CN1309192C - Method for tracing carrier frequency of orthogonal frequency division multiplexing system in multipath fading channel - Google Patents

Abstract

The present invention relates to a carrier wave frequency tracking method of an orthogonal frequency division multiplexing system in a multipath fading channel, which belongs to the technical field of digital communication. A signal reconstruction theory is used for providing a carrier tracking algorithm of a new OFDM system, which obtains a new carrier tracking technical scheme. The scheme belongs to a blind tracking method which does not reduce the spectrum effectiveness of a system and does not use virtual carrier waves. At present, blind tracking methods for OFDM carrier synchronization do not exist in China, and compared with the blind tracking methods which are proposed at abroad, the method has the advantages of low computational complexity, high tracking precision, high tracking speed, easy realization, etc. The present invention can be used for wireless network cards using OFDM technologies, (fourth generation) mobile communication systems and xDSL modulator-demodulators, and is especially suitable for the conditions of multipath fading channels.

Description

Carrier frequency tracking method of orthogonal frequency division multiplexing system in multipath fading channel

(I) technical field

The invention relates to a frequency tracking method, and belongs to the technical field of digital communication.

(II) background of the invention

With the increasing demand for high-speed digital communication technology, Orthogonal Frequency Division Multiplexing (OFDM) technology is gaining attention due to its extremely high spectral efficiency and good adaptability to scattered channels, and is considered as a supporting technology for the fourth generation mobile communication system in the future. OFDM is a multi-carrier transmission technique that divides the entire channel into N sub-channels and transmits information in parallel.

The OFDM technique is mainly used for: (fourth generation) mobile communication system, broadband wireless access. In digital cellular mobile communication applications, OFDM is one of the hot technologies currently studied, such as MC-CDMA (multi-carrier code division multiple access), OFDMA (orthogonal frequency division multiple access), etc.; in broadband wireless access applications, IEEE802.11a and IEEE802.16 are both standards (or drafts) based on OFDM, and HiperLAN II of ETSI is also a standard based on OFDM technology; DMT, a special form of OFDM, has found widespread use in broadband cable access technologies, such as xDSL (high speed digital subscriber line) technologies.

Not only each sub-channel of the OFDM system has no guard band, but also the main lobes of the frequency spectrum of the signals between the adjacent channels are mutually overlapped, so that the OFDM has very high frequency spectrum utilization rate. Moreover, OFDM implementation is very simple. When QAM or MPSK modulation is used for the sub-channel, IFFT (Inverse Fast Fourier transform, IFFT) and FFT (Fast Fourier transform, FFT) can be used for the modulation and demodulation process, and a corresponding DSP chip or dedicated integrated circuit can be directly used for hardware implementation. Because the general OFDM system adopts the cyclic prefix mode, the OFDM system can completely eliminate the intersymbol interference caused by the multipath propagation of signals under certain conditions and completely eliminate the damage of the orthogonality among carriers caused by the multipath transmission, so the OFDM system has strong multipath interference resistance and strong fading resistance. In addition, the subcarriers of OFDM divide the entire channel into many narrow channels, making the fading on each subchannel approximately flat, so that equalization of a subchannel of an OFDM system requires only a one-tap equalizer.

