CN101945060B - Channel estimation method based on pilot frequency signal in 3GPP LTE downlink system - Google Patents

Abstract

The invention provides a channel estimation method based on the pilot frequency signal in a 3GPP LTE downlink system aiming at the 3GPP LTE downlink system, which is the channel estimation method using the frequency two dimension combining Weiner iterative filtering and based on the Linear Minimum Mean-square Error (LMMSE) algorithm. In the invention, the Weiner iterative filtering is performed in the frequency domain based on the LMMSE algorithm firstly, then the value estimated by the frequency domain is used to perform the first Weiner iterative filtering in the frequency domain. The experimental test indicates that the channel estimation method can effectively enhance the property of Bit Error Rate (BER), and is closer to the ideal channel estimation compared with the traditional methods. In addition, the method of the invention reduces the operation complexity of the algorithm through the Singular Value Decomposition (SVD) of the LMMSE algorithm, therefore the channel estimation method provided by the invention can not increase too much operation complexity.

Description

In a kind of 3GPP LTE downlink system based on the channel estimation methods of pilot signal

Technical field

The invention belongs to the channel estimation technique field in the communication system, more specifically say, relate in a kind of 3GPP LTE downlink system the channel estimation methods based on pilot signal.

Background technology

LTE (Long Term Evolution) is third generation partner program (3rd GenerationPartnership Project, 3GPP) under " mobile communication broad band " trend, in order to resist global inserting of microwave interoperability (World interoperability for Microwave Access, the market challenges of mobile broadband access technology such as WiMAX), " accurate 4G " technology of developing on the technological reserve basis of super 3G of more than ten years (Beyond 3rd Generation, B3G) research.LTE is interface aspect frequency division multiple access (Frequency Division Multiple Access aloft, FDMA) substituted long-term code division multiple access (the Code Division Multiple Access that uses of 3GPP, CDMA) as multiple access technology, and adopted in a large number multiple-input and multiple-output (multiple-input multiple-output, MIMO) technology and adaptive technique improve data rate and systematic function, and the air interface transmission ability is reached more than the 100Mbit/s.3GPP LTE has been subject to the in the world favor of most operators, has been acknowledged as the mobile communication system that can support at 2010～the year two thousand twenty WorldCom's industry.

Because OFDM (Orthogonal Frequency Division Multiplexing, OFDM) technology has very high data transmission rate and spectrum efficiency, it can resist the advantages such as multidiameter delay expansion effectively simultaneously, and the physical layer of 3GPP LTE downlink system has adopted the OFDM technology.Because the wireless channel in the wide-band mobile communication system has frequency domain selectivity and time variation, so receiver is before carrying out coherent demodulation to ofdm signal at a high speed, and multidiameter fading channel is carried out dynamic estimation and tracking is very important.In the OFDM technology, the Pilot-assisted channel estimation method that is based on that channel estimating is the most frequently used.

In based on the Pilot-assisted channel estimation algorithm, transmitting terminal inserts all known pilot signals of transmitting terminal and receiving terminal in fixing position, and then receiving terminal goes out the channel response of each OFDM symbol by the reception Signal estimation of processing these positions.Channel estimation methods based on pilot signal in the 3GPP LTE downlink system roughly can be divided into least square (Least-Squares, LS) and least mean-square error (Minimum Mean-Square Error, MMSE) two class methods.These two class methods all are on frequency domain each subcarrier to be carried out channel estimating, and wherein, the LS channel estimation methods does not need channel information, realize the simplest; The MMSE channel estimation methods has utilized the channel statistical informations such as correlation between subcarrier and signal to noise ratio, and estimated performance is better, is widely used in the channel estimating of OFDM.

Summary of the invention

The object of the invention is to overcome the deficiency of existing channel method of estimation, and the channel estimation methods based on pilot signal is provided in the 3GPP LTE downlink system that a kind of error rate is low, computational complexity is not high.

For achieving the above object, based on the channel estimation methods of pilot signal, it is characterized in that in the 3GPP LTE downlink system of the present invention, may further comprise the steps:

(1), the dimension of frequency domain is received iterative filtering

At first obtain the channel frequency domain response estimation value of reference signal position on the same OFDM symbol:

${\hat{H}}_{P,\mathrm{LMMSE}}^{\left(1\right)}=R_{H_P{H}_P}{(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LS}}---\left(1\right)$

In the formula (1),

The autocorrelation matrix that represents same OFDM symbol internal reference signals virgin carrier wave, β are the constants that is determined by signal constellation (in digital modulation) figure, and SNR is average signal-to-noise ratio, I _PBe unit matrix; Channel frequency domain response H for the reference signal position _PThe LS estimated value, subscript () _PThe position of expression transmission of reference signals.

