WO2007112681A1

WO2007112681A1 - An ofdm integer multiple frequency offset estimation method

Info

Publication number: WO2007112681A1
Application number: PCT/CN2007/001038
Authority: WO
Inventors: Ming Ding; Yun Wu; Hanwen Luo; Haibin Zhang; Suoming Pu
Original assignee: Shanghai Jiao Tong University; Sharp Kabushiki Kaisha
Priority date: 2006-03-30
Filing date: 2007-03-30
Publication date: 2007-10-11
Also published as: CN1822584A; CN100477655C

Abstract

This invention relates to an OFDM integer multiple frequency offset estimation method in the technical field of telecommunications. The method is based on a N-point synchronous training symbol, which is equally divided into L subdivisions. The receiving end reconstructs M OFDM symbols from the N-point symbols, and each constructed OFDM symbol has L points. Based on the fractional multiple frequency offset value estimated from forgoing modules, a new FFT algorithm structure capable of compensating fractional multiple frequency offset is designed to demodulate part of the reconstructed OFDM symbols by introducing the fractional multiple frequency offset modifying processing to the conventional FFT structures. The part of the reconstructed OFDM symbols are extracted in a way of random-selecting or equal-interval-selecting or continued-selecting from the M reconstructed OFDM symbols. The results of the demodulation are processed using square-combination, then the frequency point corresponding to the peak of the combined demodulated results is the integer multiple frequency offset estimation value. To further reduce computational complexity of this invention, an adaptive algorithm based on reliable decision feedback is employed. This invention has the advantages of less computational complexity and lower estimation error rate, so it is of high value for application in OFDM systems.

Description

Method for estimating OFDM integer multiple frequency offset

The present invention relates to a method in the field of communication technologies, and in particular, to a method for estimating an integer multiple of frequency offset of an OFDM. Background technique

At present, OFDM technology is applied in more and more wired and wireless communication fields, mainly because OFDM technology has many advantages: effective against multipath interference and narrowband interference, spectrum utilization, and high data transmission rate. However, OFDM is very sensitive to synchronization deviations, especially to frequency deviations. The frequency deviation is further divided into a fractional multiple of the subcarrier spacing and an integer multiple of the subcarrier spacing, which are referred to as fractional multiples and integer multiples. Among them, fractional octave bias will cause inter-subcarrier interference (ICI); integer octave bias will not cause ICI, but will cause cyclic shift of received data symbols, so that the error probability of demodulated information symbols is 50%. .

There are three common methods for estimating the OFDM integer multiple frequency offset:

(1) A synchronized training sequence based on a specific spectral pattern. This method requires fast Fourier transform (FFT) on the synchronized training symbols, and then correlates with known spectral patterns for cyclic shift, and estimates the integer frequency offset by finding correlation peaks. . See the literature: Schmidl, T. M. et al. "Low-overhead, low-complexity [burst] synchronization for OFDM", IEEE International Conference on Communications, Volume 3, June 1996, Page (s): 1301 - 1306. ("Low Data Overhead, Low Complexity OFDM Synchronization Method" IEEE International Telecommunications Technology Conference)

(2) Based on the virtual subcarrier of the 0FDM system, the 0FDM signal uses only a part of the subcarriers of the entire bandwidth during transmission, and some subcarriers are generally reserved as guard bands at the edge of the band, which are called virtual carriers. Orthogonality exists between subcarriers of OFDM. Therefore, the virtual subcarrier constitutes a "zero subspace" of the OFDM signal, and the integer octave offset can be derived by using the property that the inner product of the orthogonal subcarriers is zero. See literature: Liu, H. et al., "A high-efficient carrier estimator for OFDM communications," IEEE Communications Letters, Volume 2, Issue 4, April 1998, Page (s): 104-106. ("Efficient OFDM Frequency Offset Estimation Method" IEEE Communication Technology Newsletter) (3) OFDM synchronous training symbol structure based on A equal division, this method estimates the integer multiple frequency offset by calculating the autocorrelation of the specific delay of the training symbol and then obtaining the phase angle. Its estimation range increases with the increase, but the estimation accuracy becomes worse and the computational complexity increases accordingly. See the literature: Heiskala J et al.: OFDM Wireless LANs - A Theoretical and Practical Guide. [M] . Indianapol is USA : Pearson Education Inc, 2002. 70-73. ("0FDM Wireless LAN One-Way Theory and Practice Guide") This method is called a conventional method.

