CN110290084B - Short wave channel blind symbol synchronization method based on data frequency energy peak value - Google Patents

Abstract

Translated fromChinese

该发明公开了一种基于数据频率能量峰值的短波信道盲符号同步方法,属于无线通信中的符号同步技术领域。本发明在频率选择性衰落短波信道下，无需知道信号发送频点，利用非固定频点最大峰搜索的方法即可完成符号定时。本方案有利于对抗短波信道频率偏移等不良传输特性，实现对符号最佳采样位置的准确定位。仿真结果表明:在有多径和频偏的选择性衰落信道中，本方案对符号定时位置的估计仍能保持较高的准确率。

The invention discloses a short-wave channel blind symbol synchronization method based on a data frequency energy peak, belonging to the technical field of symbol synchronization in wireless communication. Under the frequency selective fading short-wave channel, the invention can complete the symbol timing by using the method of maximum peak search of non-fixed frequency points without knowing the signal transmission frequency point. This scheme is beneficial to counteract the bad transmission characteristics such as frequency offset of shortwave channel, and realize the accurate positioning of the optimal sampling position of the symbol. The simulation results show that in the selective fading channel with multipath and frequency offset, this scheme can still maintain a high accuracy for the estimation of symbol timing position.

Description

Short wave channel blind symbol synchronization method based on data frequency energy peak value

Technical Field

The invention belongs to the technical field of symbol synchronization in wireless communication, and provides a blind symbol synchronization method which is based on short-time Fourier transform (DSTFT) and frequency energy peak search, is resistant to multipath frequency offset, small in complexity, high in accuracy and free of signal prior information under a short-wave channel.

Background

In the communication process, the accuracy of symbol synchronization directly affects the demodulation effect of signals, and is one of the most important links in signal receiving and processing. Short wave channels have a plurality of problems such as multipath time delay, Doppler shift, frequency selective fading and the like, the quality is extremely unstable, and an FSK modulation mode with strong viability and a short burst signal form are often adopted, so that the difficulty is increased for symbol synchronization.

Common and mature signal symbol self-synchronization algorithms are classified into three categories: the first phase-locked loop synchronization method utilizes a timing error detection algorithm and a loop filter to extract phase difference information in signals, and then adjusts a sampling clock, feeds back and tracks, and continuously reduces errors until the signals are locked. However, frequency offset and phase fluctuation exist in signals transmitted in a short-wave time-varying channel generally, and multipath time delay changes rapidly and randomly, so that a receiver has difficulty in accurately estimating and tracking the phase. The second type is a synchronization method based on high-order cumulant, which utilizes a large amount of abundant information which is not available in the second-order statistics (power spectrum and correlation function), can effectively test and characterize the cyclostationarity in the signal and suppress noise, but this method requires large computational complexity and sufficient observation data, and is not suitable for symbol synchronization in short burst form of short-wave channel. The third kind is differential detection, short-time zero-crossing rate and other methods, which are sensitive to noise and have the effect of reducing the short-wave communication with frequency selective fading and multipath time delay. In addition, in recent years, symbol synchronization methods based on signal transient spectrum, such as spectrum peak value calculation, spectrum peak ratio method, maximum peak search method, etc., have appeared, which combine the characteristics of FSK signal and reduce the complexity of symbol synchronization, but they must know the frequency of the transmitted signal and are also susceptible to frequency offset and noise in short wave channel. The maximum peak searching method of the fixed frequency point determines the symbol synchronization position by using the maximum peak value at the known signal frequency point at different times, and is also an algorithm for comparison emphatically in the scheme of the invention.

