CN101534281B - Diversity channel estimate method for OFDM systems based on comb-type pilot frequency - Google Patents

Abstract

The invention discloses a diversity channel estimate method for OFDM systems based on comb-type pilot frequency, comprising the following steps: a. carrying out diversity on pilot frequencies and ensuring that the interval between adjacent pilot frequencies in each group is less than the coherence bandwidth of the channel; b. carrying out channel estimate on pilot frequencies in each group by traditional channel estimate methods and obtaining a plurality of channel estimate values of the group; c. combining the a plurality of channel estimate values of the group obtained in step b by the way of diversity combining and obtaining new combined channel estimate values. The method solves the problem of bigger error of the influence of noises on channel estimate in transform-domain channel estimate based on comb-type pilot frequency under the condition of prior information of unknown channels. According to the technical proposal of the invention, when the diversity number is L, the gain brought by the invention to MSE is 101g (1/L)dB, thereby greatly improving the property of BER.

Description

Diversity channel estimation method of OFDM system based on comb-shaped pilot frequency

Technical Field

The invention relates to the technical field of wireless mobile communication, in particular to a method for estimating a transform domain channel of an OFDM system based on comb-shaped pilot frequency.

Background

Orthogonal Frequency Division Multiplexing (OFDM) technology can transmit high-speed data streams on different subcarriers, can resist frequency selective fading and narrowband interference, is easily combined with MIMO technology, and has become the core technology of WLAN, WiMAX and LTE.

However, since the transmission signal experiences a dual selective fading channel, whether an accurate channel frequency response can be obtained at the receiver determines the performance of coherent demodulation in an OFDM communication system; particularly, after the MIMO technology is adopted to improve the system capacity, the accurate channel frequency response determines the acquisition of the diversity gain of MIMO.

In the OFDM system, the comb-type pilot is widely used because it can quickly track the characteristics of the time-varying channel and can obtain a better frequency band utilization. At the receiver, the frequency response on the other data-carrying subcarriers can be obtained by comb pilot linear interpolation, cubic spline interpolation, or the like. These methods are done either in the time domain or in the frequency domain. In recent years, transform domain based channel estimation has received increasing attention due to its fast algorithm FFT/IFFT.

In the conventional transform domain-based channel estimation method, a training symbol or a comb pilot is first subjected to Least Square (LS) estimation or Minimum Mean Square Error (MMSE) and then transformed to a transform domain (time domain), assuming that a priori information such as the length (order) of a channel impulse response is known, then in the transform domain, a zero forcing can be performed on a part except the corresponding channel impulse response length, wherein a part of the zero forcing actually corresponds to noise, and a part after the zero forcing is a corresponding part of a real channel impulse response. The channel frequency response obtained by transforming the part into the frequency domain is the frequency response after the noise reduction of the system. However, in practical cases, the length of the channel impulse response is unknown, and if the channel estimation is performed by using the transform domain method, the transform domain noise reduction is difficult.

If the number of pilots is less than the length of the cyclic prefix CP, the conventional method is to "zero-fill" at the end of the transform domain and then transform to the frequency domain, so that noise still remains on each subcarrier. Therefore, in the channel estimation method based on the comb-shaped pilot frequency in the transform domain, the transform domain only completes the interpolation of the channel frequency response on the non-pilot frequency sub-carrier, but has great loss in noise suppression. The performance of conventional transform domain channel estimation based on comb pilots and LS estimation of full pilots are the same when the taps of the channel impulse response are sampled at equal intervals if the spacing of the pilots is less than the coherence bandwidth of the channel. If the taps of the channel impulse response are sampled at unequal intervals, the estimated performance will have a "false plateau" at large snr.

In summary, since the statistical property prior information of the channel in the actual OFDM system is unknown, and the number of pilots is generally not greater than the length of the cyclic prefix of the OFDM symbol in the system parameter design, under such a condition, the noise suppression performance of the transform domain channel estimation is not enough, and the Mean Square Error (MSE) of the channel estimation is just as good as that of the LS estimation. Therefore, it is desirable to find a channel estimation method that reduces noise well.

