AU732030B2 - Method of compensating group propagation times of filters in radio transmitter/receiver - Google Patents

Description

P/00/0 11 28/5/91 Regulation 3.2

AUSTRALIA

Patents Act 1990

ORIGINAL

COMPLETE SPECIFICATION STANDARD PATENT 0O 0 0O 0O

S

0O

S

0

S

*SeS 0 0 S S @0 @0 0 S 0 0 0@

S

000000

S

Invention Title: "METHOD OF COMPENSATING GROUP PROPAGATION TIMES OF FILTERS IN RADIO TRANS MITTER/RECEIVER 0 000 0 The following statement is a full description of this invention, including the best method of performing it known to us:- 0 00 S 0 0 @050

S

050000 0 x 0 25 GG 1 o5 9 2 This invention relates to the transmission of signals in phase quadrature, and in particular to a method for compensating for the differences in the group propagation times between the analog filters of a transmitter of signals inphase quadrature and between the analog filters of a receiver of signals in phase quadrature.

A transmitter and receiver of modulated signals in phase quadrature are usually employed for transmission of M-PM (M-state phase modulation) type signals in quadrature, M-APM (M-state amplitude and phase modulation) type signals in quadrature, etc as shown in Figures 1A and 1 B.

The transmitter shown in Figure 1A includes an interface 1 which, on the basis of a signal to be transmitted S, supplies two baseband digital signals on two processing paths A digital/analog converter 4" is connected to the output of a generally programmable digital filter 3" included in each processing path. A lowpass, analog transmission filter 5" is connected to the output of each digital/analog converter The signals provided via the processing paths 2" are mixed S respectively with a signal cos wt and a signal sin wt (where w 2nf and f is the carrier frequency of the transmitted signal) and added in a phase quadrature modulator 6.

The output signal ST of the modulator 6 is transmitted by a transmission channel, for example radio) for the attention of a receiver shown in Figure 1 B.

The receiver includes a phase quadrature demodulator 8 mixing the received signal SR with a signal cos wt and a signal sin wt to obtain two baseband signals ••0•0 applied to two processing paths Each processing path includes a low-pass, analog reception filter 10', 10", at the output of which is connected an analog/digital converter 11', 11" to which a generally programmable, and possibly auto-adaptive, digital filter 12', 12" is connected. In this latter case, each filter 12', 12" is controlled by a corresponding estimation means (not shown). The signals obtained as output from the paths 9' and 9" are applied to a processing unit 13 which provides an output signal S. The signals provided by the analog/digital converters 11' and 11" are also applied to the inputs of a clock recovery circuit 14 whose output signal controls the •o sampling instants of the analog/digital converters 11' and 1 1".

The problem is that on account of the tolerances of the components forming the analog filters 5' and 5" on the one hand, and 10' and 10" on the other hand, the filters are not absolutely identical in pairs (filter 5' is different to filter 5" and filter 10' is 0 *0 o 0 0000 *oo00 0@

S

*0 0 o oo oo00o @000 0 00 0 0 00 different to filter The result is that there are different group propagation times in the processing paths 2' and 2" on the one hand and 9' and 9" on the other hand. This difference is represented by an error on the closure of the curve of the feedback control loop in the samplers forming part of the analog/digital converters 11' and 11".

One solution of the prior art to eliminate this problem is to use filters matched respectively to 10', 10", which requires either very low tolerance components, or providing accurate control of the filters 10', 10". However, low tolerance components are expensive and accurate control is costly in time.

Another solution of the prior art is to compensate for the shift with equalisers at the reception end. An equaliser is constituted by an adaptive filter and estimation means which calculate the coefficients of the adaptive filter. In this case, digital filters 12' and 12" can be used while adding the estimation means. However, estimation means then have to be added in each receiver so as to control the digital filters and to have identical group propagation times in both paths of each receiver.

It is an object of the present invention to eliminate these disadvantages by providing a method for compensating for the differences between the group propagation times of the analog filters of a transmitter and a receiver of signals in phase quadrature, the method not requiring the addition of expensive components, manual adjustment or addition of estimators in each receiver.

