A multi-antenna system transmission method with time-frequency-space dispersion of symbol energy

Abstract

A multi-antenna system transmission method with symbol energy time-frequency-space dispersion, which belongs to the field of wireless communications. The present invention solves the problem that the existing full-diversity full-rate space-time code is designed for high-complexity maximum likelihood detection or spherical detection, has poor performance under low-complexity receiving methods, and has limited application. The present invention uses a linear precoding method at the transmitting end to disperse the energy of the constellation symbol in the entire time-frequency plane, and completes the spatial energy distribution adjustment through the cyclic downward shift of the multi-antenna symbols, so that the energy of the constellation symbols that have been dispersed in the time-frequency plane continues to disperse in the spatial domain, and most of the energy of each constellation symbol can be retained regardless of the deep fading in which dimension of the channel; the energy of each constellation symbol is re-concentrated at the receiving end, thereby improving the signal transmission quality without reducing the communication rate. The method of the present invention can be applied to the field of wireless communications.

Description

Technical Field

[0001] The invention belongs to the field of wireless communications, and in particular relates to a multi-antenna system transmission method with symbol energy time-frequency-space dispersion. Background Art

[0002] With the growing demand for high-speed data communication services and the increasing shortage of spectrum resources, Multiple-Input Multiple-Output (MIMO) technology has become one of the key technologies for current 4G, 5G and even future 6G mobile communication systems. As an important part of the multi-antenna transmission system, space-time coding introduces inter-symbol correlation in both spatial and temporal dimensions, which can effectively weaken the impact of channel fading and improve the quality of received signals. However, in the early stage of space-time coding research, such as Alamouti coding, the advantage of diversity gain is obtained at the expense of symbol transmission rate, which does not match the demand for high-speed business transmission, prompting the research on space-time coding to shift towards full diversity and full rate. In the research on full-diversity full-rate space-time codes, a series of achievements have emerged, such as rotating QAM space-time codes, golden space-time codes, perfect space-time codes, and space-time codes based on cyclic divisible algebra, but the above space-time codes are designed for maximum likelihood detection or spherical detection, and the reception complexity increases exponentially with the number of transmitting antennas, making it difficult to apply in practice. Therefore, the design of full-rate, high-reliability space-time coding methods for linear equalization or low-complexity nonlinear detection has research value. Summary of the invention

[0003] The purpose of the present invention is to solve the problem that the existing full-diversity full-rate space-time code is designed for high-complexity maximum likelihood detection or spherical detection, has poor performance under low-complexity receiving methods and is limited in application, and a multi-antenna system transmission method with symbol energy time-frequency-space dispersion is proposed.

[0004] The technical solution adopted by the present invention to solve the above technical problems is:

[0005] According to one aspect of the present invention, a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy is provided. In the method, a multi-antenna transmitter at a transmitting end is configured with T transmitting antennas, and each transmitting antenna exclusively uses a radio frequency chain; a multi-antenna linear receiver at a receiving end is configured with R receiving antennas, and each receiving antenna exclusively uses a radio frequency chain;

[0006] The workflow of the transmitter is:

[0007] Step 1: Convert the bit data generated by the digital signal source into serial-to-parallel data to obtain M parallel bit streams, where M≤min(T,R), where min(T,R) represents the smaller value of T and R.

[0008] Step 2: digitally modulate each bit stream in step 1 to obtain M constellation symbols;

[0009] Step 3, grouping each constellation symbol in step 2, that is, for any constellation symbol, grouping the constellation symbols from the 1st moment to the Nth moment in the constellation symbol into the 1st group, using the 1st group of constellation symbols as the 1st time data block, and then grouping the constellation symbols from the N+1th moment to the 2Nth moment into the 2nd group, using the 2nd group of constellation symbols as the 2nd time data block, until all constellation symbols in the constellation symbol are grouped, wherein N≥M, and N is an integer multiple of M;

[0010] Similarly, the other constellation symbols are grouped and processed respectively;

[0011] Step 4: The first time data block of the mth path It is expressed as:

[0012]

[0013] Among them, s m1 ,s m2 ,s m3 ,…,s mN are respectively the constellation symbols at the 1st, 2nd, 3rd, …, Nth time in the mth constellation symbol, m = 1, 2, …, M;

[0014] Linear precoding is performed on the first time data block of each channel respectively, and the linear precoding matrix of the first time data block of the mth channel is represented as W (m) , then the precoding result of the first time data block of the mth channel is for:

[0015]

[0016] Among them, "×" represents matrix multiplication, and the superscript "T" represents matrix transposition;

[0017] The precoding results corresponding to the first time data block of each channel together form a space-time processing block X 1 :