Disclosure of Invention

In view of this, the main problem to be solved by the present invention is to reduce noise in channel estimation based on the transform domain of comb pilot, and a diversity channel estimation method is adopted, which can reduce noise well.

The technical scheme adopted by the invention is as follows:

a diversity channel estimation method based on comb pilot frequency in OFDM system includes following steps:

step a: carrying out frequency diversity processing on a received signal positioned at a comb-shaped pilot frequency position in an OFDM symbol to obtain L groups of pilot frequency groups, and ensuring that the interval of adjacent pilot frequencies in each pilot frequency group is smaller than the coherent bandwidth of a channel;

step b: b, performing channel estimation on the pilot frequency in each pilot frequency group obtained in the step a to obtain frequency domain sampling values of L groups of channels;

step c: carrying out inverse Fourier transform on the frequency domain sampling values of the groups obtained in the step b to obtain the impulse response of the L groups of channel time domains;

step d: for the L groups of impulse responses obtained in step c, first, it is determined whether the number of pilots in a group is greater than the length of the cyclic prefix. If the number of pilot frequencies in the grouping is larger than the length of the cyclic prefix, truncating and zero-forcing signals on a time domain according to the length of the cyclic prefix, and then filling zero to the length of the number of subcarrier points; if the number of pilot frequencies in the grouping is smaller than the length of the cyclic prefix, zero padding is directly carried out in the time domain, and the zero padding length is equal to the length of the number of the sub-carrier points;

step e: transforming the L groups of signals subjected to zero padding in the step d into a frequency domain by performing Fourier transform to obtain L groups of channel estimation values of transform domains;

step f: and e, carrying out diversity combination on the L groups of channel estimation values obtained in the step e according to a combination criterion to obtain the estimation of the channel frequency response of the whole OFDM symbol after noise suppression.

For quasi-static channel or slowly-varying channel, when step a is executed, the frequency diversity is adjusted to be time diversity, and the pilot frequency of response of several continuous OFDM symbols is processed with time-domain diversity type joint channel estimation.

The invention has the following beneficial effects: if the pilot frequency interval in the grouping is smaller than the coherent bandwidth of the channel, the invention shows good noise suppression characteristic, compared with the traditional transform domain channel estimation method, the method can well reduce the estimated MSE, and the method does not need any prior information of the statistics of the wireless multipath channel. Although this approach increases the complexity of the system to some extent, the resulting improvement in performance is significant. The method is suitable for the base station end to complete the pre-equalization of the transmitted signal, thus reducing the complexity of the system in the aspect of mobile terminals.

Drawings

FIG. 1 is a flow chart illustrating a conventional implementation of comb-pilot transform domain channel estimation;

FIG. 2 is a flow chart illustrating a channel estimation implementation of the present invention;

fig. 3 is a diagram illustrating the MSE (mean square error) computer simulation results for different frequency diversity channel estimation and conventional transform domain channel estimation at different SNRs (signal to noise ratios) with the number of pilots not greater than the length of the CP, according to an embodiment of the present invention;

FIG. 4 is a graph showing the results of computer simulations of BER (bit error rate) for different frequency diverse channel estimates and conventional transform domain channel estimates at different SNRs for a pilot number not greater than the length of the CP according to an embodiment of the present invention;

fig. 5 is a diagram illustrating MSE computer simulation results for different frequency diversity channel estimation and conventional transform domain channel estimation at different SNRs with a pilot number greater than the length of the CP according to an embodiment of the present invention;

FIG. 6 is a graph showing the results of a BER computer simulation for different frequency diversity channel estimation and conventional transform domain channel estimation at different SNRs for a pilot number greater than the length of the CP according to an embodiment of the present invention;

fig. 7 is a diagram illustrating MSE computer simulation results for frequency diversity channel estimation at different SNRs, L2, L4, L8, according to an embodiment of the present invention;

fig. 8 is a graph showing BER computer simulation results for frequency diversity channel estimation of L-2, L-4, L-8 at different SNRs according to an embodiment of the present invention;

Detailed Description

The invention is described in more detail below with reference to the figures and specific embodiments.

