Combined detection method and device combined with frequency offset compensation
Technical Field
The present invention relates to a method and apparatus for improving joint detection performance of a mobile communication system, and more particularly, to a novel joint detection method and apparatus for combining frequency offset compensation in a TD-SCDMA mobile communication system.
Background
For the joint detection algorithm, refer to "Linear approximated data estimation IN mobile radio systems applying CDMA", published by A.Klein IN 1993 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS. To illustrate the need of the present invention, the existing joint assay methods are now briefly introduced:
firstly, the existing joint detection determines a data model as follows:
e=(e 1 ,e 2 ,e 3 ,....,e N*Q+W-1 ) T =Ad+n (1)
a is the system transmission matrix, d is the original data symbol sequence, e is the received data sequence, and n is additive noise. The covariance matrix of data d is:
the covariance matrix of the noise n is:
the structure of the system transmission matrix a is described as shown in fig. 1. Wherein V i And i is more than 1 and less than N, the channel impulse response is called a combined channel response block and is generated by convolution of a spread spectrum code c and a channel impulse response h, Q is the length of the spread spectrum code, W is the length of a channel impulse response window, and N is the length of an equalization block.
Fig. 2 is a block diagram of a method of jointly detecting ZF-BLE, which is composed of a matched filter 210, a whitening filter 220, an isi & MAI canceller 230. The ZF-BLE method is based on MLSE (maximum likelihood sequence) criterion, minimizing the following quadratic equation:
the final ZF-BLE derived data was estimated as:
FIG. 3 shows a structure diagram of jointly detecting MMSE-BLE, which comprises a ZF-BLE detector 200 and a wiener filter 310, and the MMSE-BLE method is to minimize the following equation according to the MMSE (minimum mean square error) criterion
The data estimate from MMSE-BLE is:
therefore, the system transmission matrix A plays a decisive role in the joint detection algorithm, and the joint detection effect is greatly improved by modifying the matrix A structure to make the matrix A more accord with the actual transmission environment. As can be seen from the prior art method for constructing the transmission matrix A of the system, each combined channel response block V is within an equalization block of length N i I is more than 1 and less than N, corresponding to a data symbol, which are generated by convolution of a spread spectrum code c and a channel impulse response h,so N combined channel response blocks V are within the entire equalization block i The same values were taken.
The above prior art method can be applied to the situation that the mobile station is stationary and the channel fading changes slowly, but because the mobile communication channel usually has the characteristic of fast fading, the actual combined channel response usually changes greatly within an equalization block length N, and then the processing performance of the prior art method is reduced.
An important factor for a mobile communication channel to have a fast fading characteristic is the presence of a frequency offset between the transmitting and receiving parties. In a mobile communication system, a transmitting side modulates information to a carrier fc, which must be regenerated in order for a receiving side to accurately receive the information. Although this carrier is nominally known a priori by both the sender and receiver, there are two main reasons why there is a frequency offset between the sender and receiver: 1. the local clock precision of the mobile station is not high; 2. doppler shift due to the movement of the mobile station position. In order to achieve stable and reliable receiving performance, the receiving end must effectively estimate and compensate the frequency offset. The existing frequency compensation method of the TD-SCDMA system is as follows: the demodulated data is remodulated, the modulated signal is reproduced at the receiving end, and then the phase difference between the modulated signal and the corresponding received signal is calculated, thereby estimating and compensating the frequency offset. The method is used for frequency offset compensation after data demodulation, but the method is poor in simulation real condition and low in detection precision.
Disclosure of Invention
The invention aims to solve the technical problems that: the problem that a system transmission matrix A in the prior art does not well approximate to the real situation and is low in precision in a channel environment with frequency offset in high-speed change is solved, a novel combined detection technology combined with frequency offset compensation is provided, the system transmission matrix A is corrected, a system model is enabled to be closer to the real situation, and the combined detection precision is higher. Meanwhile, the early-stage frequency offset compensation is carried out in the data detection stage, so that the precision is further improved, and the receiving performance is improved.
