CN101667982A - Removing method of WiMAX fast fading ICI based on plane spreading kalman filtering wave - Google Patents

Abstract

The invention discloses a removing method of WiMAX fast fading ICI based on plane spreading kalman filtering wave. The method comprises the steps of: separating an imaginary part from a real part of asignal by using the circular short training sequence of a WiMAX signal; and iteratively modifying and evaluating the doppler frequency deviation of the WiMAX signal by the relationship of the imaginary part and the real part of the signal with the plane spreading kalman filtering wave to remove the interference of the fast fading ICI to a WiMAX system. The method solves the defects that the ICI interference of the high-speed movement WiMAX system can not be effectively removed in the prior art, and has the advantages of high evaluation precision, low complexity, no need to evaluate signal initial phase deviation and the like.

Description

WiMAX rapid fading ICI removing method based on plane spreading kalman filtering

Technical field

The present invention relates to the signal of communication processing method in the 4G radio communication, a kind of WiMAX rapid fading ICI removing method based on plane spreading kalman filtering particularly is provided.

Background technology

Now, the third generation and the development of the 4th algebraically word mobile communication are rapidly, WiMAX (World Interoperability for Microwave Access, World Interoperability for Microwave Access, WiMax) wireless access wide band technology based on IEEE 802.16e standard has obtained the approval of global mobile industry as a kind of effective broadband wireless access solution.Along with the continuous evolution of future mobile communication system, wireless communication system will provide more and more higher data rate and communication quality guarantee more reliably.In order to satisfy the requirement of people's and high-speed mobile growing to transmission rate, WiMAX provides the high-speed wideband access mobile communication business of " last kilometer " for mobile communications network, and the translational speed of its supported portable terminal reaches as high as 120km/h.But the WiMAX system is very responsive to carrier frequency shift, and Doppler effect can cause frequency dispersal, brings inter-carrier interference ICI (Inter Carrier Interference).Under the vehicle environment of high-speed mobile, when portable terminal moves with the above speed of 120km/h, high speed between portable terminal and base station relatively moves and can make the WiMAX network cause the performance severe exacerbation of mobile communication system owing to the influence of channel rapid fadings such as multipath Doppler effect, the inter-carrier interference that is caused by the Doppler frequency expansion is more serious in the WiMAX system, causes communicating by letter being close to interrupting.In most of practical applications, balancing technique can't effectively be resisted the influence that ICI produces.Accelerate along with the receiver translational speed, how to eliminate owing to become to disturb between the subcarrier that is brought during channel and just become very meaningful.For reducing the influence of disturbing (ICI) between WiMAX system sub-carriers, a lot of methods have been proposed both at home and abroad, existing ICI removing method mainly is divided into two big classes, and a class is to handle sending symbol at transmitting terminal, reduces the susceptibility of WiMAX system to frequency shift (FS); Another kind of then is frequency deviation is estimated and to be proofreaied and correct at receiving terminal.In the reduction system typical method aspect the susceptibility of frequency deviation there are time-domain windowed, oneself elimination of subcarrier interference etc.Time-domain windowed can cause the reduction of system signal noise ratio, and subcarrier interference can reduce the availability of frequency spectrum from eliminating.Improve performance in employing reception antenna diversity, people have proposed to adopt linear model to describe the time-varying characteristics of channel, simultaneously the part coefficient in the channel matrix is ignored the amount of calculation of estimating to reduce, the loss of orthogonality that this method alters in the time of can compensating by the multipath of channel when relative Doppler frequency shift is less effectively, but be not suitable for the occasion of high-speed mobile.In addition, blind processing method does not need to insert supplementary, can save frequency resource, but is subjected to certain restriction at aspects such as complexity and accuracy.Can support high-speed mobile for solving WiMAX, and obtain transmission quality preferably, ICI removing method technology will come into one's own efficiently.

Summary of the invention

The objective of the invention is at the deficiencies in the prior art, a kind of estimated accuracy height, the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering that complexity is low are provided, the rapid fading ICI that is used to eliminate the WiMAX system disturbs.For achieving the above object, the present invention adopts following technical scheme: a kind of WiMAX rapid fading ICI removing method based on plane spreading kalman filtering, it is characterized in that utilizing the circulation short training sequence of WiMAX signal, to signal carry out actual situation part from, and the relation of imaginary part by signal and real part utilization plane spreading kalman filtering carries out the iteration correction to the Doppler frequency deviation of WiMAX signal and estimates, and then eliminate the interference of rapid fading ICI to the WiMAX system, its operating procedure is as follows:

(1) information source information is transformed into valid data to be transmitted: with information source information carry out source encoding, scrambling, chnnel coding, interweave, go here and there and S/P conversion and modulation mapping after become valid data X to be transmitted on N the subcarrier _i, wherein, channel coding method adopts LDPC or Turbo or Woven convolution code or RS-CC coding, and the modulation mapping is conciliate mapping and is adopted QPSK or 16QAM mapping;

(2) pilot tone is inserted and data multiplex: produce and channel frequency domain sampling position corresponding subcarrier pilot data X _Pilot, and be that pilot data inserts pilot tone with pilot data according to the Comb Pilot form among the IEEE802.16e at p place at interval with L, carry out data multiplex with the data to be transmitted structure, the length of generation pilot data equals 1/10th of WiMAX system-frame length;

(3) add Cyclic Prefix: after the IFFT that the data after multiplexing orders through N handles, composition WiMAX Frame x _k, and have relation $X_i=\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{x}_k{e}^{-\frac{j2\mathrm{\πki}}{N}}$ (0≤i≤N-1), add cyclic prefix CP before the symbol head of each code element therein then, the length that each code element adds Cyclic Prefix is N _CpIndividual sample point promptly has: x (i)=x _N-G+i, 0≤i≤G-1, G equals N here for protection length at interval in the formula _Cp, N _CpValue decide by the time delay largest extension value of transmission wireless channel, be four times of time delay largest extension value.；

(4) add the circulation short training sequence: add the cocycle short training sequence foremost at WiMAX Frame X ', this training sequence signal is known at receiving terminal, its length is the length of a symbol, this training sequence signal adopts the linear FM signal design, it both had been used for carrying out Domain Synchronous and also had been used for the ICI elimination, reduced the symbol expense.

