JP2005191662A - Method of demodulating ofdm signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a transmission path equalization algorithm which is strong to Rayleigh Fading and suitable for high-speed mobile reception in a demodulator of an OFDM modulating system. <P>SOLUTION: In a time region equalizer 11, equalization processing in a time region is applied to a received symbol signal. After performing time equalization, the symbol signal is subjected to FFT operation in an FFT operator 12 and becomes a received OFDM signal. Next, in a frequency region equalizer 14, equalization processing in the frequency region is applied to the received OFDM. In concrete, equalization processing for a symbol direction and a carrier direction are performed using an FIR filter. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a receiving method in OFDM digital transmission.

  In Japanese terrestrial digital television broadcasting, an OFDM (Orthogonal Frequency Division Multiplexing) system is adopted as a transmission system. The OFDM scheme is one of multicarrier transmission schemes in which a transmission signal is divided into a plurality of carriers and transmitted, and the spectrum of each subchannel can be densely arranged, which is strong against frequency selective fading of a multipath transmission path. There are advantages such as high frequency utilization efficiency.

  On the other hand, since the OFDM signal is generated by IDFT conversion, resistance to a time-varying transmission line such as a Rayleigh fading transmission line is weak. In particular, in mode 3 with a long symbol length, an OFDM signal that is carrier-modulated with 64QAM cannot be demodulated normally when the Doppler frequency exceeds 20 Hz. According to the experimental results of NHK, even in diversity reception of sub-carrier level MRC (Maximum Ratio Combining) combining of four antennas, the environment where demodulation is possible is an environment where the Doppler frequency is up to 45 Hz. In 19 channels (509 MHz), a Doppler frequency of 45 Hz corresponds to a moving speed of 95 km / h. This is a speed that is less than 100 km / h, which is the speed limit of Japanese highways, and does not meet the need for digital broadcast reception while driving on highways.

  The main cause of the weakness of the OFDM signal in the Rayleigh fading transmission path is that Rayleigh fading has a multiplicative property, and the orthogonality is destroyed by the time-varying transmission path, so that inter-carrier interference (ICI; Inter Channel) Interference) occurs. In the case of 1-path Rayleigh, the received OFDM signal is expressed by Equation 1.

  In Equation 1, a (n) is an OFDM signal, r (n) is Rayleigh fading, and v (n) is additive Gaussian noise (AWGN). When this OFDM signal is demodulated by the FFT operation in the demodulator, it becomes a signal expressed by equation (2).

  However, in Equation 2, Y (k), A (k), R (k), V (k) are FFT of y (n), a (n), r (n), v (n), respectively. It is the result of operation, and * is a convolution integral code. A (k) * R (k, k) is a normal received signal, and a term with a sigma code is an ICI component. The cause of ICI is that r (n) changes over time. When r (n) does not change with time, that is, when r (n) is a constant, R (k) is a delta function, R (k, m) = 0 (k ≠ m), and ICI Does not occur. Therefore, in order to correctly demodulate the OFDM signal, a condition that the transmission path within a period of one symbol does not change with time is necessary.

JP 2000-286817 A

  However, in a long symbol such as Japanese standard mode 3, the transmission line response fluctuates even during one symbol period. In the case of 1-path Rayleigh fading, a technique based on time-domain equalization has been reported at academic societies as a technique for compensating for transmission path response fluctuations with one symbol. However, the time domain equalization method is less effective for multipath Rayleigh fading such as TU6 (Typical Urban 6-path Rayleigh fading channel model). The reason for this is that in the case of multipath Rayleigh fading, the received signal is expressed by Equation 3, and fluctuations in the time domain are not simple linear functions.

  In Equation 3, i is the number of each path in multipath Rayleigh fading, and M is the total number of multipaths. Therefore, in the case of TU6, M = 6. Further, a (n−τi) indicates an OFDM signal, ri (n−τi) indicates Rayleigh fading, βi is an attenuation coefficient of each path, and τi is a delay time of each path.

  Also, the transmission path function multiplied by the signal demodulated by FFT is not a R (k, k) in the case of the one-path Rayleigh shown in the equation 2, but a very complicated transfer function. For this reason, in the case of multipath Rayleigh fading, accurate transmission path estimation becomes even more important.

