CN101056302B - UKF-based channel and carrier frequency deviation estimating method in the OFDM system - Google Patents

Abstract

An UKF based channel and carrier wave frequency shift estimation method in OFDM system is applied in the technique field of wireless transmission. Said channel estimation method starts from a frequency domain, to get cursory estimated values at pilot channels firstly by using a LS method, and then to get accurate estimated values by utilizing path capturing method and UKF method in time domain. According to the UKF based carrier wave frequency shift estimation method in OFDM system in accordance with the present invention, wave frequency shift is estimated in time domain through UKF filtering method, and then ICI is eliminated through carrier wave frequency shift correction. Aiming at OFDM system in which disturbances between sub-carriers exist and which passes through time-varying wireless channel, the present invention further provides an UKF based channel and carrier wave frequency combination estimation method which is capable of eliminating ICI and improving veracity of channel estimation and exhibits excellent astringency and robustness by utilizing specific pilot frequency structure and convolutional change forms of a matrix.

Description

In the ofdm system based on channel and the carrier frequency deviation estimating method of UKF

Technical field

The present invention relates to a kind of channel of wireless communication technology field and the method for frequency offset estimating, specifically is based on the channel and the carrier frequency deviation estimating method of UKF (Unscented kalman filtering device) in a kind of ofdm system.

Background technology

Signal is subjected to the influence of factors such as fading channel, multidiameter delay expansion and Doppler frequency expansion in the radio communication in communication process, usually adopt coherent demodulation at receiving terminal in order to recover the transmission signal preferably, and coherent demodulation needs channel parameter information, it obtains by channel estimating, so the performance of channel estimator directly influences systematic function.OFDM (OFDM) system can effectively resist the intersymbol interference that the multipath expansion causes, and makes that carrying out transfer of data under abominable wireless fading channel becomes possibility.In ofdm system, channel estimation methods based on two-dimentional minimum mean square error criterion (MMSE) is the channel estimation methods of best performance in the theory analysis, but these class methods not only complexity are high, also need to know part or all of channel prior information, so the engineering using value are not high.In addition, based on the channel estimation methods of least square criterion (LS), though computational complexity is not high, decreased performance is very fast when signal to noise ratio descends, the occasion of low signal-to-noise ratio in not being suitable for.Therefore, seek the quality that the channel estimation methods that is suitable on the engineering is directly connected to receiver detection and decoding performance, play crucial effects improving the wide-band mobile communication system performance.

OFDM is based upon on the basis of strict orthogonal between subcarrier, and the carrier signal waveform distortion all can influence the orthogonality between the subcarrier in any transmission course.The increase in data symbol cycle can make ofdm system more responsive more to carrier frequency shift and Doppler's expansion than common single-carrier system, increases thereby make to disturb greatly between the subcarrier, introduces and disturbs (ICI) between subcarrier, reduces the performance of system greatly.Two class methods are used for eliminating the ICI that causes because of carrier frequency shift at present: a class is to proofread and correct after frequency shift (FS) is estimated; Another kind of is the sensitiveness of reduction system to frequency shift (FS) of taking measures.In (SC) method of elimination, in adjacent sub-carrier, add redundant data and come the sensitiveness of reduction system frequency shift (FS), reach the purpose of eliminating ICI; Maximal possibility estimation (MLE) method is estimated to proofread and correct then to frequency shift (FS).When the value of frequency shift (FS) was smaller, the SC method had better performance, and this is that the ICI in the SC method between adjacent sub-carrier can effectively offset because the difference between the ICI that produces between adjacent carrier this moment is very little.When frequency offseting value is bigger, can not eliminate ICI fully between the adjacent sub-carrier in the SC method, and this moment, the MLE method there was more accurate estimation to frequency shift (FS).SC and MLE method all need one times redundancy bandwidth, and bandwidth efficiency is low.Therefore to seek a kind of bandwidth efficiency height, simultaneously frequency shift (FS) be had accurate estimation approach.Be subjected to the influence of multipath effect, time variation and the carrier frequency shift of channel in the practical wireless communication systems simultaneously, therefore also must seek a kind of method that can solve channel estimating and two problems of Carrier frequency offset estimation simultaneously.

Find through literature search prior art, people such as Dieter Schafhuber in the signal processing academic conference (IEEE Workshop on Signal ProcessingAdvances in Wireless Communication) of IEEE international wireless in 2003 communication, publish an article " Adaptive Identification of double selectivity fading channel and tracking in radio MIMO-ofdm system (Adaptive Identification and Trackingof Doubly Selective Fading Channels for Wireless MIMO-OFDM Systems) ", this article proposes a kind of channel estimation methods based on minimum mean square self-adaption filter (LMS) that is applicable to many antennas ofdm system, need not to know on the basis of channel statistical information, utilize frequency pilot sign and LMS filter, time varying channel is followed the tracks of and estimated, but this method have individual significant disadvantages be exactly estimated accuracy be not very high, especially under the environment of middle low signal-to-noise ratio, be far smaller than channel estimation methods based on two-dimentional minimum mean square error criterion.