(5) rise sampling filter and pass through channel: WiMAX Frame X ' process is risen sampling FIR filtering, and obtain signal waveform information s (t) after string P/S conversion and the digital-to-analogue D/A conversion, t is a time variable, then s (t) is launched at the WiMAX transmitting terminal, four times of over-samplings are adopted in the liter sampling that rises in the sampling FIR filtering operation, the FIR filter that rises in the sampling FIR filtering operation is the root-raised cosine filter on 64 rank, rolloff-factor is: 27/128, the present invention adopts the FIR on 4 16 rank to come Parallel Implementation, behind experience wireless fading channel h (t) and the additive noise n (t), the signal that receiving terminal receives is r (t), promptly have: r (t)=s (t) * h (t)+n (t), wherein * represents the time domain convolution, and channel output is specialized to be had: $y_i=\underset{k=0}{\overset{G}{\mathrm{\Σ}}}{h}_k^{\left(i\right)}{x}_{{\left((i-k)\right)}_N}+w_i$ 0≤i≤N-1, h _k ⁽ⁱ⁾Be that k tap is in the code element channel time domain response of i constantly;

(6) remove Cyclic Prefix with synchronously: extract carrying out Domain Synchronous after received signal r (t) process modulus A/D conversion, string and S/P conversion and the down-sampled FIR filtering, the Domain Synchronous extraction algorithm is:

If the linear frequency modulation training sequence signal that receives is C (n), correlation formula is arranged promptly:

When having time delay d, receive signal and be: z (k)=C (k-d)+w (k)

Obtaining cross-correlation is:

$\mathrm{corr}\left(k\right)=y\left(k\right)*x(N-k)=\underset{n=0}{\overset{N}{\mathrm{\Σ}}}y(k-n)x(N-n)$

$=\underset{n=0}{\overset{N}{\mathrm{\Σ}}}x(k-n+d)x(N-n)+\underset{n=0}{\overset{N}{\mathrm{\Σ}}}w(k-n)x(N-n)$

Because known R _XX(k), so: $\mathrm{corr}\left(k\right)=R_{\mathrm{XX}}\left(k\right)+\underset{n=0}{\overset{N}{\mathrm{\Σ}}}w(k-n)x(N-n)\≈\mathrm{N\δ}(k-N+d)+\mathrm{\ξ}\left(k\right)$

Because ξ (k) is very little value,, promptly just extract Domain Synchronous at corr (k) maximum place so detect N δ (k-N+d).

Obtain a set of symbols then, the length of each code element is N _Cp+ N.The sequence that promptly obtains foremost is the circulation short training sequence of experience wireless fading channel, it is used for ICI eliminates.Simultaneously the symbol of back is removed cyclic prefix CP and done N point FFT processing, obtain received signal frequency domain data frame Y, Y is expressed as:

$H_{i,d}=\frac{1}{N}\underset{m=0}{\overset{G}{\mathrm{\Σ}}}{F}_m\left(d\right)e^{-\frac{j2\mathrm{\πm}(i-d)}{N}},0\≤i,d\≤N-1$

F _mBe the FFT conversion coefficient of m channel tap, W _iFFT conversion for noise;

(7) Doppler frequency deviation is estimated: the circulation short training sequence that utilizes the WiMAX signal, to signal carry out actual situation part from, and the relation of imaginary part by signal and real part utilization plane spreading kalman filtering carries out the iteration correction to the Doppler frequency deviation of WiMAX signal and estimates, butt joint is collected mail to cease and is carried out the ICI elimination, obtains data symbols Y '.The ICI removing method is as follows:

If the normalization frequency deviation meter is shown: ε=Δ f T, Δ f is a Doppler frequency deviation, and T is the systematic symbols sample time, and the circulation short training sequence of transmission is x (n), and the sequence that receives of its correspondence is y (n), promptly has: $y\left(n\right)=x\left(n\right)e^{j\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ}}+w\left(n\right),$ θ is unknown carrier wave skew.The frequency domain symbols information that obtains after signal y (n) the process FFT conversion is:

Wherein, $S(l-k)=\frac{\sin \left(\mathrm{\π}(l+\mathrm{\ϵ}-k)\right)}{N\sin (\mathrm{\π}(l+\mathrm{\ϵ}-k)/N)}\exp \left(\mathrm{j\π}(1-\frac{1}{N})(l+\mathrm{\ϵ}-k)\right)$ Represent the ICI interference coefficient, C (k) is the useful signal on the carrier wave k, and ICI (k) is the interference signal on the carrier wave k.

The first step: receiving sequence y (n) is obtained nonlinear function f (n) divided by x (n), and f (n) is a complex function, is expressed as: $f\left(n\right)=(A_{\mathrm{real}}\cos (\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ})+w_{\mathrm{real}})+j(A_{\mathrm{imag}}\sin (\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ})+w_{\mathrm{imag}}),$ w _RealAnd w _ImagBe the real part and the imaginary part of the noise after the divide operations, $w=\left[\begin{array}{c}{w}_{\mathrm{real}}\\ w_{\mathrm{imag}}\end{array}\right];$

Second step: the f (n) that obtains is carried out imaginary part separate, obtain: f=r with real part _c+ jr _s, obtain matrix $r=\left[\begin{array}{c}{r}_c\\ r_s\end{array}\right],$ Promptly be two row sequences, its length is equal to the length of circulation short training sequence;

