CN101741783A - Method for tracking channel in single-carrier frequency domain equalization system - Google Patents

Abstract

The invention discloses a method for tracking a channel in a single-carrier frequency domain equalization system, which comprises the following steps of: acquiring an initial frequency domain equalization coefficient W0=W1=1/H according to a precursor block sequence in a data frame; tracking a channel by using a recursive least square algorithm when a first packet of data r1 is received; judging whether variance sigma (Wn-Wn-1) of the difference of the frequency domain equalization coefficients Wn and Wn-1 is less than a preset threshold sigma m or not according to W0 and W1 when the nth packet of data rn is received, if so, acquiring the frequency domain equalization coefficient Wn+1 according to the least-mean-square algorithm and tracking the channel, and otherwise, acquiring the frequency domain equalization coefficient Wn+1 according to the recursive least square algorithm and tracking the channel. In the invention, rapid convergence is carried out on the received first packet of data by using the RLS algorithm, and the channel is tracked by using the LMS algorithm after the convergence is achieved. Combining the advantages of the two algorithms, the invention has high tracking speed and high convergence accuracy, maintains smaller stable-state error and better tracking performance, and reduces calculation complexity in comparison with the RLS algorithm.

Description

The method of channel tracking in a kind of single-carrier frequency domain equalization system

Technical field

The present invention relates to wireless communication field, particularly relate to the method for channel tracking in a kind of single-carrier frequency domain equalization system.

Background technology

According to the division of the CCIR of Consultative Committee on International Radio (CCIR) (Consultative Committee of InternationalRadio), shortwave is meant that (frequency is the electromagnetic wave of 3MHz～30MHz) to wavelength at 100m～10m.The radio communication that radiothermy carries out is called short wave communication.

Short wave communication all is the form that adopts the sky wave transmission, just relies on ionospheric one or many to reflect into row communication, therefore exists serious multipath effect.The sky wave channels could of shortwave is a variable-parameter channel, and stable signal transmission is poor.Hf radio communication mainly is to rely on ionosphere to carry out the distant signal transmission, and ionosphere is that the changeability of parameter is very big as the weakness of signal reflex medium.Its feature is path loss, time delay expansion, noise and interference, all with round the clock, frequency, place and constantly change.Ionospheric variation makes signal produce decline, and the amplitude and the frequency of decline constantly change.Therefore the time-varying characteristics of channel are quite strong.

A new generation's Shortwave Communication System mainly is a high speed shortwave digital communication system, has parallel and two kinds of systems of serial.The main thought of parallel system is exactly that short wave channel is divided into some parallel subchannels, transmits a subcarrier on each subchannel, adopts the frequency division quadrature modulation, and the method raising speed with a plurality of subcarrier parallel transmissions is called multitone again.The serial system adopts the send mode of single carrier wave, is also referred to as single-tone, will reach than higher speed under the serial system, and intersymbol interference will be very serious, considers the time-varying characteristics that channel is stronger simultaneously, should adopt suitable adaptive equalization technique.Single carrier frequency domain equalization (SC-FDE in the serial system, Single Carrier-Frequency Domain Equalization) system is a kind of effectively method that anti-multipath is disturbed in the broadband wireless transmission, is fit to adopt adaptive frequency domain balancing technique antagonism channel time-varying characteristics.

The key problem of balancing technique is exactly a compensate for channel distortions, and frequency domain equalization is conceived to the distortion of compensate for channel frequency domain response, so its key problem estimates that exactly current channel frequency domain response is to obtain channel compensation coefficient (equalizing coefficient).The adaptive frequency domain balancing technique is based on the proposition of SC-FDE technology, and its core concept comes self adaptation adjustment equalizing coefficient to come compensate for channel distortions by following the tracks of channel estimating exactly.The most frequently used adaptive equalization algorithm mainly contains least mean square algorithm (LMS) or recursive least squares (RLS).

LMS algorithm thought source is in minimum mean square error criterion, and the LMS algorithm the convergence speed is slower, but short wave channel is had certain follow-up control, and the steady-state error during convergence is less; The RLS algorithm is based on criterion of least squares, and the characteristics of RLS algorithm are fast convergence rates, but steady-state error is big, and tracking performance is very low, easily imbalance.In the tracing process, if though the steady-state error during convergence is less, there is the slower shortcoming of convergence rate in single LMS algorithm that adopts; If single adopt the LMS algorithm, though fast convergence rate exists that steady-state error is big, tracking performance is very low, the shortcoming of easy imbalance.

