TWI532340B - Inverse-channel-based blind channel estimation method - Google Patents

Description

Reverse channel blind channel estimation method

The present invention relates to a wireless communication technology, and more particularly to a reverse channel type blind channel estimation method for an orthogonal frequency division multiplexing system.

The blind channel estimation method has been widely used for block transmission of Cyclically Prefixed Orthogonal Frequency-Division Multiplexing (CP-OFDM) systems with cyclic prefixes. Provides tolerable estimation performance without inserting pilot symbols or training symbols into subcarriers within a specific time interval. Save bandwidth to increase bandwidth efficiency.

The Subspace-based Blind Channel Estimation (SBCE) method is a well-known technique that does not generate, for example, a finite alphabet or a white Gaussian assumption technique on a transmitted signal. Additional restrictions. However, when the subspace blind channel estimation method is used in an orthogonal frequency division multiplexing system (OFDM) with a virtual subcarrier (VC), the orthogonal frequency division multiplexing system needs to be utilized. Redundancy between the virtual subcarrier (VC) and the Cyclic Prefix (CP), and requires a large number of receiving blocks to make accurate blind channel estimation.

Another conventional Repetition-based Blind Channel Estimation (RPBCE) method is implemented by repeatedly operating in each receiving block to form a rank deficiency matrix and The channel information is extracted therefrom, so that the number of receiving blocks of the subspace blind channel estimation (SBCE) can be greatly reduced, but the computational complexity of the repeated blind channel estimation (RPBCE) method (computational complexity) ) is very high, especially when there are a large number of repeated operations.

In order to reduce the implementation complexity, a person in the art has proposed a Remodulation-based Blind Channel Estimation (RMBCE) method, which utilizes two consecutive blocks (consecutive blocks) and A composite block of cyclic prefixes between two consecutive blocks to produce less computational complexity than the Repetitive Blind Channel Estimation (RPBCE) method. Although the operation of the Modulated Blind Channel Estimation (RMBCE) method is similar to the Repetitive Blind Channel Estimation (RPBCE) method, when the number of receiving blocks is limited, for example, the number of receiving blocks is not greater than Channel estimation cannot be performed for the number of subcarriers in the crossover multiplex system.

In order to relax the restrictions, people in the field have proposed a Repetition and Remodulation-based Blind Channel Estimation (RRBCE) method based on repeated blind channel estimation. The RPBCE method repeatedly processes each composite block and improves the estimated performance when the number of received blocks is limited.

The above methods of the prior art are all designed under a mechanism, that is, by processing a large number of received signals to estimate a predetermined zero-nuclear subspace corresponding to a vector subspace of a received signal, and using a channel impulse response (CIR) The vector is orthogonal to the zero-nuclear subspace, such that the channel impulse response can be separated from the received signal after projecting the vector of received signals into the estimated zero-nuclear subspace, and by minimizing the received signal The quadratic cost function established by the vector is identified. However, the estimated zero-nuclear subspace is affected by the noise of the noise and fading, and the efficiency of the channel estimation is significantly reduced, especially in the receiving block. When the number of Signal-to-Noise Radio (SNR) is small.

Therefore, how to overcome the problems of the above-mentioned prior art has become a problem that is currently being solved.

The present invention provides a reverse channel blind channel estimation method, which includes: generating a transmission signal having a symmetric cyclic prefix; and parsing a predetermined zero nuclear subspace and its corresponding base vector from the transmitted signal; The transmission signal is transmitted to the receiving end via the channel; the first matrix is calculated according to the base vector, and the second matrix is calculated according to the received signal of the first matrix and the receiving end; the non-negative limiting matrix is calculated according to the second matrix; and the non-negative limiting matrix is performed The eigenvalue decomposition is performed to generate a eigenvector; and the information of the channel gain of the channel is calculated according to the eigenvector.

