CN1234219C - Symbolic ynchroni zing and carrier-wave synchronizing method based on modification system of circulation prefix - Google Patents

Abstract

The invention discloses a symbol synchronization method of a modulation system based on a cyclic prefix, comprising: continuously performing time-frequency conversion on a digital signal received by a receiver, and obtaining a group of frequency-transformed sampling values of the signal corresponding to different times to form two Time-dimensional spectrum; use a predetermined flat region search method to search the flat region for the two-dimensional time spectrum to obtain the start and end boundaries of the flat region; then use any time value within the start and end boundaries of the flat region as the symbol synchronization value to perform symbol synchronization. The present invention also discloses a carrier synchronization method based on a cyclic prefix modulation system. On the basis of obtaining the flat area, the sample values at more than one time point within the start and end boundaries of the flat area are used to calculate and average the carrier frequency deviation estimation value to Smooth noise, carrier synchronization using averaged carrier frequency offset estimates. Using the present invention can greatly improve the accuracy of symbol synchronization and carrier synchronization, and has high reliability.

Description

A Symbol Synchronization and Carrier Synchronization Method for Modulation System Based on Cyclic Prefix

technical field

The invention relates to a demodulation technology of a communication system, in particular to a method for symbol synchronization and carrier synchronization of a modulation system based on a cyclic prefix.

Background technique

At present, with the rapid growth of demand for new communication services, higher requirements are placed on the transmission rate of wireless communication systems and wireless local area networks, and the increase in transmission rate brings inter-symbol interference (ISI) and The problem of deep frequency selective fading. There are currently two ways to solve this problem, one is to use orthogonal frequency division multiplexing, that is, to disperse high-speed data to several sub-carriers for parallel transmission at a low rate; the other is to use a single carrier that simply introduces a cyclic prefix system. Both methods need to insert a cyclic prefix and use frequency domain equalization. Since both methods transmit signals in a symbol block structure, this requires not only sampling clock synchronization, but also symbol timing synchronization and carrier synchronization. There are also two methods for symbol timing synchronization and carrier synchronization, one is to use the training sequence, and the other is to use the periodic signal structure introduced by the cyclic prefix for blind synchronization.

The general process of the entire blind synchronization is introduced below by taking an Orthogonal Frequency Division Multiplexing (OFDM) modulation system as an example. The baseband model represented by figures in Fig. 1 shows the basic functional block diagram of the OFDM system. The entire signal transmission of the OFDM system generally goes through the stages of sending processing of the transmitter, channel transmission and receiving processing of the receiver. As shown in Figure 1, the transmission processing of the transmitter is mainly to modulate the signal, including encoding the signal, constellation mapping and inverse discrete Fourier transform (IDFT), and transform it into a time domain signal. After the transmission processing of the transmitter, After the signal is transmitted through the channel, it is received by the receiver, and the receiver demodulates the signal, which mainly includes several processes such as symbol synchronization, carrier synchronization, sample value synchronization and discrete Fourier transform (DFT). The signal processing process before symbol synchronization and carrier synchronization will be described in more detail below.

经过并/串转换之后形成串行时域数据s(n)。In the OFDM system, the data stream is transmitted in blocks, and each data block d(k) forms a vector of length N after a certain coding process and constellation mapping $\stackrel{\&Right\; Arrow;}{x}=\left\{x_k\right\},k=0,...,N-1,$ Through the inverse discrete Fourier transform, the vector becomes a time-domain data vector, and a cyclic prefix of length L is added to obtain $\stackrel{\&Right\; Arrow;}{\mathrm{the\; s}}=\left\{{\mathrm{the\; s}}_k\right\},$ Where k=0, . . . , N+L-1, and when j=0, . . . , L-1, s _j =s _j+N . This time domain data vector after adding the cyclic prefix

Serial time-domain data s(n) is formed after parallel/serial conversion.