Summary of the invention

The invention provides the method for channel tracking in a kind of single-carrier frequency domain equalization system, use LMS algorithm or RLS algorithm all can't satisfy the problem of instructions for use separately in order to solve to exist in the prior art.

For reaching above-mentioned purpose, the invention provides the method for channel tracking in a kind of single-carrier frequency domain equalization system, said method comprising the steps of:

According to the leader block sequence in the Frame, obtain initial frequency domain equalization coefficient W ₀=W ₁=1/H;

Receive the first bag data r ₁The time, adopt recursive least squares to carry out channel tracking;

When receiving n bag data r _nThe time, according to W ₀, W ₁, judge frequency domain equalization coefficient W _nWith W _N-1The variances sigma (W of difference _n-W _N-1) whether less than the threshold value σ m that sets in advance, if then obtain frequency domain equalization coefficient W according to least mean square algorithm _N+1, carry out channel tracking; Otherwise, obtain frequency domain equalization coefficient W according to recursive least squares _N+1, carry out channel tracking; Wherein, n is the numbering of packet, and n is more than or equal to 1.

Further, described leader block sequence is the UW sequence, u=[u (1) ..., u (M)], its Fourier changing value is: U _Fft=[U (1) ... U (M)].

Further, according to the leader block sequence in the Frame, obtain frequency domain equalization coefficient W ₀=W ₁=1/H specifically comprises:

The reception data of the leader block correspondence that receiving terminal receives are r=[r (1) ... r (M)], vectorial r is carried out Fourier transform, obtain R _Fft:

R _fft＝[R(1)，...，R(M)]＝FFT(r)；

According to R _Fft, U _Fft, obtain channel estimating frequency domain response H:

$H=[H\left(1\right),...,H\left(M\right)]=R_{\mathrm{fft}}/U_{\mathrm{fft}}=\underset{n=1}{\overset{M}{\mathrm{\Σ}}}R\left(n\right)/U\left(n\right);$

According to H, obtain initial frequency domain equalization coefficient W ₀, W ₁:

$W_0=W_1=1/H=\frac{H^{*}}{{\left|H\right|}^2};$

Wherein, H ^*Conjugation for H; | H| is for to ask mould to vectorial H.

Further, when receiving n bag data r _nThe time, at first, according to W _nAnd r _n, obtain error e _n

Further, according to W _nAnd r _n, obtain error e _n, specifically may further comprise the steps:

To r _nDo fast Fourier transform and obtain sequence R _n:

R _n＝FFT(r _n)；

According to R _n, obtain balanced output sequence Z _n:

Z _n＝W _n·R _n；

To balanced output sequence Z _nCarry out inverse fast Fourier transform, obtain time domain output sequence z _n:

z _n＝IFFT(Z _n)；

To time domain output sequence z _nDecode, obtain adjudicating sequence d _n:

d _n＝Dec[z _n]；

According to judgement sequence d _nWith time domain output sequence z _n, obtain error sequence e _n:

e _n＝d _n-z _n。

Further, obtain frequency domain equalization coefficient W according to least mean square algorithm _N+1, specifically may further comprise the steps:

To error sequence e _nDo fast Fourier transform, obtain E=FFT (e _n);

According to E, obtain frequency domain equalization coefficient W _N+1:

W _n+1＝W _n+μE·R _n

Wherein, μ is the gain constant that is used for controlling adaptive speed and stability.

Further, μ=0.03.

Further, obtain frequency domain equalization coefficient W according to recursive least squares _N+1, specifically may further comprise the steps:

To error sequence e _nDo fast Fourier transform, obtain sequence E=FFT (e _n);

Calculate kalman gain vector K:

$K=\frac{\mathrm{\Δ}\·{R_n}^{*}}{\mathrm{\λ}+\mathrm{\Δ}\·{\left|R_n\right|}^2};$

Wherein, Δ is a correlated series, and λ is the gain constant that is used for controlling adaptive speed and stability, R ^*Conjugation for R;

Upgrade the Δ sequence, Δ carried out iterative computation:

$\mathrm{\Δ}=\frac{\mathrm{\Δ}}{\mathrm{\λ}+\mathrm{\Δ}\·{\left|R_n\right|}^2};$

According to W _n, K, E, obtain frequency domain equalization coefficient W _N+1:

W _n+1＝W _n+K·E

Further, λ=0.8, Δ sequence initial value is:

Further, σ m=0.1.

