CN101184069A - Single carrier frequency domain equalization based selection feedback detecting method - Google Patents

Abstract

The invention provides a selective feedback detecting method based on the single-carrier wave frequency domain equalization, in particular to a mobile communication system for transferring high-speed data via a plurality of sending antennas and receiving antennas, belonging to the technical field of single-carrier wave frequency domain equalization (SC-FDE) used for alleviating intersymbol interference in broadband system. The invention is characterized in that: on the receiving end of the multi-antenna transmission system or the receiving machine of the multi-antenna sending multi-antenna receiving single-carrier wave transmission system, a time domain feedback filter can be arranged on the tap feedback position minimizing the mean minimum mean square error by obtaining the mean minimum mean square errors in all feasible feedback modes, according to certain channel mode and time domain feedback order, and based on the rule of minimum mean square error; the tap coefficients in frequency domain and time domain can be obtained based on the rule of minimum mean square error to the detecting signals, and then the filter formed by the tap coefficients is used to detect the symbolic information of the current data block. The invention can be divided into a non-adaptive feedback detecting method and an adaptive feedback detecting method according to whether the structure of the time domain filter is stationary or not.

Description

Selection feedback detecting method based on single carrier frequency domain equalization

Technical field

The present invention relates to a kind of by using a plurality of transmitting antennas/accept mobile communication system that antenna comes transmitting high speed data, and relate to single carrier frequency domain equalization (SC-FDE) technology of alleviating intersymbol interference problem in the broadband system, relate in particular to a kind of signal detecting method that is used for multi-aerial radio communication system.Related wireless communication system has one or above reception antenna and one or above transmitting antenna.

Background technology

Information theory studies show that: enrich in existence under the wireless channel conditions of scattering, multiple-input and multiple-output (MIMO) system can obtain to receive the higher capacity of system than single-shot list, at transmitting terminal, different transmitting antennas sends different symbols streams, at receiving terminal, recover the transmission symbol according to the symbol of accepting that is formed by stacking by transmission symbols streams and noise by suitable detection method.

With respect to OFDM (OFDM) technology, single carrier frequency domain equalization (SC-FDE) technology has low peak-to-average force ratio and to the insensitive characteristics of frequency deviation.Simultaneously, under the identical condition of detection complexity, SC-FDE can reach identical performance with OFDM.Based on the single carrier frequency domain linear equalization (FD-LE) of least mean-square error (MMSE) criterion, the only detection that achieves a butt joint and collect mail number by linear equalization at frequency domain, performance is relatively poor relatively.Frequency domain equalization time domain decision-feedback (FD-DFE) detects, and by the decision-feedback to time domain, can further eliminate intersymbol interference.Performance is excellent than FD-LE.But under the MMSE criterion, the frequency domain tap coefficient of FD-DFE depends on the time domain tap coefficient, thereby causes computation complexity higher.By contrast, based on the noise prediction of frequency domain equalization (FDE-NP) technology of noise prediction by time domain, reach with the identical performance of the FD-DFE of exponent number, simultaneously based on the MMSE criterion, the FDE-NP structure can realize that the independent of frequency domain and time domain tap coefficient calculates, and complexity obviously reduces.

Summary of the invention

Technical problem: based on the noise prediction model, the present invention proposes a kind of selection feedback detecting method based on single carrier frequency domain equalization, the basic thought of this method is to select best time domain feedback signal to predict the interference noise of current symbol to be detected, by calculating with the similar tap coefficient of traditional F DE-NP and balanced, under with the complexity situation, can significantly improve performance.

