 |
Claims  |
|
|
We claim:
1. A method of synchronization for code division multiple access (CDMA)
radiotelephone communications in which each of a plurality of base
stations establishes a pilot channel in addition to a plurality of CDMA
transmission channels each defined by a respective spreading sequence
modulating data bits to be transmitted, and transmits over said pilot
channel a periodic pseudorandom reference sequence synchronized with said
spreading sequences, having a chip rate greater than the bit rate of said
data bits to be transmitted and belonging to a predetermined set of
reference sequences, the base band signals transmitted over said pilot
channel and said CDMA transmission channels of each base station being
combined and then modulated on a carrier frequency to form a radio signal
transmitted by said base station to mobile stations, said method
comprising the following steps at a mobile station:
(i) calculating the quantities E{s(t).multidot.s*(t-.tau.)} and
E{s(t).multidot.s*(t)}, in which E{.multidot.} designates a mean value
calculated over a predetermined integration time, .tau. designates a
predetermined delay, and s(t)=r(t).multidot.c(t) designates the product of
(a) the received base band signal and (b) the value of a candidate
sequence corresponding to a time-offset version of a reference sequence
belonging to said predetermined set;
(ii) calculating the ratio
.vertline.E{s(t).multidot.s*(t-.tau.)}.vertline./E{s(t).multidot.s*(t)};
(iii) comparing said ratio with a predetermined threshold;
(iv) if said comparing of step (iii) determines that said ratio exceeds the
predetermined threshold, selecting said candidate sequence as
corresponding to and being synchronized with a reference sequence
transmitted over the pilot channel of one of said plurality of base
stations and received by the mobile station over a propagation path; and
(v) if said comparing of step (iii) shows that said ratio does not exceed
the predetermined threshold, repeating steps (i) through (v) for another
candidate sequence.
2. A method according to claim 1, wherein said predetermined integration
time is of the same order as the period of the reference sequences.
3. A method according to claim 1, wherein said predetermined delay is of
the same order as the inverse of the chip rate of the reference sequences.
4. A method according to claim 1, wherein a carrier frequency, used for
obtaining the base band signal by demodulating the radio signal received
by the mobile station, is modified by applying a correction thereto, which
correction is proportional to the argument of the complex quantity
E{s(t).multidot.s*(t-.tau.)} calculated in step (i).
5. A method of synchronization for code division multiple access (CDMA)
radiotelephone communications in which each of a plurality of base
stations establishes a pilot channel in addition to a plurality of CDMA
transmission channels each defined by a respective spreading sequence
modulating data bits to be transmitted, and transmits over said pilot
channel a periodic pseudorandom reference sequence synchronized with said
spreading sequence, having a chip rate greater than the bit rate of said
data bits to be transmitted and belonging to a predetermined set of
reference sequences, the base band signals transmitted over said pilot
channel and said CDMA transmission channels of each base station being
combined and then modulated on a carrier frequency to form a radio signal
transmitted by said base station to mobile stations, said method
comprising the following steps at a mobile station:
(i) calculating the quantities E{s'(t).multidot.s'*(t-.tau.)} and
E{s'(t).multidot.s'*(t)}, in which E{.multidot.} designates the mean value
calculated over a first integration time, .tau. designates a predetermined
delay, and s'(t) designates a signal obtained by integrating, over a
second integration time shorter than said first integration time, the
product of (a) the received base band signal and (b) the value of a
candidate sequence corresponding to a time-offset version of a reference
sequence belonging to said predetermined set;
(ii) calculating the ratio
.vertline.E{s'(t).multidot.s'*(t-.tau.)}.vertline./E{s'(t).multidot.s'*(t)
};
(iii) comparing said ratio with a predetermined threshold;
(iv) if said comparing of step (iii) determines that said ratio exceeds the
predetermined threshold, selecting said candidate sequence as
corresponding to and being synchronized with a reference sequence
transmitted over the pilot channel of one of said plurality of base
stations and received by the mobile station over a propagation path; and
(v) if said comparing of step (iii) determines that said ratio does not
exceed the predetermined threshold, repeating steps (i) through (v) for
another candidate sequence.