Disclosure of Invention

The invention creatively combines the analysis means of short-time Fourier transform and the time-varying frequency spectrum characteristic and utilizes the maximum peak search of non-fixed frequency points to complete the symbol timing synchronization. The sliding times are set as the number of sampling points of one symbol, and the symbol must be crossed in the sliding process, so that the scheme only needs to track the frequency components appearing after the signal transmission frequency point is not needed, and the corresponding peak position is searched. Due to the complex channel effect, the frequency components of the signals at the receiving end may not be the frequency points of the signals in real transmission but have larger offset, and the symbol synchronization of the scheme does not depend on the frequency points of the transmitted signals, so that the method has better anti-frequency offset capability compared with a fixed frequency point maximum peak search method.

The simulation channel of the scheme is modeled by the combination of a fading process representing a direct path and a fading process representing a reflection path, a received signal comprises two path components, and the simulation channel is suitable for MFSK modulation and a short burst signal form of a frequency selective fading short-wave channel, has small algorithm complexity and certain frequency offset resistance, and the required information only comprises a symbol rate and a sampling frequency, and is a symbol synchronization method without knowing other prior information. Simulation results show that under a multipath frequency selective fading channel, the algorithm can effectively realize accurate timing of symbol synchronization, and the performance is better than that of a fixed frequency point maximum peak searching method.

The technical scheme of the invention is as follows:

the sliding reference position is determined for 5 times, and the adjacent interval of each time of reference position is required to be one symbol length. And under each reference position, keeping the window length as a symbol, and sliding the observation window by taking the sampling point as the step length until the total sliding length is a symbol. And (4) obtaining the time-varying frequency spectrum of the data in the window by using short-time Fourier transform every time of sliding, and recording the peak value and the corresponding frequency point. And under the condition that the peak frequency point is not unique, taking the maximum peak time corresponding to the peak frequency point appearing later in the sliding process as the sliding frequency at the current reference position. And calculating the average value of the 5 sliding times, wherein the reference bit plus the average value of the sliding times is the final symbol timing position.

The invention relates to a short-wave channel time-varying spectrum maximum peak search symbol synchronization method based on short-time Fourier transform, the algorithm flow is shown as the attached figure 2, the technical scheme is a short-wave channel blind symbol synchronization method based on data frequency energy peak, the method comprises the following steps:

step 1, sampling a received signal r (t) by a sampling frequency Fs to obtain r (l), wherein l is 1,2, and N represents the data length of each path, and then obtaining the number of sampling points Nsamp in one symbol according to a known symbol rate Symr and the sampling frequency Fs; taking any position pst (n) in the signal segment, wherein n is 1 as a first symbol synchronization reference, and n represents the number of times of the current reference position, wherein n is more than or equal to 1 and less than or equal to 5, and taking 5 symbol synchronization references in total;

and 2, starting from the reference position, selecting analysis data by using a window with the length fl ═ Nsamp:

{r₁(m),r₂(m),..., r_fn(m), wherein m is 1,2, as, fl }, each time, sliding is performed by taking a sampling point as a step length, the sliding times of the sampling point are represented by i, wherein i is more than or equal to 1 and less than or equal to Nsamp, the total sliding times of the sampling point are Nsamp, and the total window number fn is Nsamp;

step 3, respectively carrying out fast Fourier transform of fl points on the data in each window, and obtaining a modulus value to obtain an energy spectrum Y_i(k) 1,2, ·, fn;k 1, 2., fl, and obtaining a set of frequency spectral lines f (k),k 1, 2., fl; due to symmetry, only half is taken: energy spectrum of Y_i(k),i＝1,2,...,fn；

The set of frequency spectral lines is f (k),

step 4, in the energy spectrum Y_i(k) Find the maximum peak Mp in_i,i＝1,2,...,fn,

And the corresponding spectral line number fm_iF, calculating the frequency corresponding to the maximum peak value