Then right Carry out receiving iterative filtering based on the dimension of LMMSE algorithm and come noise reduction and disturb impact on the locational channel frequency response estimated value of reference signal, improve bit error rate performance, dimension is received the locational channel frequency response estimated value of reference signal behind the iterative filtering

${\hat{H}}_{P,\mathrm{LMMSE}}^{\left(2\right)}=R_{H_P{H}_P}{(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LMMSE}}^{\left(1\right)}---\left(2\right)$

Utilize at last through dimension and receive the channel frequency domain response estimation value of the reference signal position behind the iterative filtering

Estimate the channel frequency domain response value at all subcarrier places on the same OFDM symbol

${\hat{H}}_{F,\mathrm{LMMSE}}=R_{{\mathrm{HH}}_P}{(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LMMSE}}^{\left(2\right)}---\left(3\right)$

In the formula (3),

The cross-correlation matrix that represents the subcarrier at subcarriers all in the same OFDM symbol and reference signal place;

(2), the dimension of time domain is received iterative filtering

To the channel frequency domain response estimation value on n the subcarrier

Carry out receiving iterative filtering based on the dimension of LMMSE algorithm, to reduce relevant residual noise and to improve bit error rate performance:

${\tilde{H}}_{F,\mathrm{LMMSE}}^{\left(n\right)}=R_{H_P{H}_P}^{\left(n\right)}{(R_{H_P{H}_P}^{\left(n\right)}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_L)}^{-1}{\hat{H}}_{F,\mathrm{LMMSE}}^{\left(n\right)}---\left(4\right)$

In the formula (4), That all comprise the autocorrelation matrix between the OFDM symbol of reference signal, I on n subcarrier in the subframe _LBe unit matrix, L represents to comprise in each subframe the OFDM number of symbols of reference signal;

Utilize dimension to receive channel frequency response estimated value after the iterative filtering Estimate the channel frequency response value at all n subcarrier places of OFDM symbol in each subframe

${\hat{H}}_{T,\mathrm{LMMSE}}^{\left(n\right)}=R_{{\mathrm{HH}}_P}^{\left(n\right)}{(R_{H_P{H}_P}^{\left(n\right)}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_L)}^{-1}{\tilde{H}}_{F,\mathrm{LMMSE}}^{\left(n\right)}---\left(5\right)$

Wherein,

All OFDM symbols and comprise cross-correlation matrix between the OFDM symbol of reference signal on n the subcarrier;

To the channel frequency domain response estimation value on all subcarriers

Carry out above-mentioned calculating, finally receive iterative filtering through the dimension of frequency-domain and time-domain and obtained the channel transfer function value at all resource particle places.

In the present invention, in the superincumbent arthmetic statement, no matter be that there have to be right at frequency domain or in the dimension of time domain is received iterative filtering:

$(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}I)$

Inversion operation, if the number of reference signal is many, this inversion operation will become very complicated so.

In order to lower the computational complexity of the channel estimation methods that the present invention proposes, to autocorrelation matrix

Carry out singular value decomposition, and then calculate inverse matrix

${(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}I)}^{-1};$

Inverse matrix is decomposed into:

${(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}I)}^{-1}=\frac{\mathrm{SNR}}{\mathrm{\β}}(I-\frac{\mathrm{SNR}}{\mathrm{\β}}\×U{({\mathrm{\Λ}}^{-1}+\frac{\mathrm{SNR}}{\mathrm{\β}}I)}^{-1}{U}^H)---\left(6\right)$

Wherein U is an orthogonal moment array, and Λ is diagonal matrix.

Owing to calculating

Do not need matrix inversion, can calculate easily, simultaneously in the 3GPP LTE downlink system, each subframe only has 14 OFDM symbols, and the shared OFDM symbol of reference signal only has 4, so autocorrelation matrix

Singular value decomposition also relatively easy, so greatly simplified the computational complexity of algorithm.