6. A method according to claim 5, wherein said first integration time is of
the same order as the period of the reference sequences.
7. A method according to claim 5, wherein said predetermined delay is of
the same order as the inverse of the chip rate of the reference sequences.
8. A method according to claim 5, wherein a carrier frequency, used for
obtaining the base band signal by demodulating the radio signal received
by the mobile station, is modified by applying a correction thereto, which
correction is proportional to the argument of the complex quantity
E{s'(t).multidot.s'*(t-.tau.)} calculated in step (i).
9. A method of synchronization for code division multiple access (CDMA)
radiotelephone communications in which each of a plurality of base
stations establishes a pilot channel in addition to a plurality of CDMA
transmission channels each defined by a respective spreading sequence
modulating data bits to be transmitted, and transmits over said pilot
channel a periodic pseudorandom reference sequence synchronized with said
spreading sequences, having a chip rate greater than the bit rate of said
data bits to be transmitted and belonging to a predetermined set of
reference sequences, the base band signals transmitted over said pilot
channel and said CDMA transmission channels of each base station being
combined and then modulated on a carrier frequency to form a radio signal
transmitted by said base station to mobile stations, said method
comprising the following steps at a mobile station:
(i) calculating the quantities E{s(t).multidot.s*(t-.tau.)} and
E{s(t).multidot.s*(t)}, in which E{.multidot.} designates the mean value
calculated over a first integration time shorter than the period of the
reference sequences, .tau. designates a predetermined delay, and
s(t)=r(t).multidot.c(t) designates the product of (a) the received base
band signal and (b) the value of a candidate sequence corresponding to a
time-offset version of a reference sequence belonging to said
predetermined set;
(ii) calculating the first ratio
.vertline.E{s(t).multidot.s*(t-.tau.)}.vertline./E{s(t).multidot.s*(t)};
(iii) comparing said first ratio with a predetermined threshold;
(iv) if said comparing of step (iii) determines that said ratio exceeds the
predetermined threshold, verifying the candidate sequence through the
following sub-steps:
(iv-a) calculating the quantities E'{s(t).multidot.s*(t-.tau.)} and
E'{s(t).multidot.s*(t)}, in which E'{.multidot.} designates the means
value calculated over a second integration time longer than said first
integration time;
(iv-b) calculating the second ratio
.vertline.E'{s(t).multidot.s*(t-.tau.)}.vertline./E'{s*(t)};
(iv-c) comparing said second ratio with the predetermined threshold;
(iv-d) if the comparing step (iv-c) shows that said second ratio exceeds
the predetermined threshold, selecting said candidate sequence as
corresponding to and being synchronized with a reference sequence
transmitted over the pilot channel of one of said plurality of base
stations and received by the mobile station over a propagation path; and
(iv-e) if the comparing step (iv-c) determines that said ratio does not
exceed the predetermined threshold, repeating steps (i) through (v) for
another candidate sequence;
(v) if the comparing step (iii) determines that said first ratio does not
exceed the predetermined threshold, repeating steps (i) through (v) for
another candidate sequence.
10. A method according to claim 9, wherein said second integration time is
of the same order as the period of the reference sequences.
11. A method according to claim 9, wherein said predetermined delay is of
the same order as the inverse of the chip rate of the reference sequences.
12. A method according to claim 9, wherein a carrier frequency, used for
obtaining the base band signal by demodulating the radio signal received
by the mobile station, is modified by applying a correction thereto, which
correction is proportional to the argument of the complex quantity
E{s(t).multidot.s*(t-.tau.)} calculated in step (i).
