63 elements each having a duration of 10 ms.
In correlator 26, samples of signal R.sub.p at 200 Hz are applied directly
to a sampling circuit 28 and via a delay circuit 29 to a sampling circuit
30. The delay circuit 29 produces a delay equal to (1/200) s; the two
sampling circuits 28 and 30 are operated by a sampling signal h with a
frequency of 100 Hz which is synchronous with the sampling signal s so
that these sampling circuits 28 and 30 respectively supply the even
samples R.sub.p (p) and odd samples R.sub.p (i) of the signal R.sub.p at a
rate of 100 Hz. These even and odd samples are respectively applied to
elementary correlators 31 and 32. The latter also receive the baseband
signal B and use a memory 33 in common. The configuration of elementary
correlators 31 and 32 will be described below with the aid of FIG. 4. As
will emerge from that description, during step II extending from instant
t.sub.1 to instant t.sub.2, elementary correlator 31 forms 63 even samples
of the correlation function between the baseband signal B and the even
samples of signal R.sub.p. During this step II, elementary correlator 32
forms 63 odd samples of the correlation function between the baseband
signal B and the odd samples of signal R.sub.p. The 63 even samples formed
in correlator 31 are available at instant t.sub.2 at even addresses in
memory 33 and the 63 odd samples formed by correlator 32 are available at
instant t.sub.2 at odd addresses in memory 33. Under the control of a
read-out signal .rho. for memory 33, the 126 samples stored in memory 33
can be made to appear in series at the output of this memory, the even and
odd samples being interlaced. These 126 samples represent the in-phase
component C.sub.p of the wanted correlation function, with a sample rate
of 200 Hz.
The correlator 27 which processes signal R.sub.q is formed of elements 29'
to 33' which are identical to elements 29 to 33 of correlator 26 and
connected in the same fashion. Sampling circuits 28' and 30' are operated
by a sampling h' which is synchronous with sampling signal s'. The
operation of correlator 27 is identical to that of correlator 26 and under
the control of a read-out signal .rho. there can appear in series, at the
output of memory 33', 126 samples which are representative of the
quadrature-phase component C.sub.q of the wanted correlation function,
with a sample rate of 200 Hz.
Memories 33 and 33' in which the samples of components C.sub.p and C.sub.q
are formed during step II are read out, preferably at a high speed, by the
same read-out signal at the end of this step II. The corresponding samples
are applied simultaneously to a calculation circuit 34 which calculates
the modulus C of the correlation function, i.e. the quantity
.sqroot.C.sub.p.sup.2 +C.sub.q.sup.2. It is generally possible to make do
with approximate values of this quantity, calculated with the aid of
well-known methods. The 126 successive samples of the modulus C of the
correlation function formed at the sample rate of 200 Hz, are stored in
locations of a memory 36 in order to be used in step III of the method
according to the invention. Given that the read-out of samples in memories
33 and 33' and the calculation of the 126 samples of modulus C in circuit
33 can be performed very quickly, it will be assumed henceforth that this
step III begins practically at instant t.sub.2.
Elementary correlators 31, 32, 31' and 32' are identical and can be
designed, for example, in accordance with the diagram of FIG. 4. In this
FIG. 4, elementary correlator 31 is shown as using part 33.sub.p of memory
33. This elementary correlator 31 uses a 63-element memory 60, which may
be common to the other elementary correlators. In this memory 60 are
written the successive elements of the baseband signal B which appear at a
rate of 100 Hz. Memory 60 is arranged so as to supply at its 63 outputs
B.sub.1 -B.sub.63 the last 63 written-in elements of signal B. The
elements of signal B appearing at these outputs B.sub.1 to B.sub.63 are
applied to an input of multipliers M.sub.1 to M.sub.63. At another input,
all of these multipliers receive a same signal constituted by the even
samples of the signal R.sub.p, i.e. R.sub.p (p). The products formed at
the outputs of multipliers M.sub.1 to M.sub.63 are led to an input of
adders A.sub.1 to A.sub.63 via gates P.sub.1 to P.sub.63. These gates are
controlled by a signal AC and are only conductive for the duration of step
II extending from instant t.sub.1 to instant t.sub.2. Adders A.sub.1 to
A.sub.63 are connected to memory elements m.sub.1 to m.sub.63 so as to
form 63 accumulators in which the products formed at the output of
multipliers M.sub.1 to M.sub.63 during the period extending from instant
t.sub.1 to instant t.sub.2 are accumulated. The memory elements m.sub.1 to
m.sub.63 are reset to zero by a reset pulse RAZ occurring at instant
t.sub.1. Thus, starting from instant t.sub.2, there are available in the
memory elements m.sub.1 to m.sub.63 the 63 even samples of the correlation
function between signal B and the even samples of signal R.sub.p. These
elements m.sub.1 to m.sub.63 together form part 33.sub.p of memory 33,
which is associated with elementary correlator 31. In part 33.sub.i of
memory 33 it is possible to form in the same manner, with the aid of
elementary correlator 32 (not shown in FIG. 4), the 63 odd samples of the
correlation function between signal B and the odd samples of signal
R.sub.p. As already explained, these even and odd samples available at
instant t.sub.2 in memory 33 can be made to appear in series and
interlaced at the output of memory 33 by means of the read-out signal
.rho. in order to form samples at a rate of 200 Hz of the inphase
component C.sub.p of the wanted correlation function.
Step III of the method according to the invention performed in circuit 19
is a processing of the samples stored in memory 36 in order to deduce from
them the delay .tau. of the distant echo.
In the case that a multiphase data sequence has been used as the
transmission side, this processing operation can be particularly simple;
in fact, the character of the periodic autocorrelation function for this
type of sequence implies that all the samples stored in the memory have,
apart from the noise, a value zero, excepting those generated by the near
and distant echoes. Processing during step III (from instant t.sub.2 to
instant t.sub.3) may then consist of the following operations:
read-out from memory 36 of the samples C which are calculated in step II
from instant t'.sub.1 to instant t.sub.2 (see FIG. 2b), t'.sub.1 being the
instant up to which, at the most, the samples C generated by the near echo
can extend;
determining among the samples C read out, which therefore include only the
contribution of the distant echo, the sample with maximum value which is
generated by the distant echo;
determining the delay time .tau. of the distant echo as a function of the
rank of the sample with maximum value.
In the case that a maximum-length pseudo-random sequence length has been
used at the transmission side, the processing of sample C is slightly more
complicated by the fact that the periodic autocorrelation function of this
type of sequence has side-lobes which are not zero. The problem which this
case presents and the means of resolving it are explained with the aid of
time diagrams in FIG. 5.
Diagram 5a shows schematically signal V.sub.1 representing the correlation
function which would be obtained during the time interval (t.sub.1,
t.sub.2) of step II if the received signal R were demodulated with a
recovered carrier having the phase of the received carrier. The curve
representing V.sub.1 shows at instant t.sub.p a peak due to the near echo
and at instant t a peak due to the distant echo, th | | |