Technical scheme: the selection feedback detecting method based on single carrier frequency domain equalization of the present invention is in many antennas single carrier block transmission (MIMO-SCBT) receiver, at receiving terminal, according to channel model of determining and time domain feedback exponent number, based on minimum mean square error criterion, can be by obtaining the average least mean-square error under various feasible feedback systems, select to make the tap feedback position of average least mean-square error minimum to construct to be and feedback filter; Then based on minimum mean square error criterion to detected symbol, can obtain the tap coefficient of frequency domain and time domain, detect the symbolic information of current data block with the filter of this tap coefficient formation, whether the structure by time domain filtering is fixing, selection feedback detecting method based on single carrier frequency domain equalization can be divided into non-self-adapting selection detection feedback and self adaptation feedback detecting method, and concrete detection step is as follows:

Non-self-adapting is selected feedback detecting method:

Step 1: concrete communication channel is set up channel model, and analyze the statistical property of its frequency: f (H _k), (k=0 ..., N-1) wherein N is each data block length,

Step 2:, under the MMSE criterion,, can obtain the MMSE expression formula under various feasible feedback systems for given feedback exponent number based on the FDE-NP model.Its MMSE value is with the impulse response value H of each frequency _k(k=0 ..., N-1) change,

Step 3: respectively each MMSE is asked about H _k, (k=0 ..., expectation E{MMSE} N-1) selects to make the tap feedback position of this E{MMSE} minimum design time domain filtering,

Step 4: for each transmission data block,, under the MMSE criterion, optimize frequency domain and time domain tap coefficient according to the FDE-NP structure of design,

Step 5: this reception data block is carried out fast Fourier transform (FFT), behind the frequency domain linear equalization, carry out inverse fast Fourier transform (IFFT) and turn back to time domain,

Step 6: to above-mentioned symbol through frequency domain equalization, further eliminate interference, and judgement obtains final detection signal by the time domain noise prediction,

Detect step 1-3 and when the design time domain filtering, just finish, execution in step 4-6 only in testing process, therefore, detection complexity is identical than traditional F DE-NP,

Detect in the step 2, can be according to postponing, wherein optimum feedback system is chosen in the qualification of feedback arrangement in the limited feedback mode.

Adaptively selected feedback detecting method:

Step 1:, under the MMSE criterion,, can obtain the MMSE expression formula under various feasible feedback systems for given feedback exponent number based on the FDE-NP model.Its MMSE value is with the impulse response value H of each frequency _k, (k=0 ..., N-1) change,

Step 2: for each transmission data block, obtain the MMSE value under the various feedback systems respectively, select to make the feedback system adjustment time domain filtering of this MMSE minimum to detect current data block,

Step 3: according to the FDE-NP structure of design, under the MMSE criterion, optimize frequency domain and time domain tap coefficient,

Step 4: this reception data block is carried out FFT, behind the frequency domain linear equalization, carry out IFFT and turn back to time domain,

Step 5:, further eliminate interference by the time domain noise prediction to above-mentioned symbol through frequency domain equalization.And judgement obtains final detection signal.

Detecting step 2-5 all will be performed for each data to be tested piece.Consider the complexity of execution, in step 1, can select some to make the less relatively feedback system of E{MMSE} as be used for adjusting the alternative feedback system of time domain filtering at every turn.

Beneficial effect: what the present invention proposed selects the detection feedback algorithm to have following advantage based on single carrier frequency domain equalization:

1, compare with traditional F DE-NP, non-self-adapting selects the detection feedback algorithm to select the effective feedback position to carry out noise prediction, has reduced the average MMSE of whole communication process.

2, compare with traditional F DE-NP, under the situation of identical detection complexity, non-self-adapting selects the detection feedback algorithm can realize the detection performance gain.

3, compare with traditional F DE-NP, for each transmission data block, adaptively selected detection feedback algorithm all selects optimum feedback position to detect this data block, and average MMSE can reach minimum.

The detection method that the present invention proposes is applicable to

1, various single-carrier frequency-domain feedback equalizations.As: FD-DFE, FDE-NP, FDE-NP-SIC

2, MIMO-SCBT system.

Description of drawings

Fig. 1 is a traditional F DE-NP system block diagram.

Fig. 2 is that the non-self-adapting that the present invention proposes is selected the detection feedback system block diagram.

Fig. 3 is the adaptively selected detection feedback system block diagram that the present invention proposes.

