FR2770710A1 - METHOD AND APPARATUS FOR CORRECTING A DERIVING ERROR IN A SAMPLE DATA TRANSMISSION SYSTEM - Google Patents

Abstract

Data sampled by a receiver operating in a digital data transmission system is processed by a low-pass digital filter having a plurality of filter coefficients. A sample timing error is recovered from the processed data and is used to adjust the plurality of filter coefficients of the filter in order to correct for timing errors.

Description

 The present invention relates to a method and apparatus for correcting a timing error in a sampled data transmission system, such method and device being applicable to a digital radio receiver operable in a telemetry system, wherein Data is transmitted using a packet-based protocol.

In particular, the invention relates to data detection, timing recovery and correction, phase recovery and correction, and demodulation functions in a radio receiver.

 Figure 1, appended to this application, shows a block diagram of a known digital radio receiver incorporating a timing recovery function. A continuous waveform present at the input is converted into a discrete waveform by a sampling device 11, the frequency and timing of the sampling being determined by a clock 12 through a control line 15. The sampled waveform is converted into a digital data stream by an analog-to-digital converter 13. Errors in the sampling timing of the sampling device 11 are detected or retrieved in the signal processor 14, and a setting based on this error is returned by reaction to the clock 12.

The sampling timing error is therefore corrected so that the next sampling is executed at a correct time.

 A disadvantage in this prior art system is that a response of an adjustment to the clock 12 requires the use of a control line and an interface circuit between the digital signal processor 14 and the clock 12, which increases the cost of manufacture.

In addition, the system is characterized by a delay between the detection of the sampling timing error and the correction time of the error in the sampling device. This delay increases the period during which the data is sampled at a point in time. As a result, samples necessary for demodulation may be lost.

 According to a first aspect of the present invention, there is provided a method for correcting a sampling timing error in a digital data transmission system, wherein the sampled data is processed by a low pass digital filter having a plurality of of filter coefficients, characterized in that it consists in recovering a sampling timing error from data sampled by a receiver, and setting at least one of a plurality of coefficients of the digital filter according to the error sampled sampling timing.

 According to a second aspect of the present invention, there is provided a device for correcting a sampling timing error in a digital data transmission system, wherein the sampled data is processed by a low-pass digital filter having a plurality of filter coefficients, characterized in that it comprises timing recovery means for recovering a sampling timing error from the sampled data, by a receiver, and setting means coupled to the timing recovery means, for adjusting at least one of the plurality of coefficients of the digital filter based on a recovered sampling timing error.

 The method and the device according to the invention have the advantage that, by adjusting the coefficients of the low-pass digital filter, a feedback system for sampling timing correction can be realized entirely in the digital domain. . This allows timing corrections to be made in real time so that a high data rate can be achieved in the receiver, while maintaining the timing accuracy of the sampled data.

 In a preferred embodiment according to the first aspect of the invention, the step of retrieving a sampling timing error from data sampled by a receiver comprises correlating the sampled data with at least one data set. previously stored in the receiver.

 Preferably the transmitted data includes a predetermined mark part included in the data, preceding for example a data area.

 Suitably, the data sampled by the receiver comprises an in-phase component I and an out-of-phase Q component of n / 2. The timing recovery may then include complex correlation of the I and Q components of the sampled data with I and Q data sets previously stored in the receiver. Preferably, the previously stored I and Q data sets comprise a sequence corresponding to the marking position, which provides a distinctive feature with respect to random data. Preferably the marking portion comprises a golden code sequence since this is particularly distinctive with respect to a random noise. Preferably, the low-pass digital filter is a so-called RRC cosine root band limiting filter in the form of a finite impulse response (FIR) filter.

 According to a third aspect of the present invention, there is provided a method of detecting data in a digital radio receiver, wherein a received signal is sampled to produce a series of received signal samples, characterized in that it consists of correlating a set of received signal samples with at least one set of data previously stored in the receiver, comparing the correlation result with a predetermined threshold value, and producing a data detection signal if the result of the correlation is greater than the threshold value or repeat the process with a subsequently received set of received signal samples, if the result of the correlation is not greater than the threshold value.

 Preferably, the sampling rate of the receiver is at least twice the rate of the symbols of the transmitted data. Advantageously, the correlation is performed on each nth sample among the received signal samples.