The 3rd step: order $\mathrm{\ω}=\frac{2\mathrm{\π\ϵ}}{N},$ x ₁(n)=and cos (ω n+ θ), x ₂(n)=and sin (ω n+ θ), x ₃(n)=and cos (ω), x ₄(n)=and sin (ω), obtain state vector x (n)=[x ₁(n), x ₂(n), x ₃(n), x ₄(n)] ^TSo:

$x(n+1)=\left[\begin{array}{c}{x}_1(n+1)\\ x_2(n+1)\\ x_3(n+1)\\ x_4(n+1)\end{array}\right]=\left[\begin{array}{c}\cos (\mathrm{\ωn}+\mathrm{\ω}+\mathrm{\θ})\\ \sin (\mathrm{\ωn}+\mathrm{\ω}+\mathrm{\θ})\\ \cos \left(\mathrm{\ω}\right)\\ \sin \left(\mathrm{\ω}\right)\end{array}\right]=\left[\begin{array}{c}\cos ((\mathrm{\ωn}+\mathrm{\θ})+\mathrm{\ω})\\ \sin ((\mathrm{\ωn}+\mathrm{\θ})+\mathrm{\ω})\\ \cos \left(\mathrm{\ω}\right)\\ \sin \left(\mathrm{\ω}\right)\end{array}\right]$

The 4th step: make x (n+1)=F (x (n)), wherein F (x (n))=[F ₁(x (n)), F ₂(x (n)), F ₃(x (n)), F ₄(x (n))] ^T, obtain:

F ₁(x(n))＝x ₁(n)x ₃(n)-x ₂(n)x ₄(n)

F ₂(x(n))＝x ₂(n)x ₃(n)+x ₁(n)x ₄(n)

F ₃(x(n))＝x ₃(n)

F ₄(x(n))＝x ₄(n)

So observational equation is arranged: r=Hx+w

Wherein $H=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\end{array}\right],$ x＝[x ₁?x ₂?x ₃?x ₄] ^T。

The 5th step: utilize plane spreading kalman filtering to realize frequency offset tracking, be: initialization $\widehat{\mathrm{\ξ}}=0.1,$ The init state error is $P\left(0\right)=\mathrm{diag}(\frac{3}{2},\frac{1}{2},\frac{3}{2},\frac{1}{2}),$ Q is a covariance matrix, is initialized as null matrix, and R is the noise error vector, is initialized as $R=\left[\begin{array}{cc}0.01& 0\\ 0& 0.01\end{array}\right].$

Each goes on foot all design factor factors $H\left(n\right)=\frac{\∂F}{\∂x},$ And update mode error vector P (n)=H (n) P (n-1) H ^*(n)+Q, computer card Kalman Filtering gain K (n)=P (n-1) H then ^*(n) (H (n) P (n-1) H ^*(n)+R) ^-1, the x that estimates of update calculation again, $\hat{x}\left(n\right)=F\left(x\left(n\right)\right),$ The value of next step that estimation obtains is: $\hat{x}(n+1)=K\left(n\right)(r\left(n\right)-H\left(n\right)\hat{x}\left(n\right));$

The 6th step: carry out normalization, be about to F (x (the n)) value that each step iteration obtains and carry out following operation: A (n)=F ₁ ²(x (n))+F ₂ ²(x (n)), ${\hat{F}}_1\left(x\left(n\right)\right)=\frac{F_1\left(x\left(n\right)\right)}{A\left(n\right)},$ ${\hat{F}}_2\left(x\left(n\right)\right)=\frac{F_2\left(x\left(n\right)\right)}{A\left(n\right)}$

The 7th goes on foot: after the whole sequence iteration was finished, the number of times size of iteration operation was the training sequence signal length, obtains at last

So the normalization frequency deviation of estimating is:

Then all WiMAX subcarriers are carried out the Doppler frequency deviation compensation and can eliminate the ICI interference.

(8) channel parameter estimation: each code element of Y ' is carried out the channel frequency domain parameter estimate H ', the time-domain response of establishing channel is:

$h\left(n\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}{h}_i{e}^{j(2\mathrm{\π}/N)f_{D_i}\mathrm{Tn}}\mathrm{\δ}(\mathrm{\λ}-{\mathrm{\τ}}_i),0\≤n\≤N-1,$

Wherein, r is the maximum number of taps order, h _iBe the plural impulse response in i footpath,

Be Doppler frequency shift, λ is the time delay expansion, and T is the systematic sampling interval, τ _iIt is the multidiameter delay in i footpath.

Y(k)＝X(k)H(k)+I(k)+W(k)k＝0，1，...，N-1

Wherein, $H\left(k\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}{h}_i{e}^{\mathrm{j\π}{f}_{D_i}T}\frac{\sin \left(\mathrm{\π}{f}_{D_i}T\right)}{\mathrm{\π}{f}_{D_i}T}{e}^{-j(2\mathrm{\π}{\mathrm{\τ}}_i/N)k},$

$I\left(k\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}\underset{\underset{K\≠k}{K=0}}{\overset{N-1}{\mathrm{\Σ}}}\frac{h_{i}X\left(K\right)}{N}\frac{1-e^{j2\mathrm{\π}(f_{D_i}T-k+K)}}{1-e^{j(2\mathrm{\π}/N)(f_{D_i}T-k+K)}}{e}^{-j(2\mathrm{\π}{\mathrm{\τ}}_i/N)K}$

Owing to carried out the ICI elimination, thus ICI distracter I (k) do not exist, so at pilot tone X _PilotThe channel frequency domain response at place is: $H\left(p\right)=\frac{Y\left(p\right)}{X_{\mathrm{pilot}}\left(p\right)},$ And the channel value H that estimates to obtain is carried out linear interpolation obtain channel estimating initial coefficients on all sub-carrier positions

(9) data balancing and recovery information source information: the channel frequency domain value that utilizes estimation to obtain

Finish the data balancing on each subchannel carrier of WiMAX, and balanced data is carried out demodulation and gone here and there P/S conversion, deinterleaving, channel decoding and descrambling, last by the information source information that is restored behind the source coding.