Step 5, if Mpf_iIf the i is equal to 1, 2., fn is equal, then s (n) is equal to 0, go to step 6; if Mpf_iI is not all equal, and two different frequencies must appear, wherein the frequency of the appearance after the window sliding sequence is f; finding a frequency Mpf satisfying the maximum peak frequency_iSliding order set equal to frequency f condition phi ═ i | Mpf_i＝f&I is more than or equal to 1 and less than or equal to fn, finding a peak maximum point Mp which is max (mpi) in a sliding number set phi, wherein i belongs to the number of sliding points S (n) corresponding to phi, n is more than or equal to 1 and less than or equal to 5, namely the number of sliding sampling points with the position pst (n) as the reference;

step 6, if n is less than or equal to 4, changing the reference position to Pst (n +1) ═ Pst (n) + Nsamp, where n is less than or equal to 1 and less than or equal to 4, and using the n +1 th symbol synchronization reference position as the n +1 th symbol synchronization reference position, and turning to step 2; otherwise, turning tostep 7;

step 7, the 5-time difference symbolThe number reference position obtains the number S (n) of sliding sampling points, n is more than or equal to 1 and less than or equal to 5, and the number of average sliding sampling points is calculated

Wherein

(n) the number of times S is not 0; the timing position of the final symbol synchronization is pos (n) ═ pst (n) + S_aver,1≤n≤5。

The invention has the beneficial effects that: under the frequency selective fading short wave channel, the symbol timing can be completed by utilizing a method of searching the maximum peak of a non-fixed frequency point without knowing a signal transmission frequency point. The scheme is favorable for resisting bad transmission characteristics such as short-wave channel frequency offset and the like, and realizes accurate positioning of the optimal sampling position of the symbol. Simulation results show that in a selective fading channel with multipath and frequency offset, the estimation of the symbol timing position by the scheme can still keep higher accuracy.

Drawings

FIG. 1 is a block diagram of a short wave ionospheric channel model;

FIG. 2 is a flow chart of short wave channel blind symbol synchronization based on data frequency energy peak according to the present invention;

FIG. 3 is a diagram of absolute error comparison of symbol timing with the fixed frequency maximum peak search method according to the present invention;

fig. 4 is a comparison graph of error code performance between the scheme of the present invention and the fixed frequency point maximum peak search method.

Detailed Description

The technical scheme of the invention is detailed below by combining the accompanying drawings and the embodiment. It should be understood that the scope of the present invention is not limited to the following examples, and any techniques implemented based on the present disclosure are within the scope of the present invention.

The simulation parameters are set as follows:

taking 2FSK modulated signal as an example, let symbol rate Symr be 125sps and sampling frequency Fs be 9600Hz, and the number of sampling points Nsamp in one symbol be 77. The selected simulation channel is frequency selectivity with interference at intermediate latitudeFading channel, the channel model is shown in fig. 2. Assume that the input signal: s (t) ═ cos (2 pi ft) or s (t) ═ e^j2πft. Tau is the multipath time delay, f_dIn order to maximize the doppler shift frequency,

the fading characteristic of the short wave channel is described, the time-varying fading coefficient obeying the mean value of zero complex Gaussian distribution is described, and if the two variances are equal, the signal is output:

the symbol timing method for searching by utilizing the maximum peak of the non-fixed frequency point comprises the following steps:

step 1, sampling the received signal r (t) with a sampling frequency of 9600Hz to obtain { r (l) }, where l ═ 1, 2.. and N }, where N denotes the data length of each path, and then obtaining the number of sampling points 77 in one symbol from a known symbol rate 125sps and the sampling frequency of 9600 Hz. And taking any position pst (n) in the signal segment, wherein n is 1 as a first symbol synchronization reference, and representing the number of times of the current reference position by using n (1 is less than or equal to n is less than or equal to 5), and taking the symbol synchronization reference for 5 times in total.

And 2, starting from the reference position, selecting analysis data by using a window with the length fl ═ Nsamp:

{r₁(m),r₂(m),..., r₇₇(m), where m is 1,2,.., 77}, and the sliding is performed with one sample point as a step, the number of times of sliding of the sample point is represented by i (1 ≦ i ≦ 77), and the total number of sliding is 77 in total, so that the total window number fn is 77.