13. A method according to claim 9, wherein a carrier frequency, used for
obtaining the base band signal by demodulating the radio signal received
by the mobile station, is modified by applying a correction thereto, which
correction is proportional to the argument of the complex quantity
E'{s(t).multidot.s*(t-.tau.)} calculated in sub-step (iv-a). |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
BACKGROUND OF THE INVENTION
The present invention relates to code-division multiple access (CDMA)
radiotelephone communications.
CDMA is a method of spread spectrum digital communication in which a
plurality of transmission channels are created by using spreading
sequences for each channel that modulate the information bits to be
transmitted. The spreading sequences operate at a chip rate higher than
the data bit rate in order to achieve spectrum spreading of the radio
signal. Their self- and cross-correlation properties are adapted to enable
the various channels to be multiplexed: they are generally pseudorandom
sequences that are mutually orthogonal or quasi-orthogonal, taking chip
values of -1 or +1.
The use of CDMA in the field of cellular radiotelephony is described in
chapter I of the work "Mobile radio communications" by Raymond Steele,
Pentech Press, London 1992, and also in the article "On the system design
aspects of code division multiple access (CDMA) applied to digital
cellular and personal communications networks" by A. Salmasi and K. S.
Gilhousen, Proc. of the 41st IEEE Vehicular Technology Conference, St.
Louis, Mo., 19-22 mai 1991. The multiplexed transmission channels are
formed at the base station of each cell in the network. Each mobile
station situated within the cell uses a special spreading sequence to
recover, from the overall radio signal transmitted by the base station,
the data bits that are adressed thereto.
In the system described in the above publications, the various spreading
sequences are obtained from a common reference sequence having a chip rate
of 1.2288 MHz, and period of 32 768 chips. Since the radio modulation is
quadrature phase shift keying, the reference sequence includes an in-phase
component and a quadrature component. Sixty-four transmission channels are
formed in the base station by combining the reference sequence with each
of the sixty-four Walsh codes of length 64. The channel defined by Walsh
code W.sub.0, which comprises nothing but 1, is a pilot channel over which
no data bits are sent. The pilot channel does transmit the reference
sequence synchronously with the set of spreading sequences. The mobile
stations have a priori knowledge of the values of the reference sequence
such that, on receiving the pilot channel, they can synchronize themselves
with the base station to receive the data bits that are respectively
addressed to each of them.
Whether or not use is made of the Walsh code technique, it is always useful
to form a pilot channel enabling the mobile channels to synchronize
themselves. The pilot channel carries a reference pseudo-random periodic
sequence having the same chip rate as the spreading sequences and
synchronized therewith.
Various methods exist for obtaining the desired synchronization, all of
which make use of the advantageous self- and cross-correlation properties
of pseudo-random sequences. They are based on calculating the
cross-correlation between the sequence received on the pilot channel and a
sequence tested by the mobile station, given that the results of such an
operation will always be low except when the sequences are synchronized
and identical.
A first synchronization method uses matched filters, i.e. filters whose
coefficients are equal to the samples of the tested sequence. The result
of such filtering is thus directly the value of the looked-for
cross-correlation. Since the output rate of this filter is the same as the
input rate, a correlation peak is detected when the mobile station is
synchronized, and as a result the mean synchronization time is relatively
short.
A second method makes use of correlators that apply the same principle. The
received signal is multiplied by the tested sequence and integration over
a plurality of samples makes it possible to detect a correlation peak, if
any. When there is no correlation peak, then the operation is reiterated,
either with the same sequence subjected to a time offset, or else with a
different sequence. The mean time to obtaining synchronization is
significantly longer than it is with the first method.