Among the present invention, for 3GPP LTE downlink system, channel estimation methods based on pilot signal has been proposed in a kind of 3GPP LTE downlink system, adopt time-frequency two-dimensional to unite dimension and receive the channel estimation methods of iterative filtering, the method is based on linear minimum mean-squared error (Linear Minimum Mean-Square Error, LMMSE) algorithm.In the present invention, at first carry out receiving iterative filtering based on the dimension of LMMSE algorithm at frequency domain, and then utilize value that frequency domain estimates once to tie up in time domain and receive iterative filtering.Show through experiment test, this signal estimation method can improve bit error rate (Bit Error Rate, BER) performance effectively, compares more with traditional method and estimates close to ideal communication channel.In addition, the present invention reduces the computational complexity of algorithm by the LMMSE algorithm being carried out singular value decomposition (Singular Value Decomposition, SVD), so the channel estimation methods that the present invention proposes can not increase too many computational complexity.

Description of drawings

Fig. 1 is 3GPP LTE downlink system illustraton of model;

Fig. 2 is the distribution schematic diagram of reference signal in Resource Block;

Fig. 3 is the bit error rate performance comparison diagram of different channels method of estimation.

Embodiment

Below in conjunction with accompanying drawing the specific embodiment of the present invention is described, so that those skilled in the art understands the present invention better.What need to point out especially is that in the following description, when perhaps the detailed description of known function and design can desalinate main contents of the present invention, these were described in here and will be left in the basket.

Fig. 1 is 3GPP LTE downlink system illustraton of model, as shown in Figure 1, at transmitting terminal, input bit at first enters channel coding module, then coded-bit is modulated into complex modulation symbols again through scrambling, these complex modulation symbols are mapped on the transport layer of one or more antenna ports, every one deck is mapped on the resource particle of each antenna port through after the precoding, insert simultaneously reference signal, at last, these symbols of each antenna port comprise data and reference signal, are modulated into plural time domain OFDM signal and transmit in wireless multipath channel.

At receiving terminal, in the ideal synchronisation situation, receiving end signal Y can be expressed as after the OFDM demodulation:

Y＝XH+W (7)

Wherein H and W represent respectively channel frequency domain response and additive white Gaussian noise, and X and Y represent respectively transmitted signal and acknowledge(ment) signal.

In order to recover the transmission bit from the signal that receives, channel estimation module need to obtain the estimator of Parameter H.For this reason, 3GPP LTE downlink system inserts reference signal in fixing time-frequency two-dimensional position, can be expressed as at these locational receiving end signals:

Y _P＝X _PH _P+W _P (8)

X wherein _P, H _PAnd W _PThe subset of correlation matrix in the formula (7), subscript () _PThe position of expression transmission of reference signals.

3GPP LTE standard definition three kinds of downlink reference signals: the reference signal of the reference signal of cell-specific, Multicast Broadcast Single Frequency Network (Multicast Broadcast Single Frequency Network, MBSFN) reference signal and terminal-specific.In the present embodiment, adopt the reference signal of cell-specific to analyze.

Fig. 2 has provided the allocative decision of descending cell special reference on Resource Block that single antenna sends.As seen, reference signal (R ₀) 7 OFDM symbols in interval on same subcarrier on the time domain, 6 subcarriers in interval in the same OFDM symbol on frequency domain.

Existing channel estimation method:

Data Y by the reference signal position that receives _PWith the reference signal X that sends _P, the channel frequency domain response H of reference signal position in the formula (8) _PLS estimate and can be expressed as:

${\hat{H}}_{P,\mathrm{LS}}={X_P}^{-1}{Y}_P---\left(9\right)$

Then carry out linear interpolation by the estimated value to the channel frequency domain response of reference signal position in time domain and frequency domain, thereby obtain the channel transfer function of all data resource particles among Fig. 2.Channel estimation method based on the LS criterion is simple in structure, but does not utilize the frequency domain of channel and the correlation properties of time domain in LS estimates, has ignored the impact of noise when estimating, so channel estimation value is relatively more responsive on the impact of noise.

