CN101083644A - Channel estimation method of OFDM system - Google Patents

Abstract

The method includes following steps: (1) at transmitting terminal, designing 2D pilot frequency data structure in time domain and frequency domain; (2) at receiving end, using LS or MMSE guide lines to obtain channel estimation at symbol of pilot frequency; (3) using the said channel estimation at pilot frequency to decompose 2D channel estimation into two cascaded estimations in 1D in frequency direction and time direction to carry out interpolation filtering in order to obtain whole channel responses. Being applicable to detecting received OFDM signal, the invention reduces complexity for estimating 2D channels, system delay time, interference between sub carriers, and additive white Gaussian noise so as to improve system performance.

Description

A kind of channel estimation methods of ofdm system

Technical field

The present invention relates to a kind of channel estimation methods of wireless communication system, be specifically related to the channel estimation methods of a kind of OFDM (OFDM, Orthogonal Frequency Division Multiplexing) system.

Background technology

The OFDM technology is a kind of multi-carrier transmission technology, thereby it utilizes the orthogonal subcarrier of some to transmit the transmission that low-rate data is realized whole high-speed data concurrently.The OFDM technology has high data transmission rate, and operating factor of highband and anti-multipath decline ability have been widely used in wireless communication system.

Ofdm system has higher data speed and spectrum efficiency, but generally needs to adopt coherent demodulation technology to realize the reception of ofdm signal, and this just need carry out channel estimating.Generally speaking, channel estimation method has two kinds, and a kind of is the auxiliary channel estimation method of frequency pilot sign, and a kind of is blind algorithm for estimating.Because the operand of blind algorithm for estimating is too big, flexibility is very poor, and the application in real-time system is restricted, and therefore, present most ofdm systems adopt based on the auxiliary channel estimation method of frequency pilot sign.

Based on pilot tone assisted channel estimation method is by insert the pilot tone of some in data flow, the channel response that pilot signal and known pilot signal transmitted estimate pilot frequency locations is extracted in utilization from received signal, utilize various interpolating methods to obtain channel response on all Data Positions then.The auxiliary ofdm system of pilot tone has two kinds of pilot configurations: block pilot tone and Comb Pilot.For block pilot frequency mode, be to insert training sequence on time domain intercycle ground, each training sequence comprises all subcarriers; For the Comb Pilot pattern, be that pilot sub-carrier is inserted in the compartment of terrain in the subcarrier of each OFDM symbol.

The estimation of the channel coefficients at pilot tone place generally can be adopted least square (LS, Least Square) and least mean-square error (MMSE, Minimum Mean Square Error) method, but not the channel coefficients of pilot frequency locations can obtain by the interpolation method of one dimension or two dimension.The one dimension interpolating method is compared with two-dimensional interpolation, and it realizes simple but accuracy is not high; And the amount of calculation of traditional two-dimensional interpolation method is bigger, realizes that comparatively difficulty and system's time delay are bigger, also is difficult to suppress the interference between noise and subcarrier simultaneously.Owing to above reason, the application of these methods under high speed data transfer and abominable propagation conditions is restricted.

Summary of the invention

In view of this, the object of the present invention is to provide a kind of OFDM channel estimation methods, reduce the complexity that two dimensional channel is estimated, reduce system's time delay, reduce simultaneously to disturb and additive white Gaussian noise between subcarrier, thereby improved the ofdm system performance.

The channel estimation methods of ofdm system of the present invention may further comprise the steps:

Step 100 is in transmitting terminal structure time domain, frequency domain two-dimension pilot frequency data structure;

Step 101 obtains the channel response at frequency pilot sign place in receiving terminal estimation;

Step 102, the channel response that utilizes the frequency pilot sign place obtains the channel response of all subcarriers on the multicarrier symbol at pilot tone place carrying out the one dimension channel estimating on the frequency direction, and the channel response to above-mentioned all subcarriers on time orientation carries out the channel response that the one dimension channel estimating obtains whole frame.

The channel estimation methods of ofdm system of the present invention, in step 102, the described one dimension channel estimating of carrying out on frequency direction is to realize by the algorithm for estimating based on the FFT transform domain, specifically may further comprise the steps:

Step 102a carries out the FFT conversion to the channel response at frequency pilot sign places all on the multicarrier symbol of pilot tone place and obtains the FFT transform domain;

Step 102b, the high-frequency region component of the above-mentioned conversion of filtering in the FFT transform domain;

Step 102c adopts FFT/IFFT high-order based on zero padding to decompose the channel response of all subcarriers on the multicarrier symbol that interpolation method obtains the pilot tone place.

The channel estimation methods of ofdm system of the present invention, in step 102, the described one dimension channel estimating of carrying out on time orientation is by average method, or the first-order linear interpolation, or second order interpolation, or cubic spline interpolation, or the time domain interpolation method realizes.