The transmission signal with the symmetric cyclic prefix can be modulated by the orthogonal frequency division multiplexing modulator to transmit the transmission signal of the transmitting end, and the predetermined zero nuclear subspace and the base vector can reduce the interference on the transmitted signal. The transmission signal can be combined with the noise of the channel and varies according to the channel gain. The receiving signal of the receiving end can contain the noise of the predetermined zero-nuclear subspace, the base vector and the channel.

The base vector can generate a first matrix of ( N + N _g ) x N _g dimensions, the first matrix can be multiplied by the received signal to calculate the second matrix to remove the transmitted signal, and N is the number of subcarriers of the transmitted signal at the transmitting end. , N + N _g is the number of subcarriers of the transmitted signal with a symmetric cyclic prefix.

The first matrix is U and is expressed as follows:

Wherein, I _Ng is an N _g x N _g unit matrix, O _Ng is a N _g x N _g zero matrix, N is the number of subcarriers of the transmission signal of the transmitting end, and N _g is the number of subcarriers of the cyclic prefix of the transmitted signal, t is Transpose.

The second matrix is W and is expressed as follows:

Where U is the first matrix, R is the received signal, N _b is the number of blocks receiving the signal, and t is the transposition.

The non-negatively defined matrix is equal to the transpose of the second matrix multiplied by the second matrix. The eigenvectors correspond to the minimum eigenvalues of the non-negative finite matrix, and the eigenvectors are And expressed as follows:

Where c is a vector, † is a conjugate transpose, and W is a second matrix.

The channel gain is h and is expressed as follows:

Where T _V ^(K) is a matrix, K and V are both values, c is a vector, † is a conjugate transpose, and b is a unit vector.

It can be seen from the above that the reverse channel blind channel estimation method of the present invention mainly generates a transmission signal having a symmetric cyclic prefix, and parses out a predetermined zero-nuclear subspace and a base vector from the transmitted signal, and according to the base vector. The first matrix, the second matrix, the non-negative qualified matrix and the eigenvector are calculated in sequence, and the information of the channel gain is calculated.

Therefore, the present invention can suppress the influence of the unknown transmission signal on the channel estimation even under a small number of received signals, and maintain low computational complexity, and quickly extract reliable channel information from the received signal. In addition, it was confirmed by mathematical analysis that the present invention can provide reliable estimation results in different channel environments.

1‧‧‧Reverse channel blind channel estimation system

11‧‧‧Transport

12‧‧‧Orthogonal Frequency Division Multiplex Modulator

121‧‧‧Anti-fast Fourier Transformer

122‧‧‧Parallel to Sequence Converter

123‧‧‧Circular prefix insertion unit

13‧‧‧ Analog to digital converter

14‧‧‧ passage

15‧‧‧Time domain zero equalizer

16‧‧‧Digital to analog converter

17‧‧‧ Receiver

A1 to A3‧‧‧ Curve

B1 to B3‧‧‧ Curve

C1 to C2‧‧‧ Curve

c ‧‧‧ vector

CP‧‧‧ cycle prefix

d, s‧‧‧ transmission signal

H‧‧‧channel gain

N , N _g ‧‧‧ number of subcarriers

R‧‧‧ receiving signal

S21 to S27‧‧‧ steps

1A is a schematic diagram showing the architecture of the reverse channel type blind channel estimation system of the present invention; FIG. 1B is a schematic diagram showing the transmission signal of the first embodiment of the present invention converted into a transmission signal having a symmetric cyclic prefix. 2 is a schematic flow chart showing the reverse channel type blind channel estimation method of the present invention; FIG. 3A is a diagram showing the inverse channel type blind channel estimation (ICBCE) method and the prior art of the present invention. Overlay and remodulation blind channel estimation (RRBCE) method is a schematic diagram of the mean square error and signal noise ratio; FIG. 3B is a diagram showing the reverse channel blind channel estimation (ICBCE) method of the present invention and the repetition and remodulation of the prior art The blind channel estimation (RRBCE) method is another curve diagram of the mean square error and the signal noise ratio; and the fourth figure shows the reverse channel blind channel estimation (ICBCE) method and the prior art of the present invention. Schematic diagram of the mean square error and signal noise ratio of the Modulated Blind Channel Estimation (RMBCE) method and the Repeated and Remodulated Blind Channel Estimation (RRBCE) method.