的前L个样值，也就是去掉循环前缀，将剩下的N个样值通过离散傅立叶变换输出为 $\stackrel{\→}{y}=\left\{y_k\right\},k=0,...,N-1.$ 其后还有对该信号进行信道估计、信道去耦和信道解码等操作，这里不再详细介绍。The serial time-domain data s(n) is formed into a signal r(n) after channel transmission, and the signal r(n) received by the receiver from the channel is obtained after serial/parallel conversion $\stackrel{\&Right\; Arrow;}{r}=\left\{r_k\right\},k=0,...,N+L-1.$ Then discard according to the result of OFDM symbol synchronization

The first L samples of , that is, remove the cyclic prefix, and output the remaining N samples through discrete Fourier transform as $\stackrel{\&Right\; Arrow;}{\mathrm{the\; y}}=\left\{{\mathrm{the\; y}}_k\right\},k=0,...,N-1.$ Afterwards, operations such as channel estimation, channel decoupling, and channel decoding are performed on the signal, which will not be described in detail here.

与逆离散傅立叶变换之前的 之间的关系为：y(k)＝H(k)x(k)+N(k)，k＝0，...，N-1。其中H(k)为信道冲激响应的频域表示，N(k)是噪声v(n)的频域表示。In the above OFDM signal transmission process, when the length L of the cyclic prefix is greater than the duration M of the channel impulse response h(k), the formed after discrete Fourier transform

with the inverse discrete Fourier transform before The relationship between them is: y(k)=H(k)x(k)+N(k), k=0, . . . , N−1. Among them, H(k) is the frequency domain representation of the channel impulse response, and N(k) is the frequency domain representation of the noise v(n).

For the single-carrier system using a cyclic prefix, its principle is similar to that of the OFDM system, except that it does not undergo inverse discrete Fourier transform and discrete Fourier transform. The channeled signal received at the receiver is also the signal r(n) with a cyclic prefix. The OFDM system using cyclic prefix and the single-carrier system have one thing in common, that is, the signals they send have a certain periodic structure, and under the condition of L≥M, the signal r(n) received by the receiver also has a certain cycle structure.

的前L个样值，即去掉循环前缀，得到实际信号。在现有的盲同步算法中，都是利用循环前缀和已调制符号中部分样值的强相关性进行相关运算，通过峰值检测进行符号定时同步估计。这种方法只能估计出一个特定的时刻，容易受到已调制符号自相关特性的干扰，具有严重的地板效应。此外，在这种方法中噪声和已调制符号加窗会造成自相关峰的平顶效应，从而使峰值检测锁定在错误时刻，并因此得不到精确的符号同步。另外，这种方法的精度不稳定，随着信道冲激响应时间的变长，其精度将降低。Usually the purpose of symbol synchronization is to determine the end position of the cyclic prefix, and then discard it correctly according to the end position

The first L samples of , that is, the cyclic prefix is removed to obtain the actual signal. In the existing blind synchronization algorithms, the strong correlation between the cyclic prefix and some samples in the modulated symbols is used for correlation calculation, and the symbol timing synchronization is estimated by peak detection. This method can only estimate a specific moment, is easily interfered by the autocorrelation characteristic of the modulated symbols, and has serious floor effect. In addition, noise and modulated symbol windowing in this approach can cause a flat-topping effect of the autocorrelation peak, causing peak detection to lock on to the wrong instant, and thus not obtaining precise symbol synchronization. In addition, the accuracy of this method is unstable, and its accuracy will decrease as the channel impulse response time becomes longer.

In the current carrier synchronization, that is, the method for estimating the carrier frequency deviation, the strong correlation between the cyclic prefix and some samples in the modulated symbols is also used for correlation calculation, and the carrier frequency deviation is estimated by peak detection. This method can only estimate the carrier frequency deviation at a moment when the cyclic prefix ends. Also, due to the characteristics of severe floor effect and unstable accuracy, the accuracy of carrier frequency deviation estimation is low.

Contents of the invention

In view of this, an object of the present invention is to propose a symbol synchronization method for a cyclic prefix-based modulation system that overcomes the shortcomings of the prior art and can improve synchronization accuracy.