Despite the many advantages of OFDM, there are some difficulties currently in use for mobile communications, one of the most significant constraints being that OFDM systems have particularly high accuracy requirements for synchronous systems [1 ]. In particular, the OFDM system is particularly sensitive to frequency offset of a receiver oscillator and Doppler (Doppler) frequency shift caused by relative motion between a receiver and a transmitter, so that the performance of the system is rapidly reduced, and even the system cannot work normally, which puts high requirements on carrier frequency tracking. Theoretical analysis shows that the carrier synchronization accuracy required by the OFDM system is to make the residual (i.e. corrected) relative frequency offset (the ratio of absolute frequency offset to carrier spacing) of the system within 2% (wireless channel condition) [2], and for an Additive White Gaussian Noise (AWGN) channel, the frequency offset that the system can tolerate is slightly larger, but when the relative frequency offset exceeds 2%, the performance of the system is also greatly affected due to the inter-carrier interference caused by the frequency offset. Therefore, carrier synchronization for OFDM systems is generally divided into two steps: and (4) capturing and tracking. Acquisition is also called coarse synchronization, that is, the frequency offset of the local carrier is locked within a small range by a proper acquisition algorithm and circuit, and the relative frequency offset after acquisition is generally required to be below 10%. The tracking is also called fine synchronization, and the tracking has the functions of further reducing the synchronization error, controlling the synchronization error within the allowed range of the OFDM system, further tracking the change of the synchronization parameters and adjusting the synchronization parameters in time. The main requirements for fine synchronization are high tracking precision and high tracking speed, and the relative frequency offset is controlled within the range in which the system can normally work.

The carrier synchronization method of the currently applied OFDM system comprises the following steps: auxiliary data methods [3, 5] and blind estimation methods without auxiliary data (frequency offset estimation method based on Maximum Likelihood (ML) algorithm [4], frequency offset estimation method based on MUSIC algorithm and frequency offset estimation method based on ESPRIT algorithm [6, 7 ]).

The most significant drawback of the auxiliary data method is that it occupies an effective bandwidth for transmitting auxiliary data, which is insurmountable in mobile communication and broadband access systems requiring high spectral efficiency. Particularly, in the OFDM system, a Cyclic Prefix (CP) needs to occupy a certain bandwidth, and the channel estimation algorithm also needs to use auxiliary data to occupy the bandwidth, and if the auxiliary data is added for synchronization, the advantage of high OFDM spectrum utilization ratio cannot be reflected. Therefore, the assistance data method at the expense of spectral efficiency is not an ideal synchronization method.

The ML algorithm [4] is an earlier proposed one of OFDM synchronization parameter blind estimation algorithms. It can be used for timing acquisition and frequency offset tracking, where the range of frequency offset tracking is ± 0.5 subcarrier spacing, which is wider as a tracking algorithm, but too narrow as acquisition. However, ML suffers from a number of deficiencies [9 ]. The ML algorithm can be basically used only for AWGN channels, and it has a great difficulty in being generalized to actual multipath fading channels.

Liu and Tureli et al estimate the frequency offset using a harmonic estimation algorithm based on subspace decomposition, including the MUSIC algorithm [6] and the higher resolution ESPRIT algorithm [7 ]. The method has the advantages of large control range of frequency deviation, high precision, and capability of capturing and tracking; in particular, the tracking accuracy is difficult to achieve by many methods. However, the algorithm is complex and the calculation amount is large. Singular Value Decomposition (SVD) or a large amount of eigenvalue calculation is required, when the number of effective subcarriers is large (for example, more than 1000), the calculation amount is difficult to bear, and even if the fastest DSP chip is used, the real-time requirement is difficult to meet; due to the complexity of the algorithm, the required memory space is large, and its implementation is also problematic.

Disclosure of the invention

The invention aims to provide a carrier frequency tracking method of an orthogonal frequency division multiplexing system in a multipath fading channel, which has low cost and high precision and can meet the requirements of practical application aiming at the defects of the prior art.

The invention utilizes the signal reconstruction theory provided by the inventor to derive a new method for estimating the frequency offset in the OFDM system, thereby obtaining a new method for fine synchronization of carrier frequency, and being a new technical scheme for solving the problem of tracking (fine synchronization) of the carrier frequency of the OFDM system. The method can be used in a wireless communication system, coarse synchronization is completed by a high-precision rapid acquisition method such as an auxiliary data method, and then carrier frequency tracking is performed by the method.