Summary of the invention

The objective of the invention is to overcome deficiency of the prior art, channel and carrier frequency deviation estimating method based on UKF are provided in a kind of ofdm system, make its do not know to have under the situation of channel statistical information that estimated performance is sane, characteristics of interference between strong, the anti-subcarrier of ability of strong robustness, anti-additive white Gaussian noise, and be easy to realize.

The present invention is achieved by the following technical solutions, the present invention is that a kind of UKF that uses in ofdm system handles the method that pilot frequency information obtains channel and Carrier frequency offset estimation value, in the ofdm system of eliminating carrier wave frequency deviation, when only considering multipath channel, adopt channel estimation methods based on UKF to the influencing of system; In the length of Cyclic Prefix length greater than channel impulse response, and frequency-domain equalizer has carried out in the ofdm system of effective compensation to channel, can think that the influence of channel mainly comes self noise this moment, and ignore the influence of channel multipath effect, adopts the carrier frequency deviation estimating method based on UKF; Exist at the same time in the ofdm system of carrier shift and multipath effect, adopt channel and carrier frequency shift combined estimation method based on UKF.

Described channel estimation methods based on UKF comprises the steps:

Step 1: use pilot frequency information to obtain the initial value of frequency pilot sign place channel condition information on the frequency domain by the method for least square;

Step 2:, obtain the channel condition information under the noise contributions do not considered of time by described initial value is carried out fast adverse Fourier transform;

Step 3: do not consider that at described the utilization of channel condition information under noise contributions Adaptive Path catching method obtains the channel condition information of the active path of channel;

Step 4: as measured value, utilize the UKF filter tracks again and estimate channel condition information accurately under the consideration noise contributions of current time with the channel condition information of the active path of the described channel that captures;

Step 5: described channel condition information is accurately carried out fast fourier transform, obtain the channel transfer function of the time/frequency domain of frequency pilot sign;

Step 6: the channel transfer function to the time/frequency domain of described frequency pilot sign obtains the pairing channel transfer function of data symbol by interpolation method, gives signal detection module described transfer function and detects or decipher.

Described channel condition information comprises the number and the position of active path, and the amplitude and the phase place of the pairing channel of described number and position.The Adaptive Path catching method is number and the position of recently determining active path by the noise of path of adopting the channel that obtains according to ordering and current time in real time.Under same signal to noise ratio, adopt the number and the position of identical active path.

Described carrier frequency deviation estimating method based on UKF comprises the steps:

Step 1: pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain, and frequency pilot sign is long to be sub-carrier number;

Step 2: the time domain received signal is configured to send the form that signal and attenuation coefficient multiply each other;

Step 3: in time domain, utilization UKF filter carries out Carrier frequency offset estimation to the time domain received signal;

Step 4: utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated.

Described channel and carrier frequency shift combined estimation method based on UKF comprise the steps:

Step 1; Pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain every the OFDM of prearranged number symbol, and frequency pilot sign is long to be sub-carrier number;

Step 2: the convolution relation form that the time-domain received signal is configured to be subjected to the transmission signal and the channel of carrier frequency bias effect;

Step 3: with the described convolution relation formal transformation of time-domain received signal is the matrix relationship form;

Step 4: in time-domain, utilization UKF filter carries out channel estimating and Carrier frequency offset estimation to described matrix relationship form;

Step 5: utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated.

Step 6: the result to described channel estimating obtains the pairing channel transfer function of data symbol by interpolation method, gives signal detection module described transfer function and detects or decipher.

Below the present invention is described in further detail:

1, based on the channel estimation methods of UKF, be implemented as follows:

(1) use pilot frequency information to obtain the initial value of frequency pilot sign place channel condition information on the frequency domain by the method for least square

Suppose to adopt N _TTransmit antennas, N _RThe root reception antenna, the MIMO-OFDM system of n OFDM symbol and K subcarrier.

Be transmission signal vector, a ⁽ⁱ⁾[n, k] expression antenna i at n constantly, the transmission signal on k subcarrier.To a[n, k] make IFFT and add the long L that is _CPCyclic Prefix (CP) obtain n OFDM symbol

Each OFDM symbol duration is N=K+L _CpAll baseband transmit signals is

$s_n\left[m\right]=\left\{\begin{array}{cc}\frac{1}{\sqrt{{\mathrm{KN}}_T}}\underset{k=0}{\overset{K-1}{\mathrm{\&Sigma;}}}a[n,k]{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& m=-L_{\mathrm{CP}},...,K-1\\ 0& \mathrm{else}\end{array}\right.---\left(1\right)$

The impulse response of channel can be expressed as

$h(t,\mathrm{\&tau;})=\underset{k}{\mathrm{\&Sigma;}}{\mathrm{\&gamma;}}_k\left(t\right)c(\mathrm{\&tau;}-{\mathrm{\&tau;}}_k)---\left(2\right)$

In t frequency response constantly be

$H(t,f)={\&Integral;}_{-\&infin;}^{+\&infin;}h(t,\mathrm{\&tau;}){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j\mathrm{d\&tau;}=C\left(f\right)\underset{k}{\mathrm{\&Sigma;}}{\mathrm{\&gamma;}}_k{e}^{\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;f}{\mathrm{\&tau;}}_k}---\left(3\right)$