 This reduces the computational load for the execution of correlations. Preferably N> 1 and advantageously N = 2.

According to a fourth aspect of the present invention there is provided a method for recovering a phase error associated with data samples I and
Q received in a digital radio receiver, characterized in that it comprises the complex correlation of a set of received I and Q data samples with data sets I and Q previously stored in the receiver to produce a component correlation real and imaginary correlation component, process the real and imaginary correlation components to determine a sampling timing error, process the real and imaginary correlation components to determine an estimate of the phase error and adjust the estimating the phase error as a function of the sampling timing error.

 Preferably, the phase error determined according to the fourth aspect of the present invention is corrected in subsequent received data.

 According to a fifth aspect of the present invention, there is provided a digital data transmission system comprising a transmitter for transmitting digital data packets in a radio channel, and a receiver tuned to the transmitter and comprising means for setting implementing a method according to the first, third and / or fourth aspects of the present invention.

 Conveniently the digital radio receiver is a synchronous digital radio receiver and the digital data transmission system is a synchronous digital data transmission system.

Other features and advantages of the present invention will emerge from the following description given with reference to the accompanying drawings, in which:
FIG. 1 is a block diagram of a receiver of the prior art including a feedback circuit for correcting timing errors;
FIG. 2 is a diagram showing the structure of radio transmission packets of a radiotelemetry system associated with the invention;
FIG. 3 is a block diagram of a device according to the invention incorporating functions of digital filtering, data detection, timing recovery and data demodulation;
Fig. 4 is a graph showing complex cross-correlation values for the separated I and Q correlation components as a function of the sampling timing of the I and Q correlation data sets;
FIG. 5 is a graph showing the squared amplitude of the complex cross correlation of FIG. 4 as a function of the sampling timing of the correlation data sets;
FIG. 6 is a graph representing the amplitude of the complex cross-correlation of FIG. 4;
FIG. 7 is a graph showing the squared amplitude of three discrete complex cross-correlations performed for an early set of data, a timely set of data, and a late set of data; and
FIG. 8 is a graph showing the filter coefficients for an ideal sampling timing and for a compensated sampling timing and the corresponding impulse responses of the digital filter.

 A telemetry system includes a control center for bidirectional radio communication with a network of multiple remote radio units. The uplink and downlink for the telemetry system both operate using a packet-based protocol.

 Figure 2 shows the structure of an in-phase component I and a Q-component out of phase from w / 2 radians of a radio data packet used in the telemetry system. Each data packet contains a synchronization sequence and a data block. The synchronization sequence is coded according to a quadrature phase shift keying (QPSK) modulation. The data block of the uplink is modulated according to a QAM (QAM designating quadrature amplitude modulation), while the downlink is modulated according to the QPSK modulation. Each synchronization sequence comprises 31 symbols with a cyclic shift symbol included on either side of the synchronization sequence to maintain a correct correlation peak. For the data packet I, the synchronization sequence consists of symbols 1 to 31 of a golden code.

For the data packet Q, the synchronization sequence consists of symbols 16 to 31 of the golden code, followed by symbols 1 to 15 of the golden code. The actual data transmitted in the data packet follows the synchronization sequence and the cyclic shift symbols.

 Figure 3 illustrates the processes of data sensing, timing and phase recovery, and demodulation in the receiver of a remote radio unit. This section of the receiver receives I and Q samples in a 16-bit form, at a rate of at least twice the symbol rate, typically at a rate of about 15 kHz. These samples are temporarily stored for processing by the filter 31. Preferably, this receiver section is made using digital signal processing equipment.

Filtered
The filter 31 is a digital bandlimiting low pass filter, which acts on the sample channels in the I and Q channels. The filter is realized as a finite impulse response (FIR) filter having coefficients predetermined ones which provide a frequency response caxnue in the art under the nan RRL: (root cosine). The filter is docked on a RRC (cosine root) filter located in the transmitter that emitted the I and Q data samples.