Another object of the present invention is to provide a carrier synchronization method for a modulation system based on a cyclic prefix that can improve the estimation accuracy of carrier frequency offset.

Above-mentioned purpose of the present invention is accomplished by following technical scheme:

A symbol synchronization method based on a cyclic prefix modulation system, comprising the steps of:

a. Continuously perform time-frequency conversion on the digital signal received by the receiver, and obtain a group of frequency conversion sampling values of the signal corresponding to different times to form a two-dimensional time spectrum of the signal;

b. For the two-dimensional time spectrum, continuously calculate the mean square error of the sample value on the time axis for any frequency point. output sequence;

c. Continuously calculate a predetermined number of sliding mean square deviations and sliding mean values for the output sequence, the predetermined number is the sum of the cyclic prefix length and the effective time length of the modulation signal, and subtract the sliding mean value from the sliding mean square deviation to obtain a comparison threshold sequence;

d. compare the output sequence and the comparison threshold, if the output sequence is greater than the comparison threshold then output a non-zero positive value, otherwise output a zero value to obtain a zero-connection zone;

e. extend the predetermined value described in step b to the right on the time axis at the right boundary of the zero-continuous zone to obtain the start and end boundary of the flat zone;

f. Use any time value within the start and end boundaries of the flat area as the symbol synchronization value to perform symbol synchronization.

In the above symbol synchronization method, in step a, time-frequency transformation may be performed by short-time Fourier transformation or transversal filtering.

In the above symbol synchronization method, steps b to e may be performed for each frequency point, and then the obtained flat areas corresponding to each frequency point are combined to obtain the start and end boundaries of a combined flat area. It is also possible to perform step b for each frequency point respectively, and then combine the obtained output sequences respectively corresponding to each frequency point to obtain a combined output sequence, and then perform step c. Moreover, when the system signal-to-noise ratio is higher than a predetermined value, any frequency point in step b can be a zero frequency point, and in step a, the time-frequency transformation is completed by sliding summation or integration to obtain a two-dimensional time-frequency spectrum. The frequency points here may be all frequency points, or some frequency points greater than one.

In the above symbol synchronization method, in step f, the right boundary of the flat area can be used as the symbol synchronization value to perform symbol synchronization. It is also possible to use other arbitrary time points in the flat region except the right boundary as the symbol synchronization value, and further include removing the phase caused by using other arbitrary time points as the symbol synchronization value in channel estimation and channel decoupling in the process of symbol synchronization distortion.

A carrier synchronization method based on a cyclic prefix modulation system, comprising the steps of:

a. Continuously perform time-frequency conversion on the digital signal received by the receiver, and obtain a group of frequency conversion sampling values of the signal corresponding to different times to form a two-dimensional time spectrum of the signal;

b. For the two-dimensional time spectrum, continuously calculate the mean square error of the sample value on the time axis for any frequency point. output sequence;

c. Continuously calculate a predetermined number of sliding mean square deviations and sliding mean values for the output sequence, the predetermined number is the sum of the cyclic prefix length and the effective time length of the modulation signal, and subtract the sliding mean value from the sliding mean square deviation to obtain a comparison threshold sequence;

d. compare the output sequence and the comparison threshold, if the output sequence is greater than the comparison threshold then output a non-zero positive value, otherwise output a zero value to obtain a zero-connection zone;

e. extend the predetermined value described in step b to the right on the time axis at the right boundary of the zero-continuous zone to obtain the start and end boundary of the flat zone;

f. Use the sample values at more than one moment in the start and end boundaries of the flat area to calculate the carrier frequency deviation estimated value respectively, average the more than one carrier frequency deviation estimated value, and use the averaged carrier frequency deviation estimated value to carry out Carrier synchronization.

As can be seen from the technical solution of the present invention, the present invention uses time-frequency transformation to obtain a flat area on the time axis, and then accurately determines the start and end boundaries of this flat area through flat area search, because the present invention does not use the detection of autocorrelation peaks , thus does not depend on the autocorrelation characteristic of the modulated symbol, effectively overcomes the floor effect caused by the time domain synchronization method in the prior art, and improves the synchronization accuracy.