$C\left(f\right)={\&Integral;}_{-\&infin;}^{+\&infin;}c\left(\mathrm{\&tau;}\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;f\&tau;}}\mathrm{d\&tau;}---\left(4\right)$

The signal that reception antenna receives is by fading MIMO N _R* N _TChannel matrix H [m, l], transmission signals S[m] and noise η [m] form.η [m] for be distributed as N (0, σ _η ²) white Gaussian noise

$r\left[m\right]=\underset{l=0}{\overset{L-1}{\mathrm{\&Sigma;}}}H[m,l]S[m-l]+{\mathrm{\&sigma;}}_{\mathrm{\&eta;}}\left[m\right]---\left(5\right)$

Received signal r[m] go Cyclic Prefix to be FFT to obtain

$X[n,k]=\frac{1}{\sqrt{K}}\underset{m=0}{\overset{K-1}{\mathrm{\&Sigma;}}}r[\mathrm{nN}+m]e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{km}/K}---\left(6\right)$

If Nf _Doppler＜＜1, H[m, l] can ignore in the variation of an OFDM symbol, the I/O relation of the MIMO-OFDM system among the present invention can be expressed as

$X[n,k]=\hat{H}[n,k]a[n,k]+{\widehat{\mathrm{\&sigma;}}}_{\mathrm{\&eta;}}[n,k]---\left(7\right)$

Pilot frequency information of the present invention is that structure adopts block frequency pilot sign, each frequency pilot sign comprises the time-frequency piece on all subcarriers, the frame structure of pilot tone adopts the structure that adds several frequency pilot signs in a frame data front, and different transmit antennas is a mutually orthogonal in the pilot tone of synchronization in multiaerial system.For processing signals easily, the pilot signal on the different sub carrier of the same OFDM symbol on the antenna is identical, the pilot tone mutually orthogonal that different transmitting antennas send.The pilot matrix that P forms for the pilot frequency sequence on all antennas, Here P _nBe a string pilot frequency sequence on the antenna, P _nBe to be N in a string cycle _TCyclic sequence,

N is being arranged _TIn the MIMO-OFDM system of transmit antennas, the relation table that sends signal and received signal is shown

$X[n,k]=\sqrt{\frac{W_T}{N_T}}T[n,k]H[n,k]+{\mathrm{\&sigma;}}_{\mathrm{\&eta;}}[n,k]---\left(8\right)$

T[n, k] be the a[n that considers to send energy, k], W _TBe the T[n that transmits, k] energy.Adopt frequency domain LS channel estimating, do not consider that promptly The noise obtains

$\hat{H}[n,k]=\sqrt{\frac{N_T}{W_T}}{T}^{-1}[n,k]X[n,k]---\left(9\right)$

The channel estimating formula at pilot tone place

$\hat{H}[n,k]=\sqrt{\frac{N_T}{W_T}}{P}^{H}X[n,k]---\left(10\right)$

The Channel Transmission The Representation Equation is

$\hat{H}[n,k]=H[n,k]+\mathrm{\&psi;}[n,k]---\left(11\right)$

Therefore by right

Do the time domain state information that IFFT obtains channel

(11) formula is converted into

${\tilde{h}}_l\left[n\right]=h_l\left[n\right]+z_l\left[n\right]---\left(12\right)$

Z wherein _l[n] for satisfy N (0, σ _z ²) zero-mean that distributes, Gauss's vector.

(2), obtain the channel condition information under the noise contributions do not considered of time by described initial value is carried out fast adverse Fourier transform

Obtain the channel condition information under the noise contributions do not considered of time by fast adverse Fourier transform (IFFT).Be that receiver is done fast adverse Fourier transform to the channel estimating initial value that the LS method estimates, obtain the channel estimation value of time.

(3) do not consider that at described the utilization of channel condition information under noise contributions Adaptive Path catching method obtains the channel condition information of the active path of channel

Adopted the adaptive threshold catching method among the present invention, this method is recently determined the number L and the position of important tap in real time according to the noise of channel tap that sorts and current time, make the value of L have certain adaptivity, this method is better than traditional definite path prize law and traditional thresholding prize law.

The adaptive threshold prize law:

L is the number of important tap

L _mBe the multipath number of receptible maximum complexity, generally get the Cyclic Prefix number

Be fast adverse Fourier transform (IFFT), obtain the channel estimation value of time

For

Descending

L _iFor satisfying the multipath tap number of following formula

${\left|\hat{h}\right[l\left]\right|}^2>\mathrm{\&eta;}\&CenterDot;\frac{1}{\mathrm{\&alpha;}}\underset{l=0}{\overset{\mathrm{\&alpha;}-1}{\mathrm{\&Sigma;}}}{\left|{\hat{h}}_{\mathrm{sort}}\right[l\left]\right|}^2---\left(13\right)$

L is L _iAnd L _mMinimum value

η is a threshold value, and the value of η is decided by SNR,

(4) with the channel condition information of the active path of the described channel that captures as measured value, utilize the UKF filter tracks again and estimate channel condition information accurately under the consideration noise contributions of current time

Do not consider channel condition information under the noise contributions as measured value with current time, utilize UKF to follow the tracks of and estimate channel condition information accurately under the consideration noise contributions of current time, in iterative process, state-noise and measure noise and bring in constant renewal in.