[0018]

[0019] Step 5: Process the space-time block X output in step 4 1 The airspace energy distribution is adjusted. The specific process of the airspace energy distribution adjustment is as follows:

[0020] Process block X when empty 1In it, the data of the time data block of the m-th path at the n-th moment is used to replace the data of the time data block of the [(m + n - 2) mod M]+1-th path at the n-th moment, where mod represents the modulo operation;

[0021] Then the space-time processing block after the adjustment of the spatial domain energy distribution is expressed as:

[0022]

[0023] where represents the n-th column of the space-time processing block X 1 , n = 1, 2, …, N, Π d is an M-order cyclic shift-down matrix, and the superscripts 0, 1, …, N - 1 represent the powers of the matrix Π d ;

[0024] The M-order cyclic shift-down matrix Π d is:

[0025]

[0026] Step 6. Process each time data block in the space-time processing block after the adjustment of the spatial domain energy distribution and transmit the processed data;

[0027] The processing of each time data block in the space-time processing block after the adjustment of the spatial domain energy distribution is specifically as follows:

[0028] For each time data block in the space-time processing block , cyclic prefix addition, mapping to the transmit antenna, digital-to-analog conversion, and up-conversion processing are sequentially performed; the rule for mapping the time data block to the transmit antenna is:

[0029] If M = T, the first time data block is mapped to the first transmit antenna, the second time data block is mapped to the second transmit antenna, …, the M-th time data block is mapped to the M-th transmit antenna; otherwise, if M < T, the first time data block is mapped to the first transmit antenna, the second time data block is mapped to the second transmit antenna, …, the M-th time data block is mapped to the M-th transmit antenna, and the time data blocks mapped to the (M + 1)-th, (M + 2)-th, …, T-th transmit antennas are all any one of the time data blocks in the space-time processing block , that is, for each of the (M + 1)-th, (M + 2)-th, …, T-th transmit antennas, the time data block mapped to it can be any time data block in the space-time processing block ;

[0030] Step 7: Similarly, for the l-th time data block of each channel, execute steps 4 to 6, where l=2, 3, ..., L, and L is the number of time data blocks of each channel;

[0031] The workflow on the receiving end is:

[0032] After the first time data block of each channel processed in step 1 and step 6 passes through the channel, R receiving antennas respectively receive analog signals from free space, and then pre-process the analog signals received by each receiving antenna respectively, and obtain R digital sequences after pre-processing;

[0033] The analog signal received by each receiving antenna is preprocessed respectively, which is specifically as follows:

[0034] The analog signal received by each receiving antenna passes through a low noise amplifier, a down converter, a filter, and an analog-to-digital converter in sequence;

[0035] Step 2: Use R receiving antennas to jointly determine the starting sampling point of the receiving time data block;

[0036] Step 3, removing the cyclic prefix of the R-way digital sequence respectively according to the starting sampling point and the cyclic prefix length;

[0037] Step 4: The receiver uses the pilot to perform channel estimation and obtains a MIMO channel matrix H with R rows and M columns.

[0038] Step 5: The receiving space-time processing block consisting of the R-way digital sequence after removing the cyclic prefix is ​​recorded as Y in , and then to Y in Perform linear equalization to obtain the space-time processing block Y after linear equalization r ;

[0039] Step 6: Process the equalized space-time block Y output from step 5 r Perform spatial energy distribution recovery to obtain a space-time processing block Y after the spatial energy distribution is restored. The spatial energy distribution recovery is specifically:

[0040] After equalization, block Y is processed in time r In the example, the data of the time data block of the mth path at time n is used to replace the data of the time data block of the [(mn) mod M] + 1th path at time n. The space-time processing block Y after the spatial energy distribution is restored is expressed as:

[0041]

[0042] Among them, y r,.n Denotes the space-time processing block Y after equalization r The nth column of u is an M-order cyclic up-shift matrix, and the superscripts 0, 1, ... N-1 represent the matrix Πu The power of

[0043] The M-order cyclic up-shift matrix Π u for:

[0044]

[0045] Step 7, linearly decode each time data block in the space-time processing block Y output in step 6 to restore the initial space-time processing block r;

[0046]

[0047] Among them, y m. Represents the row vector composed of N symbols of the m-th time data block in the space-time processing block Y, r m. W represents the row vector composed of N symbols of the m-th time data block in the initial space-time processing block r. (-m) Denotes the precoding matrix W (m) The inverse matrix of

[0048] Step 8, digitally demodulate each time data block in the initial space-time processing block r of step 7 to recover the 0 and 1 bit sequences;

[0049] Step 9: Perform parallel-to-serial conversion on the 0 and 1 bit sequences obtained in step 8 to recover the source information and complete the linear reception detection of a space-time processing block;

[0050] Step 10: For the data of the lth time data block of each channel after the processing in step 6 at the transmitting end after passing through the channel, the process from step 1 to step 9 is performed, where l=2, 3, ..., L.