In addition, since the flat area obtained by the present invention through time-frequency conversion does not have any intersymbol interference, any moment in the flat area can be used as the timing of symbol synchronization. Compared with the prior art, only one timing moment can be determined, the present invention The probability of successful single synchronous detection is improved, so the present invention not only can further improve the synchronous accuracy, but also is easier to use.

The flat area obtained through the flat area search can be further used for carrier synchronization because there is no inter-symbol interference. By calculating an estimated value of carrier frequency deviation for samples at multiple moments in the flat region, and then averaging these estimated values, compared with the prior art, only one estimated value of carrier frequency deviation can be calculated at one timing moment , the invention smoothes the noise and greatly improves the calculation accuracy.

In addition, the method of the present invention is less affected by the level of the signal-to-noise ratio, and the effect of the present invention will not be reduced correspondingly with the decrease of the signal-to-noise ratio, so the present invention can effectively combat low signal-to-noise ratio environments. Using the flat area search of the present invention can also estimate the length of the channel impulse response through the length of the cyclic prefix and the width of the flat area, which is beneficial to improving the channel estimation accuracy. In addition, the present invention is hardly affected by the windowing function and has high reliability.

Description of drawings

Fig. 1 shows the basic functional block diagram of Orthogonal Frequency Division Multiplexing (OFDM) system;

Fig. 2 shows a functional block diagram of the present invention;

Figure 3A shows a time-domain diagram of a signal received by a channel receiver;

Figure 3B shows a two-dimensional schematic diagram of the two-dimensional time-frequency spectrum after the short-time Fourier transform of the signal in Figure 3A;

Fig. 4 shows an example of a three-dimensional display of the two-dimensional time spectrum of the amplitude spectrum only given in the present invention;

Fig. 5 shows a schematic diagram of a flat region search algorithm of the present invention;

Figure 6 shows an example sample of a flat region search;

Fig. 7 shows a schematic diagram of a carrier frequency offset estimation method of the present invention;

Fig. 8 shows the functional block diagram of the second method of determining the flat region of the present invention;

FIG. 9 shows a functional block diagram of the third method for determining a flat region of the present invention.

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. Firstly, referring to FIG. 2, the specific flow of the symbol synchronization method and carrier synchronization of the present invention will be described.

As shown in Figure 2, when the signal r (t) received by the receiver in the OFDM system is converted into a digital signal r (n) by an analog/digital (A/D) converter, the flow process of the present invention is started, and sampling The rate is the sample rate when the transmitter sends, and ignores relatively small deviations.

In step 201, time-frequency analysis is first performed on the signal r(n). The time-frequency analyzer is also a sliding Fourier transform analyzer, which includes a buffer that can store N samples for continuous input r(n) and an N-point Fourier transform. Take out N input sample values in the buffer at any moment j, and then perform short-time Fourier transform on them to obtain N frequency point sample values at that moment. The sampling values of N frequency points at all times constitute a two-dimensional time spectrum.

For the signal r(t), its short-time Fourier transform can be expressed as:

${\mathrm{<span\; class="notranslate">\; STFT</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&gamma;</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; \&Integral;</span>}_{<span\; class="notranslate">\; -</span><span\; class="notranslate">\; \&infin;</span>}^{<span\; class="notranslate">\; \&infin;</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="j"\; filenames="C0310207100173.png"\; state="\{\{state\}\}">j</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;f\&tau;</span>}}\mathrm{<span\; class="notranslate">\; d\&tau;</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In the above formula, r(t) is the time analysis window function, which can be selected arbitrarily according to actual needs. When r(t) takes a rectangular window whose width is the effective time of the OFDM symbol, its discrete digital form is:

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; j\&phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; DSTFT</span>}{<span\; class="notranslate">\; \{</span>\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span>}_{{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}}^{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\sqrt{\mathrm{<span\; class="notranslate">\; N</span>}}}\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; W</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}}^{\mathrm{<span\; class="notranslate">\; ik</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

表示向量

的第i个元素，k为频率变量，0≤k≤N-1，j为时间变量，-∞＜j＜∞。in,

representation vector

The i-th element of , k is a frequency variable, 0≤k≤N-1, j is a time variable, -∞<j<∞.