According to the invention there is provided a method for compensating for the differences in group propagation times between the analog filters of a transmitter of signals in phase quadrature and between the analog filters of a receiver of signals in phase quadrature, the transmitter comprising a phase quadrature signal modulator providing a transmission signal based on first and second baseband signals from first and second processing paths, the first and second processing paths including first and second analog transmission filters respectively, the receiver including a demodulatorof signals in phase quadrature which, on the basis of the received signal, provides third and fourth baseband signals to third and fourth processing paths respectively, the third and fourth processing paths including third and fourth analog reception filters respectively which provide a third filtered signal X and a fourth filtered signal Y respectively, the receiver further including a clock recovery circuit which provides a clock signal on the basis of the third and fourth filtered signals X and Y respectively, S. S 0@ 0* 0 0S 00 S@ 0 0 0ee@

S

0 0

S

0 0 @0 0 *5@ @005 0 SS 0 @0

S

000 0 0* 0 5055 000@@0

S

4 the method consisting in: a) applying the first and second signals respectively to the third and fourth analog filters, the output signals of the third and fourth analog filters constituting respectively the signals X1 and Y1 applied to means for estimating the time difference which provides a difference A1 (Tx Rx)- (Ty Ry) where Tx, Rx, Ty and Ry are respectively the signal propagation times in the first, second, third and fourth filters, and the difference Al being representative of the time difference between the optimum sampling instants between the signals X1 and Y1, b) applying the first and second signals to the fourth and third analog filters respectively, the output signals of the third and fourth analog filters constituting respectively the signals X2 and Y2 applied to the estimation means for estimating the time difference which provides a difference A2 (Ty Rx)- (Tx Ry) the difference A2 being repiresentative of the time difference between the optimum sampling instants between the signals X2 and Y2, determining, from the differences Al and A2, weighting coefficients applied to fifth, sixth, seventh and eighth digital filters inserted in the first, second, third and fourth processing paths, the digital filters being inoperative during the steps a) and so as to compensate for the differences in group propagation times on the one hand between the first and second analog filters, and on theother hand between the third and fourth analog filters.

This invention thus allows the use of analog filters which have components with large tolerances, because the differences in group propagation times between the analog filters of the transmitter as well as the differences in group propagation times between the analog filters of the receiver are automatically compensated for by the digital filters.

By calculating the weighting coefficients applied to digital filters on the basis of the time differences Al and A2, adjustment of the paths can be carried out with the aim of compensating for their differential delay, in a way comparable to compensation provided by an equaliser. But unlike the solution using real equalisers, the solution of 0 0 0*

S

this invention requires neither estimation means which continuously calculate the coefficients of adaptive filters, nor additional analog/digital converters to convert the output signals of the transmitter's analog filters.

Advantageously, the symbol rate of the first and second signals is the highest symbol rate which can be accepted by a transmission system comprising the transmitter and the receiver.

The invention also relates to an estimation device for estimating the time differences Al and A2, the device comprising: first and second Gardner type circuits receiving respectively the third and fourth filtered signals X and Y, the Gardner type circuits also serving as a clock recovery circuit, a subtracter subtracting the signal provided by the second Gardner type circuit from the signal provided by the first Gardner type circuit, and an integrator integrating the signal provided by the subtracter to produce the time differences Al and A2.

Advantageously, each of the Gardner type circuits comprises a first delay circuit receiving an input signal, the output of the first delay circuit being connected to the input of a second delay circuit, a subtracter subtracting the signal being input into the first delay circuit from the signal provided by the second delay circuit, and a multiplier multiplying the signal provided by the first delay circuit by the signal provided by the subtracter.

The subtracter and integrator of the circuit estimating the differences Al and A2 can be integrated in the receiver, their cost being minimal.

In order that the invention may be readily carried into effect, embodiments thereof will now be described in relation to the accompanying drawings, in which: Figure 1 shows a transmission system for transmitting signals in phase quadrature; Figure 2 shows a step a) of the method of this invention; Figure 3 shows a step b) of the method of this invention; Figure 4 shows a clock recovery circuit to which is added a circuit for estimating group propagation time difference according to the invention; 0*

S

Se.

OSS@

55 S @0 00 S

S

S.

S S S C 6 Figure 5 shows a finite impulse response filter.

In all these figures, identical elements have the same reference numbers.

Figure 1 has been previously described in connection with the prior art.

