CN1567762A - A channel estimation method adapted for OFDMA system - Google Patents

Abstract

The invention discloses a channel estimating method applied in orthogonal frequency division multiplex system, including the steps: at a receiving end, drawing a guide frequency sequence of some user, containing channel information, from the received multiuser information, and using least-squares estimation to obtain initial channel estimation value of the guide frequency part; using Fourier transform to obtain frequency domain sequence of initial channel estimation; making amplitude limit and force-to-zero operation on the frequency domain sequence; using inverse Fourier transform to obtain accurate channel estimation value of the guide frequency sequence; applying the channel estimation value of the guide frequency part to channel characteristic estimation of non-guide frequency part, thus making channel compensation and data detection. It effectively reduces the influence of Gauss white noise and thus obtains each accurate user channel estimation value. It not only realizes lower complexity and does not increase time delay but obviously improves system property.

Description

A kind of channel estimation methods that is applicable to orthogonal frequency division multiple access system

Technical field

The present invention relates to the channel estimation methods of a kind of OFDM (OFDMA) system.

Background technology

Can predict multimedia services such as audio frequency, video, image and the Internet will become the dominant traffic of future mobile communications, and these business obviously improve (requiring peak traffic speed greater than 20Mbps) for the requirement of transmission of radio links ability.In this case, OFDM (OFDM) technology is shown up prominently.The OFDM technology is that a plurality of parallel low rate data streams that frequency spectrum overlaps sends with data stream modulates at a high speed.Because the OFDM symbol period significantly increases; therefore improved the ability of OFDM symbol anti-multipath time delay; and by increasing at the front end of OFDM symbol greater than the protection of maximum multipath time delay (GI) at interval; then can eliminate the intersymbol interference (ISI) that causes by multidiameter delay fully, simplify the burden of receiving terminal equalizer.OFDM is also adopted by a plurality of standards at present, and is acknowledged as one of physical layer basic fundamental of B3G substantially.Different sub carrier is distributed to different user, and the OFDM technology has then naturally and understandably been distinguished the user by the FDMA mode, just the OFDMA system.The OFDMA system distributes to different users neatly by the subcarrier that quadrature is adjacent, has reduced the interference in the sub-district, has improved power system capacity.And OFDMA changes by user's sub-carrier number purpose, is easy to provide the message transmission of variable Rate, so OFDMA is one of multiple access scheme of very likely adopting of B3G system.

Because coherent detection is compared the raising of the performance that 2.3-3dB is arranged with incoherent detection, so generally adopt coherent detection technology at receiving terminal.If how the employing detection technique so just must send signal to channel and estimate, promptly channel is estimated.For the channel estimation method in the OFDMA system, roughly can be divided into auxiliary and blind Channel Estimation two classes of pilot signal.Because the channel estimation method superior performance that frequency pilot sign is auxiliary, therefore present most B3G system has adopted the auxiliary channel estimation scheme of frequency pilot sign.For the auxiliary channel estimation method of frequency pilot sign, the algorithm of simple possible is young waiter in a wineshop or an inn (LS) algorithm, and for example channel estimating is asked average method after just having adopted the young waiter in a wineshop or an inn of two pilot frequency sequences to estimate in the WLAN (wireless local area network) Hiperlan2 standard.But, this algorithm is subjected to the influence of Gaussian noise obvious, this algorithm performance worsens significantly when low signal-to-noise ratio, especially for the OFDMA system, each user has only the subcarrier of part, the pilot frequency sequence yardstick is far smaller than the yardstick of inverse-Fourier transform (IFFT), thereby this algorithm can not be applicable to the OFDMA system.

Because the impulse response of wireless channel can equivalence be the filter of limited tap, and the maximum multipath time delay of channel is far smaller than the OFDM symbol period, therefore can think the impulse response of wireless channel equivalent be low pass filter.Pilot frequency sequence channel after can estimating young waiter in a wineshop or an inn like this adopts further algorithm to improve the precision of its estimation.The scholar's suggestion that for example has is carried out the IFFT conversion to known array, adopts the method for filtering to eliminate the adjacent inter-carrier interference that time varying channel causes then, improves precision of channel estimation.Some scholar's suggestion is carried out the IFFT processing to signal in addition, utilizes the correlation matrix of sequence then, carries out filtering, improves precision of channel estimation etc.But, because the complexity of these algorithms in filtering limited its possibility of its application.The proposition of especially above-mentioned algorithm all is at ofdm system, and does not consider multi-user's problem, therefore also has suitable distance from being applied to the B3G system.