However, all of the above three methods have the disadvantage of excessive computational complexity. Let the number of subcarriers of the 0FDM system be /V. The method (1) needs to calculate the W point FFT, and at least l ₀ g ₂ N complex multiplications are needed; the method (2) needs to calculate the inner product between the signals, at least W complex numbers are needed. Multiplication; Method (3), the need to calculate the autocorrelation between delayed signals requires at least one complex multiplication. Nowadays, some typical values of W are 512, 1024 or 2048, etc., which makes the above three methods encounter great difficulties in practical applications. Summary of the invention

The purpose of the present invention is to provide a method for estimating the integer frequency offset of 0FDM, which is to perform the estimation of the integer multiple frequency offset by the method of reconstructing symbols and the efficient FFT algorithm, and fully utilize the 0FDM system. The performance of the frequency offset estimation is guaranteed, and the computational complexity is low.

The present invention is implemented by the following technical solutions, and specifically includes the following steps:

The sender generates an L-divided /V-point 0FDM synchronous training symbol, and each aliquot of the symbol contains points (= V/厶 and yV, both are integers);

Reconstructing the synchronized training symbols and reconstructing them into point 0FDM signal symbols;

In the reconstructed symbols obtained through the reconstruction step, the reconstructed symbols (ο ι, and integers) are extracted in a random, equally spaced or continuous manner, and the fractional octave bias estimates obtained from the pre-stage of the system are obtained. In the structure of the traditional FFT algorithm, the fractional frequency offset correction term is introduced to design a new structure of the FFT algorithm which can complete the fractional multiple offset compensation, which is used to demodulate the above reconstructed symbols and obtain a length of Λ. Frequency domain sequence

The obtained frequency domain sequences are combined according to the sum of squares, and the frequency points corresponding to the combined peaks are estimated values of integer multiple frequency offsets.

After the merging step is completed, an adaptive iterative algorithm based on the decision reliability feedback can also be entered, that is, the ratio of the mean value to the peak value of the frequency domain sequence obtained by the merging step is calculated, and the estimated value obtained in the merging step is reliable. The sex indicator adaptively increases the repeated demodulation step and the combining step according to the comparison result between the reliability index and a certain threshold value, and finally obtains a relatively reliable integer multiple frequency offset estimation value.

The invention also provides a receiving terminal, comprising:

a receiving unit, configured to receive the OFDM signal symbols sent by the transmitting terminal, where the L-point OFDM signal symbols are generated by reconstructing an equally-divided /V-point OFDM synchronized training symbol, each symbol An aliquot containing points;

a demodulation unit, configured to demodulate a reconstructed symbol selected from the received reconstructed symbols by using an FFT algorithm to obtain a frequency domain signal sequence of the reconstructed symbol; and

a merging unit for combining the frequency domain signal sequences obtained by the FFT operation, wherein the frequency points corresponding to the combined peak values are estimates of integer multiple frequency offsets;

The integers are all greater than 0, M = N / L, the 1 is an integer, 0 < K1. The invention also proposes a communication system, comprising:

a transmitting terminal, configured to generate a halved W-point OFDM synchronous training symbol, each of the symbols is equally divided, and reconstructing the synchronous training symbol to reconstruct a point OFDM signal symbol;

A receiving terminal according to the above aspect of the present invention.

DRAWINGS

Figure 1 OFDM baseband modulation and demodulation block diagram

Figure 2 is a schematic diagram of the frequency domain and time domain correspondence of OFDM synchronous training symbols with 1024 and Z=8. Figure 3 Schematic diagram of OFDM synchronization training symbol reconstruction with 〃=1024, =8

Figure 4 is a block diagram of the implementation of the present invention

Figure 5 Unbiased estimation for estimating integer multiple frequency offset 8 point FFT algorithm structure

Fig. 6 System performance diagram when selecting OFDM reconstructed symbols by random, equally spaced or continuous method Fig. 7 Performance diagram of the present invention for selecting OFDM reconstructed symbols by using equal interval method and having fractional octave bias estimation error

FIG. 8 is a performance comparison diagram of an adaptive algorithm based on decision reliability feedback and a conventional method when the number of iterations is 0, when the OFDM reconstruction symbol is selected by the equal interval method, and the fractional octave bias estimation is completely correct.