The channel estimation methods of ofdm system of the present invention, in step 102, the described one dimension channel estimating of carrying out on time orientation can also realize by the algorithm for estimating based on the FFT transform domain, specifically may further comprise the steps:

Step 400 estimates that to all on each subcarrier the data that obtain and the channel response at frequency pilot sign place carry out the FFT conversion to obtain the FFT transform domain;

Step 401, the high-frequency region component of the above-mentioned conversion of filtering in above-mentioned FFT transform domain;

Step 402 adopts FFT/IFFT high-order based on zero padding to decompose interpolation method and obtains each subcarrier all channel response on time orientation.

The channel estimation methods of ofdm system of the present invention, in step 100, described two-dimension pilot frequency data structure is square or diamond structure.

The channel estimation methods of ofdm system of the present invention, in step 101, described to be used for the algorithm that the channel response at frequency pilot sign place is estimated be least square, least mean-square error channel estimation method.

From the above as can be seen, OFDM channel estimation methods provided by the invention, by two dimensional channel is estimated to be converted to the one dimension channel estimating of two-layer cascade, thereby reduce the complexity that two dimensional channel is estimated, reduced complexity and system's time delay that system realizes; This method is passed through the high fdrequency component zero setting in the FFT transform domain, and the partial additive white Gaussian noise component in the filtered signal reduces simultaneously to disturb between subcarrier, thereby improved systematic function.

Description of drawings

A kind of square two-dimension pilot frequency structural representation that Fig. 1 uses for the present invention;

A kind of rhombus two-dimension pilot frequency structural representation that Fig. 2 uses for the present invention;

Fig. 3 is a kind of channel estimation methods schematic diagram based on the FFT transform domain.

Embodiment

The present invention will be further described below in conjunction with drawings and the specific embodiments.

Because the signal distributions of multicarrier system in time domain and frequency domain, is necessary in time domain and frequency domain insertion two-dimension pilot frequency with carrying out the two dimensional channel estimation therefore.The channel estimating of time-frequency two-dimensional is carried out according to the frame of multi-carrier signal, so its data burst transmission also is that unit carries out with the frame.

(f, time-domain and frequency-domain discrete representation t) is H with channel transfer function H herein _{N, i}, n=1 wherein, 2 ..., N _c, i=1,2 ..., N _s, N wherein _cBe the subcarrier number of each multicarrier symbol, N _sBe the multicarrier symbol number that every frame comprised.The valuation of discrete channel transfer function is expressed as

, n=1 wherein, 2 ..., N _c, i=1,2 ..., N _s, frequency pilot sign is expressed as N in the spacing of frequency direction _f, be expressed as N in the spacing of time orientation _t, i.e. N _fBe illustrated respectively in the every N of frequency direction _fA frequency pilot sign is arranged, N in the distance _tBe expressed as at the every N of time orientation _tIn the distance frequency pilot sign is arranged.

A kind of OFDM channel estimation methods of the present invention based on the FFT transform domain, its specific implementation may further comprise the steps:

Step 200 is in the data structure of transmitting terminal structure time domain, frequency domain two-dimension pilot frequency.

The two-dimension pilot frequency signal all is uniformly-spaced to insert on time domain and frequency domain direction in the present embodiment, can adopt square or diamond structure, referring to illustrated in figures 1 and 2, show respectively can be used for a kind of square and rhombus of the present invention the two-dimension pilot frequency structural representation for example.Circle is represented the subcarrier of OFDM among the figure, and each row is represented a multicarrier symbol, and solid circles is represented pilot frequency information, empty circles representative data information.Wherein frequency pilot sign is spaced apart N among Fig. 1 on frequency direction _f, on time orientation, be spaced apart N _t, 4 frequency pilot signs are formed a square structure, as shown in phantom in FIG..In addition, frequency pilot sign also can be a rectangle structure.Similarly, the pilot configuration of diamond structure as shown in Figure 2,4 frequency pilot signs are formed a diamond structure.

All frequency pilot signs in one frame can be expressed as set P, the number N of frequency pilot sign _FridFor:

$N_{\mathrm{frid}}=\left[\frac{N_c}{N_f}\right]\left[\frac{N_s}{N_t}\right]=\left|\right|P\left|\right|---\left(1\right)$

Step 201 at receiving terminal, utilizes LS criterion, MMSE criterion that the channel coefficients at frequency pilot sign place is estimated, supposes that pilot tone place channel estimation results is

K=0,1 ..., M-1 represents the 1st channel coefficients to M frequency pilot sign in the multicarrier symbol respectively, wherein $M=\left[\frac{N_c}{N_f}\right]$ The number of representing the pilot tone of inserting in the multicarrier symbol, for example number of pilot tone in first row among Fig. 1.