Here, a comb pilot OFDM system model is first given. After the binary information is interleaved, coded and mapped, comb-shaped pilot frequency is inserted into the frequency domain at equal intervals. The interval between two adjacent pilot carriers is <math><mrow><mi>&Delta;P</mi><mo>=</mo><mfrac><mi>N</mi><msub><mi>N</mi><mi>p</mi></msub></mfrac><mo>,</mo></mrow></math> Where N is the number of subcarriers in an OFDM symbol, N_pIs the corresponding number of subcarriers. The virtual subcarrier and the DC subcarrier are not considered here. Thus, the subcarriers and corresponding pilots that carry data in the frequency domain can be expressed as:

where X (k) is a signal carried by a subcarrier in the frequency domain, including data information and pilot, X_DCorresponding data, X_PCorresponding to the pilot. Then, the OFDM transmitter modulates x (k) with N-point idft (ifft). The time domain signal x (n) after modulation can be expressed as:

<math><mrow><mi>x</mi><mrow><mo>(</mo><mi>n</mi><mo>)</mo></mrow><mo>=</mo><mfrac><mn>1</mn><mi>N</mi></mfrac><munderover><mi>&Sigma;</mi><mrow><mi>k</mi><mo>=</mo><mn>0</mn></mrow><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></munderover><mi>X</mi><mrow><mo>(</mo><mi>k</mi><mo>)</mo></mrow><mi>exp</mi><mrow><mo>(</mo><mi>j</mi><mn>2</mn><mi>&pi;</mi><mfrac><mi>kn</mi><mi>N</mi></mfrac><mo>)</mo></mrow><mo>,</mo><mn>0</mn><mo>&le;</mo><mi>n</mi><mo>&le;</mo><mi>N</mi><mo>-</mo><mn>1</mn><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>2</mn><mo>)</mo></mrow></mrow></math>

where n is the index of the OFDM symbol time domain. After the OFDM symbols are formed, a cyclic prefix CP is added in front of each OFDM symbol, mainly to avoid inter-OFDM symbol interference. When designing a system, the length of the CP is required to be much longer than the length of the maximum multipath delay of the wireless multipath channel. The OFDM symbols then traverse a multipath channel, which is modeled here by means of a tapped delay line. For simplicity, it is assumed here that the taps of the channel are equally spaced samples. If the spacing of the channel is non-equally spaced samples, the tap gains corresponding to equally spaced sample points may be obtained by interpolation of the non-equally spaced sample taps.

<math><mrow><mi>h</mi><mrow><mo>(</mo><mi>n</mi><mo>)</mo></mrow><mo>=</mo><munderover><mi>&Sigma;</mi><mrow><mi>i</mi><mo>=</mo><mn>0</mn></mrow><mrow><mi>r</mi><mo>-</mo><mn>1</mn></mrow></munderover><msub><mi>h</mi><mi>i</mi></msub><mi>exp</mi><mrow><mo>(</mo><mi>j</mi><mfrac><mrow><mn>2</mn><mi>&pi;</mi></mrow><mi>N</mi></mfrac><msub><mi>f</mi><mi>Di</mi></msub><mi>Tn</mi><mo>)</mo></mrow><mi>&delta;</mi><mrow><mo>(</mo><mi>&lambda;</mi><mo>-</mo><msub><mi>&tau;</mi><mi>i</mi></msub><mo>)</mo></mrow><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>3</mn><mo>)</mo></mrow></mrow></math>

Where r is the total number of resolvable multipaths in the channel, h_iIs the complex gain, f, corresponding to the i-th propagation path_DiIs the Doppler shift of the ith path due to relative movement of the transceiver, T is the duration of an OFDM symbol, λ is the index of the corresponding delay spread, τ_iIs the delay time corresponding to the ith path, where τ_iIs the time normalized by the sampling interval.