The other embodiments of the present invention will be readily understood by those skilled in the art from this disclosure.

It is to be understood that the structure, the proportions, the size and the like of the present invention are intended to be in accordance with the disclosure of the specification, and the understanding and reading of those skilled in the art are not intended to limit the invention. The conditions are limited, so it is not technically meaningful. Any modification of the structure, change of the proportional relationship or adjustment of the size should remain in this book without affecting the effects and the objectives that can be achieved by the present invention. The technical content disclosed in the invention can be covered.

In the meantime, the terms "upper", "one", "first" and "second" are used to describe the scope of the invention, and are not intended to limit the scope of the invention. Changes or adjustments to the relative relationship are considered to be within the scope of the invention without departing from the scope of the invention.

The elements or mathematical symbols of the present invention include: b for a unit vector, c for a vector, d, d ^(l) , s and s ^(l) for transmitting signals, e for error vector, h for channel gain, I _K represents a K × K unit matrix, K represents a numerical value, N and N _g represent the number of subcarriers, n represents an integer, O _{N, M} represents an N × M all zero matrix, 0 _K represents a K × 1 all zero matrix, and r represents reception Signal, t denotes transpose, T n ^{( K )} (x) denotes a matrix of ( K + n -1) × n having the first row [x ^t , 0 _{n -1} ^t ] ^t , † denotes conjugate Hermitian transpose, U represents a matrix of dimensions ( N + N _g ) × Ng , V represents the length of the channel impulse response (CIR), and X represents the minimum eigenvalue, x Represents the smallest integer greater than or equal to x (Gaussian notation), [x _k ; k Z _K ] represents a vector having the kth term entity x _k as K × 1, [x _m ; m Z _M ] represents a matrix having the mth row vector x _m , Z _K represents a set of {0, 1, ..., K -1}, and Z _K ⁺ represents a set of {1, 2, ..., K } , δ _{1, k} represents the Kronecker Delta function.

1A is a schematic diagram showing the architecture of a reverse channel type blind channel estimation system according to the present invention, and FIG. 1B is a schematic diagram showing the transmission of the transmission signal of FIG. 1A of the present invention into a transmission signal having a symmetric cyclic prefix. .

As shown in FIG. 1A and FIG. 1B, the transmitting end 11 has N subcarriers, and each subcarrier generates a composite data symbol or a transmission signal at the first signal time. ]. The transmission signal d ^(l) is sequentially composed of an inverse fast Fourier Transform (IFFT) 121 and a parallel transcoder (Parallel-to) of an OFDM Modulator 12 -Serial Converter) 122 performs processing and inserts or generates a cyclic prefix CP having a number of N _g subcarriers at the end of the transmission signal d ^(l) by a Cyclic Prefix Insertion Unit 123, so that The transmission signal d ^(l) forms a transmission signal s having a symmetric cyclic prefix CP.

The analog-to-digital converter 13 converts the transmitted signal s into a carrier waveform and sequentially passes through channel 14 (eg, Multipath Fading Channel), time domain zero-forcing equalizer (Time-domain Zero). The -forcing Equalizer (TZE) 15 and the digital to analog converter 16 transmit the transmission signal s to the receiving end 17.

At the receiving end 17, the waveform of the received signal r is processed and sampled to generate N + N _g complex-valued samples or received signals. , as shown in the following formula (1).

among them, The transmission signal s ^(l) corresponds to the transmission signal d ^(l) . When n <Z _Ng, transmission signal s ^(l) having _{^{s n (l) = S N}} + n (l); n in When Z _N , the transmission signal s ^(l) has ; H _v is the channel impulse response based (CIR) of the v-th channel gain , z _n ^(l) are independent, identically distributed, and have E{z _n ^(l) }=0 and The cyclically symmetric-symmetric complex Gaussian (CSCG) noise sample, E{‧} represents the expected value, and σ represents the correlation coefficient of the noise.