Therefore the effect of both filters is to produce a cosine frequency response in the system.

dans laquelle h(k), k=0,1,..., N-l sont les coefficients de réponse impulsionnelle du filtre, N est la longueur du filtre, c'est-à-dire le nombre de coefficients du filtre, et y(n) est l'échantillon de sortie actuel, qui est fonction de valeurs antérieure et présente des échantillons d'entrée x(n) . La figure 8 représente schématiquement les coefficients h(0) à h(8) associés aux éléments de retardement d'un filtre FIR numérique. La courbe en trait plein A indique la réponse impulsionnelle du filtre. Dans la pratique, on peut utiliser plus de 9 coefficients et on peut utiliser des nombres différents de coefficients dans la liaison montante et dans la liaison descendante.The finite impulse response filter is characterized by the following relationship

in which h (k), k = 0.1, ..., Nl are the impulse response coefficients of the filter, N is the filter length, ie the number of filter coefficients, and y (n) is the current output sample, which is a function of previous values and has input samples x (n). FIG. 8 schematically represents the coefficients h (0) to h (8) associated with the delay elements of a digital FIR filter. The solid line curve A indicates the impulse response of the filter. In practice, more than 9 coefficients can be used and different numbers of coefficients in the uplink and in the downlink can be used.

Data detection
With reference to FIG. 3, the samples I and
Q from the RRC filter 31 are sent in the data detection means 32. The data detection means applies a complex correlation to the I and Q samples received with gold codes I and Q previously stored in the receiver. The previously stored I and Q codes have a length of 31 symbols and are consistent with the I and Q synchronization sequences of a radio data packet. When the receiver upsamples the I and Q samples at a rate at least twice the symbol rate, the data detection process according to the present invention receives a sequence of 62 I and Q samples and performs a complex correlation using another pair of I and Q samples. This reduces the computational load applied to the data detecting means with respect to performing a complex correlation of any received I and Q samples. In turn this allows the data detecting means to calculate in real time a new complex correlation since each pair of I and Q samples is received.

dans lesquelles I(2n+N) et Q(2n+N) sont les (2n+N)-èmes échantillons de la séquence de 62 échantillons I et Q reçus, et Ig < n) et Qg(n) sont les n-èmes symboles dans les codes d'or I et Q préalablement mémorisés. N est un facteur qui dépend du temps et qui augmente de un chaque fois qu'un nouveau couple d'échantillons I et Q est reçu.Each complex correlation calculation involves four cross-correlations between the received I and Q samples and the previously stored I and Q gold codes. These cross-correlations are summarized in the following relationships

where I (2n + N) and Q (2n + N) are the (2n + N) -ths samples of the sequence of 62 samples I and Q received, and Ig <n) and Qg (n) are the n Symbols in the I and Q codes previously memorized. N is a time dependent factor that increases by one each time a new pair of I and Q samples are received.

 The real part of the complex correlation is equal to the sum of the cross correlation i * Ig and the cross correlation Q * Qg, while the imaginary part of the complex correlation is equal to the difference between the cross correlation I * Qg and the cross correlation Q * Qg.

Using the Pythagorean theorem, the square of the amplitude of the complex correlation is provided by the following relation
amplitude2 = (I * Ig + Q * Qg) 2 + (I * Qg - Q * ig) 2
Figure 6 shows how the amplitude of the complex correlation produces a peak when the correct timing is reached between samples I and
Q received from a synchronized sequence and the golden codes I and
Q previously stored. A threshold test is performed to establish the existence of the correlation peak, and therefore the presence of a received data packet. In practice, the threshold test is applied to the square values of the magnitude of the complex correlations to reduce the computational loads in the data detecting means.

 Figure 5 shows the square of the amplitude of the complex correlations plot as a function of time, together with a threshold level set at 3000. However, one can choose a lower threshold to avoid the absence of a packet with a mistake of The timing equals approximately half of the sample duration.

Consequently, for FIG. 5, a threshold slightly greater than 2500 may be chosen. It can be seen that the square of the amplitude of the complex correlation exceeds this threshold value for the sampling instant 4. Therefore, the complex correlation for this sampling instant indicates the beginning of a data packet.

 The threshold level is optimized to ensure that all valid packets are detected while avoiding the detection of spurious data. The presence of cyclic shift symbols on either side of the I and Q timing sequences ensures that the correlation point shifted from the peak does not exceed the ideal values, even when random data samples are attached to the sequence.