Then, the received signal may be expressed as:

y(n)＝x(n)*h(n)+w(n) (4)

due to the addition of the CP, the linear convolution of the transmit signal x (n) and the channel impulse corresponding to h (n) becomes a circular convolution after the CP is removed at the receiving end. At the receiving end, assuming complete synchronization, after removing the CP, the received signal is demodulated using N-point dft (fft). Then, the signal on each subcarrier in the frequency domain can be expressed as:

Y(k)＝X(k)H(k)+W(k) (5)

where H (k) is the channel frequency response corresponding to the k-th sub-carrier, W (k) is the additive white Gaussian noise corresponding to the k-th sub-carrier, the mean is 0, and the variance is σ²。

As shown in fig. 1, a specific process of the channel estimation method based on the conventional transform domain includes the following steps:

step 101: performing LS or MMSE channel estimation for all pilots;

step 102: performing IFFT on the LS channel estimation value of the obtained pilot frequency to transform the LS channel estimation value into a time domain to obtain the impact response of the channel time domain;

step 103: judging whether the pilot number is larger than the length of the cyclic prefix CP; if the number of pilots is greater than the length of the cyclic prefix CP, go to step 104; if the pilot frequency number is less than the length of the cyclic prefix CP, executing step 106;

step 104: truncating the time domain signal after IFFT according to the CP length to force zero, and then filling zero to the length of the subcarrier points;

step 105: and performing FFT (fast Fourier transform) on the processed time domain signal to obtain a channel estimation value of a transform domain. This step is ended;

step 106: zero padding is carried out until the point length of the subcarrier is reached;

step 107: and performing FFT (fast Fourier transform) on the processed time domain signal to obtain a channel estimation value of a transform domain. This step is ended;

fig. 2 is a flow chart showing a channel estimation implementation according to an embodiment of the present invention, and the channel estimation flow according to an embodiment of the present invention is described in detail below with reference to fig. 2.

As shown in fig. 2, first, in step 201, a frequency diversity process is performed on a received signal located at a comb-shaped pilot position in an OFDM symbol to obtain L groups of pilot packets, and it is ensured that an interval between adjacent pilots in each pilot packet is smaller than a coherence bandwidth of a channel. Here, the principle of frequency diversity is to perform phase offset sampling on the pilots of the whole OFDM symbol, and each diversity phase offset is 1/L, so as to obtain L groups of pilots.

In step 202, LS or MMSE channel estimation is performed on the pilots in each packet obtained in step 201 to obtain frequency domain sample values of the channel, and then the frequency domain sample values are IFFT-transformed to the time domain to obtain the impulse response of the channel time domain. First, it is determined whether the number of pilots in a packet is greater than the cyclic prefix length. If the number of pilot frequencies in the grouping is larger than the length of the cyclic prefix, truncating and zero-forcing signals on a time domain according to the length of the cyclic prefix, and then filling zero to the length of the number of subcarrier points; if the number of pilot frequencies in the grouping is smaller than the length of the cyclic prefix, zero padding is directly carried out in the time domain, and the zero padding length is equal to the length of the number of the sub-carrier points; and finally, carrying out FFT (fast Fourier transform) on the signal subjected to zero padding to obtain a channel estimation value of a transform domain.

The LS channel estimation algorithm is only taken as an example here. The channel frequency response on the pilot subcarriers that can be obtained by the LS channel estimation algorithm can be written as:

(6)

<math><mrow><mo>=</mo><mi>H</mi><mrow><mo>(</mo><mi>k</mi><mo>)</mo></mrow><mo>+</mo><msubsup><mi>W</mi><mi>p</mi><mo>&prime;</mo></msubsup><mrow><mo>(</mo><mi>k</mi><mo>)</mo></mrow></mrow></math>

wherein, the subscript p represents pilot subcarrier, W'_p(k) It is also additive white Gaussian noise with a mean of 0 and a variance of σ². In addition, the pilot modulation scheme is BPSK modulation, i.e. | X_p(k)|²＝1。

In fact, if the channel impulse response taps are sampled at equal intervals and the spacing of adjacent pilots is less than the coherence bandwidth of the channel, then the performance exhibited by conventional transform domain channel estimation and frequency domain LS channel estimation of comb pilots is equivalent. Here, the MSE corresponding to the frequency domain LS channel estimation method may be written as:

<math><mrow><mi>MSE</mi><mo>=</mo><mfrac><mi>&beta;</mi><mi>SNR</mi></mfrac><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>7</mn><mo>)</mo></mrow></mrow></math>

wherein SNR is E { | x_k|²}/σ²Is the averaged signal to noise power ratio. Here, we assume that all subcarriers are independentCo-distributed and modulated using the same constellation. β is a constellation modulation factor, defined herein as β { | x ═ E { | x_k|²}E{|x_k|^-2}. For BPSK and QPSK, β ═ 1. For 16QAM, β 17/9.