All transmitted signals d _n ^(l) are assumed to be uncorrelated, mean zero, and have unit variance, and all transmitted signals s _n (l) at n Z _N+Ng -Z _{Ng is} also assumed to be uncorrelated, the mean is zero, and has a unit square difference. The transmission signal s _n ^(l) , the channel gain h _v and the noise sample z _n ^(l) are assumed to be independent of each other. When the I-ary subcarrier is modulated, the average received signal-to-noise ratio (SNR) of each bit is When operating on the received signal r ^(l) , the channel impulse response (CIR) can be identified by the reverse channel blind channel estimation method of the present invention.

FIG. 2 is a schematic flow chart showing the back channel blind channel estimation method of the present invention.

As shown in step S21, a transmission signal having a symmetric cyclic prefix is first generated, and the transmission signal having the symmetric cyclic prefix can be modulated by the orthogonal frequency division multiplexing modulator to transmit the transmission signal of the transmitting end. Then it proceeds to step S22.

As shown in step S22, a predetermined zero-nuclear subspace and its corresponding base vector are parsed from the transmitted signal, and the predetermined zero-nuclear subspace and the base vector can reduce interference with the transmitted signal, and then proceeds to step S23.

As shown in step S23, the transmission signal is transmitted to the receiving end via the channel, and the transmission signal is combined with the noise of the channel and varies according to the channel gain. The receiving signal of the receiving end contains the predetermined zero-nuclear subspace, the base vector and the channel. Then, proceed to step S24.

As shown in step S24, the first matrix is calculated according to the base vector, and the second matrix is calculated according to the received signals of the first matrix and the receiving end.

The base vector generates a first matrix of ( N + N _g ) x N _g dimensions, the first matrix is multiplied by the received signal system to calculate the second matrix to remove the transmission signal, and N is the number of subcarriers of the transmission signal of the transmitting end. , N + N _g is the number of subcarriers of the transmitted signal with a symmetric cyclic prefix.

The first matrix is U and is expressed as follows:

Wherein, I _Ng is an N _g x N _g unit matrix, O _Ng is a N _g x N _g zero matrix, N is the number of subcarriers of the transmission signal of the transmitting end, and N _g is the number of subcarriers of the cyclic prefix of the transmitted signal, t is Transpose.

The second matrix is W and is expressed as follows:

Where U is the first matrix, R is the received signal, N _b is the number of blocks receiving the signal, and t is the transposition. Next, the process proceeds to step S25.

As shown in step S25, a non-negative limiting matrix is calculated according to the second matrix, and the non-negative limiting matrix may be ( W † W ), and W is the second matrix. Next, the process proceeds to step S26.

As shown in step S26, the non-negative limiting matrix is subjected to eigenvalue decomposition to generate a feature vector, thereby reducing the error vector of the channel gain and the feature vector after convolution.

The eigenvectors correspond to the minimum eigenvalues of the non-negative finite matrix, and the eigenvectors are And expressed as follows:

Where c is a vector, † is a conjugate transpose, and W is a second matrix. Next, the process proceeds to step S27.

As shown in step S27, the channel gain information of the channel is calculated based on the feature vector.

The channel gain is h and is expressed as follows:

Where T _V ^(K) is a matrix, K and V are both values, c is a vector, † is a conjugate transpose, and b is a unit vector.

In the reverse channel blind channel estimation method (ICBCE) of the present invention, the inverse channel information is extracted by operating a predetermined zero subspace corresponding to the transmitted signal. The predetermined zero-nuclear subspace and time-domain zero-forcing equalizer (TZE) are described below.