Setting a frame of OFDM data to be transmitted as X (0), X (1), …, X (N-1), N is subcarrier number, which is a group of complex numbers calculated by bit data to be transmitted according to a specific subcarrier modulation mode (such as MQAM, MPSK, etc.) and a constellation diagram, and obtaining OFDM signals in time domain (discrete) form after IFFT

$s\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}X\left(k\right)\exp [\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C0213538700121.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N],(n=\mathrm{0,1},...,N-1)---\left(1\right)$

Wherein, $j=\sqrt{-1}$ is a virtual unit; n is the number of subcarriers, i.e., the number of FFT points; 1/N is the normalized subcarrier spacing. Setting the sampling impulse response of the slow fading channel as h (n), when the system has normalized relative frequency deviation epsilon₀The OFDM signal received by the receiving end is

r(n)＝(h*s)(n)exp(j2πε₀N/N) + * (N) (N is 0, 1, …, N-1) (2) where * (N)1 is the channel noise at the nth sampling instant of the OFDM signal, which is a random variable subject to Gauss distribution; * denotes a circular convolution.

FFT was performed on r (N) (0, 1, …, N-1) to obtain

$\hat{X}\left(k\right)=H\left(k\right)S(k,{\mathrm{\&epsiv;}}_0)+\hat{N}\left(k\right)(k=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;,N-1)---\left(3\right)$

Wherein, h (k) is the frequency domain characteristic of the kth Gauss channel, which can be estimated by a proper channel estimation algorithm; s (k, ε)₀) Is the output of the kth Gauss channel, which contains inter-carrier interference due to frequency offset;is the noise output on the kth Gauss channel.

Let us estimate H (k) according to a certain channel estimation algorithm, thenAnd H (k) we can get the pair S (k, ε)₀) Is estimated value of

The accuracy of the estimation is affected by the channel noise and the performance of the channel estimation algorithm

IFFT to obtain * (k, epsilon)₀) Which is an OFDM time domain signal that is unaffected by fading channels but affected by frequency offset and noise. ByAccording to the decision rule, we can get the estimation of the OFDM data information of the frame

，(k＝0，1，…，N-1)。

In the normal case of the operation of the device,(k-0, 1, …, N-1) is sufficiently close to x (k), (k-0, 1, …, N-1), that the bit error rate of the system is low (e.g., 10)^-2When below), the decided OFDM dataAnd (k is 0, 1, …, N-1) eliminates most of the influence of frequency offset and noise. Thus according to(k-0, 1, …, N-1) reconstructed OFDM signal

$\tilde{s}\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\tilde{X}\left(k\right)\exp [\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C0213538700121.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;nk}/N],(n=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;,N-1)---\left(4\right)$

Will be very close to s (N) (0, 1, …, N-1), if there is no error (which is often the case) after the current frame OFDM symbol decision, there will be $\tilde{s}\left(n\right)=s\left(n\right)(n=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;,N-1)$ . If the frequency offset estimation of the system is epsilon, the estimation of the OFDM time domain signal containing the frequency offset is

$s(n,\mathrm{\&epsiv;})=\tilde{s}\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C0213538700121.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;\&epsiv;n}/N)(n=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;,N-1)---\left(5\right)$

If ε is ═ ε₀In the absence of bit errors, * (n, ε)₀) The difference in s (n, ε) will only be affected by the channel estimation accuracy and channel noise, i.e., * (n, ε)₀) Including the effects of channel noise and channel estimation errors, and s (n, epsilon) is substantially unaffected by them, so epsilon ═ epsilon₀Time * (n, ε)₀)

And s (n, ε) will be very close, as ε and ε₀Increase in the difference of * (n, ε)₀) And s (n, epsilon) will also become larger and larger.