Step 3, respectively performing 77-point Fast Fourier Transform (FFT) on the data in each window, and obtaining a modulus value to obtain an energy spectrum:

Y_i(k) 1, 2.., 77;k 1,2, 77, and obtaining a set of frequency spectral lines f (k),k 1,2, 77. Due to symmetry, only half is taken: energy spectrum of Y_i(k) 1,2, ·, fn; 1,2, 39, set of frequency spectral lines f (k), 1,2, 39.

Step 4, in the energy spectrum Y_i(k) Find the maximum peak Mp in_i1,2, and the corresponding spectral line number fm_i,i＝1, 2.. 77, calculating the frequency corresponding to the maximum peak value

Step 5, if Mpf_iIf i is equal to 1,2,., 77, then s (n) is equal to 0, go to step 6; if Mpf_iI 1,2, 77 are not all equal, two different frequencies must occur, and the frequency of occurrence after the window sliding sequence is f. Finding a frequency Mpf satisfying the maximum peak frequency_iSliding order set equal to frequency f condition phi ═ i | Mpf_i＝f&I is more than or equal to 1 and less than or equal to 77, the maximum peak point Mp is max (mpi) in the sliding number set phi, i belongs to the sliding point number S (n) corresponding to phi, n is more than or equal to 1 and less than or equal to 5, namely the sliding sampling point number taking the position pst (n) as the reference.

Step 6, if n is less than or equal to 4, changing the reference position to Pst (n +1) ═ Pst (n) +77, where n is less than or equal to 1 and less than or equal to 4, and using the changed reference position as the n +1 th symbol synchronization reference position, and turning to step 2; otherwise go tostep 7.

Step 7, obtaining sliding sampling point number S (n) from 5 different symbol reference positions, wherein n is more than or equal to 1 and less than or equal to 5, and calculating the average sliding sampling point number

Wherein

(n) is a number other than 0. The timing position of the final symbol synchronization is pos (n) ═ pst (n) + S_aver,1≤n≤5。

The comparison results of the symbol timing absolute error and the error code performance of the scheme of the invention and the fixed frequency point maximum peak searching method are respectively shown in fig. 3 and fig. 4, and the results show that: under the short wave frequency selective fading channel, when the signal-to-noise ratio EbN0 is 20dB, the estimation of the symbol timing absolute error can still be kept about 1/10 symbols by using the method of the invention. The absolute error and error code performance of the symbol timing of the scheme of the invention are superior to those of the maximum peak search method of fixed frequency points.

Claims (1)

Translated fromChinese

1.一种基于数据频率能量峰值的短波信道盲符号同步方法，该方法包括：1. A shortwave channel blind symbol synchronization method based on data frequency energy peak, the method comprising:步骤1、对接收信号r(t)以采样频率Fs采样后得r(l),l＝1,2,...,N，N表示每一路的数据长度，再由已知的符号速率Symr和采样频率Fs求得一个符号内的采样点数Nsamp；以信号段内任意位置Pst(n),n＝1作为第一次符号同步基准，用n表示当前基准位置的次数，其中1≤n≤5，总共取5次符号同步基准；Step 1. Sampling the received signal r(t) with the sampling frequency Fs to obtain r(l), l=1,2,...,N, where N represents the data length of each channel, and then use the known symbol rate Symr Calculate the number of sampling points Nsamp in a symbol with the sampling frequency Fs; take any position Pst(n) in the signal segment, n=1 as the first symbol synchronization reference, and use n to represent the number of times of the current reference position, where 1≤n≤ 5. A total of 5 symbol synchronization benchmarks are taken;步骤2、从基准位置开始，用长度fl＝Nsamp的窗口选取分析数据:{r₁(m),r₂(m),...r_fn(m),m＝1,2,...,fl}，每次以一个采样点为步长进行滑动，采样点滑动次数用i表示，其中1≤i≤Nsamp，总共滑动Nsamp次，总窗口数fn＝Nsamp；Step 2. Starting from the reference position, use a window of length fl=Nsamp to select analysis data: {r₁ (m),r₂ (m),...r_fn (m),m=1,2,... ,fl}, each time a sampling point is used as a step for sliding, and the number of sampling points sliding is represented by i, where 1≤i≤Nsamp, a total of Nsamp times, and the total number of windows fn=Nsamp;步骤3、分别对每次窗内数据做fl点的快速傅里叶变换，取模值得能量谱:Y_i(k),i＝1,2,...,fn；k＝1,2,...,fl，并得到频率谱线集f(k),k＝1,2,...,fl；由于对称性，只取一半即可：能量谱为Y_i(k),i＝1,2,...,fn；