State equation:

x _n+1＝F(x _n)+v _n (14)

Measure equation:

y _n＝H(x _n)+n _n (15)

Wherein $x_n=\left[\begin{array}{c}{h}_0\left[n\right]\\ .\\ .\\ .\\ h_{L-1}\left[n\right]\\ \mathrm{vec}\left\{T\right\}\end{array}\right],$ $G\left(x_n\right)=\left[\begin{array}{c}T{h}_0\left[n\right]\\ .\\ .\\ .\\ T{h}_{L-1}\left[n\right]\\ \mathrm{vec}\left\{T\right\}\end{array}\right]$ $H\left(x_n\right)=\left[\begin{array}{c}{h}_0\left[n\right]\\ .\\ .\\ .\\ h_{L-1}\left[n\right]\\ h_{L-1}\left[n\right]\end{array}\right]$

v _n, n _nBe respectively the state-noise vector sum and measure noise vector.

LM * 1 dimension is measured vector

M=N _T* N _R, N _TBe number of transmit antennas, N _RBe the reception antenna number.

(5) described channel condition information is accurately carried out fast fourier transform, obtain the channel transfer function of the time/frequency domain of frequency pilot sign

Above step is tried to achieve the exact value of the channel estimating of time, what but input needed is the state information of the time/frequency domain of channel, therefore, to the channel estimation value of time do the fast Fourier transform (FFT) conversion obtain time/the frequency domain channel estimation value.

(6) channel transfer function to the time/frequency domain of described frequency pilot sign obtains the pairing channel transfer function of data symbol by interpolation method, gives signal detection module described transfer function and detects or decipher

2, based on the carrier frequency deviation estimating method of UKF, be implemented as follows:

(1) pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain, and frequency pilot sign is long to be sub-carrier number

Introduced the situation of frequency shift (FS) in the signal of considering to receive.Frequency shift (FS) can be regarded as of introducing in the channel and takes advantage of sex factor, supposes the MIMO-OFDM systems that adopt 22 to receive,

$r_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">n</figure-callout>}}^1={\mathrm{<figure-callout\; id="1"\; label="t\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">t</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+{\mathrm{<figure-callout\; id="2"\; label="t\; n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">t</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+w_n^1---\left(16\right)$

$r_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">n</figure-callout>}}^2={\mathrm{<figure-callout\; id="1"\; label="t\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">t</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+{\mathrm{<figure-callout\; id="2"\; label="t\; n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">t</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+w_n^2---\left(17\right)$

t _n ⁱBe the transmission signal of n moment i transmit antennas, r _n ^jBe the received signal of n moment j root reception antenna, ε _n ^IjBe the normalized frequency shift value of transmitting antenna i to the reception antenna j transmission path, w _n ^jBe the additive white Gaussian noise of introducing on the transmission path, N is a sub-carrier number, L _CPIt is the Cyclic Prefix number.

Insert an OFDM symbol as equivalent pilot tone in time-domain.

$p_n^i=1,$ k＝1，2...N+L _CP-1，i＝1，2

Wherein, N is a sub-carrier number, L _CPIt is the Cyclic Prefix number.

(2) the frequency shift (FS) model that the time domain received signal is configured to send the form pilot tone place that signal and attenuation coefficient multiply each other is:

$r_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">n</figure-callout>}}^1={\mathrm{<figure-callout\; id="1"\; label="p\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">p</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+{\mathrm{<figure-callout\; id="2"\; label="p\; n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">p</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+w_n^1---\left(18\right)$

$r_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">n</figure-callout>}}^2={\mathrm{<figure-callout\; id="1"\; label="p\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">p</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+{\mathrm{<figure-callout\; id="2"\; label="p\; n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">p</figure-callout>}}_n^{}\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+w_n^2---\left(19\right)$

(3) in time domain, utilization UKF filter carries out Carrier frequency offset estimation UKF estimating carrier frequency offset to the time domain received signal:

State equation:

x _n+1＝F(x _n)+v _n

Measure equation:

y _n＝H(x _n)+n _n

Wherein

F (x _n)=x _n,

Be observation vector, v _n, n _nBe respectively state-noise and measure noise vector.

Measuring equation as can be seen is nonlinear equation, needs to adopt nonlinear filter to handle.Adopt UKF filter process (18), (19) formula.

(4) utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated

3, based on channel and the carrier frequency shift combined estimation method of UKF, be implemented as follows:

(1) pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain every the OFDM of prearranged number symbol, and frequency pilot sign is long to be sub-carrier number;

Suppose to adopt N _TTransmit antennas, N _RThe root reception antenna, the MIMO-OFDM system of n OFDM symbol and N subcarrier.a ⁱ(n, k) expression antenna i at n constantly, the transmission signal on k subcarrier.To a ⁱ(n k) makes IFFT and add the long L that is _CPCyclic Prefix (CP) obtain the OFDM symbol s _n ⁱ(m), each OFDM symbol duration is N=K+L _Cp

Channel model is:

${s_{\mathrm{\&epsiv;n}}}^{\mathrm{ij}}\left(m\right)={s_n}^i\left(m\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">e</figure-callout>}}^{\frac{j}{}}---\left(20\right)$

Wherein

For being the normalized frequency shift value of transmitting antenna i when to reception antenna j transmission path, transmitting n OFDM symbol, f _d ^IjBe frequency shift (FS), f _s=1/T _sBe data symbol s _n ⁱ(m) transmission rate, Δ f=f _s/ N is a subcarrier spacing.