We try to estimate ε₀. To solve the following minimization problem

$\underset{\mathrm{\&epsiv;}}{\min }\mathrm{\&Phi;}\left(\mathrm{\&epsiv;}\right)=\underset{\mathrm{\&epsiv;}}{\min }\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\mathrm{\&phi;}(n,\mathrm{\&epsiv;})---\left(6\right)$

Wherein

φ(n，ε)＝|*(n，ε₀)-s(n，ε)|²，(n＝0，1，…，N-1) (7)

From which the frequency offset epsilon can be estimated₀。

The minimization problems (6), (7) are solved below.

Is provided with $\hat{s}(n,{\mathrm{\&epsiv;}}_0)=a\left(n\right)+\mathrm{jb}\left(n\right),\tilde{s}\left(n\right)=\tilde{a}\left(n\right)+j\tilde{b}\left(n\right),$ Then

$\mathrm{\&phi;}(n,\mathrm{\&epsiv;})=|\tilde{s}(n,{\mathrm{\&epsiv;}}_0)-s(n,\mathrm{\&epsiv;}){|}^2=|a\left(n\right)+\mathrm{jb}\left(n\right)-[\tilde{a}\left(n\right)+j\tilde{b}\left(n\right)]\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C0213538700121.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;\&epsiv;n}/N){|}^2---\left(8\right)$

When ε < 1 (this condition is generally satisfied and a coarse synchronization guarantee for the carrier frequency is obtained by the acquisition circuit), the exponential function of the above equation is Taylor expanded, only the linear term part is retained, so that the above equation can be approximately expressed as

$\mathrm{\&phi;}(n,\mathrm{\&epsiv;})\&ap;|a\left(n\right)+\mathrm{jb}\left(n\right)-[\tilde{a}\left(n\right)+j\tilde{b}\left(n\right)](1+\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C0213538700121.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;\&epsiv;n}/N){|}^2$

$=[a\left(n\right)-\tilde{a}\left(n\right)+2\mathrm{\&pi;}\tilde{b}\left(n\right)\mathrm{\&epsiv;n}/N{]}^2+[b\left(n\right)\widetilde{-b}\left(n\right)-2\mathrm{\&pi;}\tilde{a}\left(n\right)\mathrm{\&epsiv;n}/N{]}^2---\left(9\right)$

The above equation is a quadratic function with respect to epsilon, so that the objective function phi (epsilon) in equation (6) is the sum of N quadratic functions with respect to epsilon, and also a quadratic function with respect to epsilon, with obviously minimal values. To find its minimum, derivative of phi (epsilon) and order

$\frac{\mathrm{d\&Phi;}\left(\mathrm{\&epsiv;}\right)}{\mathrm{d\&epsiv;}}=\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\frac{\mathrm{d\&phi;}(n,\mathrm{\&epsiv;})}{\mathrm{d\&epsiv;}}=0---\left(10\right)$

Solving the formula (10) to obtain:

$\mathrm{\&epsiv;}=\frac{\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}[\tilde{a}\left(n\right)b\left(n\right)-a\left(n\right)\tilde{b}\left(n\right)]}{\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}[({\tilde{a}}^2\left(n\right)+{\tilde{b}}^2\left(n\right))2\mathrm{\&pi;n}/N]}---\left(11\right)$

this is the frequency offset estimated from the OFDM time domain signal. Epsilon is used as the frequency offset value estimated by the frame, and the frequency offset value is converted into a voltage signal and then input into the input end of a Voltage Controlled Oscillator (VCO), so that the frequency offset or Doppler frequency shift of the oscillator at the receiving end is corrected, and the frequency offset can be controlled to be close to a zero value.

Because of the influence of noise, the estimation value is a random variable, and the accuracy of the estimation algorithm can be known by calculating the mean square error of the estimation value and the actual frequency offset.

