CN1595824A - A combined detection method and apparatus with incorporated frequency deviation compensation - Google Patents

Abstract

The invention is a combined detecting method with frequency offset compensation, including: modifying a system model, supposing Delta f is a frequency offset between a receiving part and a transmitting part and Ts is a symbol space, and because of frequency offset existence, correcting a data model by e=(e1, e2, e3, ..., eN*Q+W-1)T=AFd+n, where A is a system transfer matrix, d is a symbol sequence of original data, e is a sequence of received data, n is an additive noise and F shows the influence of frequency offset to the original data. It makes the system model nearer the actual situation, and has high combined detection accuracy. Besides, it makes preceding frequency offset compensation in a data detecting stage, thus further improving the accuracy and the receiving property.

Description

A kind of associated detecting method of combining compensate of frequency deviation and device

Technical field

The present invention relates to a kind of method and apparatus that improves combined detection performance of mobile communication system, relate in particular to the associated detecting method and the device of combining compensate of frequency deviation of the novelty of TD-SCDMA mobile communication system.

Background technology

The TD-SCDMA system employs associated detection technique, please refer to " the Linear unbiased data estimation in mobile radio systemsapplying CDMA " that A.Klein delivered on the IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS in 1993 about joint detection algorithm.In order to set forth needs of the present invention, now simply introduce existing joint detection algorithm:

At first existing joint-detection specified data model is:

$e={(e_1,e_2,e_3,...,e_{N*Q+W-1})}^T=\mathrm{Ad}+n---\left(1\right)$

A is a system transmission matrix, and d is the raw data symbols sequence, and e is a receiving data sequence, and n is an additive noise.The covariance matrix of data d is:

$R_d=E[d^H\·d]={\mathrm{\σ}}_d^2\·I---\left(2\right)$

The covariance matrix of noise n is:

$R_n=E[n^H\·n]={\mathrm{\σ}}_n^2\·I---\left(3\right)$

The structure of system transmission matrix A has been described as shown in Figure 1.V wherein _{I, 1}＜i＜N is called the aggregate channel response block, is generated by spreading code c and channel impulse response h convolution, and Q is a spreading code length, and W is long for the channel impulse response window, and N is an equalization block length.

Be the structure chart of joint-detection ZF-BLE method as shown in Figure 2, by a matched filter 210, prewhitening filter 220, ISI and MAI arrester 230 are formed.The ZF-BLE method is according to MLSE (Maximum likelihood sequence) criterion, minimizes following quadratic equation:

${(e-A{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}})}^H{R}_n^{-1}(e-A{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}})----\left(4\right)$

The data estimation that last ZF-BLE obtains is:

${\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}={\left(A^H{R}_n^{-1}A\right)}^{-1}{A}^H{R}_n^{-1}e---\left(5\right)$

$=d+{(\mathrm{\Σ}\·\mathrm{\Γ})}^{-1}{({\mathrm{\Γ}}^H\·\mathrm{\Σ})}^{-1}{A}^H{R}_n^{-1}n$

Be joint-detection MMSE-BLE structure chart as shown in Figure 3, by a ZF-BLE detector 200, wiener filter 210 is formed, and the MMSE-BLE method is to minimize following equation according to MMSE (least mean-square error) criterion

$E\left({({\hat{d}}_{\mathrm{MMSE}-\mathrm{BLE}}-d)}^H({\hat{d}}_{\mathrm{MMSE}-\mathrm{BLE}}-d)\right)---\left(6\right)$

MMSE-BLE obtains data estimation:

${\hat{d}}_{\mathrm{MMSE}-\mathrm{BLE}}={(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}e$

$={(I+{\left(R_d{A}^H{R}_n^{-1}A\right)}^{-1})}^{-1}\·{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}\underset{\‾}{\underset{\‾}{\mathrm{def}}}{W}_0\·{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}---\left(7\right)$

$=\mathrm{diag}\left(W_0\right)\·d+\stackrel{\‾}{\mathrm{diag}}\left(W_0\right)\·\underset{\‾}{d}+{(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}n$

This shows system transmission matrix A decisive role in joint detection algorithm, revise the A matrix construction and make its more realistic transmission environment will improve the joint-detection effect greatly.From the building method of existing systems transmission matrix A as can be seen, be in the equalization block of N a length, each aggregate channel response block V _i, 1＜i＜N, corresponding data symbol, they all are to be generated by spreading code c and channel impulse response h convolution, thus in whole equalization block N aggregate channel response block V _iGet same value.

It is static that above-mentioned art methods goes for travelling carriage, channel fading changes situation slowly, but because mobile telecommunication channel often has the characteristic of rapid fading, in an equalization block length N, actual aggregate channel response often has bigger variation, and at this moment the handling property of existing method will descend.