Figure 2 shows the lay-out of a step a) of the method of this invention, in which the analog filter 5' is connected to the analog filter 10' and in which the analog filter is connected to the analog filter 10". Known signals are applied to the input of the analog filter 5' and to the input of the analog filter The signals provided by the analog filters 10' and 10" are applied to the inputs of the analog/digital converters 11' and 11" which provide signals X1 and Y1. The signals X1 and Y1 are shifted with respect to each other by a time difference Al equal to: Al (Tx Rx) (Ty Ry) where Tx, Rx, Ty and Ry are respectively the signal propagation times in the filters 10' and 10". The signals X1 and Y1 are applied to the inputs of a time difference estimator 14, 23, which provides the difference Al. The time difference estimator 23 S. forms the largest part of the clock recovery circuit 14 and will be described below. The time difference Al is representative of the time. difference between the optimum sampling instants of the signals Xl and Y1.

Figure 3 shows the lay-out of a step b) of the method of this invention, in which the analog filter 5' is connected to the analog filter 1 Known signals are applied to the inputs of the analog filters 5' and The signals provided by the analog filters and 10" are applied to the inputs of the analog/digital converters 11' and 11" which provide signals X2 and Y2. The signals X2 and Y2 are shifted with respect to each other by a time difference A2 equal to: 5500 00. A2 (Ty Rx)- (Tx Ry) 0 S.

where Tx, Rx, Ty and Ry are defined in the description of Figure 2. The signals Xl and Y1 are applied to the inputs of a time difference estimator 14, 23, which provides the 0 0. difference A2. The difference A2 is representative of the time difference between the optimum sampling instants of the signals X2 and Y2.

Measurements of Al and A2 are made by direct (physical) interconnection of the analog filters.

*0

OS

S

0 0 *000

SO

S.

0 @0 S The group propagation time differences between the analog filters 5' and 5" of the transmitter can be calculated from Al and A2 as follows: Tx Ty (Al -A2)/2 Likewise, the group propagation time differences between the receiver analog filters 10' and 10" can be calculated from Al and A2 as follows: Rx Ry (Al A2) 2 The method then consists in calculating the coefficients of the digital filters 3" so as to compensate for this propagation time difference, on the basis of the propagation time differences between the analog filters 5' and Similarly, the propagation time difference between the receiver analog filters 10' and 10" is used to obtain the coefficients of the digital filters 12', 12", so as to compensate for this propagation time difference. The digital filters 12', 12" will be described later.

The digital filters 12', 12" have no effect in relation to the value of the differential delay during estimation of the propagation time difference (steps a) and b)).These filters can be rendered inoperative by applying the same coefficients to the filters 3' and 3" (12' and 12" respectively). They are thus rendered inoperative in relation to the differential delay value.

Figure 4 shows a device for estimating time difference 23 in accordance with the invention, the device being constructed around a clock recovery circuit 14 forming part of a receiver of signals in phase quadrature.

The clock recovery circuit 14 receives the filtered and digitised signals from the baseband signal processing paths 9" of the receiver, particularly the signals Xi, Yi (where i takes the value 1 or 2 depending on whether the design of Figure 2 or Figure 3 is made use of) provided by the analog/digital converters 11' and 11". The clock recovery circuit 14 includes a Gardner type circuit 19', 19" for each of the signals Xi and Yi.

Reference can be made to the article entitled "A BPSK/QPSK Detector For Sampled Receivers" by Floyd M. Gardner, IEEE Transactions On Communications, vol.

COM-34, n 0 5, May 1986, pp 423 to 429.

A Gardner type circuit consists for example in a first delay circuit 1 5' or 1 receiving the signal Xi or Yi, the output of the first delay circuit being connected to the es 0 0

S

*OSS

0 0 00 00 0e

S

*0~ 00Y 06 *60 0 *0 S 0

S

input of a second delay circuit 16' or 16". A subtracter 17' or 17" subtracts the signal being input into the first delay circuit 15' from the signal being output from the second delay circuit 16' The delay provided by each of the delay circuits 15', 16' and 16" has a value of Ts/2, Ts being the symbol time. A multiplier 18', (18") multiplies the signal being output from the first delay circuit 16', by the signal provided by the subtracter 17' The signals provided by the multipliers 18' and 18" constitute the output signals of the Gardner type circuits 19' and 19" respectively.