The present invention has following conspicuous outstanding substantial characteristics and remarkable advantage compared with prior art:

The rapid fading ICI that described inventive method utilizes plane spreading kalman filtering to eliminate the WiMAX system disturbs, and the inventive method system has that computation complexity is low, expense is few, estimated accuracy is high and need not advantage such as estimated signal initial phase offset.

Description of drawings

Fig. 1 is for using the WiMAX communication system workflow schematic diagram of one embodiment of the invention.

Fig. 2 is the linear frequency modulation training sequence signal amplitude figure of the inventive method.

Fig. 3 is a WiMAX Comb Pilot format chart of the present invention.

Embodiment

Below in conjunction with accompanying drawing a preferred embodiment of the present invention is elaborated.Referring to Fig. 1, this is as follows based on the WiMAX rapid fading ICI removing method workflow of plane spreading kalman filtering: (1) information source information is transformed into valid data to be transmitted: information source information is carried out source encoding, scrambling, chnnel coding, interweaved, go here and there and S/P conversion and modulation become valid data X to be transmitted on N the subcarrier after shining upon _i, wherein, channel coding method adopts LDPC or Turbo or Woven convolution code or RS-CC coding, and the modulation mapping is conciliate mapping and is adopted QPSK or 16QAM mapping; (2) pilot tone is inserted and data multiplex: produce and channel frequency domain sampling position corresponding subcarrier pilot data X _Pilot, and be that pilot data inserts pilot tone with pilot data according to the Comb Pilot form among the IEEE802.16e at p place at interval with L, carry out data multiplex with the data to be transmitted structure, the length of generation pilot data equals 1/10th of WiMAX system-frame length; (3) add Cyclic Prefix: after the IFFT that the data after multiplexing orders through N handles, composition WiMAX Frame x _k, and have relation $X_i=\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{x}_k{e}^{-\frac{j2\mathrm{\πki}}{N}},(0\≤i\≤N-1),$ Add cyclic prefix CP before the symbol head of each code element therein then, the length that each code element adds Cyclic Prefix is N _CpIndividual sample point promptly has: x (i)=x _N-G+i, 0≤i≤G-1, G equals N here for protection length at interval in the formula _Cp, N _CpValue decide by the time delay largest extension value of transmission wireless channel, be four times of time delay largest extension value. (4) add the circulation short training sequence: add the cocycle short training sequence foremost at WiMAX Frame X ', this training sequence signal is known at receiving terminal, its length is the length of a symbol, this training sequence signal adopts the linear FM signal design, it both had been used for carrying out Domain Synchronous and also had been used for the ICI elimination, reduced the symbol expense.(5) rise sampling filter and pass through channel: WiMAX Frame X ' process is risen sampling FIR filtering, and obtain signal waveform information s (t) after string P/S conversion and the digital-to-analogue D/A conversion, t is a time variable, then s (t) is launched at the WiMAX transmitting terminal, four times of over-samplings are adopted in the liter sampling that rises in the sampling FIR filtering operation, the FIR filter that rises in the sampling FIR filtering operation is the root-raised cosine filter on 64 rank, rolloff-factor is: 27/128, the present invention adopts the FIR on 4 16 rank to come Parallel Implementation, behind experience wireless fading channel h (t) and the additive noise n (t), the signal that receiving terminal receives is r (t), promptly have: r (t)=s (t) * h (t)+n (t), wherein * represents the time domain convolution, and channel output is specialized to be had: $y_i=\underset{k=0}{\overset{G}{\mathrm{\Σ}}}{h}_k^{\left(i\right)}{x}_{{\left((i-k)\right)}_N}+w_i,0\≤i\≤N-1,$ h _k ⁽ⁱ⁾Be that k tap is in the code element channel time domain response of i constantly; (6) remove Cyclic Prefix with synchronously: extract with carrying out Domain Synchronous after received signal r (t) process modulus A/D conversion, string and S/P conversion and the down-sampled FIR filtering, the Domain Synchronous extraction algorithm is: establishing the linear frequency modulation training sequence signal that receives is C (n), and correlation formula is promptly arranged:

When having time delay d, receive signal and be: z (k)=C (k-d)+w (k); Obtaining cross-correlation is: $\mathrm{corr}\left(k\right)=y\left(k\right)*x(N-k)=\underset{n=0}{\overset{N}{\mathrm{\Σ}}}y(k-n)x(N-n)=\underset{n=0}{\overset{N}{\mathrm{\Σ}}}x(k-n+d)x(N-n)+\underset{n=0}{\overset{N}{\mathrm{\Σ}}}w(k-n)x(N-n)$

Because known R _XX(k), so: $\mathrm{corr}\left(k\right)=R_{\mathrm{XX}}\left(k\right)+\underset{n=0}{\overset{N}{\mathrm{\Σ}}}w(k-n)x(N-n)\≈\mathrm{N\δ}(k-N+d)+\mathrm{\ξ}\left(k\right)$

Because ξ (k) is very little value,, promptly just extract Domain Synchronous at corr (k) maximum place so detect N δ (k-N+d).