As shown in fig. 1, r (n) is a signal received by the system receiving end, and is first sent to the FFT module 1 for Fast Fourier Transform (FFT) to obtain a frequency domain signal(ii) a The channel estimation is carried out by a channel estimation module 2, the channel characteristic H (k) is obtained by a channel estimation algorithm, and a signal for removing the influence of a fading channel is obtained after equalization

The signal containing a frequency offset epsilon₀(ii) a The signal is divided into two paths: one path is input into a decision module 3, and a signal after decision is obtained by a decision rule

The signal is substantially unaffected by the frequency offset; the other path is transferred to an IFFT module A4 to carry out inverse fast Fourier transform to obtain a reconstructed time domain signal * (n, epsilon) containing real frequency offset₀) The signal is not affected by fading channel; post-decision signalWhen the signal is output, the signal is transferred to an IFFT module B5 to perform inverse fast Fourier transform, and a time domain signal which is not affected by frequency deviation basically is obtained

(ii) a The reconstructed two paths of signals are simultaneously input into a frequency deviation calculation module 6, and the frequency deviation epsilon is calculated by a frequency deviation formula, namely the frequency deviation epsilon is a pair epsilon₀And e is used as the frequency offset value estimated by the frame, and the frequency offset value is converted into a voltage signal and then input into the input end of a Voltage Controlled Oscillator (VCO), so that the frequency offset or Doppler frequency shift of the oscillator at the receiving end is corrected, and the frequency offset is controlled to be close to a zero value.

The frequency offset calculation process can be described by adopting a software flow, the system software adopts a mode that a main program calls an interrupt subprogram, the main program completes the functions of system initialization, system self-check, interrupt initialization and the like, and the whole frequency offset calculation and tracking process including software judgment, IFFT and frequency offset calculation is completed in the interrupt service subprogram; is judged to obtain

After that, the air conditioner is started to work,

andthe IFFT of (a) can be implemented serially, and the interrupt service subroutine flow is shown in fig. 5.And

the IFFT of (1) can also be implemented in parallel, and the flow of the interrupt service subroutine implemented in parallel computing is shown in fig. 6.

All the calculations in the above-mentioned flow are performed in software, but it should be noted that most of the operations of the two IFFT calculations and the frequency offset calculation (calculation of denominator and numerator of the frequency offset calculation formula) can also be performed in hardware. In the latter case, the software of the system performs only a small amount of calculations, and more of them performs system control, synchronization, and the like. This allows more flexibility in system implementation.

The above-mentioned implementation process can be implemented in one of the following ways:

(1) an FPGA (field programmable gate array) is combined with a DSP (digital signal processor) chip;

(2) a Digital Signal Processor (DSP) is adopted;

(3) an ASIC circuit design is adopted.

The invention has the outstanding advantages that: the blind estimation tracking method has the advantages of low calculation complexity, small calculation amount, high estimation precision, simple realization, large trackable range (12% relative frequency offset) and no reduction of the frequency spectrum efficiency of the system, and belongs to a blind estimation tracking method in the true sense. For fading channels, the amount of computation to estimate the frequency offset is mainly two IFFTs (excluding the amount of computation for channel estimation). Meanwhile, the tracking algorithm adopted by the invention can be used for parallel calculation with the OFDM demodulator, and the real-time performance is stronger.

(IV) description of the drawings

Fig. 1 is a block diagram of a carrier tracking implementation based on signal reconstruction.

Figure 2 is the tracking performance of the present algorithm at a signal-to-noise ratio of 21 dB.

Figure 3 is a residual frequency offset mean square error.

Fig. 4 is a block diagram of a system architecture for DSP implementation.

FIG. 5 is a flow diagram of an interrupt service routine implemented in a serial computing manner.

FIG. 6 is a flow diagram of an interrupt service routine implemented in parallel computing.

The system comprises an FFT module 1, a channel estimation module 2, a decision module 3, an IFFT module A4, an IFFT module B5, a frequency deviation calculation module 6, a channel output 7, a front-end processing module 8, an A/D analog-to-digital converter 9, an FPGA 10, a 5V voltage source 11, a 3.3V voltage source 12, an SRAM 13, a reset watchdog 14, a JTAG module 15, a TMS320C54X chip 16, a 1.6V voltage source 17, a voltage stabilizing block 18, a 10MHz crystal oscillator 19, a frequency multiplier 20, a D/A digital-to-analog converter 21 and a voltage controlled oscillator 22.