Beneficial effect of the present invention is as follows:

The present invention restrains fast to the first bag The data RLS algorithm that receives, after reaching convergence, adopt tracking accuracy height, the strong LMS algorithm of stability to follow the tracks of, so just combine advantage between the two, not only tracking velocity is fast, and the convergence precision height, keep less steady-state error and tracking performance preferably, compare the RLS algorithm and also reduced computation complexity.

Description of drawings

Fig. 1 is the flow chart of the method for channel tracking in a kind of single-carrier frequency domain equalization system of the present invention;

Fig. 2 is the structural representation of frame data.

Embodiment

Below in conjunction with accompanying drawing and embodiment, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, does not limit the present invention.

Aspect channel estimating, for changing channel faster, all wish to find a kind of fast convergence rate, and when convergence steady-state error is little, tracking performance is high tracking.Follow the tracks of and mainly be divided into two stages, sequencing is converged state and steady-state process, because the stronger time-varying characteristics of short wave communication, require channel estimation method that fast convergence rate is arranged, therefore need RLS algorithm, but consider the tracking performance of the excellence that the LMS algorithm shows in the meaning of mean square deviation and imbalance, the present invention proposes and adopt RLS allow the estimator convergence earlier with fast convergence rate, use the tracking of LMS track channel change again, i.e. RLS and LMS unified algorithm.

As shown in Figure 2, the SC-FDE technology sends signal with the symbolic blocks structure, need insert the Cyclic Prefix line frequency territory equilibrium of going forward side by side in data block.Wherein frame data are become with several data packet group by 1 leader block.

Present embodiment is at above-mentioned Frame, relates to the method for channel tracking in a kind of single-carrier frequency domain equalization system of the time-varying characteristics at short wave channel, may further comprise the steps:

Step S101 according to the leader block sequence in the Frame, obtains initial frequency domain equalization coefficient W ₀=W ₁=1/H.

The leader block sequence is the UW sequence, u=[u (1) ..., u (M)], its Fourier changing value is: U _Fft=[U (1) ... U (M)].The UW sequence: unique word (Unique Word) sequence is generally chu sequence, frank-zadaff sequence, the PN sequence of IEEE802.16a standard code.

According to the leader block sequence in the Frame, obtain frequency domain equalization coefficient W ₀=W ₁=1/H specifically comprises:

The reception data of the leader block correspondence that receiving terminal receives are r=[r (1) ... r (M)], vectorial r is carried out Fourier transform, obtain R _Fft:

R _fft＝[R(1)，...，R(M)]＝FFT(r)；

According to R _Fft, U _Fft, obtain channel estimating frequency domain response H:

$H=[H\left(1\right),...,H\left(M\right)]=R_{\mathrm{fft}}/U_{\mathrm{fft}}=\underset{n=1}{\overset{M}{\mathrm{\Σ}}}R\left(n\right)/U\left(n\right);$

According to H, obtain initial frequency domain equalization coefficient W ₀, W ₁:

$W_0=W_1=1/H=\frac{H^{*}}{{\left|H\right|}^2};$

Wherein, H ^*Conjugation for H; | H| is for to ask mould to vectorial H.

Step S102 receives the first bag data r ₁The time, adopt recursive least squares (RLS) to carry out channel tracking.Adopt the RLS algorithm to restrain fast earlier, can reach the fast advantage of tracking velocity.

Step S103 is when receiving n (wherein, n is the numbering of packet, and n is more than or equal to 1) bag data r _nThe time, according to W ₀, W ₁, judge frequency domain equalization coefficient W _nWith W _N-1The variances sigma (W of difference _n-W _N-1) whether less than the threshold value σ m that sets in advance, if then change step S104; Otherwise, then change step S105.

$\mathrm{\σ}(W_n-W_{n-1})=\frac{1}{M-1}\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{(w_n\left(i\right)-w_{n-1}\left(i\right))}^2$

Wherein, W _nBe current frequency domain equalization coefficient, W _N-1Frequency domain equalization coefficient when upgrading for the last time, frequency domain equalization coefficient are vectors, W _n=[w _n(1) ..., w _n(M)], i=1,2 ..., M.Variance yields is big more to show that the variation of twice equalizing coefficient is big more,

The channel tracking function is to W _nFollow the tracks of, and bring in constant renewal in, because the RLS fast convergence rate is worked as W _nAfter the convergence, the variation of the W value before and after upgrading will be very little rapidly, therefore by the W before and after relatively upgrading _nWhether the variation of value is judged needs to change track algorithm.And W before and after upgrading _nThe variance of difference can accurately indicate W _nThe variation of value.General σ m value is σ m=0.1.