As mentioned above, no matter it is an OFDM system or a single carrier system, the signal r(n) received by the receiver has a certain periodic structure, which can be seen from Fig. 3A. Figure 3A is a time-domain diagram. Each horizontal unit is a vector composed of N samples starting from a certain moment. The ordinate indicates the moment when the first element of the vector is located, and the abscissa indicates the relative time of the other elements of the vector. The relative coordinates of the first element. In Fig. 3A, a certain part 301 of the received signal r(n) is a repetition of another part 302 of r(n), the two parts differ in time by N samples, and the length of each part is L-M+1 , what is shown in FIG. 3A is the case of M=1.

Fig. 3B is a two-dimensional representation of a two-dimensional time-frequency spectrum. It can be seen from FIG. 3B that formula (2) is equivalent to a sliding Fourier analysis window with width N. Through the formula (2), a set of frequency-domain transformed sampling values R(j, 0), R(j, 1), R(j, N-1) at time j is obtained, that is, in Fig. 3B In this way, the two-dimensional time spectrum of the input signal r(n) is obtained. In the two-dimensional diagram of the two-dimensional time spectrum shown in FIG. 3B , the horizontal axis represents frequency f, and its range is 0...N-1, and the vertical axis represents time t, and its range is from negative infinity to positive infinity. In FIG. 3B , 303 ′, 304 ′, 305 ′, etc. are the results of performing Fourier transformation on 303 , 304 , 305 , etc. in FIG. 3A , respectively.

When the equivalent duration M of the channel impulse response is less than the length L of the cyclic prefix, the received OFDM signal still has certain cyclic characteristics. As shown in Figure 4, this characteristic is reflected as a local flat area on the time axis on the three-dimensional display of the two-dimensional time-frequency spectrum of the signal, and its width is L-M+1; for any frequency k, all moments in this flat area The magnitude of the frequency-domain transform samples is equal. Then by searching and locating the flat area on the time axis, the starting position of the OFDM symbol can be obtained. Since the OFDM symbols of length N starting at any point in this flat region do not contain any adjacent symbol information, the obtained flat region is an interval without ISI.

According to the previous time-frequency analysis, the modulus value of the two-dimensional time spectrum, that is, the amplitude value of the two-dimensional time spectrum, will have a constant value area with equal value every N+L samples on the time axis, that is, a The flat area, and for any frequency k, if the influence of the error is not considered, the start and end positions of the flat area are the same, that is, the left and right boundaries are the same.

In step 202, a flat area search is performed using a flat area search algorithm to identify flat areas. A possible flat area search algorithm based on calculating the sliding mean square error is given below, which includes the following steps:

Step 501 , as shown in FIG. 5 , sets the window size W of the sliding mean square error to 4≤W<L, and its function is to continuously calculate the mean square error of W samples in the input window. Since the actual window size of the flat region is L-M+1, when the noise is not considered, the output sequence {std(j) _w } of the sliding mean square error 401 is a non-negative sequence, and every N+L samples will be A zero-continuous zone with a width of LM-W+1 appears.

Step 502, for the output sequence of sliding mean square error with window size W, use the output of sliding mean square error with window size N+L to subtract the sliding mean value with window size N+L to form a comparison threshold {V _th (j) _{N +L} }.

Step 503, the comparator compares {std(j) _w } and {N _th (j) _N+L }, if {std(j) _w } is greater than {V _th (j) _X+L }, then output a non-zero Positive value, otherwise zero value is output, so that the output of the comparator can obtain a zero-continuous area with a width of LM-W+1, so that it can be further judged whether the current sample value falls within the flat area.