Obtain a set of symbols then, the length of each code element is N _Cp+ N.The sequence that promptly obtains foremost is the circulation short training sequence of experience wireless fading channel, it is used for ICI eliminates.Simultaneously the symbol of back is removed cyclic prefix CP and done N point FFT processing, obtain received signal frequency domain data frame Y, Y is expressed as:

$H_{i,d}=\frac{1}{N}\underset{m=0}{\overset{G}{\mathrm{\Σ}}}{F}_m\left(d\right)e^{-\frac{j2\mathrm{\πm}(i-d)}{N}},0\≤i,d\≤N-1,$ F _mBe the FFT conversion coefficient of m channel tap, W _iFFT conversion for noise;

(7) Doppler frequency deviation is estimated: the circulation short training sequence that utilizes the WiMAX signal, to signal carry out actual situation part from, and the relation of imaginary part by signal and real part utilization plane spreading kalman filtering carries out the iteration correction to the Doppler frequency deviation of WiMAX signal and estimates, butt joint is collected mail to cease and is carried out the ICI elimination, obtains data symbols Y '.The ICI removing method is as follows: establish the normalization frequency deviation meter and be shown: ε=Δ f T, and Δ f is a Doppler frequency deviation, and T is the systematic symbols sample time, and the circulation short training sequence of transmission is x (n), and the sequence that receives of its correspondence is y (n), promptly has: $y\left(n\right)=x\left(n\right)e^{j\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ}}+w\left(n\right),$ θ is unknown carrier wave skew.The frequency domain symbols information that obtains after signal y (n) the process FFT conversion is:

Wherein, $S(l-k)=\frac{\sin \left(\mathrm{\π}(l+\mathrm{\ϵ}-k)\right)}{N\sin (\mathrm{\π}(l+\mathrm{\ϵ}-k)/N)}\exp \left(\mathrm{j\π}(1-\frac{1}{N})(l+\mathrm{\ϵ}-k)\right)$ Represent the ICI interference coefficient, C (k) is the useful signal on the carrier wave k, and ICI (k) is the interference signal on the carrier wave k.

The first step: receiving sequence y (n) is obtained nonlinear function f (n) divided by x (n), and f (n) is a complex function, is expressed as: $f\left(n\right)=(A_{\mathrm{real}}\cos (\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ})+w_{\mathrm{real}})+j(A_{\mathrm{imag}}\sin (\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ})+w_{\mathrm{imag}}),$ w _RealAnd w _ImagBe the real part and the imaginary part of the noise after the divide operations, $w=\left[\begin{array}{c}{w}_{\mathrm{real}}\\ w_{\mathrm{imag}}\end{array}\right];$ Second step: the f (n) that obtains is carried out imaginary part separate, obtain: f=r with real part _c+ jr _s, obtain matrix $r=\left[\begin{array}{c}{r}_c\\ r_s\end{array}\right],$ Promptly be two row sequences, its length is equal to the length of circulation short training sequence;

The 3rd step: order $\mathrm{\ω}=\frac{2\mathrm{\π\ϵ}}{N},$ x ₁(n)=and cos (ω n+ θ), x ₂(n)=and sin (ω n+ θ), x ₃(n)=and cos (ω), x ₄(n)=and sin (ω), obtain state vector x (n)=[x ₁(n), x ₂(n), x ₃(n), x ₄(n)] ^TSo:

$x(n+1)=\left[\begin{array}{c}{x}_1(n+1)\\ x_2(n+1)\\ x_3(n+1)\\ x_4(n+1)\end{array}\right]=\left[\begin{array}{c}\cos (\mathrm{\ωn}+\mathrm{\ω}+\mathrm{\θ})\\ \sin (\mathrm{\ωn}+\mathrm{\ω}+\mathrm{\θ})\\ \cos \left(\mathrm{\ω}\right)\\ \sin \left(\mathrm{\ω}\right)\end{array}\right]=\left[\begin{array}{c}\cos ((\mathrm{\ωn}+\mathrm{\θ})+\mathrm{\ω})\\ \sin ((\mathrm{\ωn}+\mathrm{\θ})+\mathrm{\ω})\\ \cos \left(\mathrm{\ω}\right)\\ \sin \left(\mathrm{\ω}\right)\end{array}\right]$

The 4th step: make x (n+1)=F (x (n)), wherein F (x (n))=[F ₁(x (n)), F ₂(x (n)), F ₃(x (n)), F ₄(x (n))] ^T, obtain:

F ₁(x(n))＝x ₁(n)x ₃(n)-x ₂(n)x ₄(n)

F ₂(x(n))＝x ₂(n)x ₃(n)+x ₁(n)x ₄(n)

F ₃(x(n))＝x ₃(n)

F ₄(x(n))＝x ₄(n)

So observational equation is arranged: r=Hx+w

Wherein $H=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\end{array}\right],$ x＝[x ₁?x ₂?x ₃?x ₄] ^T。

The 5th step: utilize plane spreading kalman filtering to realize frequency offset tracking, be: initialization $\widehat{\mathrm{\ξ}}=0.1,$ The init state error is $P\left(0\right)=\mathrm{diag}(\frac{3}{2},\frac{1}{2},\frac{3}{2},\frac{1}{2}),$ Q is a covariance matrix, is initialized as null matrix, and R is the noise error vector, is initialized as $R=\left[\begin{array}{cc}0.01& 0\\ 0& 0.01\end{array}\right].$ Each goes on foot all design factor factors $H\left(n\right)=\frac{\∂F}{\∂x},$ And update mode error vector P (n)=H (n) P (n-1) H ^*(n)+Q, computer card Kalman Filtering gain K (n)=P (n-1) H then ^*(n) (H (n) P (n-1) H ^*(n)+R) ^-1, the x that estimates of update calculation again, $\hat{x}\left(n\right)=F\left(x\left(n\right)\right),$ The value of next step that estimation obtains is: $\hat{x}(n+1)=K\left(n\right)(r\left(n\right)-H\left(n\right)\hat{x}\left(n\right));$

The 6th step: carry out normalization, be about to F (x (the n)) value that each step iteration obtains and carry out following operation: A (n)=F ₁ ²(x (n))+F ₂ ²(x (n)), ${\hat{F}}_1\left(x\left(n\right)\right)=\frac{F_1\left(x\left(n\right)\right)}{A\left(n\right)},$ ${\hat{F}}_2\left(x\left(n\right)\right)=\frac{F_2\left(x\left(n\right)\right)}{A\left(n\right)}$

The 7th goes on foot: after the whole sequence iteration was finished, the number of times size of iteration operation was the training sequence signal length, obtains at last

So the normalization frequency deviation of estimating is:

Then all WiMAX subcarriers are carried out the Doppler frequency deviation compensation and can eliminate the ICI interference.