The conventional transform domain channel estimation of the comb pilot can be expressed as:

$H_{\mathrm{DFT}}=\sqrt{N_p/N}\mathrm{FD}{F}_p^H{\tilde{H}}_{\mathrm{LS}}---\left(8\right)$

where F is the unitary matrix of FFT with element e^-j2πik/N，F_p ^HIs a unitary matrix of IFFT, corresponding to the elements therein of

Is a channel estimation value of LS corresponding to the pilot subcarrier,

is the corresponding normalization part. D is a top half diagonal matrix, used by N in the interpolation process_ppoint-to-N zero-padding, D can be expressed as:

<math><mrow><mi>D</mi><mo>=</mo><msub><mfenced open='[' close=']'><mtable><mtr><mtd><msub><mi>I</mi><mrow><msub><mi>N</mi><mi>p</mi></msub><mo>&times;</mo><msub><mi>N</mi><mi>p</mi></msub></mrow></msub></mtd><mtd><mn>0</mn></mtd></mtr><mtr><mtd><mn>0</mn></mtd><mtd><mn>0</mn></mtd></mtr></mtable></mfenced><mrow><mi>N</mi><mo>&times;</mo><mi>N</mi></mrow></msub><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>9</mn><mo>)</mo></mrow></mrow></math>

wherein,

is N_p×N_pThe identity matrix of (2). If there is a priori information on the actual channel order in the receiver, then the identity matrix dimension can be further reduced to achieve some noise suppression.

Here, we define the autocorrelation matrix of the channel as:

R_HH＝E{HH^H}＝E{Fh(Fh)^H}＝FR_hhF^H (10)

where H is the channel frequency response vector in the frequency domain, H is the impulse response vector in the time domain, and R is the impulse response vector in the time domain_hhThe method is characterized in that a corresponding autocorrelation matrix is impacted on a time domain channel, the autocorrelation matrix is a Hermitian matrix, and then Singular Value Decomposition (SVD) is carried out on the Hermitian matrix to obtain:

R_hh＝UΣU^H (11)

wherein U is a unitary matrix and Σ corresponds to R_hhAfter eigenvalue decomposition, a diagonal matrix is obtained, which can be written as: rho_iJ is 0, 1,.., N-1, and ρ₀≥ρ₁…≥ρ_N-1. Here, if the channel taps are sampled at equal intervals, then the Hermitian matrix R_hhThe number of the corresponding non-zero eigenvalues of the SVD is equal to the number of effective taps in the channel impulse response. However, this assumption is often not true, which results in leakage of energy between taps of the impulse response of the channel, with a degree of correlation between taps. Therefore, when SVD decomposition is performed, the number of non-zero eigenvalues will be greater than the number of taps in the actual channel.

We now directly give an explicit representation of the comb-pilot conventional transform-domain channel estimation MSE:

<math><mrow><mi>MSE</mi><mo>=</mo><mfrac><mi>&beta;</mi><mi>SNR</mi></mfrac><mo>+</mo><munderover><mi>&Sigma;</mi><mrow><mi>j</mi><mo>=</mo><msub><mi>N</mi><mi>p</mi></msub></mrow><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></munderover><msub><mi>&rho;</mi><mi>j</mi></msub><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>12</mn><mo>)</mo></mrow></mrow></math>

we can consider (12) in two parts. The first part is

Including the modulation factor and the signal-to-noise ratio, which are determined at the time the system is designed. The second part is