Both of those two methods suffer from the drawback of significantly
degraded performance in the event of frequency deviation occurring in the
received radio signal. Under such circumstances, it is no longer possible
to associate a correlation peak with genuine synchronization. Frequency
deviation may be due to the Doppler effect, to Rayleigh fading, or to
differences in the characteristics of the local oscillators of the
stations in communication. To cope with this problem, the correlations
must be performed by making an assumption about the value of the frequency
deviation to which the received signal has been subjected, and by using a
battery of matched filters whose respective frequencies correspond to the
various different possible values of deviation. The results of the various
different filterings are compared and the largest correlation peak is used
as a basis for determining which filter can be used to obtain the
looked-for time and frequency synchronization. Such a solution is not
optimal in terms of performance. In addition, it significantly increases
the complexity of the receiver.
An object of the present invention is to remedy the above difficulties, by
proposing a time synchronization method that is not very sensitive to
possible frequency deviation in the radio signal.
SUMMARY OF THE INVENTION
The invention thus provides a method of synchronization, for CDMA
radiotelephone communications in which a base station establishes a pilot
channel in addition to a plurality of CDMA transmission channels each
defined by a respective spreading sequence modulating data bits to be
transmitted, a periodic pseudorandom reference sequence synchronized with
the spreading sequences and having a chip rate greater than the bit rate
of the data bits to be transmitted being transmitted over the pilot
channel, the base band signals transmitted over said channels being
combined and then modulated on a carrier frequency to form the radio
signal transmitted by the base station to mobile stations. The
synchronization method consists in selecting, at a mobile station, at
least one periodic pseudorandom sequence which corresponds to and is
synchronized with the reference sequence transmitted over the pilot
channel of a base station and received by the mobile station over a
propagation path. At the mobile station, various possible sequences are
tested by calculating the quantities E[s(t).multidot.s*(t-.tau.)] and
E[s(t).multidot.s*(t)], in which E [.multidot.] designates the mean value
calculated over a predetermined integration time (T), .tau. designates a
predetermined delay, and s(t)=r(t).multidot.c(t) designates the product of
the received base band signal by the value of the sequence under test, and
by calculating the ratio
.vertline.E[s(t).multidot.s*(t-.tau.)].vertline./E[s(t).multidot.s*(t)].
The sequence under test is selected if said ratio exceeds a predetermined
threshold.
The robustness of this synchronization method in the presence of frequency
deviation can be explained as follows: The base band complex signal
received over the pilot channel is of the form
r(t)=A(t).multidot.e.sup.2.pi.j.DELTA.ft c'(t), where c'(t) is the
reference sequence received with a time offset, A(t) represents the
disturbances due to the propagation channel, and the exponential term
represents frequency deviation of the signal. The following is thus
obtained for a tested sequence c(t):
##EQU1##
If the tested sequence corresponds to the reference sequence and is
synchronized, then c(t)=c'(t), whence:
E[s(t).multidot.s.sup..multidot. (t-.tau.)]=E[A.sup.2
(t)].multidot.e.sup.2.tau.j.DELTA.ft ( 1)
Otherwise:
E[s(t).multidot.s*(t-.tau.)]=E[A.sup.2
(t)].multidot.E[c'(t).multidot.c(t).multidot.c'(t-.tau.).multidot.c(t-.tau
.)].multidot.e.sup.2.pi.j.DELTA.ft
It can thus be seen that the modulus of the quantity
E[s(t).multidot.s*(t-.tau.)] and also the quantity E[s(t).multidot.s*(t)],
are independent of the frequency deviation in the received radio signal.
The ratio of these two quantities is close to unity when synchronization
is achieved, and otherwise it is significantly smaller since the
correlation properties of the sequences then imply that
E[c'(t).multidot.c(t).multidot.c'(t-.tau.).multidot.c(t-.tau.)]<<1. By
comparing the ratio with a threshold, it is possible to determine whether
the tested sequence is properly synchronized, even in the presence of
frequency deviation.
Another important advantage of the method is that when time synchronization
is performed, it is possible to discover the value of frequency deviation
to which the radio signal has been subjected from the argument of the
complex quantity E[s(t).multidot.s*(t-.tau.)], and therefore a priori, to
correct the frequency of the local oscillator, so as to make the signal
understandable by the receiver.