Figure 9 The present invention selects an OFDM reconstructed symbol by using an equal interval method, and the fractional octave bias estimation is completely Performance comparison of adaptive algorithm based on decision reliability feedback and traditional method with the number of iterations=1

FIG. 10 is a performance comparison diagram of an adaptive algorithm based on a decision reliability feedback method with a number of iterations = 2 when the OFDM reconstructed symbol is selected by an equal interval method, and the fractional octave bias estimation is completely correct.

The present invention will be further described below in conjunction with Figs. 1 through 10. First, the basic principle of the present invention will be described in detail.

(1) Generate halved / point OFDM sync training symbols

Equally divided OFDM synchronization training symbols are a fairly common training sequence structure and are often used in bit synchronization algorithms and fractional multiple frequency offset estimation algorithms. The method of generating it is as follows:

Let the number of subcarriers of the OFDM system be M effective symbol period is T, and the frequency of the first subcarrier is f _k =k/T (0≤k≤N-\), which is the first symbol and is loaded on the Ath subcarrier. Frequency domain data; δ _Λ , first symbol, baseband time domain data of the Jth sample point. If the first symbol is a synchronous training symbol, the pilot is inserted according to equation (1), and an octet OFDM synchronization training symbol can be generated.

a _jk ≠ 0 kQ (mod L) (丄) a _ik - 0 k≠0 (mod L) In the case of equal division, each halved training symbol contains the number of points

As shown in equation (2):

, = K _i+I , M (Z = 0, 1, 2,..., M-1, " = 0, 1, 2,..., — 1 ) ₍₂₎ Usually, a signal consisting of points is called One time slot ( _S lot).

(2) Reconstruct the synchronization training symbols

In the synchronous training symbol described in (1), each time slot contains a dot, and they are labeled 1, 2, 3, ..., M. The OFDM symbol reconstruction referred to in the present invention extracts the time domain points with the same label in each time slot to form a new 0FDM symbol, and each reconstructed 0FDM symbol includes a time domain point, as follows - Let ^, (0 ≤ w ≤ N - 1) be the first symbol of the receiving end, the first time domain sampling point, then the reconstructed symbol can be expressed as a vector form r _m (l ≤ w ≤ M), see equation (3) ) :

^Γ 1 2 ( , 0, ^f ), M > ^V i, 2M, · · , , ^r i, (L~\)M ) Γ3

^Γ Μ ⁼ yiM-\, ^r iM+M-\, 1, · · ·, , (L-\)M+Ml ) For the sake of simplicity, note:

r _m (n)=r _(illM+m (0≤m≤Ml, 0≤n≤Ll) (4) Then, r _ra =(r _ra (0), r _m (l), ..., r _m (n), ... , r _m ( -l)) , The above reconstruction process makes the data-like data obtain the advantage of time diversity, which makes the estimation algorithm more robust.

(3) Demodulating the reconstructed symbols to obtain the corresponding frequency domain sequence

Under ideal channel conditions, the time domain points in the reconstructed OFDM symbol r _m should be identical. However, due to the frequency offset, the phase of these points appears to be increasing or decreasing, as if modulated to a certain frequency. Thus, for r _m , the effect of the frequency offset can be equivalent to the baseband modulation process of an OFDM system containing subcarriers. Therefore, by demodulating these reconstructed symbols and looking for their spectral amplitude peaks, an integer multiple of the frequency offset can be estimated. The method of demodulating OFDM reconstructed symbols is a point FFT algorithm, and A is generally a relatively small number, such as 4, 8, and so on. The smaller number of FFT algorithms is the core of the present invention with low computational complexity.

Using the FFT algorithm for ^, you get:

Thus, the estimate of the integer multiple frequency offset is expressed as:

The simple L-point FFT algorithm is not enough to solve the problem of integer octave bias estimation completely. Especially when the fractional octave is around 0.5, the amplitude of the two frequencies in R _m is relatively large, which leads to the decision. Sex Can drop sharply. Therefore, the present invention combines fractional octave bias estimation values to design a new FFT algorithm structure, and ensures that the algorithm is always in the case of correct fractional octave bias estimation without adding any computational credits. Partial estimate.

The fractional multiple frequency offset estimation can be easily implemented by the prior art, for example, a method based on phase angle correlation for cyclic prefix correlation, a phase angle based correlation method based on 2 equal division symbols, and a method based on ML criterion search. The above method is generally located at the front stage of the scheme of the present invention, and the present invention can efficiently realize the task of integer octave bias unbiased estimation according to the fractional octave offset estimation value φ given therein.