(8) channel parameter estimation: each code element of Y ' is carried out the channel frequency domain parameter estimate H ', the time-domain response of establishing channel is:

$h\left(n\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}{h}_i{e}^{j(2\mathrm{\π}/N)f_{D_i}\mathrm{Tn}}\mathrm{\δ}(\mathrm{\λ}-{\mathrm{\τ}}_i),0\≤n\≤N-1,$

Wherein, r is the maximum number of taps order, h _iBe the plural impulse response in i footpath,

Be Doppler frequency shift, λ is the time delay expansion, and T is the systematic sampling interval, τ _iIt is the multidiameter delay in i footpath.

Y(k)＝X(k)H(k)+I(k)+W(k)k＝0，1，...，N-1

Wherein, $H\left(k\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}{h}_i{e}^{\mathrm{j\π}{f}_{D_i}T}\frac{\sin \left(\mathrm{\π}{f}_{D_i}T\right)}{\mathrm{\π}{f}_{D_i}T}{e}^{-j(2\mathrm{\π}{\mathrm{\τ}}_i/N)k},$

$I\left(k\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}\underset{\underset{K\≠k}{K=0}}{\overset{N-1}{\mathrm{\Σ}}}\frac{h_{i}X\left(K\right)}{N}\frac{1-e^{j2\mathrm{\π}(f_{D_i}T-k+K)}}{1-e^{j(2\mathrm{\π}/N)(f_{D_i}T-k+K)}}{e}^{-j(2\mathrm{\π}{\mathrm{\τ}}_i/N)K}$

Owing to carried out the ICI elimination, thus ICI distracter I (k) do not exist, so at pilot tone X _PilotThe channel frequency domain response at place is: $H\left(p\right)=\frac{Y\left(p\right)}{X_{\mathrm{pilot}}\left(p\right)},$ And the channel value H that estimates to obtain is carried out linear interpolation obtain channel estimating initial coefficients on all sub-carrier positions

(9) data balancing and recovery information source information: the channel frequency domain value that utilizes estimation to obtain Finish the data balancing on each subchannel carrier of WiMAX, and balanced data is carried out demodulation and gone here and there P/S conversion, deinterleaving, channel decoding and descrambling, last by the information source information that is restored behind the source coding.

Fig. 2 is the linear frequency modulation training sequence signal amplitude figure of the inventive method, this sequence has 1024 time domain sample points, has good autocorrelation, and this training sequence signal is known at receiving terminal, it is used for carrying out Domain Synchronous and ICI eliminates, and reduces the symbol expense.

Fig. 3 is a WiMAX Comb Pilot format chart of the present invention, and the pilot bits information in its code element is continuous in time domain, and certain interval is arranged on the frequency domain, and the time selective fading to channel has the good restraining effect like this.

In sum, the invention provides a kind of WiMAX rapid fading ICI removing method based on plane spreading kalman filtering, this method utilization plane spreading kalman filtering carries out iteration correction estimation to the Doppler frequency deviation of WiMAX signal, and then eliminates the interference of rapid fading ICI to the WiMAX system.The ICI that the inventive method can effectively be eliminated the WiMAX system disturbs, and has the estimated accuracy height, complexity is low and need not advantage such as estimated signal initial phase offset, for the mobile WiMAX wireless access technology provides important practicability technology.

Claims (7)

1. WiMAX rapid fading ICI removing method based on plane spreading kalman filtering, it is characterized in that utilizing the circulation short training sequence of WiMAX signal, to signal carry out actual situation part from, and the relation of imaginary part by signal and real part utilization plane spreading kalman filtering carries out the iteration correction to the Doppler frequency deviation of WiMAX signal and estimates, and then eliminate the interference of rapid fading ICI to the WiMAX system, its operating procedure is as follows:

(1) information source information is transformed into valid data to be transmitted: with information source information carry out source encoding, scrambling, chnnel coding, interweave, go here and there and S/P conversion and modulation mapping after become valid data X to be transmitted on N the subcarrier _i

(2) pilot tone is inserted and data multiplex: produce and channel frequency domain sampling position corresponding subcarrier pilot data X _Pilot, and with L be pilot data at interval at p place with pilot data according to the insertion of the Comb Pilot form among IEEE802.16e pilot tone, carry out data multiplex with the data to be transmitted structure;

(3) add Cyclic Prefix: after the IFFT that the data after multiplexing orders through N handles, composition WiMAX Frame x _k, and have relation $X_i=\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{x}_k{e}^{\frac{j2\mathrm{\πki}}{N}}(0\≤i\≤N-1),$ Add cyclic prefix CP before the symbol head of each code element therein then, the length that each code element adds Cyclic Prefix is N _CpIndividual sample point promptly has: x (i)=x _N-G+i, 0≤i≤G-1, G equals N here for protection length at interval in the formula _Cp

(4) add the circulation short training sequence: add the cocycle short training sequence foremost at WiMAX Frame X ', this training sequence signal is known at receiving terminal, its length is the length of a symbol, this training sequence signal adopts the linear FM signal design, it both had been used for carrying out Domain Synchronous and also had been used for the ICI elimination, reduced the symbol expense.