  The OFDM demodulator estimates a transmission path using a pilot signal embedded in data and equalizes received data. Due to the pilot arrangement of Japanese standards, the transmission path estimation needs to be estimated together with the symbol direction (estimating temporal fluctuation of the transmission path) and the carrier direction (estimating frequency fluctuation of the transmission path). As a general estimation method, an estimation method of straight line estimation in symbol direction + FIR estimation in carrier direction is often used. With such an estimation method, sufficient performance can be obtained in stationary reception, and the circuit scale can be reduced.

  However, in the case of mobile reception, transmission path fluctuations in the symbol direction also increase due to the influence of Rayleigh fading, and transmission path estimation cannot be performed correctly by linear interpolation. Therefore, FIR interpolation that can cope with the assumed maximum Doppler frequency is required. In linear interpolation in the symbol direction, pilots for 7 symbols may be buffered. When interpolation is performed using an FIR filter, the number of symbols is 48 in order to balance circuit scale and performance. Is said to be necessary, and the capacity of the memory for buffering becomes seven times.

  In the above-mentioned Patent Document 1, it is mentioned that the interpolation accuracy is improved by performing equalization processing on the received signal in both the time domain and the frequency domain, but an FIR filter is used in the frequency domain. In some cases, no solution has been shown for the problem that the circuit scale becomes large.

  In view of the above problems, the present invention is an OFDM modulation demodulator used in terrestrial digital television and the like, which is resistant to Rayleigh fading without increasing the circuit scale, and is suitable for high-speed mobile reception. It is an object to provide an optimization algorithm.

  In order to solve the above-mentioned problem, the invention described in claim 1 is a demodulation method in OFDM (Orthogonal Frequency Division Multiplex) transmission, wherein a first step of receiving an OFDM transmission signal, A second step of performing equalization processing in the region, a third step of performing FFT conversion on the symbol signal after time equalization processing, and a symbol direction using an FIR filter for the received OFDM signal after FFT conversion And a fourth step of performing an equalization process in the carrier direction.

  The invention according to claim 2 is the OFDM signal demodulating method according to claim 1, wherein the number of FIR filters used in the fourth step is less than the number of taps required when time equalization processing is not performed. It is characterized by the number of taps.

  According to a third aspect of the present invention, in the OFDM signal demodulation method according to the first or second aspect, the second step includes a second step of calculating a Rayleigh fading function r (n), a symbol, A step of dividing the signal by the Rayleigh fading function r (n), and the step 2.1 includes, for a plurality of signals within the guard interval, signals within the guard interval (first signal). And the step 2.1.1 for obtaining the slope of a straight line connecting the signal in the effective symbol (second signal) that is the copy source of the signal, and the plurality of the steps obtained in the step 2.1.1 And 2.1.2 step of calculating the average of the slopes as the slope of the Rayleigh fading function r (n).

  The invention according to claim 4 is a demodulation method in OFDM (Orthogonal Frequency Division Multiplex) transmission, in which a first step of calculating a Rayleigh fading function r (n), and a received OFDM signal as a Rayleigh fading function r ( a second step of performing equalization in the time domain by dividing by n), wherein the first step includes, for a plurality of signals in the guard interval, signals in the guard interval (first signal) and The 1.1 step for obtaining the slope of the straight line connecting the signal (second signal) in the effective symbol that is the copy source of the signal, and the average of the plurality of slopes obtained in the 1.1 step And 1.2th step of calculating as the gradient of the fading function r (n).

  Since the present invention performs equalization processing in the symbol direction and the carrier direction using an FIR filter in the frequency domain, the accuracy of channel estimation is high. Furthermore, since equalization processing is performed in the time domain, the number of taps of the symbol direction interpolation FIR filter can be reduced, and the circuit scale can be reduced.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

{Reception process flow}
As a transmission system for terrestrial digital broadcasting, an OFDM system for multiplexing and transmitting hundreds to thousands of subcarriers within one channel band is adopted in Japan, Europe, and the like. This OFDM scheme is a multi-carrier modulation scheme that divides transmission data into a plurality of subcarriers and transmits it. Therefore, the frequency utilization efficiency is very high and it is resistant to frequency selective fading interference that occurs during mobile reception. In addition, under the condition that the total bit rate transmitted in the 6 MHz bandwidth is the same, the symbol period of each carrier is longer by the number of carriers (several hundreds to thousands of minutes) than that of the normal single carrier modulation scheme, By providing a protection period called a guard interval between each effective symbol, the influence of multipath (ghost) can be reduced, so that there is an advantage that image quality deterioration can be suppressed.