Summary of the invention

At problems of the prior art, the invention provides a kind of pilot frequency sequence that in the OFDMA system, utilizes and provide the method for accurate channel estimating for the multi-user, the method comprising the steps of:

(1) from the multi-user information that receives, extracts a certain user's the pilot frequency sequence that comprises channel information at receiving terminal, utilize young waiter in a wineshop or an inn to estimate to obtain the channel estimation value of pilot portion;

(2) utilize Fourier transform, obtain the frequency domain sequence of initial channel estimation;

(3) described frequency domain sequence is carried out amplitude limit, ZF operation;

(4) utilize inverse-Fourier transform to obtain the accurate channel estimation value of described pilot frequency sequence;

(5) channel estimation value of pilot portion is applied to the channel characteristic estimation of non-pilot portion, thereby carries out channel compensation and Data Detection.

In the present invention, utilize the FFT conversion, obtained the frequency domain characteristic of each user's initial channel characteristic sequence, by amplitude limit and ZF to frequency domain channel characteristic sequence, reduce the influence of white Gaussian noise effectively, thereby obtained every accurate user channel estimation value.The present invention not only implementation complexity is lower, does not increase system's time delay, and has improved the performance of system significantly.

Description of drawings

Further describe the present invention with reference to the accompanying drawings and in conjunction with example.Wherein:

Fig. 1 shows a kind of time-frequency structure of the OFDMA system of hypothesis;

Fig. 2 shows according to receiver principle figure of the present invention;

Fig. 3 shows the more detailed block diagram according to channel estimation methods of the present invention;

When Fig. 4 showed signal to noise ratio and is 5dB, the characteristic of channel that the characteristic of channel of ideal communication channel pilot frequency locations is the result after the Fourier transform and is done the pilot frequency locations after the young waiter in a wineshop or an inn was the result after the Fourier transform;

Bit error rate was along with the curve of amplitude limit factor variations when Fig. 5 showed high speed and low speed;

Fig. 6 (a) and Fig. 6 (b) show channel estimation methods according to the present invention and the existing channel estimation methods performance curve of system's bit error rate under the different signal to noise ratio conditions at a high speed with low speed the time respectively.

Specific implementation

Fig. 1 shows a kind of time-frequency structure of OFDMA system of our hypothesis.Total sub-carrier number is 1024 in the supposing the system, wherein is used for 768 subcarriers that have of transfer of data; And insert a pilot frequency sequence every 3 OFDM symbols, form a data block, encode and detect by 13 OFDM symbols.Supposed to have simultaneously 3 users in the system, promptly each user has 256 subcarriers.A certain user's pilot tone, data mark out with oblique line among Fig. 1.

Fig. 2 is the schematic diagram of receiver.At first, from the data block that receives, solve each user's pilot frequency information respectively, carry out channel estimating then according to shown in Figure 1.Be without loss of generality, we are example with one of them user only in the following analysis.If the pilot frequency sequence that on n OFDM symbol, sends a certain user of transmitting terminal be P (n)=[p[n, 0), p (n, 1) ... p (n, K-1)] ^T, wherein (n k) is k frequency pilot sign to p, and total number of pilot symbols is K=256.At receiving terminal, can be expressed as so through the received signal behind the FFT demodulating unit:

r(n，k)＝p(n，k)H(n，k)+w(n，k) k＝0，1...，K-1(1)

Wherein (n is the white Gaussian noise on k frequency pilot sign on n the OFDM symbol k) to w, and its average is zero, and variance is σ ²(n k) is channel value on n OFDM symbol, the k frequency pilot sign to H, and it is the Fourier transform of time domain channel impulse response.

Usually, can utilize young waiter in a wineshop or an inn's algorithm to estimate channel, promptly

The mean square error of this algorithm for estimating is:

Wherein β is the ratio of frequency pilot sign power and data symbol power, and SNR is the signal to noise ratio of current system.Although we can promptly increase β value by improving the transmitted power of frequency pilot sign, improve the precision of system channel estimation to a certain extent, this method can be brought problems such as peak-to-average ratio increase, so the increase β that we can not be simply.The precision of therefore young waiter in a wineshop or an inn's algorithm for estimating depends primarily on the white Gaussian noise size that system is subjected to, and promptly white Gaussian noise is more little, and the SNR value is high more, and channel estimation errors is just more little, otherwise then evaluated error is big more.Below we will propose a kind of channel estimation scheme of DFT base, this channel estimation scheme can improve channel estimated accuracy significantly.