Generally, the subcarrier of the OFDM system is an integer power of 2, such as: 256, 512, 1024, and the like. To ensure that it is an integer, Z can only be an integer power of 2, set = 2 ^ (for a positive integer). 2 ^ point

The FFT algorithm can be divided into two levels of butterfly operations, and the first stage butterfly operation contains two different complex multiplication coefficients. The p-th level in the original 2-point FFT algorithm, the <? multiplication coefficient is:

The ρ level corresponding to the present invention, the first? The multiplication factor is -

As long as the mathematical results of each stage are written as they are in accordance with the butterfly operation structure, the correctness of the present invention can be proved. For the 1^ using the modified FFT algorithm, we can get the formula (9):

R _m (^) = DFT(r _m | ^) =∑r _m («) _e " ^ (9)

=0

Equation (9) shows that the structure of the FFT algorithm provided by the present invention is equivalent to the phase compensation of the fractional frequency offset of the reconstructed symbol, and then the original FFT operation, and the frequency observation point is moved by a fractional multiple. The purpose of the frequency offset is such that the integer multiple frequency offset estimation of the present invention is an unbiased estimate. As long as the estimate of the fractional octave offset is more accurate, the new FFT algorithm will yield an unbiased estimate. Moreover, the correction term in the FFT algorithm provided by the present invention adds only on the complex phase, and does not destroy the inherent characteristics of the FFT algorithm. It should be noted that the function of the FFT algorithm of the present invention can be implemented by the following alternative method: According to the fractional multiple frequency offset estimation ^, the time domain phase compensation is performed on r _m first, and then the original FFT operation is performed. Although the alternative method achieves the purpose of designing a new FFT algorithm structure of the present invention, its computational complexity The algorithm provided by the present invention is used. Specifically, for each r _m , the alternative method requires the number of complex multiplications to be

Z + log ₂ (Z), whereas the present invention requires only log ₂ ( :) times. Based on equation (9), the estimate of the integer multiple frequency offset is expressed as:

f, =argmax{ R _m (y) \ye[-L/2,L/2) Rysz] (10)

(4) Combine the frequency domain sequence, estimate the integer multiple of the frequency offset by finding the peak value

In theory, the present invention only needs to estimate the integer multiple frequency offset for an FFT algorithm that uses unbiased estimation. However, due to the effects of multipath fading and noise of the channel, multiple joint decision methods are generally used. The specific operation method is as follows: Take out at random, equal interval or continuous sampling method

(θ ^Ι, and Α is an integer). The random selection method refers to randomly extracting different reconstructed symbols from the OFDM reconstructed symbols. The equal interval sampling method refers to randomly determining the first reconstructed symbol from the reconstructed symbol, and then extracting the different reconstructed symbols at equal intervals. The continuous drawing method refers to the weight from the OFDM. In the construction symbol, the first reconstructed symbol is randomly determined first, and then A different reconstructed symbols are successively taken out.

For the FFT algorithm with unbiased estimation, a series of frequency domain data _{m is obtained} , and then combined according to the sum of squares, the frequency point corresponding to the combined peak is the estimated value of the integer multiple frequency offset. f I ∑ Rm( ) y&[-LI2,LI2) and yeZ (11)

Other merging methods are: Modal value merging, as shown in equation (12), f 1 ye[-L/2, L/2) yeZ (12)

And the combination of the absolute value of the real part and the absolute value of the imaginary part, as shown in equation (13). f, ys[-L/2,L/2) RyeZ

(13) Since the signal-to-noise ratio can be considered to be constant within 1 OFDM symbol, Equation (11) is equivalent to maximum ratio combining, which is the optimal combining method. Equations (12) and (13) are sub-optimal mergers, which are replaced by a decrease in computational complexity. (5) Adaptive algorithm based on decision reliability feedback

When only one fixed A is taken, the foregoing scheme does not have much advantage over the conventional method. Therefore, the present invention further proposes an adaptive algorithm based on decision reliability feedback.