Fig. 2 and fig. 3 are the results of tracking simulation of the MATLAB software on a 64-carrier OFDM system, in which the subcarriers employ a 16QAM modulation scheme. Simulation results show that the algorithm has high tracking speed and high precision, completely meets the requirement of an OFDM system on carrier tracking precision, and has very high practical value.

(V) detailed description of the preferred embodiments

The invention aims to solve the problem of carrier tracking in the field of wireless communication, and a system for special requirements of wireless communication has the characteristics of quick operation, strong real-time performance, small size, flexibility, energy conservation and low power consumption. As a generally accepted implementation, the digital signal processor and the FPGA chip have the advantages of simple design, flexibility, convenience and easy upgrading. The following embodiments based on DSP and FPGA chips in a parallel computing mode are given:

in the present embodiment, 64 subcarriers are used, and each subcarrier is modulated by 16 QAM.

Since the TMS320C54X series of the TI company is widely used as a terminal processing device in the field of wireless communication, this embodiment employs this series of DSP chips, and selects the FLEX10k series FPGA chip of the Altera company.

The software flow of the system is given in the third part, but in order to optimize the system performance, the FPGA computing capability and the DSP control capability are fully utilized, most of the computation amount can be given to the FPGA, and the DSP completes the system control. The treatment was as follows: signals r (n) received by a system receiving end enter an FPGA module to be subjected to Fast Fourier Transform (FFT) to obtain frequency domain signals(ii) a Then, after channel estimation, equalization is carried out according to the channel characteristics H (k) to obtain the signal without the influence of fading channel

The signal containing a frequency offset epsilon₀(ii) a Will be provided withStoring the signals into SRAM according to real part and imaginary part and making software decision on the signals

The resulting signal is substantially unaffected by the frequency offset. Then to

Inverse fast Fourier transform is carried out to obtain a reconstructed time domain signal * (n, epsilon) containing real frequency offset₀) (ii) a Since the channel estimation has the effect of gain compensation, the following principle can be adopted for decision: calculating the distance (Euclidean distance, but not necessarily square) between the actually received signal point and the ideal signal of the constellation point, judging according to the principle of minimum distance, and judging the signal after judgment

Also performs an inverse fast Fourier transform, which is

The IFFT is parallel to obtain a time domain signal which is not influenced by frequency offset(ii) a The results of the two inverse fast fourier transforms are stored in SRAM. From the reconstructed two-way signal * (n, ε)₀) And

and respectively calculating the numerator and the denominator of the frequency offset formula, and simultaneously sending an interrupt request signal to the DSP chip. The DSP calls an interrupt service program after receiving the interrupt request, reads the final calculation result of the FPGA, and calculates the frequency deviation epsilon, namely the frequency deviation epsilon₀Is estimated. Epsilon is used as the frequency offset value estimated by the frame, and the frequency offset value is converted into a voltage signal and then input into the input end of a Voltage Controlled Oscillator (VCO), so that the frequency offset or Doppler frequency shift of the oscillator at the receiving end is corrected, and the frequency offset can be controlled to be close to a zero value. The dotted frame part of the attached figure 1 is the key part of the invention.

In the embodiment, the FPGA completes the main part of a design target, including two times of inverse Fourier transform and preparation calculation for frequency offset estimation; the DSP chip plays the roles of CPU commanding, interface and unified pace and completes the final calculation of the frequency offset.