The frame structure of pilot tone adopts the structure that adds a frequency pilot sign in a frame data front.

(2) the time-domain received signal is configured to be subjected to the convolution relation form of the transmission signal and the channel of carrier frequency bias effect

Taking into account system the channel of process be slow fading channel, in the time of an OFDM symbol of transmission, channel information is constant.

If Nf _Doppler＜＜1, channel can be ignored in the variation of an OFDM symbol, and the input/output relation of MIMO-OFDM of the present invention system can be expressed as

$r_n^j=\underset{i=1}{\overset{N_T}{\mathrm{\&Sigma;}}}{h}_n^{\mathrm{ij}}*s_{\mathrm{\&epsiv;n}}^{\mathrm{ij}}+{\mathrm{\&eta;}}_n^j---\left(21\right)$

R wherein _n ^j=[r _n ^j(1) ..., r _n ^j(N)] ^TBe the received signal vector on the reception antenna j, h _n ^Ij=[h _n ^Ij(1) ..., h _n ^Ij(L)] ^TBe the channel impulse response vector of transmitting antenna i to reception antenna j transmission path, s _{ε n} ^Ij=[s _{ε n} ^Ij(1) ..., s _{ε n} ^Ij(N)] ^TFor having the transmission signal vector of carrier wave frequency deviation, η _n ^j=[η _n ^j(1) ..., η _n ^j(N)] ^TBe the additive white Gaussian noise column vector, L is the maximum length of channel impulse response.

(3) the described convolution relation formal transformation with the time-domain received signal is the matrix relationship form

The form that the convolution form can be expressed as matrix multiple has

$r_n^j=\underset{i=1}{\overset{N_T}{\mathrm{\&Sigma;}}}{S}_{\mathrm{\&epsiv;n}}^{\mathrm{ij}}{h}_n^{\mathrm{ij}}+{\mathrm{\&eta;}}_n^j---\left(22\right)$

Through wireless channel, go the received signal of Cyclic Prefix to be expressed as:

$r_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">n</figure-callout>}}^1=[C_{\mathrm{\&epsiv;}}^{11}\&CenterDot;p_n^1]*{\mathrm{<figure-callout\; id="11"\; label="h\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">h</figure-callout>}}_n^{}+[C_{\mathrm{\&epsiv;}}^{21}\&CenterDot;p_n^2]*h_n^{21}+{\mathrm{\&eta;}}_n^1---\left(23\right)$

$r_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=[C_{\mathrm{\&epsiv;}}^{12}\&CenterDot;p_n^1]*{\mathrm{<figure-callout\; id="12"\; label="h\; n"\; filenames="H2007100414596E00021.png,H2007100414596E00032.png"\; state="\{\{state\}\}">h</figure-callout>}}_n^{}+[C_{\mathrm{\&epsiv;}}^{22}\&CenterDot;p_n^2]*h_n^{22}+{\mathrm{\&eta;}}_n^2---\left(24\right)$

P wherein _n ⁱBe the pilot tone column vector

$r_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">n</figure-callout>}}^1=\left[P_{\mathrm{\&epsiv;n}}^{11}\right]*{\mathrm{<figure-callout\; id="11"\; label="h\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">h</figure-callout>}}_n^{}+\left[p_{\mathrm{\&epsiv;n}}^{21}\right]*h_n^{21}+{\mathrm{\&eta;}}_n^1---\left(25\right)$

$r_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=\left[P_{\mathrm{\&epsiv;n}}^{12}\right]*{\mathrm{<figure-callout\; id="12"\; label="h\; n"\; filenames="H2007100414596E00021.png,H2007100414596E00032.png"\; state="\{\{state\}\}">h</figure-callout>}}_n^{}+\left[p_{\mathrm{\&epsiv;n}}^{22}\right]*h_n^{22}+{\mathrm{\&eta;}}_n^2---\left(26\right)$

Wherein $p_{\mathrm{\&epsiv;n}}^{\mathrm{ij}}=C_{\mathrm{\&epsiv;}}^{\mathrm{ij}}{p}_n^i$

The convolution form is expressed as matrix form, has

$r_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">n</figure-callout>}}^1=P_{\mathrm{\&epsiv;n}}^{11}{\mathrm{<figure-callout\; id="11"\; label="h\; n"\; filenames="H2007100414596E00013.png"\; state="\{\{state\}\}">h</figure-callout>}}_n^{}+P_{\mathrm{\&epsiv;n}}^{21}{h}_n^{21}+{\mathrm{\&eta;}}_n^1---\left(27\right)$

$r_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=P_{\mathrm{\&epsiv;n}}^{12}{\mathrm{<figure-callout\; id="12"\; label="h\; n"\; filenames="H2007100414596E00021.png,H2007100414596E00032.png"\; state="\{\{state\}\}">h</figure-callout>}}_n^{}+P_{\mathrm{\&epsiv;n}}^{22}{h}_n^{22}+{\mathrm{\&eta;}}_n^2---\left(28\right)$

(4) in time-domain, utilization UKF filter carries out channel estimating and Carrier frequency offset estimation to described matrix relationship form

At first come processing formula (27) with UKF

x _n+1＝F(x _n)+v _n (29)

y _n＝H(x _n)+n _n (30)

Wherein F (x _n)=x _n,

Be observation vector, v _n, n _nBe respectively state-noise and measure noise vector.