Due to the error of the fractional octave bias estimation or the non-ideal channel condition, the R (;?) in equation (11) is significantly lower than that of the other R(J), resulting in an unreliable estimation of equation (11). The ratio of the average value of R(y) to R^, ) is taken as the reliability index V of the formula (11). According to the comparison with a certain threshold value η, the multi-level joint decision method is used to improve the estimation of the present invention. First, perform symbol reconstruction according to equation (3) to obtain an OFDM reconstructed subsymbol r _m , denoted as set S, set the threshold, and let = 1 (/7 denotes the /7 decision). Entering the adaptive algorithm based on decision reliability feedback with the number of iterations: Firstly, ^ Μ (0< ≤1-∑::: 1. and 1„ is an integer) r _m are randomly extracted from S to form a set S„ _; S = S - S,,. For each r _m in 8 „, use equation FFT to calculate equation (9), get l„M _m . Then put Α,, Μ^^ according to equation (11) Merging, and adding the result of the previous stage, we get U„, which contains L elements U,, ( ):

U„ ) (14)

Find the frequency point corresponding to the peak of equation (14) and make an estimate for /, f, =argmax{U _f! (.y)} (15) y

The ratio of the average value of U„ (y) to U„ (J ₁ ) is taken as the reliability index of the formula (15) V": ∑u,, W

(16) When the threshold is less than or equal to η = Χ, the algorithm ends; otherwise, let “ = " + 1, repeat the above process. From equations (14) and (15), the range of ?; is [1/ , 1), the smaller the 77, the greater the probability of invoking the final decision. As a result, the computational complexity increases, but the estimated performance is improved.

It should be pointed out that equation (16) is not the only expression of reliability indicators. Other reliability indicators are: W

The ratio of the second largest value of U„ (y) to U„(f ₁ ) is used as the reliability index V„, as shown in equation (17).

If it is less than the threshold, the estimate is considered reliable, and the range of α is (0, 7).

U „ (f, ) The ratio of the average of the two adjacent values to the IJ „ (/ _; ) is used as the reliability index V _n , as shown in equation (18).

If ^ is less than the threshold, the estimate is considered reliable, and the range of values is ( ). After the adaptive algorithm based on the decision reliability feedback is adopted, the computational complexity of the present invention is further reduced, and the estimated performance is significantly improved.

The invention has the advantages of: utilizing OFDM symbol reconstruction, fully utilizing the received data, and having the characteristics of time diversity; using the FFT algorithm to estimate the integer multiple frequency offset, which greatly reduces the computational complexity of the system; The new FFT algorithm is designed to make the estimation of the present invention an unbiased estimation, which improves the reliability of the system. In addition, the present invention adopts an adaptive combining algorithm based on decision reliability feedback, so that the present invention has the characteristics of low computational complexity and superior estimation performance. Moreover, the simulation shows that the present invention is insensitive to the error of the fractional multiple frequency offset estimation and has good robustness. A specific OFDM parameter configuration is given below to illustrate the implementation steps of the present invention. It should be noted that the parameters in the following examples do not affect the generality of the present invention.

Document of the 3GPP organization: TR 25. 892 V6. 0. 0, "Feasibility Study for Orthogonal Frequency Division Multiplexing (OFDM) for UTRAN enhancement (Release 6)", given a set of OFDM parameters, as follows:

System bandwidth B 6. 528MHz

Number of subcarriers N 1024

Number of valid subcarriers N„ 705

Effective bandwidth 4. 495MHz Subcarrier spacing 6.375kHz

Cyclic expansion CP 64 (9.803us)

Symbol period T _s 156.85+9.81=166.66us

Under the above parameters, take =8, then #=128, use the equal interval sampling method and the square sum combination method of equation (11), and use equation (16) as the estimated reliability index, using reliable decision-based The 3-level adaptive algorithm of sexual feedback, the implementation steps of the present invention are as follows:

(1) According to equation (1), the subcarrier numbers are 160, 168, 176, 184, 192, 200, 208, 216, 224, 232, 240, 248, 256, 264, 272, 280, 288, 296, 304, 312, 320, 328, 336, 344, 352, 360, 368, 376, 384, 392, 400, 408, 416, 424, 432, 440, 448, 456, 464, 472, 480, 488, 496, 504, 520, 528, 536, 544, 552, 560, 568, 576, 584, 592, 600, 608, 616, 624, 632, 640, 648, 656, 664, 672, 680, 688, 696, 704, 712, 720, 728, 736, 744, 752, 760, 768, 776, 784, 792, 800, 808, 816, 824, 832, 840, 848, 856, 864 are loaded with pilot data on the frequency, other subcarriers Zero data is placed thereon, and such frequency domain data is passed through the serial-to-parallel conversion module 1, the IDFT module 2, and the parallel-to-serial conversion module 3 of FIG. 1, thereby generating an 8-equivalent synchronous training symbol of the form (2). 2 is a schematic diagram of a frequency domain and a time domain correspondence relationship of the synchronization training symbol. Then, the module of FIG. 1 is inserted into the cyclic prefix module 4, the insertion synchronization information module 5, the D/A conversion module 6, and the transmission filter processing module 7 are sent to the receiving end.