The technical scheme of the invention is as follows:
a method of joint detection incorporating frequency offset compensation, the method comprising: modifying the system model, setting delta f as the frequency offset between a receiver and a sender, setting Ts as a symbol interval, and modifying the system model as follows due to the existence of the frequency offset:
e=(e 1 ,e 2 ,e 3 ,....,e N*Q+W-1 ) T =AFd+n
wherein, A is a system transmission matrix, d is an original data symbol sequence, e is a received data sequence, N is additive noise, N is a balanced block length, Q is a spreading code length, and W is a channel impact response window length; the F matrix reflects the effect of the frequency offset on the original data.
The joint detection method, wherein the F matrix is:
is the phase offset corresponding to the first symbol in the equalization block.
The joint detection method comprises the following main steps:
step 1: taking i =1, 2.. And N, i to represent the serial numbers of different combined channel response blocks, and repeating the following steps 2 to 4;
step 2: determining a correction factor: w is a
i =e
-j{ +2πΔf(i-1)Ts} When the joint detection is performed on the
data block 1,
(ii) a When the joint detection is performed on data block 2,
(ii) a M is a numerical value obtained by dividing the length of the midamble by the length of the spreading code, a
data block 1 is N symbols before the midamble in the TD-SCDMA data burst structure of the time division synchronous code division multiple access, and a data block 2 is N symbols after the midamble in the TD-SCDMA data burst structure;
and step 3: taking k = (i-1) × Q + 1.., i × Q + W-1, and repeating the step 4;
and 4, step 4: if v is
k i Is not equal to 0, and is,
v k i a k element of an ith combined channel response block representing the system transmission matrix A, wherein k represents an element index of the combined channel response block;
and 5: by newly obtaining v k i Generating a new system transmission matrix A according to the structure of the original system transmission matrix;
and 6: and (3) applying a formula (5) to carry out ZF-BLE joint detection:
data estimation, R, representing a zero forcing data block linear equalizer ZF-BLE method
n Representing a covariance matrix of the noise n, gamma representing an upper triangular matrix with all diagonal elements of 1, and sigma representing a diagonal matrix;
or applying formula (7) to perform MMSE-BLE joint detection:
data estimation, R, representing the MMSE-BLE method of a minimum mean square error data Block Linear equalizer
d Covariance matrix representing data d, I representing identity matrix, W
0 Equivalent to (I + (R)
d A
H R
n -1 A)
-1 )
-1 。
A joint detection device combined with frequency offset compensation comprises a channel estimator, an A matrix generator, an A matrix corrector, a joint detector and a frequency offset estimator;
the midamble sequence part of the input signal is sent to the channel estimator, a channel impulse response h (t) is generated by using a deconvolution or fft/ifft method, and the midamble sequence part is sent to an A matrix generator;
a matrix generator forms a system transmission matrix A, wherein
The ith combined channel response block of the system transmission matrix A is generated by convolution of a spread spectrum code c and a channel impulse response h (t), the generated system transmission matrix A is sent to the matrix A modifier, Q is the length of the spread spectrum code, and W is the length of a channel impulse response window;
the matrix A corrector corrects the system transmission matrix A by using the frequency deviation delta f sent by the frequency deviation estimator, and the corrected system transmission matrix A is sent to the joint detector;
the joint detector realizes joint detection by using a ZF-BLE method or an MMSE-BLE method, and the generated modulation symbols are sent to a frequency offset estimator;
and the frequency offset estimator estimates the frequency offset by using the information carried by the modulation symbol, and the generated frequency offset delta f is sent to the A matrix corrector.
The combined detection method and the combined detection device combined with the frequency offset compensation modify the system transmission matrix A, so that a system model is closer to the real situation, and the combined detection precision is higher. Meanwhile, the frequency offset compensation is performed in the early stage in the data detection stage, so that the precision is further improved, and the receiving performance is improved.
Drawings
FIG. 1 is a prior art architecture of a system transmission matrix A;
figure 2 is a prior art joint detection ZF-BLE structure;
figure 3 is a prior art joint detection MMSE-BLE structure;
FIG. 4 is a TD-SCDMA data burst structure of the present invention method;
FIG. 5 is a block diagram of a joint detection implementation of the method of the present invention incorporating frequency offset compensation;
fig. 6 is a block diagram of a method for correcting a system transmission matrix according to the present invention.