Utilization UKF filter is to the time domain estimated value of channel

Obtain the frequency domain estimated value through the FFT conversion

Be used for Channel Detection.

$\hat{H}(n,k)=\frac{1}{N_T}\underset{l=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{\hat{h}}_0\left(l\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2007100414596E00032.png,H2007100414596E00042.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{kl}/N}---\left(31\right)$

(5) utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated

(6) result to described channel estimating obtains the pairing channel transfer function of data symbol by interpolation method, gives signal detection module described transfer function and detects or decipher.

The invention has the advantages that: adopted the Adaptive Path acquisition algorithm to greatly reduce computation complexity in the channel estimation methods based on UKF and improved estimated accuracy; With UKF channel estimation value is carried out Filtering Processing, improve estimated accuracy.

Method based on the Carrier frequency offset estimation of UKF is eliminated ICI, disturbs self elimination method to compare with traditional SC, is handling under the bigger situation of frequency shift (FS), and the performance of better elimination ICI is arranged, and the bandwidth availability ratio height.It is similar to the complexity of ML method to eliminate the method for ICI based on the method for the Carrier frequency offset estimation of UKF among the present invention, but the ML method need repeat to send code element, need one times redundancy, has the same bandwidth efficiency with the SC method, so based on the method for UKF bandwidth availability ratio height than ML method.

Channel and carrier frequency shift combined estimation method based on UKF can estimate frequency shift (FS) between channel information and subcarrier simultaneously.Because UKF is more suitable for handling nonlinear problem than EKF, so this combined estimation method based on UKF can be eliminated the influence of carrier wave frequency deviation to channel estimating better, has satisfactory performance.

Description of drawings

Fig. 1 is based on channel and the carrier frequency deviation estimating method theory diagram of UKF in the ofdm system

Fig. 2 is channel estimating pilot configuration figure in the MIMO-OFDM system

Fig. 3 is based on the channel estimation methods block diagram of UKF in the MIMO-OFDM system

Fig. 4 is the MSE performance comparison diagram of Adaptive Path catching method and definite number of path channel estimating in the channel estimating

Fig. 5 is based on the channel estimating BER performance comparison diagram of UKF, LS and LMS in the channel estimating

Fig. 6 is based on the Carrier frequency offset estimation schematic diagram of UKF in the MIMO-OFDM system

Fig. 7 is that the BER letter of MIMO-OFDM system when having frequency shift (FS) can be schemed

Fig. 8 is channel and a carrier frequency shift associating estimating pilot frequency structure chart in the MIMO-OFDM system

Fig. 9 is that channel and the carrier frequency shift based on UKF united estimation principle figure in the MIMO-OFDM system

Figure 10 is based on the channel of UKF and the BER performance map that carrier frequency shift is united estimation

Embodiment

Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed execution mode and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

Its theory diagram of the present invention in the ofdm system of eliminating carrier wave frequency deviation, when only considering multipath channel to the influencing of system, adopts the channel estimation methods based on UKF as shown in Figure 1; In the length of Cyclic Prefix length greater than channel impulse response, and frequency-domain equalizer has carried out in the ofdm system of effective compensation to channel, can think that the influence of channel mainly comes self noise this moment, and ignore the influence of channel multipath effect, adopts the carrier frequency deviation estimating method based on UKF; Exist at the same time in the ofdm system of carrier shift and multipath effect,

Employing is based on channel and the carrier frequency shift combined estimation method of UKF.

Provide a concrete OFDM parameter configuration below, set forth performing step of the present invention.Need to prove that the parameter in the following example does not influence generality of the present invention.

The present invention adopts 22 the MIMO-OFDM systems of receiving, i.e. N _T=2, N _R=2. adopt the modulation system of QPSK, space time coding scheme sampling quadrature space-time block code (O-STBC), code check is 1/2, decoding sampling maximum likelihood (ML) decoding scheme when empty, carrier frequency 4GHz, bandwidth 6M, the subcarrier number is 64, consider that channel estimation methods of the present invention relatively is suitable for the frequency selectivity slow fading channel, therefore channel model adopts the multipath Rayleigh fading channel that ITU proposes, and the channel multi-path number is 5, and time delay is [0 260 520 780 1040] ns, power is [1.78 0-7.47-10-12.62] dB, movement velocity 3km/h.