Calculate frequency domain equalization coefficient W _N+1The time, need error e _n, therefore, at first, need be according to W _nAnd r _n, obtain error e _n, specifically may further comprise the steps:

To r _nDo fast Fourier transform and obtain sequence R _n:

R _n＝FFT(r _n)；

According to R _n, obtain balanced output sequence Z _n:

Z _n＝W _n·R _n；

To balanced output sequence Z _nCarry out inverse fast Fourier transform, obtain time domain output sequence z _n:

z _n＝IFFT(Z _n)；

To time domain output sequence z _nDecode, obtain adjudicating sequence d _n:

d _n＝Dec[z _n]；

According to judgement sequence d _nWith time domain output sequence z _n, obtain error sequence e _n:

e _n＝d _n-z _n。

Step S104 then obtains frequency domain equalization coefficient W according to lowest mean square (LMS) algorithm _N+1, carry out channel tracking, specifically may further comprise the steps:

To error sequence e _nDo fast Fourier transform, obtain E=FFT (e _n);

According to E, obtain frequency domain equalization coefficient W _N+1:

W _n+1＝W _n+μE·R _n

Wherein, μ is the gain constant that is used for controlling adaptive speed and stability.Present embodiment, μ=0.03.

Step S105 obtains frequency domain equalization coefficient W according to recursive least squares _N+1, carry out channel tracking; Specifically may further comprise the steps:

To error sequence e _nDo fast Fourier transform, obtain sequence E=FFT (e _n);

Calculate kalman gain vector K:

$K=\frac{\mathrm{\Δ}\·{R_n}^{*}}{\mathrm{\λ}+\mathrm{\Δ}\·{\left|R_n\right|}^2};$

Wherein, Δ is a correlated series, and λ is the gain constant that is used for controlling adaptive speed and stability, R ^*Conjugation for R;

Upgrade the Δ sequence, Δ carried out iterative computation:

$\mathrm{\Δ}=\frac{\mathrm{\Δ}}{\mathrm{\λ}+\mathrm{\Δ}\·{\left|R_n\right|}^2};$

According to W _n, K, E, obtain frequency domain equalization coefficient W _N+1:

W _n+1＝W _n+K·E。

In the present embodiment, λ=0.8, λ is big more, and convergence rate is fast more, but unstable more, easy more imbalance.Δ sequence initial value is:

In the said method, RLS algorithm and LMS algorithm are prior aries, are basic adaptive iteration algorithms, and the present invention mainly is at above-mentioned algorithm, chooses suitable parameters, and then obtains more excellent effect.

Owing to can calculate W earlier according to leader block ₀, W ₁, therefore, when n=1,, can obtain W according to recursive least squares ₂, and, can calculate e ₂When n=2, promptly at first need to judge W ₂With W ₁The variances sigma (W of difference ₂-W ₁) whether less than the threshold value σ m that sets in advance, if then obtain frequency domain equalization coefficient W according to least mean square algorithm ₃, otherwise, then obtain frequency domain equalization coefficient W according to recursive least squares ₃, and the like, obtain the frequency domain equalization coefficient W when receiving each packet _n

RLS algorithm the convergence speed piece, but tracking accuracy is low, poor stability, and the LMS algorithm the convergence speed is slow, but the tracking accuracy height, stability is high.Restrain fast with the RLS algorithm earlier, follow the tracks of reaching convergence laggard line trace precision high stability strong LMS algorithm, the advantage that has so just possessed LMS and RLS simultaneously, when guaranteeing the channel estimating fast convergence rate, channel estimated accuracy be can also guarantee, less steady-state error and tracking performance preferably kept.Compare the RLS algorithm and also reduced computation complexity.And judge that the criterion whether the RLS algorithm restrains is exactly to utilize the variance of the difference value of equalizing coefficient to judge.If the variance of the difference value of adjacent twice equalizing coefficient is big, then show to be in quick converged state, when the difference value of adjacent twice equalizing coefficient is little, then show to be in steady-state process.

The method also is applicable to the channel that time-varying characteristics are stronger not only at short wave channel.

Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.