  FIG. 1 is a functional block diagram schematically showing an OFDM receiving apparatus according to the present embodiment. An RF (Radio Frequency) signal 1 transmitted from an OFDM transmitter (not shown) is received by a receiving antenna 2 through a transmission path. The received RF signal is frequency-converted by the tuner 3 into an IF (Intermediate Frequency) signal. The IF signal is input to the mixer 5 via the BPF (band pass filter) 4, multiplied by the signal supplied from the carrier wave oscillator 6, and then output to the LPF (low pass filter) 7. The signal from which the high frequency component has been removed in the LPF 7 is output to the A / D converter 8, and the signal input to the A / D converter 8 is converted into a digital signal (symbol signal) at a predetermined sampling frequency. The symbol signal output from the A / D converter 8 is multiplied by a frequency correction coefficient in the multiplier 9 and then output to the serial / parallel converter 10. The symbol signal input as a serial signal is converted into a parallel signal by the serial / parallel converter 10 and output to the time domain equalizer 11.

  Although details of specific processing in the time equalizer 11 will be described later, the time domain equalizer 11 executes equalization processing of symbol signals in the time domain. Then, the equalized symbol signal is output to an FFT (Fast Fourier Transform) calculator 12.

  The FFT calculator 12 performs Fourier transform on the input time-domain symbol signal to a frequency-domain signal (this signal will be referred to as a received OFDM signal). The received OFDM signal is output to frequency shift detector 13 and frequency domain equalizer 14. The frequency shift detector 13 detects a frequency shift based on the input received OFDM signal and outputs a frequency correction coefficient to the multiplier 9. As described above, the symbol signal after A / D conversion is multiplied by a frequency correction coefficient, and correction processing for frequency deviation is performed.

  On the other hand, the received OFDM signal output from the FFT calculator 12 is output to the frequency domain equalizer 14 that performs equalization processing of the received signal in the frequency domain. The received OFDM signal is equalized in the frequency domain equalizer 14, and the received OFDM signal after the equalization is converted from a parallel signal to a serial signal by the parallel-serial converter 15, and then the channel decoder 16. Viterbi decoding and Reed-Solomon decoding are performed, then source decoder 17 performs decoding such as MPEG (Moving Picture Experts Group) -2 method, and then analogizes and outputs by D / A converter 18 Is done.

{Time domain equalization processing}
Next, the time domain equalization process performed in the time domain equalizer 11 will be described. As shown in FIG. 2, the time domain symbol signal has a guard interval (GI) period. The guard interval is a copy of a part of the rear part of the effective symbol. By using this guard interval signal, the slope of Rayleigh fading can be obtained. Since the Rayleigh fading in one symbol section can be regarded as a linear function, the Rayleigh fading function in the effective symbol section can be calculated if the slope is known. The symbol signal is corrected using the function.

  FIG. 3 shows signals at each time in one symbol signal. g1, g2,... gL are signals in the guard interval (L is the number of signals in the guard interval), and p1, p2,... pL are copy source signals of the guard interval. That is, gi and pi are originally the same signal (i = 1, 2,... L). Gi changes to pi through a transmission line that changes with time.

  When g1 and p1 are applied to the coordinates as shown in FIG. 4, the equation gp1 (n) of the g1-p1 straight line can be obtained as shown in Equation 4.

  Since the slope of the straight line obtained by Equation 4 depends on the data, gi and pi (i-1,2, ... L) are based on the same signal as shown in Equation 5. Remove data dependencies.

  Therefore, the slope of the Rayleigh fading function r (n) is given by equation (6).

  The same calculation is performed for i = 2, 3,... M (1 <M <L). That is, the inclination αi is obtained for each of g2 to p2, g3 to p3... GM to pM. When there is no interference such as noise, all the inclinations αi are the same, but taking into account the presence of noise, the average value α of αi is taken. Therefore, α is expressed by Equation 7.

  As a result, the slope of the Rayleigh fading function r (n) for each symbol is obtained. When the Rayleigh fading function r (n) is obtained, signal equalization (signal correction) is performed for each symbol in the time domain by dividing the received signal by the Rayleigh fading function r (n).