(2) The receiving end is processed by the receiving filter processing module 8 and the A/D conversion module 9 in Fig. 1, and combined with the bit synchronization information given by the system in Fig. 4, a complete ^, 0τ=0, 1, ... ..., 1023). According to equation (3), „, is reconstructed into 128 8-point OFDM symbols r _m (w= 2, . . . , 128). FIG. 3 is a schematic diagram of the reconstruction process.

(3) 3⁄4 = 0.25, A ₂ = 0.25, A ₃ = 0.5, = 3. Set the threshold 77 = 0.5, let " =1

(77 represents the level 77 judgment).

(4) After taking the integer, it is |_ t„M”. Therefore, according to the equal interval method, take out |_Α„Μ” r _m and combine with the fractional octave bias estimation value p given by the previous stage. ^^^ performs the operation shown in Fig. 5 to obtain „M“ _m . (5) Combine the above-mentioned |_A„M“ according to equation (14) and make an estimate of the integer multiple frequency offset according to equation (15). Calculate according to equation (16). When less than or equal to the threshold /, or = , the algorithm ends; otherwise,

^■n = n + \ , back to (4).

The channel is an 8-channel Rayleigh fading channel, as follows:

Delay ( ns ) relative power (dB)

Path 1 0 0

Path 2 153 -6. 7

Path 3 306 - 13. 3

Path 4 459 -19. 9

Path 5 612 -27. 6

Path 6 765 -33. 3

Path 7 919 -39. 9

Path 8 1072 -46. 6

6 is a system performance diagram when the OFDM reconstructed symbol is selected by a random, equally spaced or continuous method according to the present invention, and the graph shows that when A is small (such as 0.1 or 0.2), the system performance using the continuous extraction method is shown. Inferior to the other two methods, when A is large, the system performance tends to be consistent when using random, equal interval or continuous methods. Overall, systems with equally spaced extraction methods have the best performance.

7 is a performance diagram of the present invention in which an OFDM reconstructed symbol is selected using an equally spaced method, and there is a fractional octave bias estimation error, and the figure shows that the present invention is not sensitive to the error of the fractional octave bias estimation. It is beneficial to improve the robustness of the entire system.

FIG. 8 is a performance comparison diagram of an adaptive algorithm based on decision reliability feedback and a conventional method when the fractional octave bias estimation is completely correct when the OFDM reconstructed symbol is selected by the equal interval method. In the conventional method, setting the delay to a time slot, correlating, and finding the phase angle, it is possible to estimate the integer multiple offset, which requires (N - M) = 896 complex multiplications. In the present invention, when 1 = 0.2, the multiplication is required.

300 times, which is equivalent to 1/3 of the traditional method, but achieves a signal-to-noise ratio gain of 0.5 dB over the conventional method; when A = 0.6, the present invention requires complex multiplication by about 900 times, and its computational complexity and conventional methods Equal, but achieved a signal-to-noise ratio gain of around 3dB.

Figure 9 is an iteration of the present invention when the fractional octave bias estimation is completely correct when using the equal interval method. Performance comparison diagram of adaptive algorithm based on decision reliability feedback and traditional method with number of times = 1. In the simulation, 1^^ = 0.25 is set, and three cases of 77 are examined - 0.5, 0.75, and 0.775. When = 0.5, the present invention obtains a signal-to-noise ratio gain of 3 dB over the conventional method, and an adaptive algorithm simulation of = l shows that the average number of complex multiplications required is only 57% to 64% of the conventional method; when; 7 = At 0.75, the present invention achieves a signal-to-noise ratio gain of 2. 2 dB over the conventional method, but the average number of complex multiplications required is only 57% of the conventional method.