Suppose under the frequency domain form that the transmission symbol of ofdm system transmitting terminal is X _{N, i}, the additive Gaussian noise of introducing in the transmission course is N _{N, i}, then the signal of the multicarrier frame that receives of receiving terminal is:

Y _n，i＝H _n，iX _n，i+N _n，i (n＝1，2，…，N _c，i＝1，2，…，N _s) (2)

Suppose that first frequency pilot sign is arranged in first subchannel of first OFDM symbol of frame structure, then the frequency pilot sign of square structure can be expressed as

X _n′，i′＝X _{(p-1)Nf+1，(q-1)Ni+1}， $p=\mathrm{1,2},\·\·\·,\frac{N_c}{N_f},q=\mathrm{1,2},\·\·\·,\frac{N_s}{N_t}---\left(3\right)$

Wherein:

$n^{\′}=(p-1)N_f+1,p=\mathrm{1,2},\·\·\·,\frac{N_c}{N_f}$

$i^{\′}=(q-1)N_t+1,q=\mathrm{1,2},\·\·\·,\frac{N_s}{N_t}$

Utilize LS criterion, MMSE criterion can estimate to obtain the channel coefficients at frequency pilot sign place.

Step 202 is utilized the channel estimating at frequency pilot sign place, is obtained the channel response of whole frame by algorithm for estimating.Among the present invention, estimate to be converted on the frequency direction two dimensional channel and the one dimension channel estimating of two-layer cascade on the time orientation, as shown in Figure 3, its concrete steps are as follows:

Step 202a utilizes the channel estimating at frequency pilot sign place, to the OFDM symbol that comprises frequency pilot sign at the enterprising row interpolation of frequency direction.In order to reduce between subcarrier to disturb and the influence of additive white Gaussian noise, adopt a kind of interpolation method here based on the FFT transform domain, as shown in Figure 3, the FFT transform domain here is meant signal in the frequency domain is carried out the resulting space of FFT conversion.It is as follows to utilize the FFT conversion to carry out the step of the interpolation on the frequency direction:

Step 202a0 carries out the FFT conversion that M orders to the channel response at all frequency pilot sign places on the multicarrier symbol of pilot tone place and obtains FFT transform domain G _M(p), that is, the channel response at all the frequency pilot sign places on the pilot tone column among the pilot configuration figure is carried out the FFT conversion.Here $M=\left[\frac{N_c}{N_f}\right]$ The number of representing the pilot tone of inserting in the multicarrier symbol:

$G_M\left(p\right)=\underset{k=0}{\overset{M-1}{\mathrm{\Σ}}}{\hat{H}}_{\mathrm{pilot}}\left(k\right)e^{-j2\mathrm{\πkp}/M},p=\mathrm{0,1},\·\·\·,M-1---\left(4\right)$

Step 202a1 is with the G in above-mentioned FFT transform domain medium-high frequency zone _M(p) component zero setting is to realize the HFS in low pass filter filtering and the filtering transform domain.That is:

Wherein, G ' _M(p) G of expression filtering high fdrequency component _M(p), p _C1And p _C2Be the cut-off frequency of low pass filter, p _C1And p _C2Size can be provided with according to empirical value, also can carry out drawing after the spectrum energy analysis to concrete signal.

Because G _M(p) signal energy mainly is distributed in the low-frequency range (0≤p≤p in the FFT transform domain _C1, M-p _C2The part of≤p≤M-1), noise then is distributed in whole frequency, is p with transform-domain signals by cut-off frequency therefore _C1And p _C2Low pass filter, can filtering partial noise component, keep main signal component, through after the filter, noise component(s) is reduced to the (p of former noise _C1+ p _C2)/M.Simultaneously, theory according to random process is analyzed signal, can obtain the interference (ICI between subcarrier, Inter-Carrier Interference) and additive white Gaussian noise (AWGN, Additive White GaussianNoise) all be the random process of zero-mean Gaussian Profile, and channel response is with respect to noise, its variation is slowly, therefore can be separately by low pass filter with channel response and noise, thus play the effect that reduces interference and white noise between subcarrier.