Recovery of timing
Figures 5 and 6 show the correlation peaks with a correlation value of the detected (timely) data located at the peak of the correlation peak. This situation occurs when samples are taken at the correct timing. Retrieving the timing involves processing the correlation values to retrieve a timing error for sampling at an incorrect time. With reference to FIG. 3, the timing recovery means 33 receive the correlation values from the data detection means 32.

 When a data packet is detected based on a particular complex correlation, the timing recovery process controls the data detecting means to recalculate the complex correlation for the set of samples I and Q associated with the previous sample and to calculate the complex correlation for this set of samples I and Q associated with the next sample. These calculations are made possible by the correlator (data detection means) which temporarily stores the I and Q samples when received.

To recover the timing error, the shape of the correlation peak in FIG. 5 is modulated in the form of a single triangular peak, as shown in FIG. 7, the base having a width equal to twice the period of the symbol. which is equal to four times the sampling period T. Figure 7 also shows the square of the amplitude of complex correlations calculated a sample before, in a timely manner and a sample later. The timing error te during the timely correlation is calculated as a fraction of the sampling period T using the following relation
amplitude [early] 2 - amplitude [late] 2
T amplitude [early] 2 + amplitude [late] 2
Therefore, the timing error can be calculated using only the squares of the amplitudes that require less computation.

Recovery of the phase
If the I and Q samples are correctly in phase with each other and if there is no similarity between the I and Q samples and if the sampling timing is correct, then the imaginary components of the complex cross correlation can be zero. The non-similarity between the components I and Q of the synchronization sequence of a radio data packet is established for the Q component of the gold code shifted cyclically by 15 symbols with respect to the component I of the gold code, as shown in Figure 2.

When phase errors occur, the magnitude of the imaginary components of the complex cross correlation increases. The actual components of the complex cross correlation undergo a corresponding amplitude reduction. An estimate of the phase error is calculated using the following relationship
(I * Qg - Q * ig)
phase error = tg1
(I * Ig + Q * Qg)
Otherwise, the sine or cosine phase error can be recovered separately using the following relationships
(I * Ig + Q * Qg) (phase error) in cosine = ----------------------------
amplitude [in a timely manner]
(I * Qg - Q * Ig) (phase error) in sine = -----------------------------
amplitude [in a timely manner]
This is easier from a computational point of view than the calculation of the phase error itself.

 This estimate of the phase error does not take into account additional errors that result from an incorrect sampling timing. These additional errors occur because of the non-ideal nature of cross-correlation of I and Q timing sequences caused by RRC filtering and transmit and receive interpolation.

 With reference to FIG. 4, these additional errors disappear only at the precise instant of appearance of the symbol. However, according to the invention, a linear adjustment can be made to the phase error to take account of the sampling timing error recovered by the recovery means. This setting is shown in Figure 3 by the data flow marked "phase adjustment".

 The recovered phase is delivered by the phase recovery means 34 to the demodulation means 36 to correct phase errors during the demodulation.

Correction of timing
With reference to FIG. 3, the timing correction process is enabled by the feedback of the timing error recovered from the timing recovery means 33 to the RRC filter 31.

As part of the timing correction process, the retrieved timing error is delivered to you as an input signal to the RRC coefficient setting means as indicated by the command marked "packet timing correction". . The means 35 for adjusting the RRC coefficients are able to adjust the coefficients of the RRC filter 31 to compensate for timing errors in the sampling device. A look-up table is provided in the RRC coefficient setting means 35 and contains 16 sets of values for the coefficients h (0) to h (k). Each set of values for the coefficients provides an impulse response for the filter, suitable for a particular timing error. With reference to FIG. 8, there is shown a set of values for the filter coefficients marked h '(0) to h' (k) (with k = 8 in the simplified example shown in FIG. with the associated impulse response curve B.

The impulse response curve B is shifted along the time axis with respect to the impulse response curve A. This duration is equal to the timing error that the filter 31 has to absorb.

 The 16 sets of coefficient values correspond to timing errors separated by 1/16 th of the sampling period T to provide maximum correction of a sampling period. The timing correction process is therefore able to correct the full range of possible timing errors. A correction of the asynchronous sampling in the receiver is executed once for each packet of received radio data.

demodulation
The actual data contained in the radio data packet is demodulated after detection of a valid synchronization sequence (data detection).