(1) Schedule zero-core subspace: the transmission signal is When n In Z _Ng , since the transmission signal s _n ^(l) is equal to the transmission signal s _N+n ^(l) such that s _n ^{(l) -} s _N+n ^(l) =0, the transmission signal s ^(l) will have N _g redundant symbols, and a predetermined zero-nuclear subspace of the N _g dimension exists in the transmitted signal s ^(l) . Therefore, the present invention defines a first matrix U of ( N + N _g ) x N _g dimensions in the transmitted signal s ^(l) , and the first matrix U includes N _g basis vectors corresponding to the predetermined zero-nuclear subspace, All transmission signals s ^(l) satisfy U ^t U = I _Ng and U ^t s ^(l) = 0 _Ng as shown in the following formula (2):

The first matrix U is an independent data symbol, and the row vector can span a predetermined zero-core subspace for the transmission signal s ⁽¹⁾ , and the first matrix U can also be used by the receiving end 16 to form a Extraction subspace.

(2) Time Domain Zero Forcing Equalizer (TZE): It compensates for channel impulse response (CIR) and contains reverse channel information. The time domain zero-forcing equalizer has an infinite impulse response and infinite impulses. The response can be obtained by the K- dimensional finite impulse response (Finite Impulse Response, FIR) vector Approximate, where K is set to K >> V and the time domain zero-forcing equalizer can achieve channel gain with channel impulse response (CIR) The convolve with the vector c is close to the unit vector b at the square distance, as shown in the following formula (3):

Among them, the value Unit vector

In addition, in the reverse channel blind channel estimation method (ICBCE) of the present invention, according to the above formula (3) and any channel gain h, the least-square (LS) solution of the vector c can be as follows Formula (4):

Similarly, the channel gain h can also be calculated from the vector c of the least mean square solution, as shown in the following equation (5):

The error vector of T _K ^{( v )} ( h ) c and unit vector b according to the vector c of the least mean square solution of the above formula (4) It can be substituted and characterized by the following formula (6):

Wherein, the vector c and the channel gain h are associated by the following formula (7):

Where m Z _{K + V -1} ,h _v =0, v Z v .

As shown in the above formula (1), the transmission signal s _n ^{(l) is} convolved with the channel gain h _v , so when the transmission signal s _n ^(l) is unknown, the estimation of the channel gain h is subject to The interference of the transmitted signal s _n ^(l) . Therefore, the present invention changes the channel gain h of the truncated inverse by a predetermined zero-nuclear subspace (eg, the vector c satisfies the formula (7)), and indirectly obtains from the estimate of the vector c from the formula (5). Estimation of channel gain h. In order to estimate the vector c and not be affected by the transmitted signal s _n ^(l) , the present invention converts the received signal r _n ^(l) into the following formula (8) from equation (1):

Among them, when And Time,

The above formula (8) can also be expressed in vector form as the following formula (9):

Among them, when And Time, , , ,

The transmit signal s _n ^(l) is unknown to the receiving end 16 and can be varied by exponent 1 and exponent n . When the receiving end 16 directly estimates the channel gain h from the received signal r ^(l) , The variation of the transmitted signal s _n ^(l) interferes with the estimation of the channel gain h and significantly reduces the estimation performance of the channel gain h.

In order to suppress signal interference that is operated in a predetermined zero-core subspace, the present invention stacks N + N _g transmission signals and And rewrite the above formula (9) as the following formula (10):

Among them, when When receiving, the signal is as follows:

In the above formula (10), the left and right sides are multiplied by the transpose U ^{t of the} first matrix U and U ^t s ^(l) = 0 _Ng , thereby removing the transmission signal s _n ^(l) without interfering with the channel. Estimate the gain h and obtain the following formula (11):

Among them, when l When Z _Nb ⁺ , Is an approximate value obtained from the time domain zero-forcing equalizer (TZE), so in equation (11), Cannot be removed by increasing the signal noise ratio γ _b , but This can be reduced by setting the value K to be much greater than the length V of the channel impulse response (CIR).