[0051] According to another aspect of the present invention, the method is configured with T (T>1) transmitting antennas in the multi-antenna transmitter at the transmitting end, and each transmitting antenna has a dedicated radio frequency chain; the multi-antenna nonlinear receiver at the receiving end is configured with R receiving antennas, and each receiving antenna has a dedicated radio frequency chain;

[0052] The workflow of the transmitter is:

[0053] Step 1: Convert the bit data generated by the digital signal source into serial-to-parallel data to obtain M parallel bit streams, where M≤min(T,R), where min(T,R) represents the smaller value of T and R.

[0054] Step 2: digitally modulate each bit stream in step 1 to obtain M constellation symbols;

[0055] Step 3, grouping each constellation symbol in step 2, that is, for any constellation symbol, grouping the constellation symbols from the 1st moment to the Nth moment in the constellation symbol into the 1st group, using the 1st group of constellation symbols as the 1st time data block, and then grouping the constellation symbols from the N+1th moment to the 2Nth moment into the 2nd group, using the 2nd group of constellation symbols as the 2nd time data block, until all constellation symbols in the constellation symbol are grouped, wherein N≥M, and N is an integer multiple of M;

[0056] Similarly, the other constellation symbols are grouped and processed respectively;

[0057] Step 4: The first time data block of the mth path It is expressed as:

[0058]

[0059] Among them, s m1 ,s m2 ,s m3 ,...,s mN are respectively the first, second, third, ..., Nth moment constellation symbols in the mth path constellation symbol, m = 1, 2, ..., M;

[0060] Linear precoding is performed on the first time data block of each channel respectively, and the linear precoding matrix of the first time data block of the mth channel is represented as W (m) , then the precoding result of the first time data block of the mth channel is for:

[0061]

[0062] Among them, "×" represents matrix multiplication, and the superscript "T" represents matrix transposition;

[0063] The precoding results corresponding to the first time data block of each channel together form a space-time processing block X 1 :

[0064]

[0065] Step 5: Process the space-time block X output in step 4 1 The airspace energy distribution is adjusted. The specific process of the airspace energy distribution adjustment is as follows:

[0066] Process block X when empty 1 In the above, the data of the time data block of the mth path at the nth moment is used to replace the data of the time data block of the [(m+n-2)mod M]+1th path at the nth moment (both data refer to the data in the space-time processing block), where mod represents the remainder operation;

[0067] From the perspective of the entire space-time processing block, it is equivalent to cyclically shifting the n-th column of the space-time processing block X 1 downward by [(n - 1) mod M] times;

[0068] Then the space-time processing block after adjusting the spatial domain energy distribution is expressed as:

[0069]

[0070] wherein, the matrix multiplication symbol is omitted in the above formula without causing ambiguity, representing the n-th column of the space-time processing block X 1 , n = 1, 2,..., N, Π d is an M-order cyclic shift-down matrix, and the superscripts 0, 1,..., N - 1 represent the powers of the matrix Π d ;

[0071] The M-order cyclic shift-down matrix Π d is:

[0072]

[0073] Step 6. Process each time data block in the space-time processing block after adjusting the spatial domain energy distribution and transmit the processed data;

[0074] The processing of each time data block in the space-time processing block after adjusting the spatial domain energy distribution is specifically as follows:

[0075] For each time data block in the space-time processing block , cyclic prefix addition, mapping to the transmit antenna, digital-to-analog conversion, and up-conversion processing are sequentially performed; the rule for mapping the time data block to the transmit antenna is:

[0076] If M = T, the first time data block is mapped to the first transmit antenna, the second time data block is mapped to the second transmit antenna,..., the M-th time data block is mapped to the M-th transmit antenna; otherwise, if M < T, the first time data block is mapped to the first transmit antenna, the second time data block is mapped to the second transmit antenna,..., the M-th time data block is mapped to the M-th transmit antenna, and the time data blocks mapped to the (M + 1)-th, (M + 2)-th,..., T-th transmit antennas are all any one of the time data blocks in the space-time processing block , that is, for each of the (M + 1)-th, (M + 2)-th,..., T-th transmit antennas, the time data block mapped to it can be any one of the time data blocks in the space-time processing block ;

[0077] Step 7: Similarly, for the l-th time data block of each channel, execute steps 4 to 6, where l=2, 3, ..., L, and L is the number of time data blocks of each channel;

[0078] The workflow on the receiving end is:

[0079] After the first time data block of each channel processed in step 1 and step 6 passes through the channel, R receiving antennas respectively receive analog signals from free space, and then pre-process the analog signals received by each receiving antenna respectively, and obtain R digital sequences after pre-processing;

[0080] Step 2: Use R receiving antennas to jointly determine the starting sampling point of the receiving time data block;

[0081] Step 3, removing the cyclic prefix of the R-way digital sequence respectively according to the starting sampling point and the cyclic prefix length;

[0082] Step 4: The receiver uses the pilot to perform channel estimation and obtains a MIMO channel matrix H with R rows and M columns.