Figure 6 shows an output sample of a flat area search, where the dotted line represents the sliding mean square error {std(j) _w }, and the dotted line represents the comparison threshold {V _th (j) _N+L }, the real The lines represent their comparison results. It can be clearly seen in Fig. 6 that there is a periodic zero-continuous region in the comparison result, that is, a flat region, and its width is LM-W+1.

After the flat area is obtained, the start and end boundary estimation of the flat area can be obtained after a simple correction, that is, the window size W of the sliding mean square error is added to the right on the width of the zero-continuous area with a width of L-M-W+1, to obtain A flat area with a width of L-M+1, all samples in this flat area have no inter-symbol interference, and its right boundary corresponds to the moment when the cyclic prefix ends, that is, the end position of the flat area corresponds to the real start of the OFDM symbol start position. When any other time point in the flat region can also be used as the starting position of the Fourier transform window in the subsequent DFT process, since the corresponding phase distortion will be introduced in the DFT demodulation result, at this time, as long as the channel based on frequency domain sampling Estimation and linear interpolation and channel decoupling can completely cancel this phase distortion, so any position in this flat region can be used as the basis for symbol synchronization. The process of canceling the phase distortion here is well known to those skilled in the art and will not be repeated here.

Step 203, sending the results of the flat area search in step 202 to the combiner, and the combiner combines these results to reduce noise interference. The flat area is essentially a sample value interval without inter-symbol interference, which is timing information for system synchronization. The timing information is fed to both the carrier offset estimator and a serial-to-parallel converter. The present invention does not specify a specific merging algorithm, and those skilled in the art can easily implement this merging process.

The flat area obtained in the above steps can be used not only for symbol synchronization, but also for estimation and tracking of carrier frequency, phase and sampling clock deviation. The following explains how to use this flat region to improve the accuracy of carrier frequency offset estimation in carrier synchronization.

Carrier frequency deviation estimation in the present invention is based on the L-M+1 timing information output in the symbol synchronization process, respectively using the sample value at that time and the conjugate of the sample value after N sampling moments to perform product operations to estimate L -M+1 carrier frequency deviation Its average value is then sent to the phase corrector. FIG. 7 shows a schematic diagram of such a method for estimating carrier frequency deviation, and the method specifically includes the following steps:

In step 701, first perform a product operation on the conjugate of the sample value r _F (n) after N sampling moments and the original sample value r _F (nN). In the case of subcarrier bandwidth normalization, when there is a frequency deviation ε at the transceiver end, if the channel effect is not considered, the received signal can be expressed as r(n)=s(n-θ)e ^j2πεn/N , where Noise is not included, and θ is the time shift. Then when s(n-θ)=s(nN-θ), there are:

${\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="e\; j"\; filenames="C0310207100173.png"\; state="\{\{state\}\}">e</figure-callout></span><figure-callout\; id="2"\; label="e\; j"\; filenames="C0310207100173.png"\; state="\{\{state\}\}">\; </figure-callout>}}^{\mathrm{<span\; class="notranslate">\; </span>}}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="j"\; filenames="C0310207100173.png"\; state="\{\{state\}\}">j</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;\&epsiv;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="e\; j"\; filenames="C0310207100173.png"\; state="\{\{state\}\}">e</figure-callout></span><figure-callout\; id="2"\; label="e\; j"\; filenames="C0310207100173.png"\; state="\{\{state\}\}">\; </figure-callout>}}^{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

即：Step 702, calculate the phase angle to the value obtained in step 701, and obtain the carrier frequency deviation

Right now:

值，对这些值进行平均，即可得到本发明的载波频率偏差估计值。当然，在实际情况中也可以不使用平坦区的每个时刻计算载波频率偏差估计值，而是在其中抽取一部分值，只要计算出的精度能达到系统要求即可。Step 703, use the formula (4) to traverse each moment obtained by using the flat area search, so as to obtain a set of L-M+1

value, and these values are averaged to obtain the carrier frequency deviation estimation value of the present invention. Of course, in actual situations, it is also possible not to use each moment in the flat zone to calculate the estimated value of the carrier frequency deviation, but to extract a part of the value, as long as the calculated accuracy can meet the system requirements.