In addition, the vector c is for l Z _Nb ⁺ is static, so R ^(l) , R ⁽²⁾ , ..., R ^{( Nb )} can be combined to improve the estimation of vector c . Therefore, the above formula (11) is at 1 Z _Nb ⁺ can be combined, and the result is that when the dimension K is much larger than the length V of the channel impulse response (CIR), Wc Zc ( W is multiplied by c to approximate Z multiplied by c ), versus For two N _b N _g x K matrix, the second matrix W is calculated based on the first received signal R and the receiving end of the matrix U, a second matrix W at the receiving end to be observed, noise including all noise Z Sample, so Z _c (noise Z multiplied by vector c ) represents the common interference from the noise and the channel. Considering the second matrix W , the eigenvector of the channel It is calculated by minimizing the mean square interference ∥Z c ∥ ² as shown in the following formula (12):

Based on the K-dimensional mean square solution of equation (12), when N _b N _g ≧ K , the eigenvector of the channel It can be calculated by EigenValue Decomposition (EVD). Since the nonnegative definite matrix is equal to the transpose of the second matrix multiplied by the second matrix ( W † W ), the non-negative qualified matrix ( W † W ) can be eigenvalue-decomposed to generate a eigenvector ,that is =eig( W † W ), where eig(X) is represented as an eigenvector corresponding to the minimum eigenvalue X. According to formula (5) and eigenvector , replace c _k with And h _{v is} replaced by , the estimated value of the channel impulse response (CIR) can be calculated Estimated value This is the information of the channel gain h. By the above method, the reverse channel blind channel estimation (ICBCE) method of the present invention can be obtained.

FIG. 3A is a diagram showing the mean square error and signal noise ratio of the reverse channel blind channel estimation (ICBCE) method of the present invention and the conventional method of repeating and remodulating blind channel estimation (RRBCE). Schematic diagram of the curve, FIG. 3B is a diagram showing the mean square error of the reverse channel blind channel estimation (ICBCE) method of the present invention and the repetitive modulation and remodulation blind channel estimation (RRBCE) method of the prior art. Another schematic diagram of signal noise compared to the other.

As shown in FIGS. 3A and 3B, the reverse channel blind channel estimation (ICBCE) method of the present invention is based on a curve drawn by a signal noise ratio (SNR) and a mean square error (Mean-Square Error). The curve A1 to curve A3 are respectively the conventional technique and the repeated modulation and blind modulation channel estimation (RRBCE) method are curve B1 to curve B3, respectively. At the same time, the curves A1 to A3 of the present invention are compared with the curves B1 to B3 of the prior art, respectively.

As shown in FIG. 3A, the number of subcarriers N is equal to 64, the number of subcarriers N _g is equal to 8, the channel impulse response (CIR) of a length equal to V under condition 4, the curve A1 to the present invention were lower than the conventional curve A3 Knowing the curve B1 to B of the technology, the invention can still suppress the influence of the unknown transmission signal on the channel estimation under a small number of receiving signals.

As shown in FIG. 3B, the number of subcarriers N is equal to 64, the number of subcarriers N _g is equal to 8, the channel impulse response (CIR) of a length equal to V under condition of 8, the signal to noise ratio γ _b is about 34 or less The curve A1 to the curve A3 of the present invention are lower than the curve B1 to the curve B of the prior art, which means that the invention can suppress the influence of the unknown transmission signal on the channel estimation under a small amount of receiving signals.

Figure 4 is a diagram showing the reverse channel blind channel estimation (ICBCE) method of the present invention and the remodulation blind channel estimation (RMBCE) method and the repeated and remodulated blind channel estimation method of the prior art. (RRBCE) method is a schematic diagram of the mean square error and signal noise ratio curve.