(5) rise sampling filter and pass through channel: obtain signal waveform information s (t) after WiMAX Frame X ' process is risen sampling FIR filtering and the P/S that goes here and there conversion and digital-to-analogue D/A conversion, t is a time variable, then s (t) is launched at the WiMAX transmitting terminal, behind experience wireless fading channel h (t) and the additive noise n (t), the signal that receiving terminal receives is r (t), promptly have: r (t)=s (t) * h (t)+n (t), wherein * represents the time domain convolution, channel output is specialized to be had: $y_i=\underset{k=0}{\overset{G}{\mathrm{\Σ}}}{h}_k^{\left(i\right)}{x}_{{\left((i-k)\right)}_N}+w_i$ 0≤i≤N-1, h _k ⁽ⁱ⁾Be that k tap is in the code element channel time domain response of i constantly;

(6) remove Cyclic Prefix with synchronously: extract carrying out Domain Synchronous after received signal r (t) process modulus A/D conversion, string and S/P conversion and the down-sampled FIR filtering, obtain a set of symbols, the length of each code element is N _Cp+ N.The sequence that promptly obtains foremost is the circulation short training sequence of experience wireless fading channel, it is used for ICI eliminates.Simultaneously the symbol of back is removed cyclic prefix CP and done N point FFT processing, obtain received signal frequency domain data frame Y, Y is expressed as:

$H_{i,d}=\frac{1}{N}\underset{m=0}{\overset{G}{\mathrm{\Σ}}}{F}_m\left(d\right)e^{-\frac{j2\mathrm{\πm}(i-d)}{N}}0\≤i,d\≤N-1$

F _mBe the FFT conversion coefficient of m channel tap, W _iFFT conversion for noise;

(7) Doppler frequency deviation is estimated: the circulation short training sequence that utilizes the WiMAX signal, to signal carry out actual situation part from, and the relation of imaginary part by signal and real part utilization plane spreading kalman filtering carries out the iteration correction to the Doppler frequency deviation of WiMAX signal and estimates, butt joint is collected mail to cease and is carried out the ICI elimination, obtains data symbols Y ';

(8) channel parameter estimation: each code element of Y ' is carried out the channel frequency domain parameter estimate H ', the time-domain response of establishing channel is:

$h\left(n\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}{h}_i{e}^{j(2\mathrm{\π}/N)f_{D_i}\mathrm{Tn}}\mathrm{\δ}(\mathrm{\λ}-{\mathrm{\τ}}_i)0\≤n\≤N-1,$

Wherein, r is the maximum number of taps order, h _iBe the plural impulse response in i footpath,

Be Doppler frequency shift, λ is the time delay expansion, and T is the systematic sampling interval, τ _iIt is the multidiameter delay in i footpath;

Y(k)＝X(k)H(k)+I(k)+W(k)k＝0，1，...，N-1

Wherein, $H\left(k\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}{h}_i{e}^{\mathrm{j\π}{f}_{D_i}T}\frac{\sin (\mathrm{\π}{f}_{D_i}T}{\mathrm{\π}{f}_{D_i}T}{e}^{-j(2\mathrm{\π}{\mathrm{\τ}}_i/N)k},$

$I\left(k\right)=\underset{i=0}{\overset{r-1}{\mathrm{\Σ}}}\underset{\begin{array}{c}K=0\\ K\≠k\end{array}}{\overset{N-1}{\mathrm{\Σ}}}\frac{h_{i}X\left(K\right)}{N}\frac{1-e^{j2\mathrm{\π}(f_{D_i}T-k+K)}}{1-e^{j(2\mathrm{\π}/N)(f_{D_i}T-k+K)}}{e}^{-j(2\mathrm{\π}{\mathrm{\τ}}_i/N)K}$

Owing to carried out the ICI elimination, thus ICI distracter I (k) do not exist, so at pilot tone X _PilotThe channel frequency domain response at place is: $H\left(p\right)=\frac{Y\left(p\right)}{X_{\mathrm{pilot}}\left(p\right)},$ And the channel value H that estimates to obtain is carried out linear interpolation obtain channel estimating initial coefficients on all sub-carrier positions

(9) data balancing and recovery information source information: the channel frequency domain value that utilizes estimation to obtain

Finish the data balancing on each subchannel carrier of WiMAX, and balanced data is carried out demodulation and gone here and there P/S conversion, deinterleaving, channel decoding and descrambling, last by the information source information that is restored behind the source coding.

2. the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering according to claim 1, it is characterized in that: the channel coding method in the step (1) adopts LDPC or Turbo or Woven convolution code or RS-CC coding, and the length that produces pilot data in the step (2) equals 1/10th of WiMAX system-frame length.

3. the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering as claimed in claim 1 is characterized in that: N in the step (3) _CpValue decide by the time delay largest extension value of transmission wireless channel, be four times of time delay largest extension value.

4. the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering as claimed in claim 1, it is characterized in that: four times of over-samplings are adopted in the liter sampling in the liter sampling FIR filtering operation of step (5), the FIR filter that rises in the sampling FIR filtering operation is the root-raised cosine filter on 64 rank, rolloff-factor is: 27/128, and adopt the FIR on 4 16 rank to come Parallel Implementation.

5. the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering as claimed in claim 1 is characterized in that: the modulation mapping in step (1) and the step (9) is conciliate mapping and is adopted QPSK or 16QAM mapping.

6. the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering as claimed in claim 1 is characterized in that: the Domain Synchronous extraction algorithm is in the step (6):

If the linear frequency modulation training sequence signal that receives is C (n), correlation formula is arranged promptly:

K represents the sampled point number of sequence, and when having time delay d, receive signal and be: z (k)=C (k-d)+w (k) obtains cross-correlation and is:

$\mathrm{corr}\left(k\right)=y\left(k\right)*x(N-k)=\underset{n=0}{\overset{N}{\mathrm{\Σ}}}y(k-n)x(N-n)$

$=\underset{n=0}{\overset{N}{\mathrm{\Σ}}}x(k-n+d)x(N-n)+\underset{n=0}{\overset{N}{\mathrm{\Σ}}}w(k-n)x(N-n)$

Because known R _XX(k), so: $\mathrm{corr}\left(k\right)=R_{\mathrm{XX}}\left(k\right)+\underset{n=0}{\overset{N}{\mathrm{\Σ}}}w(k-n)x(N-n)\≈\mathrm{N\δ}(k-N+d)+\mathrm{\ξ}\left(k\right)$

Because ξ (k) is very little value,, promptly just extract Domain Synchronous at corr (k) maximum place so detect N δ (k-N+d).