In the above steps, step 701 and step 702 belong to the content of the prior art, and can refer to J. Beek et al. in "The Maximum Likelihood Estimation of Time and Frequency Offset in OFDM Systems" (MLEstimation of Time and Frequency Offset in OFDM Systems, IEEE TransactionsOn Signal Processing, Vol.45, No.7, pp.1800-1805, July, 1997). In step 703 of the present invention, the concept of the previous flat region is utilized. After the flat region is found, the sample values at different moments in the flat region are used to estimate the frequency deviation. Since the flat region contains more than one time number, it can be obtained by The mean value method smooths the error caused by the noise, so it can obtain a higher frequency deviation estimation accuracy than the common autocorrelation function peak detection method.

After performing symbol synchronization and carrier synchronization according to the foregoing, the serial-to-parallel converter selects N samples in the effective OFDM time range under the drive of timing information and sends them to the back-end phase corrector to output N samples. The phase corrector outputs N samples according to The essence of carrier synchronization is the frequency deviation that represents the phase difference Perform phase correction on these N samples to eliminate the influence of carrier frequency deviation, and then output y(k). Finally, y(k) is sent to modules such as channel estimation and Fourier transform, and finally the sent data is obtained This part of the content does not belong to the problem to be solved by the present invention, and can be accomplished by using appropriate prior art.

The flat region search and the general method of symbol synchronization and carrier synchronization using flat region search are given above. Two special cases for determining flat regions are given below.

FIG. 8 shows a functional block diagram of the second method for determining a flat area of the present invention. The implementation of the embodiment shown in Figure 8 is essentially the same as that in Figure 2, the difference is that the flat region searcher is placed after the combiner, and the combiner is also replaced by a summation The device, which is essentially an averager, is used to take the mean of N sliding mean square errors. That is, step 501 is executed for N frequency points, and then the average value of the obtained output sequence is taken, and step 502 is executed for the averaged output sequence, and finally a flat area is obtained. The purpose of doing this is to avoid the use of a more complex decision combiner, save cost and reduce calculation time. The time-frequency analysis here can actually be realized by transversal filtering of sinusoidal signals corresponding to each frequency point, where the transversal filter coefficients correspond to the row/column vectors of the Fourier transform matrix, and each branch has an equal status. In addition, if the signal-to-noise ratio can be estimated, the number of branches can be dynamically reduced or increased according to the level of the signal-to-noise ratio, or some fast FFT algorithms such as Goertzel algorithm can be used to calculate the Fourier transform of the corresponding point.

FIG. 9 shows a functional block diagram of the third method for determining a flat area of the present invention. When the signal-to-noise ratio is high, there is no need to combine the decision results at each frequency point as shown in Figure 8, only a few decision branches are required. This embodiment is a further simplification on this premise. At this time, time-frequency analysis can only be performed on a branch with an arbitrary frequency. In particular, for a branch with a frequency of 0, the Fourier transform vector W _N ⁰ can be used as the cross-sectional filter coefficient for filtering, and a window size of A sliding summer of N or an integrator with a duration corresponding to N for summing or integrating. This method is simpler to implement, but it is only suitable for high signal-to-noise ratio, and generally requires that the signal-to-noise ratio is greater than 20dB.

Two specific methods for determining the flat region are given here. It will be understood that the invention can be varied in many ways according to the spirit of the invention. For example, the time-frequency analysis of the present invention can also be implemented by not using short-time Fourier transform, but using the transversal filtering of sinusoidal signals corresponding to each frequency point, wherein the transversal filter coefficients correspond to the row/column vectors of the Fourier transform matrix, Therefore, the above is only a demonstration of the spirit of the present invention, and is not intended to limit the protection scope of the present invention.