The output signal of the Gardner type circuit 19' is added to the output signal of the Gardner type circuit 19" in an adder 20. The sum is integratedby an integrator 21 which provides a control signal for a voltage-controlled oscillator 22 producing the rate recovery clock signal intended for the analog/digital converters.

The difference estimating device 23 according to this invention uses the two "Gardner" circuits of the clock recovery circuit 14 and in addition includes a module 29 comprising a subtracter 24 and an integrator 25. The subtracter 24 subtracts the output signal of the "Gardner" circuit 19' from the output signal of the "Gardner" circuit 19". This difference is integrated by the integrator 25 which, at its output, provides the time difference estimation signal Al or A2, depending on the configuration (Figure 2 or Figure 3).

The method of this invention is applied before operation of a system which includes the analog filters 10" and 10", particularly during testing of cards containing the analog filters 10" and 10". The transposition of the paths for step b) is made between the mixers of the phase quadrature demodulator 8 and the analog filters 10' and 10". Thus it involves physical linkages.

The subtracter 24 and the integrator 25 of thedifference estimation circuit 23, which does not form part of the clock recovery circuit 14, can be physically in the receiver, that is to say integrated into the receiver, or alternatively can be connected to the latter.

The method of this invention can be used for transmission systems which allow different data rates. In that case, compensation for the group propagation time differences is optimum when the steps of the method previously described are carried out at the highest symbol rate, where the differences in group propagation time cause the most problems.

A system for the transmission of signals in phase quadrature according to this invention is identical to that in Figures 1A and 1 B, with the filters 12' and 12" constituted by non-adaptive filters. They have for example Nyquist square root characteristics and are preferably finite impulse response filters (FIR).

Figure 5 shows such a finite impulse response filter. It consists of a plurality of delay circuits 261 262 26, mounted in cascade. The signals provided by the delay circuits are each multiplied by a coefficient cl, c2, cn by means of the multipliers 271 272 27 and the products of these multiplications are summed in an adder 28 providing the output filtered digital signal. The coefficients cl, c2, cn of one of the digital filters 3' (or 12') differ from those of the other filter 3" in such a way that the group propagation time differences are compensated for.

Different techniques for adjusting digital filters can be used: it is possible for example to calculate, for each transmitter and each receiver to be adjusted, the coefficients c using a microprocessor which receives the differences Al and A2. The microprocessor calculates the values Tx-Ty and Rx-Ry. By considering firstly which are s. the transmitter digital filters that are corrected, the time difference Tx-Ty, called AT, is :0 divided by Ts which is the symbol time. The result of this division is a function whose value controls the change in the coefficients of the digital filter.

00 By way of example, for a time AT equal to 250 ns and for Ts equal to 1 /s, AT/Ts equals 0.25. If the coefficients of the digital filters 3' and 3" are calculated for times (before correction) t_ 3 t 2 t 1 l, to, t+l, t+ 2 and t4 3 (seven-coefficient filters), one of the filters will retain these coefficients, the other filter receiving coefficients calculated for the times t3.

2 5 t2.2 5 t_1.25, t-0.25, t+0.

75 t+1.75, and t+2.

75 It is also possible to correct the coefficients of the two digital filters, for example by correcting the coefficients of one of the filters by AT/2Ts and the coefficients of the other of these filters by -AT/2Ts The main thing is therefore that, after correction, the coefficients of one of the filters differ from the coefficients of the other filter calculated for a time difference AT/Ts The same reasoning then applies to the digital filters 12' and 12" of the receiver while taking into account a difference Rx-Ry, called AY.

Another adjustment solution is to constitute in memory a bank of a plurality of typical coefficients cti corresponding to a plurality of predetermined differences Al; and A2 i As a function of measurements Al and A2 made on a transmitter and receiver set, the coefficients ct i corresponding to the measured differences Al and A2 i closest to Al and A2 are applied to the filters 1 2' and 12". The number of typical coefficients used for example equals The digital filters used in the invention are for example Nyquist square root filters whose impulse response at the instants selected by the sampler have a value given by the equation: h(t) cos(1 sin t/Ts}.(4P.t/Ts)- (n.Ts 2 1 4. where 3 is the value of the desired roll-off factor and t designates the value of the instants when it is desired to sample the impulse response.