The IFFT same-address operation characteristic can be used for saving a storage unit. The frequency offset calculation involves multiplication to sum and fractional division. The way of respectively calculating the numerator and denominator of the frequency offset formula fully utilizes the operational capability of the FPGA chip, reduces the read-write quantity between the DSP and the FPGA, and avoids the characteristic that the FPGA chip is difficult to realize division. When the DSP assembly language is programmed, the fractional division can be realized by using a shift instruction and a subtraction instruction. The assembly language implementation of fixed point division can be referred to [10 ]. The C language can be directly programmed since it defines the divide operator.

When the FPGA is programmed by using the VHDL language, the two IFFT before frequency offset calculation can be set as a parallel calculation mode, so that the real-time performance of the system is enhanced. Since N is a power exponent of 2, it can be implemented by a shifting method when calculating the denominator of the frequency offset formula. After the computation is completed, the FPGA sends an interrupt request to the TMS320C54X, triggering an interrupt response program of the digital signal processor. The result of the calculation is read in by the TMS320C54X to complete the frequency offset estimation, which is implemented by software programming. The calculated frequency deviation is subjected to D/A conversion and voltage matching and then acts on a voltage controlled oscillator VCO, so that carrier frequency synchronization is completed. Because the method needs less data storage amount, the DSP chip memory RAM can be fully utilized to provide enough storage space.

Reference to the literature

[1]Pollet T，Bladel M and Moeneclaey M.BER sensitivity of OFDM systems to carrier frequencyoffset and Wiener phase noise.IEEE Trans.Commun.，43(2/3/4)：191-193，Feb/Mar/Apr 1995

[2]L Wei and C Schlegel.Synchronization requirements for multi-user OFDM on satellite andtwo-path Rayleigh fading channels.IEEE Trans.Commun.，43(2/3/4)：887-895，1995

[3]P H Moose.A technique for orthogonal frequency division multiplexing frequency offsetcorrection.IEEE Trans.Commun..，42(10)：2908-2914，1994

[4]J J van de Beek，M Sandell and P O Börjesson.ML estimation of time and frequency offset inOFDM systems.IEEE Trans.on Signal Processing.Vol.45(7)：1800-1805，1997

[5]Yun Hee Kim，Iickho Song and Seokho Yoon An Efficient Frequency Offset Estimator forOFDM Systems and Its performance Characteristics，IEEE Trans.On vehicular Technology，vol.50，NO.5，sep，2001

[6]H Liu and U Tureli.A High efficiency carrier estimator for OFDM communications.IEEECommunication Letters，2(4)：104-106 1998

[7]U Tureli，H Liu and M D Zoltowski.OFDM blind carrier offset estimation：ESPRIT.IEEETrans.Commun.，48(9)：1459-1461，2000

[8]X L Ma，C Tepedelenlioglu and G B Giannakis et al.Non-data-aided carrier offset estimatorfor OFDM with null subcarriers：identifiability，algorithms，and performance.IEEE J.on Select.Areas in Commn.，19(12)：2504-2515，2001

[9] Mahui, Du rock, He Bo, and Xiao Geng, analysis of ML algorithm for OFDM system synchronization, reported in "Communications

[10] Zhang Xiongwei, principle and development and application of DSP chip, electronic industry Press, 9 months, 1 st edition 1997

[11] Wangzui Xue et al, design of DSP base and application System, Beijing university of aerospace Press, 8 months, 1 st edition 2001

Claims (4)

1. A carrier frequency tracking method of an orthogonal frequency division multiplexing system in a multipath fading channel is characterized in that according to a signal frequency offset estimation formula provided by the invention, a system converts the obtained frequency offset into a voltage signal through an operation module, inputs the voltage signal into an input end of a Voltage Controlled Oscillator (VCO), corrects the oscillation frequency offset or Doppler frequency offset of a receiving end, and controls the frequency offset to be close to a zero value, and the specific operation control process is as follows:

the signal received by the system receiving end is firstly sent to an FFT module (1) for Fast Fourier Transform (FFT) to obtain a frequency domainSignalChannel estimation is carried out by a channel estimation module (2), channel characteristics H (k) are obtained by a channel estimation algorithm, and signals for removing the influence of fading channels are obtained after equalizationThe signal contains a frequency offset epsilon₀(ii) a The signal is divided into two paths: one path of input judgment module (3) obtains a judged signal according to a judgment rule