[0083] Step 5: The receiving space-time processing block consisting of the R-way digital sequence after removing the cyclic prefix is ​​recorded as Y in , for Y in Perform serial-to-parallel conversion to obtain spatial data block Y p , Y p =[y p,1 ,y p,2 ,y p,3 ,…,y p,RN ] T ,y p,1 ,y p,2 ,y p,3 ,…,y p,RN They are spatial data blocks Y p the 1st, 2nd, 3rd, …, RNth element in ;

[0084] The rules of the serial-to-parallel conversion are:

[0085] The empty time processing block Y in The 1st to the Rth rows of the nth column are used as the spatial data block Y p From row [(n-1)R+1] to row nR in column 1;

[0086] Step 6: The receiver uses the MIMO channel matrix H and the linear precoding matrix of each time data block to calculate the equivalent channel matrix H for serial interference cancellation detection with NR rows and NM columns. eq ;

[0087] The equivalent channel matrix H eqThe calculation method for the [(k-1)R+1]th row to the kRth row in the cth column is:

[0088]

[0089] Among them, H eq,k,c Denotes the matrix H eq The submatrix consisting of the elements from row [(k-1)R+1] to row kR in column c, 1≤c≤MN, 1≤k≤N, represents the k-th row and v-th column element of the linear precoding matrix of the u-th time data block. The superscript (k-1) represents the (k-1)th power of the matrix. The subscript “·u” represents all elements of the u-th column of the matrix. “·” represents the multiplication of a scalar and a matrix.

[0090]

[0091] In the formula, represents the smallest positive integer not less than c / M;

[0092] Step 7: Use spatial data block Y p and the equivalent channel matrix H eq Perform i-times of serial interference elimination based on post-detection Euclidean distance sorting and MN-i-times of serial interference elimination based on signal-to-interference-noise ratio sorting;

[0093] The specific process of step 7 is as follows:

[0094] Step 7.1: Set the detection number record variable iter = 1, and initialize the post-detection vector r p is a vector of all zeros with 1 column and MN rows;

[0095] Step 7.2: Compare iter with MN. If iter>MN, execute step 7.21. Otherwise, continue to compare iter with i. If iter≤i, execute step 7.3; otherwise, execute step 7.12.

[0096] Step 7.3: Use the equivalent channel matrix after iter-1th serial interference cancellation detection Calculate the linear equalization matrix of the iter-th detection, and use the linear equalization matrix to eliminate the spatial data block after the iter-1-th serial interference detection Perform equalization and obtain the iterth linear equalization result in,

[0097] Step 7.4: Take the iterth linear equalization result in step 7.3 Send to buffer;

[0098] Step 7.5: For the iterth linear equalization result in step 7.3 Perform digital demodulation, and then digitally modulate the demodulation result to obtain the constellation symbol vector of the iter-th digital demodulation result

[0099] Step 7.6: Read the iterth linear equalization result in the buffer of step 7.4 Calculate the iterth linear equalization result The constellation symbol vector of the iter-th digital demodulation result The Euclidean distance vector e( iter ):

[0100]

[0101] Where “|·|” means calculating the absolute value of each element of the vector respectively, and “⊙” means multiplying the corresponding elements of the vector;

[0102] Step 7.7: Search for the minimum value of the element in the Euclidean distance vector of step 7.6, and record the minimum value in the Euclidean distance vector e (iter) The index index and the minimum value in the first linear equalization result The corresponding original index ori;

[0103] Step 7.8: The indexth row element is digitally demodulated, and the demodulation result is used as an element to update the post-detection vector r in step 7.1. p The ori-th row element of ;

[0104] Step 7.9: Keep the value in step 7.5 The index-th row element of Change the other elements of to 0;

[0105] Step 7.10: Calculate the spatial data block after the iterth serial interference elimination detection and the equivalent channel matrix after the iterth serial interference cancellation detection

[0106]

[0107]

[0108] In the formula, Representation Matrix dth column of ;

[0109] Step 7.11, update iter to iter+1, and return to step 7.2;