Another kind of channel estimation methods is the MMSE algorithm, is widely used in the OFDM channel estimating, and this method is utilized signal to noise ratio (Signal to Noise Ratio, SNR) and other channel statistic information, and performance is better than the LS algorithm.The expression formula of MMSE channel estimation in frequency domain algorithm is as follows:

$H_{\mathrm{LMMSE}}=R_{{\mathrm{HH}}_P}{(R_{H_P{H}_P}+{\mathrm{\σ}}_W^2{\left({\mathrm{XX}}^H\right)}^{-1})}^{-1}{\hat{H}}_{P,\mathrm{LS}}---\left(10\right)$

Wherein, H in the representation formula (9) _PThe LS estimated value,

The cross-correlation matrix that represents the subcarrier at subcarriers all in the same OFDM symbol and reference signal place,

The autocorrelation matrix that represents same OFDM symbol internal reference signals virgin carrier wave,

Be the variance of additive white Gaussian noise, subscript () ^HThe expression conjugate transpose.If the sign map of transmission arrives same planisphere,

So in the formula (10)

Can be with its expectation Represent:

${\mathrm{\σ}}_W^2{\left({\mathrm{XX}}^H\right)}^{-1}=E\left\{{\mathrm{\σ}}_W^2{\left({\mathrm{XX}}^H\right)}^{-1}\right\}=\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P---\left(11\right)$

Wherein,

A constant that is determined by signal constellation (in digital modulation) figure, X _kFor the point on the planisphere, for QPSK, β=1, for 16QAM, β=17/9, SNR is average signal-to-noise ratio, I _PBe unit matrix.Formula (11) can further be reduced to so:

${\hat{H}}_{\mathrm{LMMSE}}=R_{{\mathrm{HH}}_P}{(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LS}}---\left(12\right)$

Then carry out linear interpolation by the channel frequency domain response estimation value to frequency domain in time domain, obtain at last the channel transfer function of all data resource particles among Fig. 2.

Can be found out by formula (12), the LMMSE channel estimation method is actually according to the MMSE criterion LS channel estimation results is carried out filtering processing, can eliminate like this impact of partial noise.In order further to eliminate noise, improve the error rate of system performance, simultaneously can not increase too many computational complexity again, consider that 3GPP LTE downlink system has inserted reference signal in time domain and frequency domain both direction simultaneously, the time-frequency two-dimensional that has proposed in the present invention a kind of 3GPP of being applicable to LTE downlink system is united dimension and is received the iterative filtering channel estimation methods, this channel estimation methods, at first at frequency domain the OFDM symbol that comprises reference signal is tieed up and received iterative filtering, the value that the recycling frequency domain estimates is tieed up all subcarriers in time domain and is received iterative filtering.

In this enforcement, with the single output of the single input in alone family (single-user single-input single-output, SUSISO) be example, tested the performance of various channel estimation methods under the VehA of 3GPP LTE downlink system (VehicularA) channel circumstance.Test environment parameter: the 1.4M bandwidth, the QPSK modulation, reference signal adopts residential quarter special reference signal, and sample frequency is 1.92MHz, and subcarrier spacing is 15KHz, and number of times is 1000 times.

Fig. 3 is the bit error rate performance comparison diagram of different channels method of estimation.

Fig. 3 has showed the bit error rate performance of different channels method of estimation under the 3GPP LTE downlink system VehA channel, wherein " LS " represents the bit error rate performance of LS channel estimation methods, the bit error rate performance of " MMSE " expression MMSE channel estimation methods, the bit error rate performance curve is followed successively by the bit error rate performance curve of LS, MMSE, the present invention and ideal communication channel method of estimation from top to bottom.As can be seen from Figure 3 the time-frequency two-dimensional that proposes of the present invention is united dimension and is received the bit error rate performance of iterative filtering channel estimation method and be better than LS and MMSE algorithm, more approachingly estimates with ideal communication channel, can not increase too many computational complexity again simultaneously.

The present invention be 3GPP LTE downlink system proposed a kind of complexity relatively not high time-frequency two-dimensional unite dimension and receive the iterative filtering channel estimation methods.The method is received iterative filtering and is reduced noise effect and improve bit error rate performance by tieing up at frequency-domain and time-domain.Test result shows that the channel estimation methods that the present invention proposes has better performance than LS and MMSE channel estimation methods.The inventive method has significantly reduced the computational complexity of algorithm also by the autocorrelation matrix in the algorithm is carried out singular value decomposition simultaneously.

Although the above is described the illustrative embodiment of the present invention; so that the technical staff of present technique neck understands the present invention; but should be clear; the invention is not restricted to the scope of embodiment; to those skilled in the art; as long as various variations appended claim limit and the spirit and scope of the present invention determined in, these variations are apparent, all utilize innovation and creation that the present invention conceives all at the row of protection.