The data detection is signaled to the demodulation means as shown in FIG. 3 via the central line marked "demodulation of a packet". The actual data samples I and Q are received by the demodulator from the filter 31 and a demodulation is applied to each of the samples.
I and Q so that the demodulation occurs at the rate of the symbols.

 The recovered phase is corrected in the demodulator so that the received I and Q samples are aligned with the overall pattern for QPSK or QAM coding.

 In view of the foregoing description, it is obvious that various modifications may be made within the scope of the invention. For example, the golden code used in the synchronization sequence may be replaced by an equivalent distinctive code, or the filter arrangement forms may be modified possibly including the replacement of the FIR filters by infinite impulse response filters ( IIR).

Claims (18)

 A method for correcting a sampling timing error in a digital data transmission system, wherein the sampled data is processed by a low-pass digital filter (31) having a plurality of filter coefficients, characterized in that it consists in recovering (at 33) a sampling timing error from data sampled by a receiver, and setting (at 35) at least one of a plurality of coefficients of the digital filter according to the error of Sampling timing retrieved.  2. Method according to claim 1, characterized in that it is applied to a radio receiver.  3. Method according to either of claims 1 or 2, characterized in that the step of recovering (at 33) a sampling timing error from data sampled by a receiver comprises the correlation sampled data with at least one set of data previously stored in the receiver.  4. Method according to any one of claims 1 to 3, characterized in that the data sampled by the receiver comprises an in-phase component I and a phase-shifted Q component of in / 2.  5. Method according to claim 4, characterized in that the step of recovering (at 33) a sampling timing error from data sampled by the receiver comprises the complex correlation of the components I and Q of the data sampled with data sets I and Q previously stored in the receiver.  6. Method according to any one of claims 1 to 5, characterized in that the data transmitted in the system include a predetermined marking portion preceding a data area.  7. Method according to claim 6, characterized in that the previously stored data sets comprise a sequence corresponding to the predetermined marking portion.  8. Method according to either of claims 6 and 7, characterized in that the marking part comprises a golden code sequence.  9. Method according to any one of claims 1 to 8, characterized in that the low-pass digital filter is a RRC (bandlimit cosine root) filter arranged in the form of a finite impulse response filter ( FIR).  Apparatus for correcting a sampling timing error in a digital data transmission system, wherein the sampled data is processed by a low-pass digital filter (31) having a plurality of filter coefficients, characterized in that it comprises timing recovery means (33) for recovering a sampling timing error from the sampled data, by a receiver, and setting means (35) coupled to the timing recovery means, for setting to minus one of the plurality of coefficients of the digital filter based on a retrieved sample timing error.  A method of detecting data in a digital radio receiver, wherein a received signal is sampled to produce a series of received signal samples, characterized by correlating a set of received signal samples. with at least one set of data previously stored in the receiver, comparing the correlation result with a predetermined threshold value, and producing a data detection signal if the result of the correlation is greater than the threshold value or repeat the process with a subsequently received set of received signal samples, if the result of the correlation is not greater than the threshold value.  12. The method of claim 11, characterized in that the sampling rate of the receiver is equal to at least twice the rate of the symbols of the transmitted data.  13. Method according to either of claims 11 and 12, characterized in that the correlation is performed on each n-th sample among the received signal samples.  14. The method of claim 13, characterized in that n is greater than 1.  15. The method of claim 14, characterized in that n = 2.  16. A method for recovering a phase error associated with I and Q data samples received in a digital radio receiver, characterized in that it comprises the complex correlation of a set of I and Q data samples received. with datasets I and Q previously stored in the receiver to produce a real correlation component and an imaginary correlation component, process the real and imaginary correlation components to determine a sampling timing error, process the real and imaginary correlation components to determine estimating the phase error and adjusting the estimate of the phase error as a function of the sampling timing error.  17. The method of claim 16, characterized in that the phase error is corrected in subsequent received data.  18. A digital data transmission system comprising a transmitter for transmitting digital data packets in a radio channel, and a receiver tuned to the transmitter and comprising means for carrying out a method according to any one of Claims 1 to 9 or 11 to 17.