Step 202a2～202a3 according to the channel variation characteristic, adopts the FFT/IFFT high-order based on zero padding to decompose the channel coefficients that interpolation method obtains all subcarriers

Promptly

${\hat{H}}_N\left(k\right)=\frac{a}{N}\underset{q=0}{\overset{N-1}{\mathrm{\Σ}}}{G}_N\left(q\right)e^{j2\mathrm{\πqk}/N},k=\mathrm{0,1},\·\·\·,N-1---\left(6\right)$

Wherein,

$G_N\left(q\right)=\left\{\begin{array}{c}{G}_M^{\&prime;}\left(q\right),0\&le;q\&le;{\mathrm{<figure-callout\; id="1"\; label="p\; c"\; filenames="A20061001208700111.png"\; state="\{\{state\}\}">p</figure-callout>}}_c\\ 0,{\mathrm{<figure-callout\; id="1"\; label="p\; c"\; filenames="A20061001208700111.png"\; state="\{\{state\}\}">p</figure-callout>}}_c\&le;q\&le;N-p_{c2}\\ G_M^{\&prime;}(q-N+M),(N-p_{c2})\&le;q\&le;N-1\end{array}\right.---\left(7\right)$

In formula (6) and (7), N=N _c, represent the subcarrier number of each multicarrier symbol, a value is by equilibrium that real system is mediated, and loss compensates to determine to the average power after the zero padding.

Step 202b carries out interpolation respectively to each subcarrier on time orientation, promptly obtain the channel response of whole frame.Interpolation method on the time orientation can utilize the average method of prior art or first-order linear interpolation, or second order interpolation, or cubic spline interpolation, or method such as time domain interpolation, and the algorithm for estimating based on the FFT transform domain that is adopted at the enterprising row interpolation of frequency direction is herein realized.

If adopt algorithm for estimating, then specifically may further comprise the steps based on the FFT transform domain:

Step 202b0 according to formula (4), carries out the FFT conversion obtaining the FFT transform domain to all channel responses of estimating the data that obtain and frequency pilot sign place on each subcarrier, in formula this moment (4), $M=\left[\frac{N_s}{N_t}\right];$

Step 202b1, in above-mentioned FFT transform domain to the high-frequency region component zero setting of above-mentioned conversion, to realize the filtering HFS in low pass filter and the filtering transform domain;

Step 202b2 according to formula (6), adopts FFT/IFFT high-order based on zero padding to decompose interpolation method and obtains each subcarrier all channel response on time orientation, in formula this moment (6), and N=N _s

From the above as can be seen, the present invention estimates to be converted into the one dimension channel estimating of two-layer cascade by utilizing a kind of channel estimation methods based on the FFT transform domain with two dimensional channel, has reduced the complexity that two dimensional channel is estimated, thereby has reduced system's time delay; The present invention simultaneously utilizes a kind of interpolation method based on the FFT transform domain, by the high fdrequency component of transform-domain signals in this transform domain of filtering, has effectively reduced interference and additive white Gaussian noise between subcarrier, thereby has improved the performance of ofdm system.

Claims (6)

1. the channel estimation methods of an ofdm system is characterized in that, may further comprise the steps:

Step 100 is in transmitting terminal structure time domain, frequency domain two-dimension pilot frequency data structure;

Step 101 obtains the channel response at frequency pilot sign place in receiving terminal estimation;

Step 102, the channel response that utilizes the frequency pilot sign place obtains the channel response of all subcarriers on the multicarrier symbol at pilot tone place carrying out the one dimension channel estimating on the frequency direction, and the channel response to above-mentioned all subcarriers on time orientation carries out the channel response that the one dimension channel estimating obtains whole frame.

2. method according to claim 1 is characterized in that, in step 102, the described one dimension channel estimating of carrying out on frequency direction is to realize by the algorithm for estimating based on the FFT transform domain, specifically may further comprise the steps:

Step 102a carries out the FFT conversion to the channel response at frequency pilot sign places all on the multicarrier symbol of pilot tone place and obtains the FFT transform domain;

Step 102b, the high-frequency region component of the above-mentioned conversion of filtering in the FFT transform domain;

Step 102c adopts FFT/IFFT high-order based on zero padding to decompose the channel response of all subcarriers on the multicarrier symbol that interpolation method obtains the pilot tone place.

3. method according to claim 1 and 2 is characterized in that, in step 102, the described one dimension channel estimating of carrying out on time orientation is by average method, or the first-order linear interpolation, or second order interpolation, or cubic spline interpolation, or the time domain interpolation method realizes.

4. method according to claim 1 and 2 is characterized in that, in step 102, the described one dimension channel estimating of carrying out on time orientation is to realize by the algorithm for estimating based on the FFT transform domain, specifically may further comprise the steps:

Step 400 estimates that to all on each subcarrier the data that obtain and the channel response at frequency pilot sign place carry out the FFT conversion to obtain the FFT transform domain;

Step 401, the high-frequency region component of the above-mentioned conversion of filtering in above-mentioned FFT transform domain;

Step 402 adopts FFT/IFFT high-order based on zero padding to decompose interpolation method and obtains each subcarrier all channel response on time orientation.

5. method according to claim 1 is characterized in that, in step 100, described two-dimension pilot frequency data structure is square or diamond structure.

6. method according to claim 1 is characterized in that, in step 101, described to be used for the algorithm that the channel response at frequency pilot sign place is estimated be least square, least mean-square error channel estimation method.

Images