[0110] Step 7.12: Use the equivalent channel matrix after iter-1th serial interference cancellation detection Calculate the linear equalization matrix G for the iterth serial interference cancellation detection (iter) ;

[0111] Step 7.13: Use the linear equalization matrix G (iter) The spatial data block after the iter-1th serial interference elimination detection Perform linear equalization and obtain the iterth linear equalization result

[0112] Step 7.14: Use the linear equalization matrix and Calculate the signal to interference noise ratio of each to-be-demodulated symbol in the iter-th linear equalization result, where the signal to interference noise ratio SINR of the lth to-be-demodulated symbol in the iter-th linear equalization result is l The calculation method is:

[0113]

[0114] in, Denotes the linear equalization matrix G (iter) The lth row of Representation Matrix The lth column of Representation Matrix The col-th column of , col≠l, “||·||” represents the 2-norm of the vector;

[0115] Step 7.15: Search for the maximum value of the signal-to-interference-noise ratio of each symbol to be demodulated in the iterth linear equalization result of step 7.14, record the index of the maximum value as index, and record the maximum value in the first linear equalization result. The corresponding original index ori;

[0116] Step 7.16: The indexth row element is digitally demodulated, and the demodulation result is used as an element to update the post-detection vector r in step 7.1. p The ori-th row element of ;

[0117] Step 7.17: Construct the constellation symbol vector of the iter-th digital demodulation result The number of rows and columns are (MN-iter+1) and 1 respectively. The indexth row element of is the constellation modulation symbol corresponding to the digital demodulation result in step 7.16, and the remaining elements are 0;

[0118] Step 7.18, calculate the spatial data block after the iter-th serial interference cancellation detection and the equivalent channel matrix after the iter-th serial interference cancellation detection

[0119] Step 7.19, compare the magnitudes of iter and MN. When iter < MN, execute Step 7.20; otherwise, execute Step 7.21;

[0120] Step 7.20, update iter to iter + 1, and return to Step 7.12;

[0121] Step 7.21, output the detected vector r p , complete i times of serial interference cancellation based on the sorted Euclidean distance after detection and (MN - i) times of serial interference cancellation based on the signal-to-interference-plus-noise ratio;

[0122] Step 8, perform a serial-to-parallel conversion on the detected vector r output in Step 7.21 p to recover the initial space-time processing block r; the serial-to-parallel conversion rule is:

[0123] Take the [(n - 1)R + 1]-th to the nR-th rows of the first column of the detected vector r p as the first to the R-th rows of the n-th column of the space-time processing block r;

[0124] Step 9, perform a serial-to-parallel conversion on the 0, 1 bit sequences of each path of the space-time processing block r in Step 8 to recover the source information, and complete the non-linear reception detection of one space-time processing block;

[0125] Step 10, for the data of the l-th time data block of each path after being processed in Step 6 at the sending end and passing through the channel, execute the processes of Steps 1 to 9, where l = 2, 3,..., L.

[0126] The beneficial effects of the present invention are:

[0127] At the transmitting end, the present invention uses a linear precoding method to disperse the energy of the constellation symbol in the entire time-frequency plane, and completes the spatial energy distribution adjustment by cyclically shifting the multi-antenna symbol downward, so that the energy of the constellation symbol dispersed in the time-frequency plane continues to disperse in the spatial domain, and most of the energy of each constellation symbol can be retained regardless of which dimension of the channel has deep fading; at the receiving end, the energy of each constellation symbol is re-concentrated, thereby improving the signal transmission quality without reducing the communication rate. Moreover, the present invention can also control the size of the noise impact on each constellation symbol after receiving detection by reasonably selecting a linear precoding matrix: from a statistical point of view, when the total noise power after receiving detection is small, the noise power is evenly distributed to all constellation symbols in a space-time code block, so that the noise impact on each symbol is within an acceptable range; when the total noise power is large, the noise power is concentrated in the individual symbols of each time data block, and the overall bit error rate is maintained at a low level at the cost of sacrificing a few symbols in each time data block. The method of the present invention can achieve full-rate transmission when the number of parallel bit streams is equal to the smaller of the number of transmitting antennas and the number of receiving antennas. The detection complexity at the receiving end is low, and it is easy to be compatible with the existing MIMO system, convenient for practical application, and has strong resistance to noise and fading channels, thereby improving the performance of the full-rate MIMO system under the low-complexity receiving method. BRIEF DESCRIPTION OF THE DRAWINGS

[0128] Figure 1 It is a transmitter functional block diagram of a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy of the present invention;