The important factor that mobile telecommunication channel has the rapid fading characteristic is exactly that transmit leg and recipient exist frequency shift (FS).In mobile communication system, transmit leg is modulated to carrier wave fc to information, and the recipient will accurately receive information this carrier wave of must regenerating.Nominally though this carrier wave transmit leg and recipient are known in advance, have two main causes to make transmit leg and recipient exist frequency shift (FS): 1. travelling carriage local clock precision is not high; 2. location of mobile station moves the Doppler frequency shift that brings.Want to obtain reliable and stable receptivity, receiving terminal must be estimated and compensating for frequency offset effectively.The compensation method of existing TD-SCDMA system frequency is: the data after the demodulation are modulated again, reappear modulation signal at receiving terminal, calculate then and the phase difference between the received signal accordingly, estimate thus and compensating for frequency offset.This method is to carry out compensate of frequency deviation later in data demodulates, but this method simulation truth is poor, and accuracy of detection is not high.

Summary of the invention

Purpose of the present invention and technical problem to be solved are: system transmission matrix A does not have the fine truth of approaching in the solution prior art, changing the low problem of precision under the channel circumstance that has frequency deviation at a high speed, a kind of associated detection technique of combining compensate of frequency deviation of novelty is provided, update the system transmission matrix A, make system model more approach truth, the joint-detection precision is higher.Simultaneously carry out the compensate of frequency deviation in early stage, further improve precision, improve receptivity in the Data Detection stage.

Technical scheme of the present invention is as follows:

A kind of associated detecting method of combining compensate of frequency deviation, this method comprises: revise system model, establish Δ f and be the frequency deviation between recipient and the transmit leg, Ts is a mark space, because the existence of frequency deviation, data model is modified to:

$e={(e_1,e_2,e_3,....,e_{N*Q+W-1})}^T=\mathrm{AFd}+n---\left(8\right)$

Wherein, A is a system transmission matrix, and d is the raw data symbols sequence, and e is a receiving data sequence, and n is an additive noise; The F matrix has reflected the influence of frequency deviation to initial data.

Described associated detecting method, wherein, described F matrix is:

 ₀Be the phase deviation of first symbol correspondence in the equalization block.

Described associated detecting method, wherein, the key step of this method is as follows:

Step 1: get i=1,2 ..., N, i represent various combination channel response piece sequence number, repeat following steps 2 to step 4;

Step 2: determine modifying factor: When data block 1 is carried out joint-detection,

When data block 2 is carried out joint-detection, Δ f is the frequency deviation between recipient and the transmit leg, and Ts is a mark space;

Step 3: get k=(i-1) * Q+1 ..., i*Q+W-1, W is the subscriber channel window width,Repeating step 4;

Step 4: if v _k ⁱBe not equal to 0, $v_k^i=v_k^i*w_i;$

Step 5: by newly obtaining v _k ⁱStructure according to the original system transmission matrix generates new system transmission matrix A;

Step 6: using formula (5) carries out the ZF-BLE joint-detection:

${\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}={\left(A^H{R}_n^{-1}A\right)}^{-1}{A}^H{R}_n^{-1}e$

$=d+{(\mathrm{\Σ}\·\mathrm{\Γ})}^{-1}{({\mathrm{\Γ}}^H\·\mathrm{\Σ})}^{-1}{A}^H{R}_n^{-1}n---\left(5\right)$

Or using formula (7) carries out the MMSE-BLE joint-detection:

${\hat{d}}_{\mathrm{MMSE}-\mathrm{BLE}}={(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}e$

$={(I+{\left(R_d{A}^H{R}_n^{-1}A\right)}^{-1})}^{-1}\·{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}\underset{\‾}{\underset{\‾}{\mathrm{def}}}{W}_0\·{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}$

$=\mathrm{diag}\left(W_0\right)\·d+\stackrel{\‾}{\mathrm{diag}}\left(W_0\right)\·\underset{\‾}{d}+{(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}n---\left(7\right)$

A kind of joint-detection device of combining compensate of frequency deviation, wherein, it comprises channel estimator, A matrix maker, A matrix corrector, combined detector and frequency offset estimator;

The midamble sequence of input signal partly is sent to described channel estimator, utilizes deconvolution or fft/ifft method to produce channel impulse response h (t), and is sent to A matrix maker;

A matrix maker forms system transmission matrix A, wherein $V_i=[v_1^i\·\·\·v_{Q+W-1}^i]$ Be i the aggregate channel response block of system transmission matrix A, generate that the system transmission matrix A of generation is sent to described A matrix corrector by spreading code c and channel impulse response h convolution;

Described A matrix corrector utilizes frequency offset estimator to send here to such an extent that frequency offset estimating Δ f revises the A matrix, and revised A matrix is sent to combined detector;

Described combined detector uses ZF-BLE method or MMSE-BLE method to realize joint-detection, and the modulation symbol of generation is sent to frequency offset estimator;

The information that described frequency offset estimator utilizes modulation symbol to carry is estimated frequency deviation, and the frequency deviation Δ f of generation is sent to A matrix corrector.