In a particular embodiment of the invention, the mean values are calculated
not on the above-defined quantities s(t) but on a signal s'(t) obtained by
integrating the quantity s(t) over a shorter duration. This implementation
is particularly suitable when in the presence of a relatively low signal
to noise ratio.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of the transmitter portion of a base station in a
radiotelephone system operating in CDMA mode.
FIG. 2 is a block diagram of the receiver portion of a mobile station in
the system.
FIGS. 3 and 4 are diagrams showing variants of the synchronization module
in the mobile station of FIG. 2.
DESCRIPTION OF PREFERRED EMBODIMENTS
In a cellular radiotelephone system, base stations are distributed over the
territory that is to be covered. In the zone or "cell" covered by a given
base station, a plurality of mobile stations may communicate
simultaneously with the base station. The method of the invention is
applicable to downlinks, i.e. to the transmission of signals from a base
station to mobile stations, for which code division multiple access (CDMA)
is used.
Each base station, such as the station shown in FIG. 1, sets up CDMA
transmission channels C1, C2, . . . Cn each defined by a respective
spreading sequence. The spreading sequences which take the values +1 or -1
are periodic, pseudorandom, and substantially decorrelated. They are at a
chip rate that is higher than the bit rate of the data bits to be
transmitted, for example they are at a rate of 1 MHz. In the example
described, the spreading sequences are very long (period 2.sup.41 -1). All
of them correspond to the same base sequence of period 2.sup.41 -1, and
they are obtained by applying respective time offsets to said base
sequence, each offset being characteristic of the user (mobile station)
with which a call is established over the corresponding channel. The
instantaneous position in the base sequence is indicated by an address
ADD. In each CDMA channel, the data bits to be transmitted M1, M2, . . .
Mn, that represent voice or data signals previously encoded by
conventional techniques, are multiplied by the spreading sequence produced
by a suitable generator which may be conventionally constituted by a
cascade of 41 bistables.
The base station also forms a pilot channel CP and a synchronization
channel CS. A reference sequence PN is transmitted over the pilot channel
CP without being modulated by data bits. The sequence PN which takes the
values +1 or -1 is periodic and pseudorandom. Its chip rate is the same as
that of the spreading sequences with which it is synchronized. The
synchronization channel CS serves to transmit the address ADD which is
updated in each period of the reference sequence PN. The address bits ADD
are modulated by a sequence PN' associated with the reference sequence PN.
The sequence PN' has the same chip rate and the same period as the
sequence PN, and it is synchronized therewith.
By way of example, the reference sequence PN and its associated sequence
PN' are Gold sequences of period 1023 (see "optimum binary sequences for
spread spectrum multiplexing" by R. Gold, IEEE Trans. Inform. Theory,
IT-13, pages 619-621, October 1967). 512 Gold sequences of period 1023 are
used, forming 256 pairs. A pair of Gold sequences is allocated to each
base station, one for the sequence PN and the other for the sequence PN'.
Different pairs of sequences are allocated to the base stations in
adjacent cells in order to prevent interference between base stations in
boundary regions.
The base band signals formed in the tranmission channels C1, C2 . . . , Cn,
the pilot channel CP, and the synchronization channel CS are combined and
then modulated on a carrier frequency (block 10 in FIG. 1). Combination is
performed by summing the base band signals, optionally using weights. By
way of example, the carrier frequency may be 2.4 GHz. The modulation may
consist of binary phase shift keying when the base band signals are real.
However, the base band signals are usually complex, in which case
quadrature phase shift keying should be used. Under such circumstances,
each of the pseudorandom sequences PN1, PN2 . . . , PNn, PN, PN' includes
two independent components used respectively for the I phase and the Q
phase of the corresponding channel.
The radio signal obtained after modulation is transmitted via the antenna
12 towards the mobile stations in the cell.