Claims (2)

1. In the 3GPP LTE downlink system based on the channel estimation methods of pilot signal, it is characterized in that, may further comprise the steps:

   (1), the dimension of frequency domain is received iterative filtering

   At first obtain the channel frequency domain response estimation value of reference signal position on the same OFDM symbol:

   ${\hat{H}}_{P,\mathrm{LMMSE}}^{\left(1\right)}=R_{H_P{H}_P}{(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LS}}---\left(1\right)$

   In the formula (1),

   

   The autocorrelation matrix that represents same OFDM symbol internal reference signals virgin carrier wave, β are the constants that is determined by signal constellation (in digital modulation) figure, and SNR is average signal-to-noise ratio, I _PBe unit matrix;

   

   Channel frequency domain response H for the reference signal position _PLeast square LS estimated value, subscript () _PThe position of expression transmission of reference signals;

   Then right Carrying out receiving iterative filtering based on the dimension of linear minimum mean-squared error LMMSE algorithm comes noise reduction and disturbs impact on the locational channel frequency response estimated value of reference signal, improve bit error rate performance, dimension is received the locational channel frequency response estimated value of reference signal behind the iterative filtering

   

   ${\hat{H}}_{P,\mathrm{LMMSE}}^{\left(2\right)}=R_{H_P{H}_P}{(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LMMSE}}^{\left(1\right)}---\left(2\right)$

   Utilize at last through dimension and receive the channel frequency domain response estimation value of the reference signal position behind the iterative filtering

   

   Estimate the channel frequency domain response value at all subcarrier places on the same OFDM symbol

   

   ${\hat{H}}_{F,\mathrm{LMMSE}}=R_{H{H}_P}{(R_{H{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_P)}^{-1}{\hat{H}}_{P,\mathrm{LMMSE}}^{\left(2\right)}---\left(3\right)$

   In the formula (3),

   

   The cross-correlation matrix that represents the subcarrier at subcarriers all in the same OFDM symbol and reference signal place;

   (2), the dimension of time domain is received iterative filtering

   To the channel frequency domain response estimation value on n the subcarrier

   

   Carry out receiving iterative filtering based on the dimension of linear minimum mean-squared error LMMSE algorithm, to reduce relevant residual noise and to improve bit error rate performance:

   ${\tilde{H}}_{F,\mathrm{LMMSE}}^{\left(n\right)}=R_{H_P{H}_P}^{\left(n\right)}{(R_{H_P{H}_P}^{\left(n\right)}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_L)}^{-1}{\hat{H}}_{F,\mathrm{LMMSE}}^{\left(n\right)}---\left(4\right)$

   In the formula (4),

   

   That all comprise the autocorrelation matrix between the OFDM symbol of reference signal, I on n subcarrier in the subframe _LBe unit matrix, L represents to comprise in each subframe the OFDM number of symbols of reference signal;

   Utilize dimension to receive channel frequency response estimated value after the iterative filtering Estimate the channel frequency response value at all n subcarrier places of OFDM symbol in each subframe

   

   ${\hat{H}}_{T,\mathrm{LMMSE}}^{\left(n\right)}=R_{H{H}_P}^{\left(n\right)}{(R_{H{H}_P}^{\left(n\right)}+\frac{\mathrm{\β}}{\mathrm{SNR}}{I}_L)}^{-1}{\tilde{H}}_{F,\mathrm{LMMSE}}^{\left(n\right)}---\left(5\right)$

   Wherein,

   

   All OFDM symbols and comprise cross-correlation matrix between the OFDM symbol of reference signal on n the subcarrier;

   To the channel frequency domain response estimation value on all subcarriers

   

   Carry out above-mentioned calculating, finally receive iterative filtering through the dimension of frequency-domain and time-domain and obtained the channel transfer function value at all resource particle places.
2. In the 3GPP LTE downlink system according to claim 1 based on the channel estimation methods of pilot signal, it is characterized in that, at first to the autocorrelation matrix in step (1) and (2)

   

   Carry out singular value decomposition, and then calculate inverse matrix

   

   Inverse matrix is decomposed into:

   ${(R_{H_P{H}_P}+\frac{\mathrm{\β}}{\mathrm{SNR}}I)}^{-1}=\frac{\mathrm{SNR}}{\mathrm{\β}}(I-\frac{\mathrm{SNR}}{\mathrm{\β}}\×U{({\mathrm{\Λ}}^{-1}+\frac{\mathrm{SNR}}{\mathrm{\β}}I)}^{-1}{U}^H),$ Wherein U is an orthogonal moment array, subscript () ^HThe expression conjugate transpose, Λ is diagonal matrix, I is the I in step (1) and (2) _pOr I _L

Images