Method for the correction of phase and amplitude errors in COFDM signals.

The invention relates to a method for correcting the phase and amplitude of a broadband received signal with the aid of reference signals according to the preamble of claim 1 and a circuit arrangement for performing the method.

From the publication "Development Trends in TV Broadcasting", published in telekom praxis, issue 12/92, pages 32 to 38, it is known that work is being done worldwide on a digital terrestrial television system. The COFDM (Coded Orthogonal Frequency Division Multiplex) method is to be used for digital transmission in this television system. This results in a higher transmission quality with more efficient use of the frequency ranges used for transmission.

The COFDM transmission method is known from WO 88/00417. The information to be transmitted is first converted from analog to digital if the information is available in analog form. The digital information is then encoded with a channel encoder, thereby adding an error protection code to the information. Finally, the channel-coded information is modulated onto a multiplicity of individual carrier frequencies, the time sequence of the information being scrambled. This provides increased protection against time and frequency selective interference.

A method for frequency control in local oscillators of a receiver is known from WO 92/10043. The digital television data received are transmitted using the OFDM (Orthogonal Frequency Division Multiplex) method. In addition to the digital television data, two pilot signals are transmitted in each transmission frame, which have a constant frequency difference. In order to regulate the frequency of the local oscillators in the receiver, the start of a transmission time slot is detected, the position of the two pilot signals in the transmission frame is then determined, and the phase difference between the two pilot signals is then calculated as a function of time. A control variable for the local oscillators can then be derived from this.

A method for frequency response correction is known from the applicant's unpublished application P 42 40 609.9. It is used for terrestrial transmission of digital signals (DAB). Here, frequency-selective amplitude and phase distortions of a differential phase-modulated COFDM transmission signal are detected by a control unit, a corresponding correction characteristic curve is calculated and the spectrum of the transmission signal is weighted with the correction characteristic curve. It is irrelevant whether the detected distortion has arisen on the transmission link or in transmitter or receiver modules. The method also enables the reliability of the received signals to be estimated.

The disadvantage of this method is that a high computational effort is required to calculate the correction characteristic in the case of distortions that fluctuate widely in time and frequency.

Another disadvantage is that the method can only correct a part of the possible distortions in the case of signals that are not phase-modulated in different phases.

It is therefore the object of the invention to specify a method for equalization of a COFDM transmission signal that is not phase-modulated by difference. The computing effort required for this should be as low as possible.

According to the invention, this object is achieved by the characterizing features of patent claim 1.

The method according to claim 1 has the advantage that, due to the reference signal transmitted in each transmission frame, the amplitude and phase distortions of the data signals can be interpolated as a function of time and frequency for the entire transmission frame. This calculation can be carried out for each individual carrier frequency or for a group of several individual carrier frequencies of a transmission frame. Another advantage is that a maximum likelihood decision is made with regard to the amplitude and phase of each individual carrier frequency. As a result, small phase or amplitude distortions can be corrected immediately, with large distortions it is recognized that the reliability of the received signal decreases.

The method also has the advantage that the amplitude correction adapts to changing reception conditions. By taking into account a number of signals received last, a steady course of the equalized received signal amplitude is achieved.

The method according to claim 2 has the advantage that an amplitude and phase correction can also be calculated according to a higher-order correction characteristic due to the known temporal course of the amplitude and phase distortion.

The method according to claim 3 has the advantage that the amplitude correction and the phase correction can be carried out by a single multiplication of the short-term spectrum of the received signal by the correction characteristic.

The method according to claim 4 has the advantage that no Fourier transformation and reverse transformation is required specifically for the frequency-selective correction of the amplitude and phase of the received signal. As a result, considerable computing and circuitry outlay can be avoided.

The use of the method according to claim 5 has the advantage that equalization of television broadcasts is made possible. It is also advantageous that the correction values ​​for amplitude correction can be easily calculated, since the amplitude of the spectrum ideally assumes discrete values ​​when transmitting digital television data using the COFDM method.

The circuit arrangement according to claim 6 has the advantage that the calculation of the correction characteristic, the correction of the individual carrier frequencies and the maximum likelihood decision can be carried out by a control unit, which must be provided for controlling the receiver anyway. The additional circuitry required to correct the received signal is therefore limited to a relatively inexpensive and highly integrated memory.

The invention is described and explained in more detail below. It shows:<dl id="dl0001"><dt>Fig. 1</dt><dd>a possible course of the single carrier frequencies used for transmission according to the COFDM method as complex pointers,</dd><dt>Fig. 2</dt><dd>a possible phase distortion of a single carrier frequency,</dd><dt>Fig. 3</dt><dd>possible phase differences between the phases of individual carriers received and the phases originally modulated onto the individual carrier frequencies,</dd><dt>Fig. 4</dt><dd>possible phase differences between the phases of individual carriers received and the phases originally modulated onto the individual carrier frequencies after a phase correction and</dd><dt>Fig. 5</dt><dd>a circuit arrangement for performing the correction.</dd></dl>

In the following exemplary embodiment, the COFDM method is used for the transmission of digital television data. The individual carrier frequencies are modulated with the 4-PSK (phase shift keying) television data. In each transmission frame, after a synchronization signal SYNC for the time synchronization of the receiver, in which the amplitudes of all individual carrier frequencies are zero, a reference signal REF with a reference phase PR and then a data block with n data signals D1 to Dn, which are phase-modulated with respect to the reference phase PR , received. Due to the 4-PSK modulation of each individual carrier frequency, two binary states, e.g. 01, i.e. a dibit, can be transmitted.