Claims (9)

1. A symbol synchronization method of a modulation system based on cyclic prefix includes the following steps:

a. continuously performing time-frequency transformation on a digital signal received by a receiver, and respectively obtaining a group of frequency transformation sampling values of the signal corresponding to different moments to form a two-dimensional time-frequency spectrum of the signal;

b. for the two-dimensional time frequency spectrum, continuously calculating the mean square error of sample values on a time axis for any frequency point, wherein the sample values are sample values with the preset number of more than or equal to 4 and less than the length of a cyclic prefix, and obtaining a non-negative output sequence;

c. continuously calculating a preset number of sliding mean square deviations and sliding mean values for the output sequence, wherein the preset number is the sum of the length of a cyclic prefix and the effective time length of a modulation signal, and subtracting the sliding mean values from the sliding mean square deviations to obtain a comparison threshold sequence;

d. comparing the output sequence with the comparison threshold, if the output sequence is larger than the comparison threshold, outputting a non-zero positive value, otherwise, outputting a zero value to obtain a zero-connecting area;

e. c, extending the preset quantity value in the step b rightwards on a time axis at the right boundary of the zero connecting area to obtain a starting and stopping boundary of the flat area;

f. and using any time value in the starting and stopping boundary of the flat zone as a symbol synchronization value to carry out symbol synchronization.

2. The symbol synchronization method according to claim 1, wherein the time-frequency transform is performed by a short-time fourier transform or a cross-section filter in step a.

3. The symbol synchronization method according to claim 1, wherein steps b to e are performed for each frequency point, and then the obtained flat regions corresponding to each frequency point are combined to obtain a start-stop boundary of a combined flat region.

4. The symbol synchronization method according to claim 1, wherein step b is performed for each frequency bin, and then the obtained output sequences corresponding to each frequency bin are combined to obtain a combined output sequence, and then step c is performed.

5. The symbol synchronization method according to claim 4, wherein when the system SNR is higher than the predetermined value, any one of the frequency points in step b is a zero frequency point, and the time-frequency transform is performed by integration or sliding summation in step a to obtain a two-dimensional time-frequency spectrum.

6. The symbol synchronization method according to claim 3 or 4, wherein the frequency points are all frequency points or more than 1 but less than all frequency points.

7. The symbol synchronization method as claimed in claim 1, wherein the symbol synchronization is performed using a right boundary of the flat region as a symbol synchronization value in step f.

8. The symbol synchronization method as claimed in claim 1, wherein other arbitrary time points except the right boundary of the flat region are used as the symbol synchronization values in step f, and further comprising removing the phase distortion caused by using the other arbitrary time points as the symbol synchronization values in the channel estimation and channel decoupling.

9. A carrier synchronization method of a modulation system based on cyclic prefix includes the following steps:

a. continuously performing time-frequency transformation on a digital signal received by a receiver, and respectively obtaining a group of frequency transformation sampling values of the signal corresponding to different moments to form a two-dimensional time-frequency spectrum of the signal;

b. for the two-dimensional time frequency spectrum, continuously calculating the mean square error of sample values on a time axis for any frequency point, wherein the sample values are sample values with the preset number of more than or equal to 4 and less than the length of a cyclic prefix, and obtaining a non-negative output sequence;

c. continuously calculating a preset number of sliding mean square deviations and sliding mean values for the output sequence, wherein the preset number is the sum of the length of a cyclic prefix and the effective time length of a modulation signal, and subtracting the sliding mean values from the sliding mean square deviations to obtain a comparison threshold sequence;

d. comparing the output sequence with the comparison threshold, if the output sequence is larger than the comparison threshold, outputting a non-zero positive value, otherwise, outputting a zero value to obtain a zero-connecting area;

e. c, extending the preset quantity value in the step b rightwards on a time axis at the right boundary of the zero connecting area to obtain a starting and stopping boundary of the flat area;

f. and respectively calculating carrier frequency deviation estimated values by using the sample values of more than 1 time in the starting and stopping boundaries of the flat area, averaging the more than 1 carrier frequency deviation estimated values, and carrying out carrier synchronization by using the averaged carrier frequency deviation estimated values.

Images