As shown in the figure, according to the curve drawn by the signal noise ratio and the mean square error, the reverse channel blind channel estimation (ICBCE) method of the present invention is curve A1 to curve A2, respectively, and the repetition of the conventional technique. The modified modulation blind channel estimation (RRBCE) method is curve B1 to curve B2, respectively, and the conventional modified modulation blind channel estimation (RMBCE) method is curve C1 to curve C2, respectively. At the same time, the curve A1 of the present invention is compared with the curve B1 and the curve C1 of the prior art, and the curve A2 of the present invention is compared with the curve B2 and the curve C2 of the prior art.

As shown in the figure, when the number of subcarriers N is equal to 64, the number of subcarriers N _g is equal to 8, and the length V of the channel impulse response (CIR) is equal to 8, when the signal noise ratio γ _b is about 29 or less, The curve A1 of the invention is lower than the curve B1 and the curve C1 of the prior art, and the curve A2 of the present invention is also lower than the curve B2 and the curve C2 of the prior art, indicating that the invention can suppress the unknown under a small amount of receiving signals. The effect of the transmitted signal on the channel estimate.

It can be seen from the above that the reverse channel blind channel estimation method of the present invention mainly generates a transmission signal having a symmetric cyclic prefix, and parses out a predetermined zero-nuclear subspace and a base vector from the transmitted signal, and according to the base vector. The first matrix, the second matrix, the non-negative qualified matrix and the eigenvector are calculated in sequence, and the information of the channel gain is calculated.

Thereby, the present invention can suppress the failure even under a small amount of received signals. Knowing the impact of the transmitted signal on the channel estimation, and maintaining low computational complexity, and quickly extracting reliable channel information from the received signal, and mathematically confirming that the present invention can provide reliable in different channel environments. Estimate the result.

The above embodiments are intended to illustrate the principles of the invention and its effects, and are not intended to limit the invention. Any of the above-described embodiments may be modified by those skilled in the art without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention should be as set forth in the appended claims.

S21 to S27‧‧‧ steps

Claims (7)

A reverse channel blind channel estimation method includes: generating a transmission signal having a symmetric cyclic prefix; and parsing a base vector having a predetermined zero nuclear subspace and a corresponding zero nuclear subspace from the transmission signal; Transmitting the transmission signal to the receiving end via the channel; calculating a first matrix according to the base vector, and calculating a second matrix according to the first matrix and the receiving signal of the receiving end; calculating a non-negative limiting matrix according to the second matrix The non-negative limiting matrix is subjected to eigenvalue decomposition to generate a feature vector; and information of channel gain of the channel is calculated according to the feature vector. The reverse channel blind channel estimation method according to claim 1, wherein the predetermined zero-nuclear subspace and the base vector are used to reduce interference with the transmitted signal. The reverse channel blind channel estimation method according to claim 1, wherein the base vector generates a first matrix of ( N + N _g ) x N _g dimensions, the first matrix and the received signal The system is multiplied to calculate the second matrix to remove the transmission signal, where N is the number of subcarriers of the transmission signal of the transmitting end, and N + N _g is the number of subcarriers having the transmission signal of the symmetry cyclic prefix. The reverse channel blind channel estimation method according to claim 1, wherein the first matrix is U and is expressed as follows: I _Ng is the N _g x N _g unit matrix, O _Ng is the N _g x N _g zero matrix, N is the number of subcarriers of the transmission signal of the transmitting end, N _{g is} the number of subcarriers of the cyclic prefix of the transmission signal, t is the rotation Set. The reverse channel blind channel estimation method according to claim 1, wherein the second matrix is W and is expressed as follows: U is the first matrix, R is the received signal, N _{b is} the number of blocks of the received signal, and t is transposed. The reverse channel blind channel estimation method according to claim 1, wherein the non-negative limiting matrix is equal to the transposition of the second matrix multiplied by the second matrix. The reverse channel blind channel estimation method according to claim 1, wherein the eigenvector corresponds to a minimum eigenvalue of the non-negative limiting matrix, and the eigenvector is And expressed as follows: c is a vector, † is a conjugate transpose, and W is the second matrix.