In a mobile station (FIG. 2), the radio signal received by the antenna 14
is demodulated into base band by the radio stage 16 by means of a carrier
FP obtained from a local oscillator 18. The digitalized base band signal
r(t) can then be processed by conventional CDMA circuits 20. As
represented symbolically in FIG. 2, the circuits 20 typically correlate
the received base band signal r(t) with the spreading sequence PNi
allocated to the mobile station to extract, where appropriate, the data
bits Mi addressed to the mobile station, with the other CDMA components of
the signal r(t) being eliminated because of the orthogonal properties of
the spreading sequences.
This processing requires the spreading sequence PNI to be synchronized
relative to at least one propagation delay between the base station and
the mobile station. Synchronization is looked for by the module 22 which
tests successive sequences of period 1023 until it finds one that
corresponds to the reference sequence PN of the transmitting base station,
and which presents a time offset relative thereto equal to the propagation
delay over a propagation path. A priori there are 256.times.1023 possible
sequences to be tested, each corresponding to one of the 256 possible Gold
sequences for the sequence PN, subjected to a time offset lying in the
range 0 to 1022 positions.
For each tested sequence, the module 22 calculates the quantities
E[s(t).multidot.s*(t-.tau.)] and E[s(t).multidot.s*(t)] in which
s(t)=r(t).multidot.c(t) is the product, as calculated by a multiplier 24,
of the instantaneous value of the received signal r(t) by the
instantaneous value of the tested sequence c(t) as provided by a
pseudorandom generator 23;
.tau. is a predetermined time delay which is preferably of the same order
as the inverse of the chip rate of the reference sequence PN; thus, for a
rate of 1 MHz, it would be possible to chose .tau.=1.mu.s; and
the notation E[.multidot.] represents a mean value calculated over a
predetermined integration time T which is preferably of the same order as
the period of the reference sequence PN (1.023 ms).
The complex conjugate s*(t) of s(t) is calculated at 26 and is then
multiplied by s(t) by the multiplier 28 whose output is applied to the
integrator 30. In parallel, the signal s(t) is delayed by .tau., at 32,
and then the complex conjugate of s(t-.tau.) is calculated at 34 and is
multiplied by s(t) by the multiplier 36 whose output is applied to the
integrator 38. At the end of each integration period T, the integrators 30
and 38 deliver the quantities E[s(t).multidot.s*(t)] and
E[s(t).multidot.s*(t-.tau.)], respectively.
The modulus E1 of the complex quantity E[s(t).multidot.s*(t-.tau.)] is
calculated at 40, after which the ratio
El/E2=.vertline.E[s(t).multidot.s*(t-.tau.)].vertline./E[s(t).multidot.s*(
t)] is calculated at 42. A decision unit 44 compares this ratio with a
predetermined threshold 1/.lambda. in order to determine whether the
sequence under test is correct and synchronized. If E1/E2<1/.lambda., then
the decision unit 44 controls the generator 23 to test another sequence
which will either be the same Gold sequence subjected to a different
offset, or else will be a different Gold sequence. If
E1/E2.gtoreq.1/.lambda., it is considered that the tested sequence
corresponds to the reference sequence of the base station with an offset
that relates to the time delay over a propagation path between the base
station and the mobile station. The value of this offset is transmitted to
the pseudorandom generator 21 of the processing circuits 20 so as to
enable them to apply the same offset to the spreading sequence PNi of the
mobile station in order to achieve time synchronization with the observed
propagation path.
In the example described, it has be seen that the spreading sequences PN1,
PN2, . . . PNn of the various mobile stations are all obtained by applying
an offset characteristic of the mobile station to a base sequence that is
very long. At the mobile station, this characteristic offset is stored in
a memory 48 of an address recovery module 50 which processes the
synchronization channel CS. Once the reference sequence PN of a base
station has been detected with an appropriate time offset, the Gold
sequence PN' being associated with said reference sequence is also known.