The associated detecting method of a kind of combining compensate of frequency deviation provided by the present invention and device, update the system transmission matrix A makes system model more approach truth, and the joint-detection precision is higher.Simultaneously carry out the compensate of frequency deviation in early stage, further improve precision, improve receptivity in the Data Detection stage.

Description of drawings

Fig. 1 is the structure of prior art system transmission matrix A;

Fig. 2 is the joint-detection ZF-BLE structure of prior art;

Fig. 3 is the joint-detection MMSE-BLE structure of prior art;

Fig. 4 is the TD-SCDMA data burst structure of the inventive method;

Fig. 5 is that the joint-detection of the combining compensate of frequency deviation of the inventive method realizes block diagram;

Fig. 6 is that the system transmission matrix modification method of the inventive method is realized block diagram.

Embodiment

Be described in further detail below in conjunction with the enforcement of accompanying drawing technical scheme:

The method of the joint-detection of combining compensate of frequency deviation of the present invention is as follows:

At first revise system model, establish Δ f and be the frequency deviation between recipient and the transmit leg, Ts is a mark space, because the data model that exists of frequency deviation is modified to:

$e={({\mathrm{<figure-callout\; id="1"\; label="e"\; filenames="A200410027885.000121.png"\; state="\{\{state\}\}">e</figure-callout>}}_1,e_2,e_3,...,e_{N*Q+W-1})}^T=\mathrm{AFd}+n---\left(8\right)$

A is a system transmission matrix, and d is the raw data symbols sequence, and e is a receiving data sequence, and n is an additive noise.The structural form of system transmission matrix A still as shown in Figure 1, wherein $V_i=\left[\begin{array}{cc}{v}_1^i\&CenterDot;\&CenterDot;\&CenterDot;& v_{Q+W-1}^i\end{array}\right],$ 1≤i≤N is i the aggregate channel response block of system transmission matrix A, generates by spreading code c and channel impulse response h convolution, and, N is an equalization block length.The F matrix has reflected the influence of frequency deviation to initial data.

 0 is the phase deviation of first symbol correspondence in the equalization block, with reference to figure 4, has described TD-SCDMA concrete data burst structure, and data block 1, data block 2 all contain N*Q chip (being equivalent to N symbol), and the midamble sequence contains M*Q chip.Because the channel estimating of TD-SCDMA utilizes midamble partly to obtain, the mid point of midamble sequence can be regarded the phase reference point as, when so data block 1 being carried out joint-detection,

When data block 2 is carried out joint-detection,

By the data model of redetermination as can be known, replace original system transmission matrix A, more approach the real channel characteristic, will in joint-detection, bring performance gain with the product of matrix A, F.

The core procedure of the joint-detection A matrix correction of combining compensate of frequency deviation of the present invention is as follows, as shown in Figure 6:

Step 1: get i=1,2 ..., N, i represent various combination channel response piece sequence number, and repeating step 2 is to step 4;

Step 2: determine modifying factor:

When data block 1 is carried out joint-detection,

When data block 2 is carried out joint-detection,

Step 3: get k=(i-1) * Q+1 ..., i*Q+W-1, W is the subscriber channel window width, repeating step 4;

Step 4: if v _k ⁱBe not equal to 0, $v_k^i=v_k^i*w_i;$

Step 5: by newly obtaining v _k ⁱStructure according to Fig. 1 generates new system transmission matrix A.

Step 6: structure and the formula (5) of utilization Fig. 2 carry out the ZF-BLE joint-detection, or structure and the formula (7) of utilization Fig. 3 carry out the MMSE-BLE joint-detection.

Be the TD-SCDMA data burst structure of the inventive method as shown in Figure 4, middle midamble partly is the M*Q chip, and data block 1, data block 2 all are N symbol, are equivalent to N*Q chip.

The joint-detection that is combining compensate of frequency deviation as shown in Figure 5 realizes block diagram, by channel estimator 510, and A matrix maker 520, A matrix corrector 530, combined detector 540, frequency offset estimator 550 is formed.

The midamble sequence of input signal partly is sent to channel estimator 510, utilize deconvolution or fft/ifft method to produce channel impulse response h (t), concrete grammar has a detailed description on " the Linearunbiased data estimation in mobile radio systems applying CDMA " that delivered on the IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS in 1993 at A.Klein.The channel impulse response h (t) that produces is sent to A matrix maker 520.