7. the WiMAX rapid fading ICI removing method based on plane spreading kalman filtering as claimed in claim 1 is characterized in that: the ICI removing method is as follows in the step (7):

If the normalization frequency deviation meter is shown: ε=Δ f T, Δ f is a Doppler frequency deviation, and T is the systematic symbols sample time, and the circulation short training sequence of transmission is x (n), and the sequence that receives of its correspondence is y (n), promptly has: $y\left(n\right)=x\left(n\right)e^{j\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ}}+w\left(n\right),$ θ is unknown carrier wave skew.The frequency domain symbols information that obtains after signal y (n) the process FFT conversion is:

Wherein, $S(l-k)=\frac{\sin \left(\mathrm{\π}(l+\mathrm{\ϵ}-k)\right)}{N\sin (\mathrm{\π}(l+\mathrm{\ϵ}-k)/N)}\exp \left(\mathrm{j\π}(1-\frac{1}{N})(l+\mathrm{\ϵ}-k)\right)$ Represent the ICI interference coefficient, C (k) is the useful signal on the carrier wave k, and ICI (k) is the interference signal on the carrier wave k.

The first step: receiving sequence y (n) is obtained nonlinear function f (n) divided by x (n), and f (n) is a complex function, is expressed as: $f\left(n\right)=(A_{\mathrm{real}}\cos (\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ})+w_{\mathrm{real}})+j(A_{\mathrm{imag}}\sin (\frac{2\mathrm{\πn\ϵ}}{N}+\mathrm{\θ})+w_{\mathrm{imag}}),$ w _RealAnd w _ImagBe the real part and the imaginary part of the noise after the divide operations, $w=\left[\begin{array}{c}{w}_{\mathrm{real}}\\ w_{\mathrm{imag}}\end{array}\right];$

Second step: the f (n) that obtains is carried out imaginary part separate, obtain: f=r with real part _c+ jr _s, obtain matrix $r=\left[\begin{array}{c}{r}_c\\ r_s\end{array}\right],$ Promptly be two row sequences, its length is equal to the length of circulation short training sequence;

The 3rd step: order $\mathrm{\ω}=\frac{2\mathrm{\π\ϵ}}{N},$ x ₁(n)=and cos (ω n+ θ), x ₂(n)=and sin (ω n+ θ), x ₃(n)=and cos (ω), x ₄(n)=and sin (ω), obtain state vector x (n)=[x ₁(n), x ₂(n), x ₃(n), x ₄(n)] ^TSo:

$x(n+1)=\left[\begin{array}{c}{x}_1(n+1)\\ x_2(n+1)\\ x_3(n+1)\\ x_4(n+1)\end{array}\right]=\left[\begin{array}{c}\cos (\mathrm{\ωn}+\mathrm{\ω}+\mathrm{\θ})\\ \sin (\mathrm{\ωn}+\mathrm{\ω}+\mathrm{\θ})\\ \cos \left(\mathrm{\ω}\right)\\ \sin \left(\mathrm{\ω}\right)\end{array}\right]=\left[\begin{array}{c}\cos ((\mathrm{\ωn}+\mathrm{\θ})+\mathrm{\ω}))\\ \sin ((\mathrm{\ωn}+\mathrm{\θ})+\mathrm{\ω})\\ \cos \left(\mathrm{\ω}\right)\\ \sin \left(\mathrm{\ω}\right)\end{array}\right]$

The 4th step: make x (n+1)=F (x (n)), wherein F (x (n))=[F ₁(x (n)), F ₂(x (n)), F ₃(x (n)), F ₄(x (n))] ^T, obtain:

F ₁(x(n))＝x ₁(n)x ₃(n)-x ₂(n)x ₄(n)

F ₂(x(n))＝x ₂(n)x ₃(n)+x ₁(n)x ₄(n)

F ₃(x(n))＝x ₃(n)

F ₄(x(n))＝x ₄(n)

So observational equation is arranged: r=Hx+w

Wherein $H=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\end{array}\right],$ x＝[x ₁?x ₂?x ₃?x ₄] ^T。

The 5th step: utilize plane spreading kalman filtering to realize frequency offset tracking, be: initialization $\widehat{\mathrm{\ξ}}=0.1,$ The init state error is $P\left(0\right)=\mathrm{diag}(\frac{3}{2},\frac{1}{2},\frac{3}{2},\frac{1}{2}),$ Q is a covariance matrix, is initialized as null matrix, and R is the noise error vector, is initialized as $R=\left[\begin{array}{cc}0.01& 0\\ 0& 0.01\end{array}\right].$

Each goes on foot all design factor factors $H\left(n\right)=\frac{\∂F}{\∂x},$ And update mode error vector P (n)=H (n) P (n-1) H ^*(n)+Q, computer card Kalman Filtering gain K (n)=P (n-1) H then ^*(n) (H (n) P (n-1) H ^*(n)+R) ^-1, the x that estimates of update calculation again, $\hat{x}\left(n\right)=F\left(x\left(n\right)\right),$ The value of next step that estimation obtains is: $\hat{x}(n+1)=K\left(n\right)(r\left(n\right)-H\left(n\right)\hat{x}\left(n\right));$

The 6th step: carry out normalization, be about to F (x (the n)) value that each step iteration obtains and carry out following operation:

$A\left(n\right)=F_1^2\left(x\left(n\right)\right)+F_2^2\left(x\left(n\right)\right),$ ${\hat{F}}_1\left(x\left(n\right)\right)=\frac{F_1\left(x\left(n\right)\right)}{A\left(n\right)},$ ${\hat{F}}_2\left(x\left(n\right)\right)=\frac{F_2\left(x\left(n\right)\right)}{A\left(n\right)}$

The 7th goes on foot: after the whole sequence iteration was finished, the number of times size of iteration operation was the training sequence signal length, obtains at last So the normalization frequency deviation of estimating is:

Then all WiMAX subcarriers are carried out the Doppler frequency deviation compensation and can eliminate the ICI interference.

Images