The decision unit 44 then controls the pseudorandom generator 52 of the
module 50 so that it delivers the associated sequence PN' with the
determined time offset. This associated sequence, which is thus
synchronized, is applied to the signal r(t) by the multiplier 54 whose
output is applied to the input of an integrator 56 having an integration
time of the order of the duration of one bit on the synchronization
channel CS. The correlation provided by the integrator 56 corresponds to
the address bits transmitted over the channel CS. The offset stored in the
memory 48 is applied, at 58, to the received address ADD so as to provide
the pseudorandom generator 21 of the circuits 20 with the position that it
needs to take up in the long base sequence. This position is then modified
to take account of the additional offset due to propagation delay, as
provided by the decision unit 44.
Circuits other than those shown in FIG. 2 can be used to calculate the
quantities E[s(t).multidot.s*(t-.tau.)] and E[(s(t).multidot.s*(t)]. For
example, the synchronization module 22 may be replaced by the module 122
shown in FIG. 2, in which identical numerical references designate
elements that are similar to elements in FIG. 2.
Given that ignoring a normalization constant, c(t).multidot.c*(t)=1 for any
tested pseudorandom sequence, the energy quantities E[s(t).multidot.s*(t)]
and E[r(t).multidot.r*(t)] are identical. The latter quantity can be
calculated by the top branch of the module 122 in FIG. 3. The complex
conjugate r*(t) is obtained at 126, then multiplied by r(t) by the
multiplier 128, whose output is applied to the input of a transversal
filter 129. The coefficients of the filter 129 are equal to 1 in its real
portion and 0 in its imaginary portion so that said filter acts as an
integrator for calculating the mean value E2. The multiplier 136 provides
the instantaneous value of r(t).multidot.r*(t-.tau.) to the input of
another transversal filter 137, said instantaneous value being obtained by
multiplying the signal r(t) by the complex conjugate (as calculated in
134) of the same signal as delayed by .tau. in 132. The coefficients of
the filter 137 are equal to the successive values of the samples
c(t).multidot.c(t-.tau.). They are provided by a multiplier 135 having one
input that receives the instantaneous value c(t) of the tested sequence as
produced by the generator 23, and another input that receives the complex
conjugate of the value of the sequence as delayed by .tau., as calculated
by the elements referenced 131 and 133. It can be verified that the output
from the filter 137 is indeed equal to the complex quantity
E[s(t).multidot.s*(t-.tau.)] that is to be calculated. This quantity is
then processed and compared with E2 by circuits identical to those of FIG.
2.
In the ideal case, where the base station transmits only over the pilot
channel CP and where there exists only one noiseless propagation path, the
received signal r(t) will have the form
r(t)=A(t).multidot.c'(t).multidot.e.sup.2.pi.j.DELTA.ft. Under such
circumstances, it has been shown that synchronization means that
E1/E2.apprxeq.1, whereas lack of synchronization means that El/E2<<1. The
sequences PN can therefore be selected on the basis of a threshold
1/.lambda. lying in the range 0 to 1. To take other CDMA channels into
account, together with multiple paths and noise, it is necessary to select
an appropriate threshold level 1/.lambda., given that the largest values
of the ratio E1/E2 always correspond to synchronization over a propagation
path, provided that the attenuation A(t) and the frequency deviation
.DELTA.f do not vary too much at the time scale over which the mean values
are calculated. In practice, the value of .lambda. may be determined
experimentally or by simulation. In the example described, satisfactory
results have been obtained using simulation with .lambda.=10 and with
.lambda.=15. To reduce the disturbance from the other CDMA channels, it is
also possible, in the base station, to combine channels by giving greater
weighting to the pilot channel CP.
The time required for the synchronization is at most 256.times.1023=261888
times the test duration of a sequence. If the integration time for
calculating the mean values is of the same order as the period of the
sequence PN, that can lead to long synchronization times, unless the
mobile station has a priori knowledge of the cell in which it is present.