The signal is substantially unaffected by the frequency offset; the other path is transferred to an IFFT module A (4) to carry out inverse fast Fourier transform to obtain a reconstructed time domain signal containing real frequency offset

The signal is not affected by fading channels; post-decision signal

When the signal is output, the signal is transferred to an IFFT module B (5) to perform inverse fast Fourier transform, and a time domain signal which is basically not influenced by frequency deviation is obtained

The reconstructed two paths of signals are simultaneously input into a frequency deviation calculation module (6), and the frequency deviation epsilon is calculated by a frequency deviation formula, namely the frequency deviation epsilon is a pair epsilon₀And e is used as the frequency offset value estimated by the frame, and the frequency offset value is converted into a voltage signal and then input into the input end of a Voltage Controlled Oscillator (VCO), so that the frequency offset or Doppler frequency shift of the oscillator at the receiving end is corrected, and the frequency offset is controlled to be close to a zero value.

2. The method for tracking carrier frequency of orthogonal frequency division multiplexing system in multipath fading channel as claimed in claim 1, wherein said frequency offset control method can be implemented by one of the following methods:

(1) combining an FPGA with a DSP chip;

(2) a Digital Signal Processor (DSP) is adopted;

(3) an ASIC circuit design is adopted.

3. The method for tracking carrier frequency of OFDM system in multipath fading channel as claimed in claim 2, wherein when the method (2) is adopted, the system software adopts the way that the main program calls the interrupt subprogram, and the main program completes the functions of system initialization, system self-check, interrupt initialization, etc.; the whole frequency offset calculation and tracking process comprises software judgment, IFFT and frequency offset calculation which are completed in an interrupt service subprogram; is judged to obtain

After that, the air conditioner is started to work,

and

the IFFT of (a) can be implemented serially, and the interrupt service subroutine implemented in a serial computing manner has the following flow:

1) starting;

2) protecting the site;

3) reading channel equalized data

Respectively storing the real part and the imaginary part of the real part and the imaginary part;

4)

determined by software

5) Through IFFT to obtain

Respectively storing the real part and the imaginary part of the real part and the imaginary part;

6)

through IFFT to obtainRespectively store the real part andan imaginary part;

7) calculating frequency deviation;

8) starting D/A conversion;

9) restoring the site;

10) returning by interruption;

when the method (1) is adopted to realize,andthe calculation of the IFFT and the numerator and denominator of the frequency deviation calculation formula can be realized by hardware in series, and the DSP chip realizes system control and synchronization; when the method (3) is used for implementation, all operations can be implemented by hardware.

4. The carrier frequency tracking method for an orthogonal frequency division multiplexing system in a multipath fading channel as claimed in claim 2, wherein when implemented using the method (2),andthe IFFT of (a) can be implemented in parallel, and the flow of the interrupt service subroutine implemented in a parallel computing manner is as follows:

10) starting;

11) protecting the site;

12) reading channel equalized data

Respectively storing the real part and the imaginary part of the real part and the imaginary part;

13)

determined by software

14) Through IFFT to obtainRespectively storing the real part and the imaginary part of the real part and the imaginary part;through IFFT to obtainRespectively storing the real part and the imaginary part of the real part and the imaginary part;

15) calculating frequency deviation;

16) starting D/A conversion;

17) restoring the site;

18) returning by interruption;

when the method (1) is adopted for realizing,

and

the calculation of the IFFT and the numerator and denominator of the frequency deviation calculation formula can be realized by hardware in parallel, and the DSP chip realizes system control and synchronization; when the method (3) is used for implementation, all operations can be implemented by hardware.

Images