Channel estimation methods based on UKF is implemented as follows:

(1) as shown in Figure 2, insert two frequency pilot signs, send 600 OFDM symbols altogether, promptly insert 200 frequency pilot signs every six OFDM symbols.The design of pilot samples orthogonalization, the pilot tone mutually orthogonal that different transmitting antennas send, transmitting antenna 1 is respectively P to the pilot tone of transmitting antenna 2 ₁=[1,1], P ₂=[1 ,-1].

(2) shown in Fig. 3 block diagram, utilize least squares method (LS) to obtain the initial estimate of the channel at each pilot tone place at frequency domain

This is actually a kind of Utopian channel estimation methods of not considering noise effect.

(3) as shown in Figure 3, each receiver is done fast adverse Fourier transform (IFFT) to the channel estimating initial value that separately LS method estimates, and obtains the channel estimation value of time

(4) in the Adaptive Path acquisition algorithm, η is a threshold value, and the value of η is decided by SNR,

Observation can span be 0.1-0.4 when the signal to noise ratio snr value is 10dB.

(5) described channel condition information is accurately carried out fast fourier transform, obtain the channel transfer function of the time/frequency domain of frequency pilot sign;

(6) channel transfer function to the time/frequency domain of described frequency pilot sign obtains the pairing channel transfer function of data symbol by the Gauss interpolation method, gives signal detection module described transfer function and detects or decipher.

As shown in Figure 4, the MSE of system (mean square error) performance relatively in, the UKF channel estimation methods that adopts the Adaptive Path acquisition algorithm is that 16 UKF channel estimation methods and definite number of path are that the property that 8 UKF channel estimation methods is obtained 3dB and 6dB is respectively made an uproar than gain than definite number of path.

As shown in Figure 5, the MSE of system (mean square error) performance relatively in, make an uproar than gain than the property that LMS channel estimation methods and LS channel estimating obtain 5dB and 10dB respectively based on the channel estimation methods of UKF.

Carrier frequency estimating method based on UKF is implemented as follows:

(1) as shown in Figure 6, adopt 2 the 2 MIMO-OFDM systems of receiving in the invention.When there was frequency shift (FS) in ofdm system, frequency shift (FS) can be regarded as of introducing in the channel and takes advantage of sex factor.

Its pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain, and frequency pilot sign is long to be sub-carrier number, in specific embodiments, promptly insert one long be that 64 value is 1 frequency pilot sign;

2) influence in the skew of time domain intercarrier frequency can be expressed as

Form, wherein ε is normalized frequency shift (FS), having 4 carrier frequency shifts in 2 the 2 receipts systems in specific embodiments is ε ¹¹Transmitting antenna 1 is to the carrier frequency shift of reception antenna 1, ε ¹²Transmitting antenna 1 is to the carrier frequency shift of reception antenna 2, ε ²¹Transmitting antenna 2 is to the carrier frequency shift of reception antenna 1, ε ²²Transmitting antenna 2 is to the carrier frequency shift of reception antenna 2, and the time domain received signal is configured to send the form that signal and attenuation coefficient multiply each other;

(3) in time domain, utilization UKF filter carries out Carrier frequency offset estimation to the time domain received signal, and to measure equation as can be seen be nonlinear equation, needs to adopt nonlinear filter to handle.Adopt the UKF filter process to come estimating carrier frequency offset.Because carrier frequency shift is a real number, so each step renewal is obtained the estimated value x of frequency shift (FS) _nGet real part, estimate to finish to obtain frequency shift (FS) ε up to time-domain pilot signal ¹¹, ε ²¹, ε ¹², ε ²²Estimated value, carry out frequency offset correction to the received signal with the Frequency offset estimation value again;

(4) utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated.

Simulation result shows, Carrier frequency offset estimation value ε ¹¹, ε ¹², ε ²¹And ε ²²Promptly converge near its actual value.

As shown in Figure 7, the BER of system (bit error rate) performance relatively in, to be subjected to the influence of frequency shift (FS) very big for the error rate of ofdm system as can be seen, increase along with frequency shift (FS), the error rate of ofdm system will increase greatly, especially as | ε | in the time of＞0.3, the improvement of signal to noise ratio is helpless to the reduction of the error rate.

As shown in Figure 8, employing approaches the system of carrierfree frequency shift (FS) based on the systematic function of the carrier frequency deviation estimating method of UKF.

Channel and carrier frequency combined estimation method based on UKF are implemented as follows:

(1) Fig. 9 provides the system model of this method.Pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain every 7 OFDM symbols, and frequency pilot sign is long to be sub-carrier number;

(2) the time-domain received signal is configured to be subjected to the convolution relation form of the transmission signal and the channel of carrier frequency bias effect;

(3) the described convolution relation formal transformation with the time-domain received signal is the matrix relationship form;

(4) in time-domain, utilization UKF filter carries out channel estimating and Carrier frequency offset estimation to described matrix relationship form, because carrier frequency shift is a real number, so each step renewal is obtained the estimated value ε of frequency shift (FS) _n ^IjGet real part, estimate to finish to obtain frequency shift (FS) up to time-domain pilot signal.After filtering obtains the estimated value of frequency shift (FS) based on UKF, carry out frequency offset correction to the received signal with this estimated value; Obtain the time domain estimated value of channel Obtain the frequency domain estimated value through the FFT conversion Be used for Channel Detection;

(5) utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated.