A matrix maker 520 forms system transmission matrix A according to the structure of Fig. 1.Wherein $V_i=\left[\begin{array}{cc}{v}_1^i\&CenterDot;\&CenterDot;\&CenterDot;& v_{Q+W-1}^i\end{array}\right]$ Be i the aggregate channel response block of system transmission matrix A, generate by spreading code c and channel impulse response h convolution.The system transmission matrix A that produces is sent to A matrix corrector 530.

The method that A matrix corrector 530 uses the present invention to describe utilizes frequency offset estimator 550 to send here to such an extent that frequency offset estimating Δ f revises the A matrix.Revised A matrix is sent to combined detector 540.

The MMSE-BLE method that combined detector 540 can use ZF-BLE method that Fig. 2 describes or Fig. 3 to describe realizes joint-detection.The modulation symbol that produces is sent to frequency offset estimator 550.

The information that frequency offset estimator 550 utilizes modulation symbol to carry is estimated frequency deviation, in initial condition, when also not estimating frequency deviation, can produce Δ f=0.The frequency deviation Δ f that produces is sent to A matrix corrector 530.

Claims (4)

1, a kind of associated detecting method of combining compensate of frequency deviation, this method comprises: revise system model, establish Δ f and be the frequency deviation between recipient and the transmit leg, Ts is a mark space, because the existence of frequency deviation, data model is modified to:

$e={(e_1,e_2,e_3,....,e_{N\&CenterDot;Q+W-1})}^T=\mathrm{AFd}+n---\left(8\right)$

Wherein, A is a system transmission matrix, and d is the raw data symbols sequence, and e is a receiving data sequence, and n is an additive noise; The F matrix has reflected the influence of frequency deviation to initial data.

2, associated detecting method according to claim 1 is characterized in that, described F matrix is:

φ _OBe the phase deviation of first symbol correspondence in the equalization block.

3, associated detecting method according to claim 2 is characterized in that, the key step of this method is as follows:

Step 1: get i=1,2 ..., N, i represent various combination channel response piece sequence number, repeat following steps 2 to step 4;

Step 2: determine modifying factor: When data block 1 is carried out joint-detection,

When data block 2 is carried out joint-detection,

Δ f is the frequency deviation between recipient and the transmit leg, and Ts is a mark space;

Step 3: get k=(i-1) * Q+1 ..., i*Q+W-1, W are the subscriber channel width, repeating step 4;

Step 4: if v _k ⁱBe not equal to 0, $v_k^i=v_k^i*w_i;$

Step 5: by newly obtaining v _k ⁱStructure according to the original system transmission matrix generates new system transmission matrix A;

Step 6: using formula (5) carries out the ZF-BLE joint-detection:

${\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}={\left(A^H{R}_n^{-1}A\right)}^{-1}{A}^H{R}_n^{-1}e$

$=d+{(\mathrm{\&Sigma;}\&CenterDot;\mathrm{\&Gamma;})}^{-1}{({\mathrm{\&Gamma;}}^H\&CenterDot;\mathrm{\&Sigma;})}^{-1}{A}^H{R}_n^{-1}n---\left(5\right)$

Or using formula (7) carries out the MMSE-BLE joint-detection:

${\hat{d}}_{\mathrm{MMSE}-\mathrm{BLE}}={(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}e$

$={(I+{\left(R_d{A}^H{R}_n^{-1}A\right)}^{-1})}^{-1}\&CenterDot;{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}\underset{=}{\mathrm{def}}{W}_0\&CenterDot;{\hat{d}}_{\mathrm{ZF}-\mathrm{BLE}}$

$=\mathrm{diag}\left(W_0\right)\&CenterDot;d+\stackrel{\&OverBar;}{\mathrm{diag}}\left(W_0\right)\&CenterDot;\underset{\&OverBar;}{d}+{(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}n---\left(7\right)$

4, a kind of joint-detection device of combining compensate of frequency deviation is characterized in that, it comprises channel estimator, A matrix maker, A matrix corrector, combined detector and frequency offset estimator;

The midamble sequence of input signal partly is sent to described channel estimator, utilizes deconvolution or fft/ifft method to produce channel impulse response h (t), and is sent to A matrix maker;

A matrix maker forms system transmission matrix A, wherein $V_i=[v_1^i\&CenterDot;\&CenterDot;\&CenterDot;v_{Q+W-1}^i]$ Be i the aggregate channel response block of system transmission matrix A, generate that the system transmission matrix A of generation is sent to described A matrix corrector by spreading code c and channel impulse response h convolution;

Described A matrix corrector utilizes frequency offset estimator to send here to such an extent that frequency offset estimating Δ f revises the A matrix, and revised A matrix is sent to combined detector;

Described combined detector uses ZF-BLE method or MMSE-BLE method to realize joint-detection, and the modulation symbol of generation is sent to frequency offset estimator;

The information that described frequency offset estimator utilizes modulation symbol to carry is estimated frequency deviation, and the frequency deviation Δ f of generation is sent to A matrix corrector.

Images