[0129] Figure 2 It is a specific implementation of a multi-antenna system transmission method for symbol energy time-frequency-space dispersion of the present invention - a schematic diagram of transmitter time-frequency domain and space domain energy distribution adjustment;

[0130] Figure 3 It is a functional block diagram of a linear receiver of a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy of the present invention;

[0131] Figure 4 It is a specific implementation of a multi-antenna system transmission method for time-frequency-space dispersion of symbol energy of the present invention - a schematic diagram of energy distribution recovery in the time-frequency domain and space domain of a linear receiver;

[0132] Figure 5 It is a flow chart of selecting the transformation order of a specific implementation mode 2 of a multi-antenna system transmission method for time-frequency-space dispersion of symbol energy of the present invention;

[0133] Figure 6 It is a functional block diagram of a nonlinear receiver of a multi-antenna system transmission method for symbol energy time-frequency-space dispersion of the present invention;

[0134] Figure 7 It is a specific implementation method of a multi-antenna system transmission method for symbol energy time-frequency-space dispersion of the present invention; a nonlinear receiver nonlinear detection flow chart;

[0135] Figure 8 It is a bit error rate performance simulation diagram of a specific implementation mode 2 and a specific implementation mode 3 of a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy of the present invention under a quasi-static flat fading channel and QPSK modulation;

[0136] Fig. 9 It is a bit error rate performance simulation diagram of specific implementation mode 7 and specific implementation mode 10 of a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy of the present invention under quasi-static flat fading channel and QPSK modulation. DETAILED DESCRIPTION

[0137] Specific implementation method 1. Combination Figure 1 , Figure 2 , Figure 3 and Figure 4 This embodiment is described. In the multi-antenna system transmission method with time-frequency-space dispersion of symbol energy described in this embodiment, the multi-antenna transmitter at the transmitting end is configured with T (T>1) transmitting antennas, and each transmitting antenna has a dedicated radio frequency chain; the multi-antenna linear receiver at the receiving end is configured with R (R>1) receiving antennas, and each receiving antenna has a dedicated radio frequency chain;

[0138] The workflow of the transmitter is:

[0139] Step 1: Convert the bit data generated by the digital signal source into serial-to-parallel data to obtain M parallel bit streams, where M≤min(T,R), where min(T,R) represents the smaller value of T and R.

[0140] Step 2: digitally modulate each bit stream in step 1 to obtain M constellation symbols;

[0141] Step 3, grouping each constellation symbol in step 2, that is, for any constellation symbol, grouping the constellation symbols from the 1st moment to the Nth moment in the constellation symbol into the 1st group, using the 1st group of constellation symbols as the 1st time data block, and then grouping the constellation symbols from the N+1th moment to the 2Nth moment into the 2nd group, using the 2nd group of constellation symbols as the 2nd time data block, until all constellation symbols in the constellation symbol are grouped, wherein N≥M, and N is an integer multiple of M;

[0142] Similarly, the other constellation symbols are grouped and processed respectively;

[0143] Step 4: The first time data block of the mth path It is expressed as:

[0144]

[0145] Among them, s m1 ,s m2 ,s m3 ,…,s mN are respectively the constellation symbols at the 1st, 2nd, 3rd, …, Nth time in the mth constellation symbol, m = 1, 2, …, M;

[0146] Linear precoding is performed on the first time data block of each channel respectively, and the linear precoding matrix of the first time data block of the mth channel is represented as W (m) , then the precoding result of the first time data block of the mth channel is for:

[0147]

[0148] Among them, "×" represents matrix multiplication, and the superscript "T" represents matrix transposition;

[0149] The precoding results corresponding to the first time data block of each channel together form a space-time processing block X 1 :

[0150]

[0151] Step 5: Process the space-time block X output in step 4 1 The airspace energy distribution is adjusted. The specific process of the airspace energy distribution adjustment is as follows:

[0152] Process block X when empty 1 In the example, the data of the time data block of the mth path at the nth time is used to replace the data of the time data block of the [(m+n-2)mod M]+1th path at the nth time (both data refer to the data in the space-time processing block), where mod represents the remainder operation; from the perspective of the entire space-time processing block, it is equivalent to replacing the space-time processing block X 1 The nth column of cyclically shifts downward [(n-1) mod M] times;

[0153] Then the space-time processing block after the spatial energy distribution is adjusted It is expressed as:

[0154]

[0155] In the above formula, the matrix multiplication symbol is omitted without causing ambiguity. Represents space-time processing block X 1 The nth column of dis an M - order cyclic shift - down matrix, and the superscripts 0, 1, …, N - 1 represent the powers of the matrix Π d ;

[0156] The M - order cyclic shift - down matrix Π d is as follows:

[0157]