Which corresponds to the eigenvalues of the autocorrelation matrix of the time-domain channel impulse response. Design N, as we discussed earlier_pThe value is greater than the number of taps of the channel. If the channel taps are sampled at equal intervals, the energy of each tap is concentrated on the respective tap. And the length of the CP is larger than the maximum delay time of the channel, we can obtain rho_j＝0，j＝N_p,.., N-1. (12) Can be written as <math><mrow><mi>MSE</mi><mo>=</mo><mfrac><mi>&beta;</mi><mi>SNR</mi></mfrac><mo>.</mo></mrow></math> Then, according to (7), we can make the performance of channel estimation of comb pilots and the performance of LS channel estimation in frequency domain equivalent under the previous assumption. However, if the channel is sampled at unequal intervals, or if the sampling time does not match the channel, this will result in a portion of ρ_jIs not equal to 0, j > r, where r is the number of separable multipath paths in the channel discussed in (3) above, i.e., the number of taps of the corresponding channel. Therefore, this results in <math><mrow><munderover><mi>&Sigma;</mi><mrow><mi>j</mi><mo>=</mo><msub><mi>N</mi><mi>p</mi></msub></mrow><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></munderover><msub><mi>&rho;</mi><mi>j</mi></msub><mo>&NotEqual;</mo><mn>0</mn><mo>.</mo></mrow></math> In this case, when under conditions of large SNR, and

in contrast to the above-mentioned results, <math><mrow><munderover><mi>&Sigma;</mi><mrow><mi>j</mi><mo>=</mo><msub><mi>N</mi><mi>p</mi></msub></mrow><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></munderover><msub><mi>&rho;</mi><mi>j</mi></msub><mo>&NotEqual;</mo><mn>0</mn></mrow></math> will be relatively large, so in (12), MSE

Occupying the major part. We define this section here as:

<math><mrow><mi>errorfloor</mi><mo>=</mo><munderover><mi>&Sigma;</mi><mrow><mi>j</mi><mo>=</mo><msub><mi>N</mi><mi>p</mi></msub></mrow><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></munderover><msub><mi>&rho;</mi><mi>j</mi></msub><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>13</mn><mo>)</mo></mrow></mrow></math>

the error floor is a constant quantity under certain channel conditions. (12) The formula can be written as:

<math><mrow><mi>MSE</mi><mo>=</mo><mfrac><mi>&beta;</mi><mi>SNR</mi></mfrac><mo>+</mo><mi>errorfloor</mi><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>14</mn><mo>)</mo></mrow></mrow></math>

therefore, from the performance point of view, we can use the frequency domain LS estimation method to obtain the channel estimation in the comb pilot transform domain, which can be written as:

H_TF＝H_LS+e＝H+W+e (15)

wherein,ethe method is an 'error flat bottom' caused by leakage caused by the non-equal-interval sampling mentioned above in the channel frequency response estimation, namely, the signal-to-noise ratio of transmission is increased, and the performance is not improved.

In step 203, diversity combining processing is performed on the channel estimation values obtained in step 202. The merging criterion adopts equal probability merging criterion or maximum ratio merging criterion or equal gain merging criterion or optimal selection merging criterion.

In step 204, an estimate of the noise-suppressed channel frequency response of the entire OFDM symbol is obtained from step 203.

It is assumed here that the channel estimate of the l-th group, whose estimated value is H_TF1And the weighted value of the corresponding channel estimation is alpha_lAnd has an alpha₀+α₁+…+α _L-11, then the channel estimate H after diversity combining_NewComprises the following steps:

H_New＝α₀H_TF0+α₁H_TF1+…α_L-1H_TFL-1 (16)

for simplicity, we illustrate L ═ 2.

Two sets of pilots can be expressed as X_{p_0}∪X_{p_1}＝X_pWherein:

$\left\{\begin{array}{c}{X}_{p\_0}\left(i\right)=X_p\left(2i\right)\\ X_{p\_1}\left(i\right)=X_p(2i+1)\end{array}\right.---\left(17\right)$

by (17), it can be found that X is actually_{p_0}Is corresponding to X_pPilot of odd position, X_{p_1}Is corresponding to X_pEven-numbered position pilots in (1). From these two sets of pilots, conventional transform domain channel estimation can be performed on their independent pilots. After obtaining two sets of estimated values, a certain diversity combining mode, such as equal probability combining, is adopted for the two sets of channel estimated values. The result can be written as:

H_New＝αH_TF0+(1-α)H_TF1 (18)

wherein H_NewIs the channel estimate of the diversity combining, H_TF0And H_TF1And alpha is a weight coefficient for estimating and combining the odd-numbered group pilot frequency. Due to the consistency (15), (18) of the comb pilot conventional transform domain channel estimation and the frequency domain LS channel estimation can be written as:

H_New＝H+e+αW₀+(1-α)W₁ (19)

wherein, W₀And W₁Is independent additive white Gaussian noise with zero mean and variance of σ². When equal probability combining is used, i.e., α ═ 1/2, (19) can be expressed as:

<math><mrow><msub><mi>H</mi><mi>New</mi></msub><mo>=</mo><mi>H</mi><mo>+</mo><munder><mi>e</mi><mo>&OverBar;</mo></munder><mo>+</mo><mfrac><mn>1</mn><mn>2</mn></mfrac><mrow><mo>(</mo><msub><mi>W</mi><mn>0</mn></msub><mo>+</mo><msub><mi>W</mi><mn>1</mn></msub><mo>)</mo></mrow><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>20</mn><mo>)</mo></mrow></mrow></math>

as can be seen from (20) above,

it is also additive white Gaussian noise with a mean of 0 and a variance of

Due to the consistency discussed above, the MSE corresponding to the transform domain frequency diversity channel estimation of the comb pilots we propose can be expressed as:

<math><mrow><mi>MSE</mi><mo>=</mo><mfrac><mn>1</mn><mn>2</mn></mfrac><mfrac><mi>&beta;</mi><mi>SNR</mi></mfrac><mo>+</mo><mi>errorfloor</mi><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>21</mn><mo>)</mo></mrow></mrow></math>

we can see from (21) that when using L ═ 2 frequency diversity, the diversity gain of MSE is 3dB if the spacing of pilots inside the packet is smaller than the coherence bandwidth of the channel; when L-4 frequency diversity is used, the MSE has a diversity gain of 6 dB; when using L-8 frequency diversity, the MSE has a diversity gain of 9 dB.

Theoretically, if complexity is not taken into account, L-diversity will result when the spacing of pilots in a packet is less than the coherence bandwidth of the channel

MSE may be expressed as:

<math><mrow><mi>MSE</mi><mo>=</mo><mfrac><mn>1</mn><mi>L</mi></mfrac><mfrac><mi>&beta;</mi><mi>SNR</mi></mfrac><mo>+</mo><mi>errorfloor</mi><mo>-</mo><mo>-</mo><mo>-</mo><mrow><mo>(</mo><mn>22</mn><mo>)</mo></mrow></mrow></math>

the gain on the MSE due to frequency diversity will greatly improve the performance of the bit error rate. And, another advantage of this diversity channel estimation method is that it does not require any a priori information of the statistics of the wireless multipath channel.

The beneficial effects of the solution of the present invention are analyzed with reference to fig. 3, fig. 4, fig. 5, fig. 6, fig. 7, and fig. 8. The computer simulation parameters of the present invention are shown in the following table:

parameter value of parameter

Bandwidth of 10MHz

FFT/IFFT 1024

CP Length 128

OFDM symbol period 115.2us

Pilot/data modulation BPSK (β ═ 1)

N_p Case 1：64

Case 2：128

Case 3：256

ΔP Case 1：1024/64＝16

Case 2：1024/128＝8

Case 3：1024/256＝4

The channel model is an ITU-R vehicular A channel model, and the multipath channel is modeled by adopting a tapped delay line structure. The simulated corresponding center frequency is 2.3 GHz.

The simulation results of fig. 3 and 4 are corresponding different frequency diversity channel estimation and conventional transform domain channel under different SNREstimated MSE and BER. The results of the asterisk connection and triangle connection simulations correspond to the two cases, Case 1 and Case 2, of table 1. In both cases, the number of pilots is not greater than the length of the CP. The circled line is the corresponding frequency diversity L-2 (N) we propose_p＝64，ΔP＝16，N_{p_2}＝32，ΔP₂32) MSE performance. The box connection is the corresponding frequency diversity L-4 (N) we propose_p＝128，ΔP＝8，N_{p_4}＝32，ΔP₄32) MSE performance.