58%; When; 7 = 0.875, the present invention achieves a signal-to-noise ratio gain of 1.4 dB over the conventional method, but the average number of complex multiplications required is only 57% of the conventional method;

FIG. 10 is a performance comparison diagram of the adaptive algorithm based on the decision reliability feedback method and the conventional method when the fractional multiple frequency offset estimation is completely correct when the equal interval method is extracted by the present invention. In the simulation, let = = 0.25, = 0.5, and consider the three cases of 77: 0. 5, 0. 75 and 0. 875. When 7 = 0.5, the present invention obtains a signal-to-noise ratio gain of 4.7 dB over the conventional method, and an adaptive algorithm simulation of = 2 shows that the average number of complex multiplications required is only 57% to 75% of the conventional method; When 7 = 0.75, the present invention achieves a signal-to-noise ratio gain of 2. 6 dB over the conventional method, but the average number of complex multiplications required is only 57 °/ of the conventional method. To 58%; when ;7 = 0.875, the present invention achieves a signal-to-noise ratio gain of 1.3 dB over the conventional method, but the average number of complex multiplications required is only 57% of the conventional method.

The simulation results show that the invention has the advantages of low computational complexity and low estimation error rate, and has high application value in OFDM systems.

Claims

Rights request

A method for estimating an integer multiple of frequency offset of an OFDM, comprising:

In the generating step, the transmitting end generates an A-divided /V-point OFDM synchronous training symbol, and each of the symbols is equally divided;

a reconstructing step, reconstructing the synchronous training symbol, and reconstructing the symbol into a ^ point OFDM signal symbol;

In the demodulating step, the receiving end demodulates the λ reconstructed symbols selected from the reconstructed symbols using an FFT algorithm to obtain a frequency domain signal sequence of the reconstructed symbols;

The merging step combines the frequency domain signal sequences obtained by the FFT operation, and the frequency points corresponding to the combined peaks are estimated values of integer multiple frequency offsets;

The L, are all integers greater than 0, M-N/L, the integer, 0 < K1.

The method for estimating an OFDM integer multiple frequency offset according to claim 1, wherein the reconstructing step comprises: sequentially averaging each OFDM symbol sample point, and then using the same number The fields are extracted in order to form a new OFDM symbol.

The method for estimating an OFDM integer multiple frequency offset according to claim 1, wherein the demodulating step comprises: combining a fractional multiple frequency offset estimation value given by a system pre-stage synchronization module in a conventional FFT algorithm The fractional frequency offset correction term is introduced into the complex multiplication coefficient of the structure, and the fractional multiple frequency offset compensation and the demodulation OFDM reconstruction symbol are completed at the same time; or, after directly subtracting the fractional frequency offset of the OFDM reconstructed symbol, The OFDM reconstructed symbols are demodulated using a conventional FFT algorithm.

The method for estimating an OFDM integer multiple frequency offset according to claim 1 or 3, wherein the demodulating step comprises: randomly extracting different reconstructed symbols from the OFDM symbols; or From a point OFDM symbol, the first reconstructed symbol is randomly determined first, and then the different reconstructed symbols are extracted at equal intervals; or the first reconstructed symbol is randomly determined from the A-point OFDM symbols. A further different reconstructed symbols are successively taken out.

5. The method of estimating OFDM integer multiple frequency offset according to claim 1, wherein the combining step comprises: combining squares and sums, or modulo and combining, or absolute values of real parts and absolute values of imaginary parts. The combination of and combines the frequency domain signals of the OFDM reconstructed symbols.

6. The method of estimating an OFDM integer multiple frequency offset according to claim 1, wherein After the merging step, adaptive iteration based on the decision reliability feedback is performed, that is, by evaluating the reliability of the estimation value obtained by the merging step, the demodulation step and the merging step are adaptively added and repeated until a more reliable integer octave bias estimation is obtained. value.

7. The method for estimating an OFDM integer multiple frequency offset according to claim 6, wherein evaluating the reliability of the estimated value obtained by the combining step comprises: calculating a ratio of an average value to a maximum value of the frequency domain signal sequence obtained by the combining step. As the reliability index of the estimated value obtained in the merging step, if the reliability index is less than the threshold η, the estimation is considered to be reliable. Otherwise, the system will increase and repeat the demodulation step and the merging step until a more reliable integer multiple is obtained. The partial value of the estimate is η, which is [// i].