The synchronization time can be reduced by storing portions of the signal
r(t) in a buffer register and by performing on said portions parallel
calculations of ratios E1/E2 so as to be able to test sequences more
rapidly.
To reduce the time required for synchronization, it is also possible to
implement the integrators 30, 38 with an integration time that is less
than the period of the reference sequence PN, e.g. a time that is a few
tens of times longer than the duration of a chip. This makes it possible
to select sequences that are subsequently subjected to synchronization
verification in which the same calculations and the same comparisons are
performed, but using longer integration times T in the integrators 30,38,
i.e. times of the same order as the period of the sequence. The tested
sequence is finally accepted only if verification is positive.
To optimize the mean duration of synchronization, reference may be made to
the article "Acquisition time performance of PN Spread-Spectrum systems"
by J. K. HOLMES, IEEE Trans. on Comm., Vol. com-25, N.degree. 8, August
1977, pages 778-783.
To optimize the performance of the method in the presence of a low signal
to noise ratio, the mean values may be calculated not on the basis of the
signals s(t), but on the basis of signals s'(t) obtained by a first
integration of s(t). Under that circumstances, a synchronization module
222 is used such as a module shown in FIG. 4. This synchronization module
is identical to the module of FIG. 2, except in that an integrator 225,
having an integration time T.sub.1 that is small compared with the period
of the reference sequence is placed at the output of the multiplier 24
providing s(t). The integrator makes it possible to reduce the power of
the noise added to the signal so that said noise interfers less during the
subsequent operations. Such a solution can only be adopted if the first
integration is feasable: If excessive frequency deviation causes the phase
of the received signal to rotate in such a manner that the results of the
integration are meaningless, then the operation is pointless. The limit
set on said integration time T.sub.1 is such that if T.sub.1
=N.multidot.T.sub.0 (while T.sub.0 is the duration of a sample of signal
s(t), then it is necessary for: 2.tau..DELTA.f.multidot.N.multidot.T.sub.0
<.tau. and thus for N<1/(2.multidot..DELTA.f.multidot.T.sub.0). The
various integration times are such that the total integration time is of
the same order as the period of the reference sequence PN.
The method of the invention makes it possible to identify a base station by
detecting its reference sequence, to achieve time synchronization with
said base station by determining a propagation delay, and also to perform
frequency correction to take frequency deviation, if any, into account.
From equation (1) it can be seen that the argument of the complex quantity
E[s(t).multidot.s*(t-.tau.)], as calculated when testing sequences, is
proportional to the frequency deviation .DELTA.f. The argument of
E[s(t).multidot.s*(t-.tau.)] is extracted to control a voltage controlled
oscillator 62. The output from the oscillator 62 is at the frequency
.DELTA.f, and it is applied to input of a mixer 64, whose other input
receives the carrier delivered by the local oscillator 18. The carrier
frequency is thus corrected, after filtering 66, in a manner that is
proportional to the argument of the E[s(t).multidot.s*(t-.tau.)] to
compensate for the frequency deviation.
However, it is not necessary to wait for such deviation to be fully
compensated before proceeding with a search for synchronization, because
the synchronization test relies on the module E1 of
E[s(t).multidot.s*(t-.tau.)] which is not influenced by the frequency
deviation .DELTA.f.
Since the argument 2.pi..multidot..DELTA.f.multidot..tau. of the quantity
E[s(t).multidot.s*(t-.tau.)] is small, the calculation of said argument by
the block 60 can be reduced merely to:
2.pi..multidot..DELTA.f.multidot..tau.=sin
(2.pi..multidot..DELTA.f.multidot..tau.)=Im
{E[s(t).multidot.s*(t-.tau.)]}/E1 ,
where Im {. . . } represents the imaginary part.
* * * * *
|
|
|
|
|
|
|
Description | |