Embodiment

Basic thought based on the selection feedback detecting method of single carrier frequency domain equalization is: according to channel model of determining and time domain feedback exponent number, by calculating and the average MMSE under various feasible feedback systems relatively, select to make the tap feedback position of average MMSE minimum to construct to be and feedback filter.Optimize the tap coefficient of frequency domain and time domain then according to the MMSE criterion, detect the symbolic information of current data block with the filter of this tap coefficient formation.Whether fixing by the structure of time domain filtering, non-self-adapting be can be divided into based on the selection feedback detecting method of single carrier frequency domain equalization and detection feedback and self adaptation feedback detecting method selected.

Here supposing the system has N _TRoot transmitting antenna, N _RThe root reception antenna.The source bits ovennodulation of flowing through is distributed to n _TThe root transmitting antenna sends with regular length N piecemeal, and inserts the circulation embedding for each data block and sew (CP), and at receiving terminal, behind the removal CP, acknowledge(ment) signal can be expressed as

$y_n=\underset{m=0}{\overset{N-1}{\mathrm{\Σ}}}{h}_m{x}_{(n-m)\mathrm{mod}N}+v_n,$ n＝0，1，...，N-1

Wherein, receiving symbol vector $y_n={[y_{n,1},\·\·\·y_{n,N_R}]}^T,$ y _{N, j}Be the received signal of j root reception antenna at moment n.For sending symbolic vector $x_n={[x_{n,1},\·\·\·x_{n,N_T}]}^T,$ x _{N, i}Be the transmission signal of i root transmitting antenna at moment n.

h _{N, ij}Be i root transmitting antenna to the impulse response of the channel of j root reception antenna at moment n.v _nCorrespond to N _RAdditive white Gaussian noise vector on the root reception antenna.And establishing noise power is σ _v ², average transmitting power is σ _x ²Corresponding frequency-domain expression is:

Y _k＝H _kX _k+V _k， k＝0，1，...，N-1

Wherein, H _kBe block diagonal matrix, each divides block size is N _R* N _T, and:

$H_{k,\mathrm{pq}}=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{h}_{n,\mathrm{pq}}{e}^{-j(2\mathrm{\π}/N)\mathrm{nk}}$

If time domain feedback exponent number is B, based on the MMSE criterion, the MMSE under the various feedback systems can be expressed as:

$\mathrm{MMSE}=\frac{{\mathrm{\σ}}_x^2{\mathrm{\σ}}_v^2}{N}\mathrm{tr}\{q_0-\underset{l=1}{\overset{B}{\mathrm{\Σ}}}{c}_l{q}_{-k_l}\}$

Wherein: $q_l=\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{e}^{j(2\mathrm{\π}/N)/k}{({\mathrm{\σ}}_x^2{H}_k^H{H}_k+{\mathrm{\σ}}_v^2{I}_{N_T})}^{-1},$ k ₁..., k _BBe illustrated under the various feedback systems position with respect to symbol to be detected of feedback tap.c ₁..., c _BExpression is with respect to k ₁..., k _BEach feedback tap on optimal coefficient, and

For given channel model, can obtain the average MMSE under various feedback systems:

$E\left\{\mathrm{MMSE}\right\}=\∫\·\·\·\∫\mathrm{MMSE}(q_0,\·\·\·q_{k_1-k_B})f\left(H_0\right)\·\·\·f\left(H_{N-1}\right)d{H}_0\·\·\·d{H}_{N-1}$

Average MMSE under the more various feedback systems just can select to make its minimum feedback system to design time domain feedback filter in the FDE-NP model.At last, based on the MMSE criterion, detect received signal with the FDE-NP model of this design.

Non-self-adapting selects the detection feedback flow process as follows:

1. concrete communication channel is set up channel model, and analyze the statistical property of its frequency: f (H _k), (k=0 ..., N-1).

2. based on the FDE-NP model, under the MMSE criterion, the feedback exponent number B for given can obtain the MMSE expression formula under various feasible feedback systems:

$\mathrm{MMSE}=\frac{{\mathrm{\σ}}_x^2{\mathrm{\σ}}_v^2}{N}\mathrm{tr}\{q_0-\underset{l=1}{\overset{B}{\mathrm{\Σ}}}{c}_l{q}_{-k_l}\}$

Its MMSE value is with the impulse response value H of each frequency _k, (k=0 ..., N-1) change.