[0158] Step 6: Process each time - data block in the space - time processing block after the adjustment of the spatial - domain energy distribution, and transmit the processed data; Specifically, for each time - data block in the space - time processing block after the adjustment of the spatial - domain energy distribution, the processing is as follows:

[0159] For each time - data block in the space - time processing block after the adjustment of the spatial - domain energy distribution, perform the operations of adding a cyclic prefix, mapping to a transmit antenna, digital - to - analog conversion, and up - conversion in sequence; the rule for mapping the time - data block to the transmit antenna is:

[0160] If M = T, the first - path time - data block is mapped to the first transmit antenna, the second - path time - data block is mapped to the second transmit antenna, …, the M - th path time - data block is mapped to the M - th transmit antenna; otherwise, if M < T, the first - path time - data block is mapped to the first transmit antenna, the second - path time - data block is mapped to the second transmit antenna, …, the M - th path time - data block is mapped to the M - th transmit antenna, and the time - data blocks mapped to the (M + 1)-th, (M + 2)-th, …, T - th transmit antennas are all any one of the time - data blocks in the space - time processing block; that is, for each of the (M + 1)-th, (M + 2)-th, …, T - th transmit antennas, the time - data block mapped to it can be any one of the time - data blocks

[0161] in the space - time processing block ;

[0162] Step 7: Similarly, for the l - th time - data block of each path, execute Steps 4 to 6, where l = 2, 3, …, L, and L is the number of time - data blocks per path;

[0163] The working process of the receiving end is as follows:

[0164] Step 1: After the first - time - data block of each path processed in Step 6 passes through the channel, R receive antennas respectively receive analog signals from free space, and then pre - process the analog signals received by each receive antenna. After pre - processing, R digital sequences are obtained;

[0165] The specific pre - processing of the analog signals received by each receive antenna is as follows:

[0166] The analog signal received by each receiving antenna passes through a low noise amplifier, a down converter, a filter, and an analog-to-digital converter in sequence;

[0167] Step 2: Use R receiving antennas to jointly determine the starting sampling point of the receiving time data block;

[0168] Step 3, removing the cyclic prefix of the R-way digital sequence respectively according to the starting sampling point and the cyclic prefix length;

[0169] Step 4: The receiver uses the pilot to perform channel estimation and obtains a MIMO channel matrix H with R rows and M columns.

[0170] Step 5: The receiving space-time processing block consisting of the R-way digital sequence after removing the cyclic prefix is ​​recorded as Y in , and then to Y in Perform linear equalization to obtain the space-time processing block Y after linear equalization r ;

[0171] Step 6: Process the equalized space-time block Y output from step 5 r Perform spatial energy distribution recovery to obtain a space-time processing block Y after the spatial energy distribution is restored. The spatial energy distribution recovery is specifically:

[0172] After equalization, block Y is processed in time r In the process, the data of the time data block of the mth path at time n is used to replace the data of the time data block of the [(mn) mod M] + 1th path at time n, and the whole space-time processing block Y r From the perspective of r The nth column of is cyclically shifted upward [(n-1) mod M] times, so that the space-time processing block Y after the spatial energy distribution is restored is expressed as:

[0173]

[0174] Among them, y r,·n Represents the space-time processing block Y after equalization r The nth column of u is an M-order cyclic up-shift matrix, and the superscripts 0, 1, ... N-1 represent the matrix Π u The power of

[0175] The M-order cyclic up-shift matrix Π u for:

[0176]

[0177] Step 7, linearly decode each time data block in the space-time processing block Y output in step 6 to restore the initial space-time processing block r;

[0178]

[0179] Among them, y m· Represents the row vector composed of N symbols of the m-th time data block in the space-time processing block Y, r m· W represents the row vector composed of N symbols of the m-th time data block in the initial space-time processing block r. (-m) Denotes the precoding matrix W (m) The inverse matrix of

[0180] Step 8, digitally demodulate each time data block in the initial space-time processing block r of step 7 to recover the 0 and 1 bit sequences;

[0181] Step 9: Perform parallel-to-serial conversion on the 0 and 1 bit sequences obtained in step 8 to recover the source information and complete the linear reception detection of a space-time processing block;

[0182] Step 10: For the data of the lth time data block of each channel after the processing in step 6 at the transmitting end after passing through the channel, the process from step 1 to step 9 is performed, where l=2, 3, ..., L.

[0183] The method of the present invention uses linear equalization and can achieve full-rate, high-reliability performance in a low-complexity receiver.