{Frequency domain equalization processing}
Next, frequency domain equalization processing performed in the frequency domain equalizer 14 will be described. The symbol signal is converted into a frequency domain signal by the FFT calculator 12. FIG. 5 is a diagram schematically showing a received OFDM signal in the frequency domain. In the figure, the horizontal axis is the symbol direction, and the vertical axis is the carrier direction. The black square points are pilot signals, and the white circle points are data signals.

  Here, the pilot signal is a signal whose amplitude and phase are known, and the frequency domain equalizer 14 stores the amplitude and phase of the pilot signal in a memory or the like. Therefore, it is possible to obtain the transmission path response by comparing the stored pilot signal with the received pilot signal. Then, the transmission path response of the data signal is estimated from the transmission path response of the pilot signal arranged in the received symbol signal as shown in FIG.

  Here, as described above, a general estimation method is a method in which the symbol direction is linearly estimated and the carrier direction is estimated using an FIR filter. An example of a method for performing straight line estimation will be described. As shown in FIG. 5, in the symbol direction, pilot signals are arranged so as to appear every four symbols. Therefore, by taking the average of the transmission path response functions of two adjacent pilot signals, it is possible to estimate the transmission path response function of the data signal located between the two pilot signals. Further, if the average of the transmission path response function of the data signal located in the middle and the transmission path response function of the pilot signal 2 symbols before is taken, the transmission path response function of the data signal 1 symbol before is estimated. The Alternatively, if the average of the transmission path response function of the data signal located in the middle and the transmission path response function of the pilot signal after 2 symbols is taken, the transmission path response function of the data signal after 1 symbol is estimated. The

  However, in the case of mobile reception, transmission path fluctuations in the symbol direction also increase due to the influence of Rayleigh fading, and transmission path estimation cannot be performed correctly by linear estimation (linear interpolation) as described above. Therefore, FIR interpolation that can cope with the assumed maximum Doppler frequency is required. Therefore, in this embodiment, both the symbol direction equalizer 141 and the frequency direction equalizer 142 execute equalization processing using an FIR filter.

  Here, in order to obtain an ideal spectrum using the FIR filter, a large number of taps are required. And the number of pilot signals that need to be stored corresponding to the number of taps also increases. As a result, a large amount of memory for the buffer rig is consumed, and the circuit scale is increased. However, in the present embodiment, by performing equalization processing on the symbol signal in the time domain before the FFT calculation, it is possible to moderate the transmission path fluctuation in the symbol direction and reduce the number of taps of the symbol interpolation FIR filter. It can be done.

  That is, by performing equalization processing using the FIR filter for both the symbol direction and the carrier direction in the frequency domain, the number of taps of the FIR filter can be reduced while enabling highly accurate transmission path estimation. The scale has also been reduced. Specifically, when equalization processing is performed only in the frequency domain without performing equalization processing in the time domain, the number of taps necessary for the FIR filter in order to normally demodulate the OFDM signal is determined by experiments or the like. Then, according to the method of the present embodiment, it is possible to normally demodulate the OFDM signal using the FIR filter having a smaller number of taps.

  FIG. 6 is a diagram illustrating the state of the symbol signal before and after performing equalization processing in the time domain. The horizontal axis in the figure is the time axis, and the vertical axis indicates the amplitude. This figure shows a case where all the signals in one symbol are transmitted as signals having the same amplitude. As can be seen from the figure, the symbol signal before the equalization processing in the time domain (shown by a solid line) shows that the signal within one symbol varies greatly due to the influence of Rayleigh fading. . On the other hand, in the symbol signal (shown by a broken line) after performing the equalization process in the time domain, it can be seen that the fluctuation of the signal within one symbol is reduced.

  FIG. 7 shows a constellation before equalization processing in the time domain, and FIG. 8 shows a constellation after equalization processing in the time domain. Constellation represents a signal for quadrature modulation with the in-phase component as the x-axis (horizontal axis) and the quadrature component as the y-axis (vertical axis). 7 and FIG. 8 show that the constellation in FIG. 7 has a larger signal point size and the signal contains more noise.

  FIG. 9 is a diagram showing transmission path fluctuations in the symbol direction before equalization processing in the time domain, and FIG. 10 is a diagram showing transmission path fluctuations in the symbol direction after equalization processing in the time domain. It can be seen that the transmission path in FIG. 10 changes more smoothly than in FIG. Thereby, it is possible to improve the accuracy of equalization processing in the frequency domain.