1 shows the transmission signal, the transmitted individual carrier frequencies being shown as complex pointers with amplitude and phase as a function of time t and frequency f. The frequencies f₁ to f_m are the single carrier frequencies of the COFDM transmission method, which are arranged equidistantly on the frequency axis f. The signals SYNC for receiver synchronization, REF for the transmission of the reference phase PR and D1 to Dn for the transmission of the television data are arranged equidistantly on the time axis t and form a transmission frame. Due to the phase modulation of the data signals D1 to Dn, the individual carrier frequencies can be arranged out of phase with the modulation-free reference phase PR of the reference signal REF, which is expressed by corresponding rotations of the complex pointers. The reference phase PR is redefined for each transmission frame by the reference signal REF. Therefore, the reference signal REF can also have no phase distortion. It is not necessary that all individual carrier frequencies of the reference signal REF have the same reference phase PR. 1, the data signals D1 to Dn mostly have phase distortions. These can be caused by the transmission link and / or by transmitter and receiver modules. For the sake of clarity, the representation of possible amplitude distortions has been omitted.

For demodulation in the receiver, the phase difference between the transmitted reference phase PR and the phases of the individual carrier frequencies of the same transmission frame is calculated. If you then decode this phase difference into binary values, you get back the transmitted information.

Due to distortion, the phase difference in 4-PSK modulation is not exactly an even multiple of 90^O, but for example 100^O. A maximum likelihood decision is carried out in order to be able to decode binary states from the phase differences even in the case of distortions which are so small.

This corrects minor phase deviations and the transmission system has a certain fault tolerance. For example, phase values ​​in an interval between 45^O and 135^O in the receiver of a phase of 90^O be assigned. 2 illustrates this graphically. The ideal phase Pi, which is recognized in the maximum likelihood decision, does not correspond to the phase position Pe of the received individual carrier frequency. Possible limits for the maximum likelihood decision are drawn in broken lines.

FIG. 3 shows the phase deviation P of the received individual carrier frequencies from one of the four possible phase positions with a 4-PSK module. By definition, the reference signal REF of a first transmission frame, which is received at time t1, has no phase distortion. The phase distortions of the data signals D1 to Dn are shown as a function of frequency. Since all the individual carrier frequencies have zero amplitude during the synchronization signals SYNC, no signal can be determined. The reference signal REF of the second transmission frame redefines the reference phase PR for the following data signals D1 to Dn, it being assumed that the reference phase PR is time-invariant. Nevertheless, the reference phases PR (t1) and PR (t2) received in different transmission frames can have a phase difference P due to phase distortions. This phase difference is shown in Fig. 3rd for the reference signal REF, which is received at the time t2 in the second transmission frame, is shown as a function of frequency.

If, in particular due to time-dependent parameter fluctuations, there are deviations Δ P of the received phases, as shown in FIG. 3, which are not due to modulation and are not within the correctable error tolerance described above, errors occur in the decoding despite error tolerance. In order to avoid this, phase equalization is carried out on the received signal. With this phase equalization, a control unit calculates the phase difference d_PR between at least two reference phases PR (t1) and PR (t2) of the same single carrier frequency. As a rule, reference signals REF of time-sequential transmission frames are evaluated according to (1).<maths id="math0001" num=""><math display="inline"><mrow><msub><mrow><mtext>d</mtext></mrow><mrow><mtext>PR</mtext></mrow></msub><mtext> = PR (t1) - PR (t2) (1)</mtext></mrow></math><img file="EP0618697A2_D0001.tif" /></maths> Is this phase difference d_PR in terms of magnitude greater than a threshold value, it is recognized that phase equalization is required, and a quotient Q from the phase difference d_PR and the time dTR, which lies between the reception of the two reference signals REF (t1) and REF (t2), according to (2).<maths id="math0002" num=""><img file="EP0618697A2_D0002.tif" /></maths> This quotient Q now indicates the linear time-dependent change in the reference phase PR. A linear phase error Pfi to be expected can thus be calculated for each data signal Di as a function of the time of its transmission in accordance with (3).<maths id="math0003" num=""><math display="inline"><mrow><mtext>Pfi = Q · (T (REF, t1) - T (Di)) (3)</mtext></mrow></math><img file="EP0618697A2_D0003.tif" /></maths> If the phase errors Pfi calculated in this way are subtracted from the phases of the individual carrier frequencies from FIG. 3, linearly time-dependent phase errors can be eliminated.