频率谱线集为f(k),

Step 3. Perform the fast Fourier transform of the fl point on the data in each window respectively, and take the modulo value energy spectrum: Y_i (k), i=1,2,...,fn; k=1,2, ...,fl, and obtain the frequency spectrum line set f(k), k=1,2,...,fl; due to symmetry, only half can be taken: the energy spectrum is Y_i (k), i= 1,2,...,fn;

The frequency spectrum line set is f(k),

步骤4、在能量谱Y_i(k)中找到最大峰值Mp_i,i＝1,2,...,fn；Step 4. Find the maximum peak value Mp_i in the energy spectrum Y_i (k), i=1,2,...,fn;以及对应的频谱线号fm_i,i＝1,2,...,fn；and the corresponding spectral line numbers fm_i , i=1,2,...,fn;计算得到最大峰值对应的频率

Calculate the frequency corresponding to the maximum peak

步骤5、如果Mpf_i,i＝1,2,...,fn全相等，则S(n)＝0，转步骤6；如果Mpf_i,i＝1,2,...,fn不全相等，必出现两个不同的频率，按窗口滑动顺序后出现的频率为f；寻找满足最大峰值频率Mpf_i等于频率f条件的滑动次数集Φ＝{i|Mpf_i＝f&1≤i≤fn}，在滑动次数集Φ中找到峰值最大点Mp＝max(Mpi),i∈Φ对应的滑动采样点数S(n),1≤n≤5，即以位置Pst(n)为基准的滑动采样点数；Step 5. If Mpf_i , i=1,2,...,fn are all equal, then S(n)=0, go to step 6; if Mpf_i ,i=1,2,...,fn are not all equal , two different frequencies must appear, and the frequency that appears after the window sliding order is f; find the set of sliding times that satisfies the condition that the maximum peak frequency Mpf_i is equal to the frequency f Φ={i|Mpf_i =f&1≤i≤fn}, Find the maximum peak point Mp=max(Mpi) in the sliding times set Φ, the number of sliding sampling points S(n) corresponding to i∈Φ, 1≤n≤5, that is, the number of sliding sampling points based on the position Pst(n);步骤6、如果n≤4，更改基准位置为Pst(n+1)＝Pst(n)+Nsamp,1≤n≤4，作为第n+1次符号同步基准位置，转步骤2；否则转步骤7；Step 6. If n≤4, change the reference position to Pst(n+1)=Pst(n)+Nsamp, 1≤n≤4, as the n+1th symbol synchronization reference position, go to step 2; otherwise, go to step 2 7;步骤7、由5次不同符号基准位置得到了滑动采样点数S(n),1≤n≤5，计算平均滑动采样点数

其中

为S(n)不为0的次数；最终符号同步的定时位置为Pos(n)＝Pst(n)+S_aver,1≤n≤5。Step 7. The number of sliding sampling points S(n) is obtained from 5 different symbol reference positions, 1≤n≤5, and the average number of sliding sampling points is calculated

in

is the number of times that S(n) is not 0; the timing position of the final symbol synchronization is Pos(n)=Pst(n)+Saver ,_1≤n≤5 .

Images