(6) result to described channel estimating obtains the pairing channel transfer function of data symbol by interpolation method, gives signal detection module described transfer function and detects or decipher.

The method that the present invention proposes can be followed the tracks of the variation of carrier frequency shift well, and the Carrier frequency offset estimation value promptly converges near the actual value, estimates carrier frequency shift quickly and accurately.Based on the system BER performance of the channel of UKF and carrier frequency shift combined estimation method as shown in figure 10.

As seen from the figure, can estimate carrier frequency shift and channel information well, eliminate ICI and channel multipath effect, make the performance of system obtain very big improvement based on channel and the carrier frequency shift combined estimation method of UKF.By simulation performance relatively, be 10 at BER ^-3The time, the UKF method among the present invention can obtain the snr gain of 0.3dB than EKF method, can embody when handling nonlinear problem, and the performance of UKF is better than EKF.

Claims (6)

1. In the ofdm system based on channel and the carrier frequency deviation estimating method of UKF, it is characterized in that, in the ofdm system of eliminating carrier wave frequency deviation, when only considering multipath channel, adopt channel estimation methods based on UKF to the influencing of system; In the length of Cyclic Prefix length greater than channel impulse response, and frequency-domain equalizer has carried out in the ofdm system of effective compensation to channel, this moment, the influence of channel came self noise, ignored the influence of channel multipath effect, adopted the carrier frequency deviation estimating method based on UKF; Exist at the same time in the ofdm system of carrier shift and multipath effect, adopt channel and carrier frequency shift combined estimation method based on UKF;

   Described channel estimation methods based on UKF comprises the steps:

   1.1): use pilot frequency information to obtain the initial value of frequency pilot sign place channel condition information on the frequency domain by the method for least square;

   1.2): by described initial value is carried out fast adverse Fourier transform, obtain the channel condition information under the noise contributions do not considered of time-domain;

   1.3): do not consider that at described the utilization of channel condition information under noise contributions Adaptive Path catching method obtains the channel condition information of the active path of channel;

   1.4):, utilize the UKF filter tracks again and estimate channel condition information accurately under the consideration noise contributions of current time as measured value with the channel condition information of the active path of the described channel that captures;

   1.5): described channel condition information is accurately carried out fast fourier transform, obtain the channel function of the frequency domain of frequency pilot sign;

   1.6): the channel function to the frequency domain of described frequency pilot sign obtains the pairing channel function of data symbol by interpolation method, gives signal detection module the pairing channel function of described data symbol and detects or decipher;

   Described carrier frequency deviation estimating method based on UKF comprises the steps:

   2.1): pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain, and frequency pilot sign is long to be sub-carrier number;

   2.2): the time domain received signal is configured to send the form that signal and attenuation coefficient multiply each other;

   2.3): in time domain, utilization UKF filter carries out Carrier frequency offset estimation to the time domain received signal;

   2.4): utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated;

   Described channel and carrier frequency shift combined estimation method based on UKF comprise the steps:

   3.1): pilot configuration is to insert a frequency pilot sign as equivalent pilot tone in time-domain every the OFDM of prearranged number symbol, and frequency pilot sign is long to be sub-carrier number;

   3.2): the convolution relation form that the time-domain received signal is configured to be subjected to the transmission signal and the channel of carrier frequency bias effect;

   3.3): with the described convolution relation formal transformation of time-domain received signal is the matrix relationship form;

   3.4): in time-domain, utilization UKF filter carries out channel estimating and Carrier frequency offset estimation to described matrix relationship form;

   3.5): utilize the carrier frequency shift that estimates to carry out carrier frequency shift and proofread and correct the system after the subcarrier interference that is eliminated;

   3.6): the result to described channel estimating obtains the pairing channel function of described data symbol by interpolation method, gives signal detection module the pairing channel function of described data symbol and detects or decipher.
2. 2. in the ofdm system according to claim 1 based on channel and the carrier frequency deviation estimating method of UKF, it is characterized in that: 1.1) described in channel condition information comprise the number and the position of active path and the amplitude and the phase place of the pairing channel of described number and position.
3. 3. based on channel and the carrier frequency deviation estimating method of UKF, it is characterized in that: in the ofdm system according to claim 1 1.3) described in the Adaptive Path catching method be number and the position that the noise of path by adopting the channel that obtains according to ordering and current time is recently determined active path in real time.
4. 4. based on channel and the carrier frequency deviation estimating method of UKF, it is characterized in that: in the ofdm system according to claim 3 1.3) in, under same signal to noise ratio, adopt the number and the position of identical active path.
5. 5. in the ofdm system according to claim 1 based on channel and the carrier frequency deviation estimating method of UKF, it is characterized in that: 1.4) in, utilizing the UKF filtering method to follow the tracks of and estimating the current time process of channel condition information accurately, bring in constant renewal in state-noise in each iteration and measure noise.
6. 6. based on channel and the carrier frequency deviation estimating method of UKF, it is characterized in that: in the ofdm system according to claim 1 2.3) in, described Carrier frequency offset estimation adopts nonlinear filter to realize by the frequency shift (FS) model in conjunction with the pilot tone place.

Images