FIG. 4 shows the phase distortions Pn adjusted for the calculated proportion, which can be corrected on the basis of the subsequent most likelihood decision, so that no transmission error occurs. Here the phase correction can have different effects. The single carrier frequency f1 is considered as an example. As shown in Fig. 3rd this single carrier frequency has no phase distortion Δ P for the data signal D1 and a low phase distortion Δ P for the data signal Dn. The phase distortion Δ P of the reference signal of the second transmission frame REF (t2) already has a medium phase distortion in comparison to the reference signal of the first transmission frame REF (t1). The phase distortion Δ P of the individual carrier frequency f1 thus increases with time t, but not linearly. Because of the linearly calculated phase correction Pfi, there is therefore a phase distortion Pn for the individual carrier frequency f1 even after the phase correction, but this is less, as shown in FIG. 4. A phase distortion Pn arises for the data signal D1, which lies within the error tolerance and therefore does not lead to a transmission error. There is no phase distortion Pn of the individual carrier frequency f1 for the data signal Dn, since the originally existing phase distortion P can be eliminated by the linear correction.

The time dependence of the phase distortion P can be analyzed by a control unit MP, so that an optimal mathematical model for the calculation of the phase correction values ​​Pfi for the interpolation of the phases of the data signals D1 to Dn can be found, which of course does not have to be linear.

It is possible to include reference phases PR of transmission frames that have already been received earlier and to carry out a time-dependent weighting of these reference phases PR. This allows for phase distortions that are not linear over time. Equations (1) to (3) must then be modified accordingly.

In order not f₁ to f for each individual carrier frequency_m Having to perform the calculation described above, it is conceivable, for example, to perform the above calculation only for every second or every 16th individual carrier frequency and to interpolate missing quotients Q from the calculated quotients Q as a function of the individual carrier frequency. As a result, channel capacity in the reference signal REF can be made free, which can be used for the transmission of additional data.

According to the phase correction described above, the amplitudes of the individual carrier frequencies are also corrected. The amplitudes of the reference signals are temporarily stored in a memory SPE, the amplitudes of the individual carrier frequencies are redefined on the basis of the interpolated amplitude correction values. The interpolation of the amplitude can take place both as a function of time and of frequency.

A complex correction characteristic curve is then calculated from the frequency-dependent amplitude and phase correction, by which the short-term spectrum SP of the received signal is multiplied. The correction characteristic is composed of the amplification value for the amplitude and the phase correction value Pfi for each individual carrier frequency. The multiplication can be carried out at a point in the signal path at which the short-term spectrum already exists, for example after the Fourier transform FFT.

Alternatively, a correction pointer can also be calculated for a single carrier frequency, which results from the difference pointer between two reference pointers of a single carrier frequency. Depending on the time of reception of the signal to be corrected, this is multiplied by a scale factor and added to the originally received pointer in order to obtain the phase and amplitude corrected pointer for the individual carrier frequency.

In order not to have to calculate this correction characteristic for each transmission frame again, it is stored in a memory SPE and can be used to correct data signals of several transmission frames. A recalculation of the correction characteristic can be initiated by the user or the control unit MP. For this purpose, the control unit MP evaluates the equalized amplitudes and the equalized phases of the individual carrier frequencies. As soon as the amplitudes or phases of the reference signals REF fluctuate greatly, the correction characteristic curve must be recalculated. 5 shows a circuit arrangement for carrying out the correction of the phase and amplitude of the received signal. After the digital received signal of a first transmission frame is available as a short-term spectrum SP in the spectral range due to a fast Fourier transformation FFT, a control unit MP evaluates this short-term spectrum SP. The reference signal REF and the data signals D1 to Dn are buffered in a memory SPE. Likewise, the signals of a second transmission frame, which is transmitted after the first, are temporarily stored in the memory SPE. The complex correction characteristic curve is then calculated by the control unit MP from the reference signals REF of the two transmission frames and is likewise stored in the memory SPE. After the individual carrier frequencies of the first transmission frame have been multiplied by the correction characteristic, the individual carrier frequencies are fed to the maximum likelihood decision by the control unit MP and finally processed in the demodulator DMOD. If the maximum likelihood decision determines the maximum likelihood decision that the phases of the received individual carrier frequencies strongly differ from the possible modulation phases (with 4-PSK modulation: 90^O, 180^O, 270^O and 360^O) differ, the information is recognized as unreliable.

The received data of the first transmission frame are then overwritten in the memory SPE by the received data of the following third transmission frame. The received data of the second transmission frame are then multiplied by the correction characteristic previously calculated and fed to the maximum likelihood decision. If it is determined by the control unit MP during this decision that the amplitude fluctuations or the phase distortions are large, a new correction characteristic curve is calculated by the control unit MP. This new correction characteristic is calculated as a function of the reference signals REF of the second and third transmission frames. The new correction characteristic curve replaces the previously stored one in the SPE and is used to correct the received data after the first transmission frame. As soon as it is necessary, a new correction characteristic is calculated again.

If nonlinear distortions of the phase are also to be corrected, reference signals REF which have already been transmitted must also be stored in the memory SPE and must be taken into account by the control unit MP in the calculation, in particular the phase correction values.