As can be seen from fig. 3 and 4, when the number of pilots is not greater than the CP length, the performance without using frequency diversity is the same as that of LS estimation, and MSE is 1/SNR. We can see that the performance of conventional transform domain channel estimation does not improve when the number of pilots increases. When using a channel estimation with frequency diversity L2, the diversity gain of MSE is 3dB compared to the channel estimation in the conventional transform domain. When using a channel estimate with frequency diversity L-4, the diversity gain of the MSE is 6 dB. As can be seen from fig. 3, the gain of MSE improves the performance of BER, and when L is 2, the BER gain is 1.5dB, and when L is 4, the BER gain is greater than 2 dB.

The simulation results of fig. 5 and fig. 6 are the MSE and BER for the corresponding frequency diversity channel estimation and conventional transform domain channel estimation at different SNRs. Under this simulation condition, the number of pilots is greater than the length of the CP. The block connection corresponds to L ═ 8 (N) we propose_p＝256，ΔP＝4，N_{p_8}＝32，ΔP₈32) of the substrate. The diamond connection corresponds to the traditional transform domain channel estimation method. Here, when the total number of pilots exceeds the length of the CP, the conventional transform domain may perform some noise suppression in the time domain, and this method performs "zero forcing" on the points exceeding the number of pilots in the time domain. We can see that the frequency diversity channel estimation method is due to this time domain "zero forcing" noise reduction estimation method. However, when L is 8, the complexity of the frequency diversity method is significantly increased. Therefore, a balance in complexity and performance is required when designing the system.

Fig. 7 and 8 are graphs of MSE and BER performance for different diversity L2, L4, L8 at different signal-to-noise ratios. It can be seen that, in the channel estimation based on frequency diversity, as the diversity number increases, the noise suppression performance further improves, and the system performance also improves. When L is 8, the BER performance of the diversity channel estimation method differs by less than 0.8dB from the ideal channel estimation.

The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.

Claims (5)

1. A diversity channel estimation method based on comb pilot frequency in OFDM system is characterized in that the method includes the following steps:

step a: carrying out frequency diversity processing on a received signal positioned at a comb-shaped pilot frequency position in an OFDM symbol to obtain L groups of pilot frequency groups, and ensuring that the interval of adjacent pilot frequencies in each pilot frequency group is smaller than the coherent bandwidth of a channel;

step b: b, performing channel estimation on the pilot frequency in each pilot frequency group obtained in the step a to obtain frequency domain sampling values of L groups of channels;

step c: carrying out inverse Fourier transform on the frequency domain sampling values of the groups obtained in the step b to obtain the impulse response of the L groups of channel time domains;

step d: for the L groups of impact responses obtained in the step c, firstly judging whether the number of the pilot frequencies in the groups is greater than the length of the cyclic prefix, if the number of the pilot frequencies in the groups is greater than the length of the cyclic prefix, truncating the signals on the time domain according to the length of the cyclic prefix, forcing to zero, and then filling zero to the number length of the subcarrier points; if the number of pilot frequencies in the grouping is smaller than the length of the cyclic prefix, zero padding is directly carried out in the time domain, and the zero padding length is equal to the length of the number of the sub-carrier points;

step e: transforming the L groups of signals subjected to zero padding in the step d into a frequency domain by performing Fourier transform to obtain L groups of channel estimation values of transform domains;

step f: and e, carrying out diversity combination on the L groups of channel estimation values obtained in the step e according to a combination criterion to obtain the estimation of the channel frequency response of the whole OFDM symbol after noise suppression.

2. A diversity channel estimation method according to claim 1, characterized in that the diversity processing in step a is sampling the pilot frequency of the whole OFDM symbol by phase shift, each diversity phase shift is 1/L, resulting in L groups of pilot frequency packets.

3. A diversity channel estimation method according to claim 1, characterized in that the channel estimation performed in step b is by using a least square algorithm or a least mean square error algorithm.

4. A diversity channel estimation method according to claim 1, characterized in that in step f, said combining criterion uses an equal probability combining criterion or a maximum ratio combining criterion or an equal gain combining criterion or a best choice combining criterion.

5. A diversity channel estimation method according to claim 1, characterized in that for quasi-static channels or slowly varying channels, the diversity process is a time diversity process, and the pilots of the responses of consecutive OFDM symbols are subjected to time domain diversity joint channel estimation.

Images