*6 *Sb

Claims (9)

1. A method of compensating differences in group propagation times between the analog filters of a transmitter of signals in phase quadrature, and between the analog filters of a receiver of signals in phase quadrature, Ssaid transmitter comprising a phase quadrature signal modulator delivering a transmission signal on the basis of first and second baseband signals from first and second processing paths, said first and second processing paths including respective first and second analog transmission filters; said receiver comprising a phase quadrature signal demodulator which, on the basis of the received signal, delivers third and fourth baseband signals to respective third and fourth processing paths, said third and fourth processing paths respectively including third and fourth analog reception filters respectively delivering a third filtered signal X and a fourth filtered signal Y; Ssaid receiver further including a clock recovery circuit delivering a clock signal on the basis of said third and fourth filtered signals, respectively X and Y; Sosaid method comprising the steps of: a 0 I 0: a) applying said first and second signals respectively to said third and fourth analog filters, the signals output by said third and fourth analog filters constituting respective signals X1 and Y1 which are applied to means for estimating 0 time difference (23) and delivering a difference: 094ba: A] (Tx Rx)- (Ty Ry) where Tx, Rx, Ty, and Ry are respective propagation times for the signals through said first, second, third, and fourth filters, said difference Al being representative of the time difference between instants of optimum sampling instants between said signals rO•O X1 and Y1; s b) applying said first and second signals respectively to said fourth and third analog filters, the signals output by said third and fourth analog filters constituting respective signals X2 and Y2 which are applied to said means for 0 0 estimating time difference which deliver a difference: 0*00 A2 (Ty Rx) -(Tx Ry) said difference A2 being representative of the time difference between the optimum sampling instants between said signals X2 and Y2; and c) determining weighting coefficients from said differences Al and A2 for application to fifth, sixth, seventh, and eighth digital filters inserted in said first, second, third, and fourth processing paths, said digital filters being inoperative during said steps a) and thereby compensating said group propagation time differences firstly between said first and second analog filters and secondly between said third and fourth analog filters.

2. A method as claimed in claim 1, wherein the symbol rate of said first and second signals is the highest symbol rate that can be accepted by a transmission system including said transmitter and said receiver.

3. A device for estimating the differences Al and A2 as claimed in 1 or 2, wherein said device comprises: first and second Gardner type circuits respectively receiving said third and fourth filtered signals X and Y, said Gardner type circuits also serving as a clock recovery circuit; frm a subtracter subtracting the signal provided by the second Gardner type circuit from the signal provided by the first Gardner type circuit; and an integrator integrating the signal provided by said subtracter to produce said differences Al and A2.

4. A device as claimed in claim 3, wherein each of said Gardner type circuits S. comprises a first delay circuit receiving an input signal, the output from said first delay 0 circuit being connected to the input of a second delay circuit, a subtracter subtracting the signal input into said first delay circuit from the signal provided by said second 16'" delay circuit, and a multiplier multiplying the signal provided by said first delay circuit .00:0. by the signal provided by said subtracter. 0:

5. A device as claimed in 4, wherein the delay of said delay circuits is of duration 0 Ts/2, where Ts is the symbol time.

6. A device as claimed in any one of claims 3 to 5, wherein said subtracter and said integrator of said circuit for estimating said differences Al and A2 are integrated in said receiver.

7. A system for transmitting signals in quadrature and including said fifth, sixth, seventh and eighth digital filters inserted in said first, second, third, and fourth paths and compensating the group propagation time differences between said analog filters 13 of said transmitter of signals in phase quadrature and between the analog filters of said receiver of signals in phase quadrature, as claimed in any one of claims 1 to the system being characterized in that said digital filters are finite impulse response filters.

8. A method substantially as herein described with reference to Figures 1-5 of the accompanying drawings.

9. A device substantially as herein described with reference to Figures 1-5 of the accompanying drawings. A system substantially as herein described with reference to Figures 1-5 of the accompanying drawings. DATED THIS SIXTEENTH DAY OF APRIL 1997 ALCATEL ALSTHOM COMPAC NE CENERALE dELE CRICITFE T 0 S S *S