  FIG. 11 shows an OFDM receiver employing the method of the present embodiment (demodulation in which equalization processing is performed in the time domain and then equalization processing is performed in the symbol direction and the carrier direction using an FIR filter in the frequency domain. ) And the performance of an OFDM receiver of another company that employs a conventional method. In the figure, the horizontal axis represents the Doppler frequency, and the vertical axis represents the required C / N ratio (Carrier to Noise Ratio). The C / N ratio is the ratio of signal power to noise power.

  For example, when compared with the case of a Doppler frequency of 40 Hz, according to 4-antenna diversity reception employing the demodulation method of the present embodiment, the required C / N ratio is 16.8 dB. On the other hand, 18.6 dB is required for the 4-antenna diversity reception of the conventional method (other company). That is, in order to normally demodulate the OFDM signal in an environment where the Doppler frequency is 40 Hz, the signal quality required by the OFDM receiver of this embodiment is lower than the signal quality required by the conventional OFDM receiver. It is good. Also, in the 4-antenna diversity reception, reception is possible with a C / N ratio of about 20 dB even in an environment where a Doppler frequency of 70 Hz, which cannot be demodulated by a conventional OFDM receiver, is received. It can also be seen that even in the case of 2-antenna diversity reception, if the demodulation method of the present embodiment is adopted, a signal having a C / N ratio of 28 dB can withstand an environment receiving a Doppler frequency of 50 Hz.

It is a block diagram of the demodulator for OFDM concerning an embodiment. It is a figure which shows the structure of an OFDM symbol signal. It is a figure which shows the state of the symbol signal in a Rayleigh fading environment. It is a figure which shows the calculation method of a Rayleigh fading function. It is a figure which shows the pilot arrangement | positioning of a received OFDM signal. It is a figure which shows the effect of time domain equalization. It is a figure which shows the constellation before time domain equalization. It is a figure which shows the constellation after time domain equalization. It is a figure which shows the transmission-line fluctuation | variation of the symbol direction before time domain equalization. It is a figure which shows the transmission-line fluctuation | variation of the symbol direction after time domain equalization. It is a figure which shows the Doppler characteristic comparison result of the demodulator of this invention and the demodulator made from other companies.

Explanation of symbols

11 Time Domain Equalizer 12 FFT Operator 14 Frequency Domain Equalizer 141 Symbol Direction Equalizer 142 Carrier Direction Equalizer

Claims (4)

A demodulation method in OFDM (Orthogonal Frequency Division Multiplex) transmission,
A first step of receiving an OFDM transmission signal;
A second step of performing equalization in the time domain on the symbol signal;
A third step of performing FFT conversion on the symbol signal after the time equalization processing;
A fourth step of performing equalization processing in the symbol direction and the carrier direction using an FIR filter on the received OFDM signal after FFT conversion;
A method for demodulating an OFDM signal, comprising: The method of demodulating an OFDM signal according to claim 1,
An OFDM signal demodulation method, wherein the FIR filter used in the fourth step is configured with a smaller number of taps than that required when time equalization processing is not performed. The OFDM signal demodulation method according to claim 1 or 2,
The second step includes
Step 2.1 for calculating the Rayleigh fading function r (n);
A step 2.2 of dividing the symbol signal by the Rayleigh fading function r (n);
Including
The step 2.1 includes
For a plurality of signals in the guard interval, a second example is obtained in which a slope of a straight line connecting a signal in the guard interval (first signal) and a signal in the effective symbol that is a copy source of the signal (second signal) is obtained. .1 step,
A 2.1.2 step of calculating an average of a plurality of the slopes obtained in the step 2.1.1 as a slope of a Rayleigh fading function r (n);
A method for demodulating an OFDM signal, comprising: A demodulation method in OFDM (Orthogonal Frequency Division Multiplex) transmission,
A first step of calculating a Rayleigh fading function r (n);
A second step of performing equalization in the time domain by dividing the received OFDM signal by the Rayleigh fading function r (n);
Including
The first step includes
For a plurality of signals within the guard interval, the first 1.1 is used to obtain the slope of a straight line connecting the signal within the guard interval (first signal) and the signal within the effective symbol that is the copy source of the signal (second signal). Process,
A first step of calculating an average of a plurality of the slopes obtained in the step 1.1 as a slope of a Rayleigh fading function r (n);
A method for demodulating an OFDM signal, comprising:

Images