At first, use

The channel estimating value sequence of representing certain user.Then this sequence is carried out Fourier transform, the frequency domain value that obtains channel estimation value is:

Wherein $W_K^{\mathrm{kl}}=e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A0314251300111.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}/K}.$

Because the impulse response of wireless channel can equivalence be the filter of limited tap, and the maximum multipath time delay of channel is far smaller than the OFDM symbol period, therefore can equivalence think that the impulse response of wireless channel is low pass filter, (n is slowly to change k) can to observe channel value H in emulation.Therefore after channel estimation sequence is carried out Fourier transform, the tap of the characteristic of channel will appear at low frequency part (two of frequency spectrum), and the fluctuating of HFS mainly is because (middle part of frequency spectrum) that additive white Gaussian noise causes.Fig. 4 (a) and Fig. 4 (b) show ideal respectively, and young waiter in a wineshop or an inn estimates that the characteristic of channel is the result after the Fourier transform, and we select for use in the emulation is the channel parameter of table three A M.1225.Obviously, it is the main tap of channel that the range value behind the channel estimation sequence Fourier transform has only the part tap, and all the other mainly are noises.Therefore we have adopted the method for ZF method to come the noise of filtering HFS, promptly

ZF thresholding α wherein ₁, α ₂Selection mainly choose according to channel condition.For effective filtering noise, keep the feature of channel simultaneously again, the α that we select in Fig. 4 (a) ₁, α ₂Parameter value be 1 and 216.

Further, we can also carry out filtering to the noise of low frequency part.Main operation is to set thresholding β ₁, β ₂(β ₁＜β ₂), for greater than β ₂Perhaps less than β ₁H ₂(n l), keeps it as effectively tap, otherwise thinks the signal fluctuation that Gaussian noise causes, it is compeled value is zero, can be expressed as

Thresholding β ₁, β ₂Value can be changeless, also can be (numerical value of current noise power can utilize synchronizing sequence to estimate to obtain) that the variation with signal to noise ratio changes, simulation result shows the better performances of the adaptive threshold value that changes with signal to noise ratio.The β that we select for use in emulation ₁, β ₂Value be ${\mathrm{\&beta;}}_1=-{\mathrm{\&theta;}}^{1-\frac{\mathrm{<figure-callout\; id="10"\; label="SNR"\; filenames="A0314251300111.png"\; state="\{\{state\}\}">SNR</figure-callout>}}{10}},$ ${\mathrm{\&beta;}}_2={\mathrm{\&theta;}}^{1-\frac{\mathrm{<figure-callout\; id="10"\; label="SNR"\; filenames="A0314251300111.png"\; state="\{\{state\}\}">SNR</figure-callout>}}{10}},$ θ is the amplitude limit factor in the expression formula.Fig. 5 show at a high speed and during low speed bit error rate along with the curve of amplitude limit factor variations, the performance of system when taking all factors into consideration high speed and low speed, we have selected amplitude limit factor θ=4.0.

At last, to H ₃(n) do corresponding inverse-Fourier transform, just obtained the accurate estimated value of the characteristic of channel, promptly

After having obtained the channel estimation value of pilot portion, the method by average method or single order, second order interpolation can calculate the characteristic of channel of whole data block, carries out channel compensation and Data Detection then.

Bit error rate was along with the change curve of signal to noise ratio (snr) when Fig. 6 showed high speed and low speed, as can be seen from the figure, this patent is than young waiter in a wineshop or an inn's algorithm for estimating, performance during low speed and ideal communication channel estimated performance almost overlap, about performance during high speed and ideal communication channel estimation performance difference 2dB.Therefore no matter this patent is when high speed and low speed, has all improved channel estimated accuracy significantly, thereby has improved the performance of system.

Claims (5)

1. channel estimation methods that is applicable to orthogonal frequency division multiple access system comprises step:

(1) from the multi-user information that receives, extracts a certain user's the pilot frequency sequence that comprises channel information at receiving terminal, utilize young waiter in a wineshop or an inn to estimate to obtain the channel estimation value of pilot portion;

(2) utilize Fourier transform, obtain the frequency domain sequence of initial channel estimation;

(3) described frequency domain sequence is carried out amplitude limit, ZF operation;

(4) utilize inverse-Fourier transform to obtain the accurate channel estimation value of described pilot frequency sequence;

(5) channel estimation value of pilot portion is applied to the channel characteristic estimation of non-pilot portion, carries out channel compensation and Data Detection.

2. according to the channel estimation methods of claim 1, also comprise step:,, carry out Fourier transform then to described initial channel estimation sequence zero padding if described user's initial channel estimation sequence curtailment is 2 or 4 power;

Sequence behind the inverse-Fourier transform is only extracted the channel information of relevant position.

3. according to the channel estimation methods of claim 1, wherein in step (3), also comprise step: recently set the threshold value of filtering noise by measuring current noise, thereby realize amplitude limit.

4. according to the channel estimation methods of claim 1, wherein also comprise step in step (3): use the ZF operation adaptively at the mid portion of described frequency domain sequence, the length of this part depends on the setting of the characteristic of channel and using system parameter.

5. according to the channel estimation methods of claim 1, wherein in step (5),, utilize the channel estimation value of described pilot portion to calculate the channel characteristic estimation of non-pilot portion by the method for average method or single order, second order.

Images