Detailed Description
The following detailed description of the embodiments is made with reference to the accompanying drawings:
the method for joint detection combined with frequency offset compensation comprises the following steps:
firstly, modifying a system model, setting delta f as frequency deviation between a receiving party and a sending party, setting Ts as a symbol interval, and modifying the system model into the following steps due to the existence of the frequency deviation:
e=(e 1 ,e 2 ,e 3 ,...,e N*Q+W-1 ) T =AFd+n (8)
a is system transmission matrix, d is original data symbol sequence, e is interfaceAnd receiving a data sequence, wherein n is additive noise. The system transmission matrix A is constructed as shown in FIG. 1, wherein
And i is more than or equal to 1 and less than or equal to N is the ith combined channel response block of the system transmission matrix A and is generated by convolution of a spreading code c and a channel impulse response h, and N is the length of the equalization block. The F matrix reflects the effect of the frequency offset on the original data.
Referring to fig. 4, a specific data burst structure of TD-SCDMA is described, where data block 1 and data block 2 both contain N × Q chips (corresponding to N symbols), and the midamble sequence contains M × Q chips. Since the channel estimation of TD-SCDMA is obtained by using midamble part, the midpoint of midamble sequence can be regarded as phase reference point, thenWhen the joint detection is performed on
data block 1,
(ii) a When joint detection is performed on data block 2,
。
according to the newly defined system model, the product of the matrix A and the matrix F is used for replacing the original system transmission matrix A, so that the real channel characteristic is more approximate, and performance gain is brought in the joint detection.
The core steps of the transmission matrix a correction of the joint detection system combined with frequency offset compensation according to the present invention are as follows, as shown in fig. 6:
step 1: taking i =1, 2.. And N, i to represent the serial numbers of different combined channel response blocks, and repeating the steps 2 to 4;
step 2: determining a correction factor: w is a
i =e
-j{ +2πΔf(i-1)Ts} . When the joint detection is performed on the
data block 1,
(ii) a When joint detection is performed on data block 2,
。
and 3, step 3: taking k = (i-1) × Q + 1., i × Q + W-1, W as the user channel window width, and repeating the step 4;
and 4, step 4: if v is
k i Is not equal to 0, and is,
and 5: by newly obtaining v k i A new system transmission matrix a is generated according to the structure of fig. 1.
And 6: performing ZF-BLE joint detection by using the structure and formula (5) of FIG. 2, or using
The structure of figure 3 and equation (7) perform MMSE-BLE joint detection.
Fig. 4 shows a TD-SCDMA data burst structure of the present invention, wherein the middle midamble portion is M × Q chips, and the data blocks 1 and 2 are N symbols, which is equivalent to N × Q chips.
Fig. 5 is a block diagram of a joint detection implementation incorporating frequency offset compensation, which is composed of a channel estimator 510, an a matrix generator 520, an a matrix modifier 530, a joint detector 540, and a frequency offset estimator 550.
The midamble sequence portion of the input signal is sent to the channel estimator 510, which generates the channel impulse response h (t) using a deconvolution or fft/ifft method, as described IN detail IN Linear approximated data estimation IN mobile system adaptive CDMA, published by A.Klein IN IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, 1993. The resulting channel impulse response h (t) is fed to an a matrix generator 520.
The a
matrix generator 520 forms a system transmission matrix a according to the structure of fig. 1. Wherein
And the ith combined channel response block of the system transmission matrix A is generated by convolution of a spreading code c and a channel impact response h. The generated system transmission matrix a is supplied to an a
matrix modifier 530.
Matrix a modifier 530 modifies system transmission matrix a using the frequency offset Δ f from frequency offset estimator 550 using the methods described herein. The modified system transmission matrix a is sent to the joint detector 540.
Joint detector 540 may implement joint detection using either the ZF-BLE method described in fig. 2 or the MMSE-BLE method described in fig. 3. The resulting modulation symbols are provided to frequency offset estimator 550.
The frequency offset estimator 550 estimates a frequency offset using information carried by the modulation symbol, and may generate Δ f =0 in an initial state when the frequency offset is not estimated yet. The resulting frequency offset Δ f is sent to an a matrix modifier 530.