3. respectively each MMSE is asked about H _k, (k=0 ..., expectation E{MMSE} N-1):

$E\left\{\mathrm{MMSE}\right\}=\∫\·\·\·\∫\mathrm{MMSE}(q_0,\·\·\·q_{k_1-k_B})f\left(H_0\right)\·\·\·f\left(H_{N-1}\right)d{H}_0\·\·\·d{H}_{N-1}$

Selection makes the tap feedback position of this E{MMSE} minimum

Design time domain filtering.

$({\hat{k}}_1,{\hat{k}}_2,\·\·\·{\hat{k}}_B)=\arg \left\{\underset{\mathrm{and}{k}_1\≠k_2\≠\·\·\·\≠k_B}{\underset{\{k_1,\·\·\·{,k}_B\}\∈\{1,\·\·\·,N-1\}}{\min }}E\left\{\mathrm{MMSE}\right\}\right\}$

4. for each transmission data block,, under the MMSE criterion, optimize frequency domain tap coefficient W according to the FDE-NP structure of design _k, (k=0 ..., N-1) with time domain tap coefficient c _l, (l=0 ..., B-1).

$W_k={\mathrm{\σ}}_x^2{H}_k^H{({\mathrm{\σ}}_x^2{H}_k{H}_k^H+{\mathrm{\σ}}_v^2{I}_{N_R})}^{-1}$

5. this reception data block is carried out FFT, behind the frequency domain linear equalization, carry out IFFT and turn back to time domain.Obtain time-domain signal a _n, (n=0 ..., N-1)

$a_n=\frac{1}{\sqrt{N}}\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{W}_k{Y}_k{e}^{j(2\mathrm{\π}/N)\mathrm{kn}}$

6. to above-mentioned symbol, further eliminate interference by the time domain noise prediction through frequency domain equalization.And judgement obtains final detection signal

(n=0 ..., N-1).

${\tilde{x}}_n=\mathrm{dec}(a_n-\underset{l=1}{\overset{B}{\mathrm{\Σ}}}{c}_l{d}_{(n-{\hat{k}}_l)\mathrm{mod}N})$

Wherein $d_{n-l}=a_{n-l}-{\tilde{x}}_{n-l}.$

Adaptively selected detection feedback algorithm flow is as follows:

1. based on the FDE-NP model, under the MMSE criterion, the feedback exponent number B for given can obtain the MMSE expression formula under various feasible feedback systems:

$\mathrm{MMSE}=\frac{{\mathrm{\σ}}_x^2{\mathrm{\σ}}_v^2}{N}\mathrm{tr}\{q_0-\underset{l=1}{\overset{B}{\mathrm{\Σ}}}{c}_l{q}_{-k_l}\}$

Its MMSE value is with the impulse response value H of each frequency _k, (k=0 ..., N-1) change.

2. for each transmission data block H _k, (k=0 ..., N-1) can obtain by channel estimating, obtain the MMSE value under the various feedback systems respectively, select to make the tap feedback position of this MMSE minimum

Adjust time domain filtering.

3. according to the FDE-NP structure of design, under the MMSE criterion, optimize frequency domain and time domain tap coefficient.Get frequency domain tap coefficient W _k, (k=0 ..., N-1) with time domain tap coefficient c _l, (l=0 ..., B-1):

$W_k={\mathrm{\σ}}_x^2{H}_k^H{({\mathrm{\σ}}_x^2{H}_k{H}_k^H+{\mathrm{\σ}}_v^2{I}_{N_R})}^{-1}$

4. this reception data block is carried out FFT, behind the frequency domain linear equalization, carry out IFFT and turn back to time domain.Get time-domain signal a _n, (n=0 ..., N-1)

$a_n=\frac{1}{\sqrt{N}}\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}{W}_k{Y}_k{e}^{j(2\mathrm{\π}/N)\mathrm{kn}}$

5. to above-mentioned symbol, further eliminate interference by the time domain noise prediction through frequency domain equalization.And judgement obtains final detection signal

(n=0 ..., N-1).