8. The method for estimating an OFDM integer multiple frequency offset according to claim 6, wherein evaluating the reliability of the estimated value obtained by the combining step comprises: calculating a second largest value and a maximum value of the frequency domain signal sequence obtained by the combining step. The ratio is the reliability index of the estimated value obtained in the merging step. If the reliability index is less than the threshold a, the estimation is considered to be reliable. Otherwise, the system will increase and repeat the demodulation step and the merging step until a more reliable integer is obtained. The frequency offset estimation value, the value range of a is (0, 1).

9. The method of estimating an OFDM integer multiple frequency offset according to claim 6, wherein evaluating the reliability of the estimated value obtained by the combining step comprises: calculating an average value of adjacent two values of the frequency domain signal sequence obtained by the combining step. The ratio to the maximum value is used as the reliability index of the estimated value obtained in the merging step. If the reliability index is less than the threshold value, the estimation is considered to be reliable. Otherwise, the system will increase the meaning and repeat the demodulation step and the merging step until the comparison is obtained. A reliable integer multiple frequency offset estimation value, the value range is (0, ).

10. A receiving terminal, comprising:

a receiving unit, configured to receive a point OFDM signal symbol sent by the transmitting terminal, where the L point 0FDM signal symbol is generated by reconstructing an L-divided OFDM synchronous training symbol, and each symbol is equally divided Point

a merging unit, configured to combine the frequency domain signal sequences obtained by the FFT operation, and the frequency point corresponding to the combined peak value is an estimated value of the integer multiple frequency offset;

The integers are all greater than 0, M = N / L, the integer is 0, A < A 1.

The receiving terminal according to claim 10, wherein the demodulation unit combines the fractional frequency offset estimation value given by the system pre-stage synchronization module to introduce a small multiplication coefficient in the traditional FFT algorithm structure. a multiple of the frequency offset correction term, and simultaneously complete the fractional multiple offset compensation and demodulate the 0FDM reconstructed symbol; or, After directly offsetting the fractional octave offset of the OFDM reconstructed symbol, the OFDM reconstructed symbol is demodulated by the conventional FFT algorithm.

The receiving terminal according to claim 10 or 11, wherein the demodulation unit randomly extracts different reconstructed symbols from the point OFDM symbols; or randomly from the point OFDM symbols Determining the first reconstructed symbol, and then extracting A different reconstructed symbols at equal intervals; or randomly determining the first reconstructed symbol from a single point OFDM symbol, and then successively taking out A different reconstructed symbols symbol.

13. The receiving terminal according to claim 10, wherein the merging unit uses a sum of squares, or a modulo value and a merging, or a combination of a real part absolute value and an imaginary part absolute value, combining OFDM weights. The frequency domain signal of the symbol.

The receiving terminal according to claim 10, further comprising: a reliability evaluation unit, configured to evaluate reliability of the estimated value generated by the merging unit, adaptively increasing until a more reliable integer multiple is obtained Frequency offset estimate. -

The receiving terminal according to claim 14, wherein the reliability evaluation unit calculates a ratio of an average value to a maximum value of the frequency domain signal sequence obtained by the merging unit as an estimated value obtained by the merging unit. Reliability index, if the reliability index is less than the threshold value ^, the estimation is considered to be reliable. Otherwise, A is increased until a more reliable integer multiple frequency offset estimation value is obtained. The value range of the η is U/L, 1 .

The receiving terminal according to claim 14, wherein the reliability estimating unit calculates a ratio of a second largest value to a maximum value of the frequency domain signal sequence obtained by the merging unit as an estimation obtained by the merging unit The reliability index of the value is considered to be reliable if the reliability index is less than the threshold value α, otherwise, it is increased until a more reliable integer multiple frequency offset estimation value is obtained, and the value range of the a is (0, 1).

The receiving terminal according to claim 14, wherein the reliability evaluation unit calculates a ratio of an average value to a maximum value of adjacent two values of the frequency domain signal sequence obtained by the merging unit as the merge The reliability index of the estimated value obtained by the unit, if the reliability index is less than the threshold value β, the estimation is considered to be reliable, otherwise, the input is increased until a more reliable integer multiple frequency offset estimation value is obtained, and the value range of the β is obtained. Yes (0, 1) ο

18. A communication system comprising:

a transmitting terminal, configured to generate an equal-point point 0FDM synchronization training symbol, each of the symbols Equally including a point, and reconstructing the synchronized training symbol to reconstruct it into an A-point OFDM signal symbol;

A receiving terminal according to any one of claims 10 to 17.