Claims (10)

1. the method for channel tracking in the single-carrier frequency domain equalization system is characterized in that, said method comprising the steps of:

According to the leader block sequence in the Frame, obtain initial frequency domain equalization coefficient W ₀=W ₁=1/H;

Receive the first bag data r ₁The time, adopt recursive least squares to carry out channel tracking;

When receiving n bag data r _nThe time, according to W ₀, W ₁, judge frequency domain equalization coefficient W _nWith W _N-1The variances sigma (W of difference _n-W _N-1) whether less than the threshold value σ m that sets in advance, if then obtain frequency domain equalization coefficient W according to least mean square algorithm _N+1, carry out channel tracking; Otherwise, obtain frequency domain equalization coefficient W according to recursive least squares _N+1, carry out channel tracking; Wherein, n is the numbering of packet, and n is more than or equal to 1.

2. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 1, and described leader block sequence is the UW sequence, u=[u (1) ..., u (M)], its Fourier changing value is: U _Fft=[U (1) ... U (M)].

3. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 2, according to the leader block sequence in the Frame, obtains frequency domain equalization coefficient W ₀=W ₁=1/H specifically comprises:

The reception data of the leader block correspondence that receiving terminal receives are r=[r (1) ... r (M)], vectorial r is carried out Fourier transform, obtain R _Fft:

R _fft＝[R(1)，...，R(M)]＝FFT(r)；

According to R _Fft, U _Fft, obtain channel estimating frequency domain response H:

$H=[H\left(1\right),...,H\left(M\right)]=R_{\mathrm{fft}}/U_{\mathrm{fft}}=\underset{n=1}{\overset{M}{\mathrm{\Σ}}}R\left(n\right)/U\left(n\right);$

According to H, obtain initial frequency domain equalization coefficient W ₀, W ₁:

$W_0=W_1=1/H=\frac{H^{*}}{{\left|H\right|}^2};$

Wherein, H ^*Conjugation for H; | H| is for to ask mould to vectorial H.

4. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 1, when receiving n bag data r _nThe time, at first, according to W _nAnd r _n, obtain error e _n

5. the method for channel tracking is characterized in that, according to W in the single-carrier frequency domain equalization system as claimed in claim 4 _nAnd r _n, obtain error e _n, specifically may further comprise the steps:

To r _nDo fast Fourier transform and obtain sequence R _n:

R _n＝FFT(r _n)；

According to R _n, obtain balanced output sequence Z _n:

Z _n＝W _n·R _n；

To balanced output sequence Z _nCarry out inverse fast Fourier transform, obtain time domain output sequence z _n:

z _n＝IFFT(Z _n)；

To time domain output sequence z _nDecode, obtain adjudicating sequence d _n:

d _n＝Dec[z _n]；

According to judgement sequence d _nWith time domain output sequence z _n, obtain error sequence e _n:

e _n＝d _n-z _n。

6. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 5, obtains frequency domain equalization coefficient W according to least mean square algorithm _N+1, specifically may further comprise the steps:

To error sequence e _nDo fast Fourier transform, obtain E=FFT (e _n);

According to E, obtain frequency domain equalization coefficient W _N+1:

W _n+1＝W _n+μE·R _n

Wherein, μ is the gain constant that is used for controlling adaptive speed and stability.

7. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 6, μ=0.03.

8. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 5, obtains frequency domain equalization coefficient W according to recursive least squares _N+1, specifically may further comprise the steps:

To error sequence e _nDo fast Fourier transform, obtain sequence E=FFT (e _n);

Calculate kalman gain vector K:

$K=\frac{\mathrm{\Δ}\·{R_n}^{*}}{\mathrm{\λ}+\mathrm{\Δ}\·{\left|R_n\right|}^2};$

Wherein, Δ is a correlated series, and λ is the gain constant that is used for controlling adaptive speed and stability, R ^*Conjugation for R;

Upgrade the Δ sequence, Δ carried out iterative computation:

$\mathrm{\Δ}=\frac{\mathrm{\Δ}}{\mathrm{\λ}+\mathrm{\Δ}\·{\left|R_n\right|}^2};$

According to W _n, K, E, obtain frequency domain equalization coefficient W _N+1:

W _n+1＝W _n+K·E。

9. the method for channel tracking is characterized in that in the single-carrier frequency domain equalization system as claimed in claim 8, λ=0.8, and Δ sequence initial value is:

10. as the method for channel tracking in each described single-carrier frequency domain equalization system of claim 1～9, it is characterized in that σ m=0.1.

Images