${\tilde{x}}_n=\mathrm{dec}(a_n-\underset{l=1}{\overset{B}{\mathrm{\Σ}}}{c}_l{d}_{(n-{\hat{k}}_l)\mathrm{mod}N})$

Wherein $d_{n-l}=a_{n-l}-{\tilde{x}}_{n-l}.$

Claims (3)

1. selection feedback detecting method based on single carrier frequency domain equalization, it is characterized in that this detection method is in the single carrier transmitting system receiver of multi-aerial transmission system or the many antennas receptions of many antenna transmission, at receiving terminal, according to channel model of determining and time domain feedback exponent number, based on minimum mean square error criterion, can be by obtaining the average least mean-square error under various feasible feedback systems, select to make the tap feedback position of average least mean-square error minimum to construct to be and feedback filter; Then based on minimum mean square error criterion to detected symbol, can obtain the tap coefficient of frequency domain and time domain, detect the symbolic information of current data block with the filter of this tap coefficient formation, whether fixing by the structure of time domain filtering, non-self-adapting be can be divided into based on the selection feedback detecting method of single carrier frequency domain equalization and detection feedback and self adaptation feedback detecting method selected.

2. the selection feedback detecting method based on single carrier frequency domain equalization according to claim 1 is characterized in that the concrete steps of non-self-adapting selection detection feedback are as follows:

Step 1: concrete communication channel is set up channel model, and analyze the statistical property f (H of its frequency _k), wherein: k=0,1 ..., N-1; N represents the length of each data block,

Step 2: based on noise prediction single carrier frequency domain equalization model, under minimum mean square error criterion, for given feedback exponent number, can obtain the least mean-square error expression formula under various feasible feedback systems, its least mean-square error value is with the impulse response value H of each frequency _kChange,

Step 3: respectively each least mean-square error is asked about impulse response value H _kExpectation E{MMSE}, select to make the tap feedback position of this E{MMSE} minimum design time domain filtering,

Step 4: for each transmission data block, the noise prediction single carrier frequency domain equalization structure according to design under the minimum mean square error criterion of symbol detection, obtains frequency domain and time domain tap coefficient,

Step 5: this reception data block is carried out fast Fourier transform, behind the frequency domain linear equalization that is undertaken by the above-mentioned frequency domain tap coefficient that obtains, carry out Fast Fourier Transform Inverse and turn back to time domain,

Step 6:, carry out the prediction of time domain feedback noise by the above-mentioned time domain tap coefficient that obtains, and the symbol that obtains is sent into decision device make judgement to above-mentioned symbol through frequency domain equalization.

3. the selection feedback detecting method based on single carrier frequency domain equalization according to claim 1 is characterized in that the concrete steps of self adaptation feedback detecting method are as follows:

Step 1: based on noise prediction single carrier frequency domain equalization model, under minimum mean square error criterion, for given feedback exponent number, can obtain the least mean-square error expression formula under various feasible feedback systems, its least mean-square error value is with the impulse response value H of each frequency _kChange,

Step 2: for each transmission data block, obtain the least mean-square error value under the various feedback systems respectively, select to make the feedback system adjustment time domain filtering of this least mean-square error minimum to detect current data block,

Step 3: the noise prediction single carrier frequency domain equalization structure according to design, based on minimum mean square error criterion, obtain frequency domain and time domain tap coefficient to detected symbol,

Step 4: this reception data block is carried out fast Fourier transform, behind the frequency domain linear equalization that is undertaken by the above-mentioned frequency domain tap coefficient that obtains, carry out Fast Fourier Transform Inverse and turn back to time domain,

Step 5:, carry out the prediction of time domain feedback noise by the above-mentioned time domain tap coefficient that obtains, and the symbol that obtains is sent into decision device make judgement to above-mentioned symbol through frequency domain equalization.

Images