[0184] Specific implementation method 2: Combination Figure 5 This embodiment is a further limitation of the first embodiment, in which the linear balance in step 5 is ZF balance, and the ZF balance matrix is:

[0185] G ZF =(H H H) -1 H H

[0186] Among them, G ZF is the ZF balanced matrix, the superscript “H” indicates the Hermitian transpose of the matrix, and the superscript “-1” indicates the inverse of the matrix;

[0187] The linear precoding method in step 4 is weighted fractional Fourier transform (WFRFT), and the calculation method of the transform order is:

[0188] Step ①, for the first time data block of each channel, the transformation order is a default order agreed upon by the transmitter and the receiver;

[0189] Step ②: The calculation method for the transformation order of the second time data block of each channel is:

[0190] Step A: The receiver calculates the critical noise power P th ; The specific process is:

[0191] Define the function P e,1 (x) is:

[0192]

[0193] Among them, x is the independent variable of the function, Q(·) represents the right tail function of the standard normal distribution, P erg To test the power;

[0194]

[0195] Where e is the base of natural logarithms;

[0196] Take 1 / 3 as the initial value of the test power, and then gradually increase the test power P by δ1 (δ1>0) erg The value of P erg / M is no longer a function P e,1 When (x) is at its minimum value, the test power P is stopped. erg The update of the final trial power P erg The critical noise power P th ;

[0197] Among them, δ1 is determined comprehensively based on the required accuracy and computational complexity;

[0198] Step B: Calculate the noise power after ZF equalization

[0199]

[0200] Among them, “||·|| F " represents the F-norm of the matrix, σ z 2 is the noise power before ZF equalization;

[0201] Step C: Compare the noise power after ZF equalization and critical noise power P th size;

[0202] If the noise power after ZF equalization Not greater than the critical noise power P th , then the weighted fractional Fourier transform order of the second time data block of each channel is: α1=α2=…α M =1, and execute step H;

[0203] Otherwise, if the noise power after ZF equalization is Greater than the critical noise power Pth , then execute step D;

[0204] Step D: Define function P e,2 (x) is:

[0205]

[0206] Solving function P e,2 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P low ;

[0207] Step E: Define function P e,3 (x) is:

[0208]

[0209] Solving function P e,3 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P high ;

[0210] Step F: Construct an ideal noise power distribution vector P ideal :

[0211]

[0212] Step G: Use ZF balance matrix G ZF and the ideal noise power distribution vector P ideal Calculate the WFRFT transform order of the second time data block of each channel; the specific process is:

[0213] Step G1, set m=1;

[0214] Step G2, initialize the mean square error value to MSE temp , let the trial order α erg =δ2-2(0<δ2<4), where δ2 is determined based on the required accuracy and computational complexity;

[0215] Step G3, (receiver) calculates the actual noise power distribution vector P of the mth time data block act,m , the calculation formula is:

[0216] P act,m =[P act,m1 ,P act,m2 ,P act,m3 ,…P act,mN ] 1×N

[0217] Among them, P act,m1 is the actual noise power of the first constellation symbol of the mth time data block, P act,m2is the actual noise power of the second constellation symbol of the mth time data block, P act,m3 is the actual noise power of the third constellation symbol of the mth time data block, P act,mN is the actual noise power of the Nth constellation symbol of the mth time data block;

[0218]

[0219] Where W qM+m,n (α erg ) indicates that the transformation order is α erg The element of the qM+mth row and nth column of the N-point weighted fractional Fourier transform matrix, q=0,1,…,N / M-1, “|·|” represents the scalar absolute value, G ZF,a· Denotes the matrix G ZF The a-th row of , a=[(m-1)modM]+1,[mmodM]+1,[(m+1)modM]+1,…,[(m+M-2)modM]+1, “||·||” represents the vector 2-norm;

[0220] Step G4: The actual noise power distribution vector P act,m The elements in are sorted from large to small, and the noise power distribution vector P of the mth time data block is obtained after sorting. order,m ;

[0221] Step G5: Calculate the noise power distribution vector P order,m With the ideal noise power distribution vector P ideal Mean square error MSE:

[0222] MSE=||P order,m -P ideal ||

[0223] Step G6: Compare the MSE calculated in step G5 with the MSE temp size;

[0224] If MSE is less than MSE temp , then execute step G7; otherwise, execute step G8;

[0225] Step G7: MSE temp Update to MSE, and change the weighted fractional Fourier transform order α of the m-th time data block m Updated to alpha erg , and then execute step G8;

[0226] Step G2: Initialize MSE temp , initialize it to positive infinity to ensure that step G7 is executed at least once;

[0227] Step G8: If α erg +δ2 is less than or equal to 2, then update α erg to α erg +δ2, and then return to Step G3;

[0228] Otherwise, if α erg +δ2 is greater than 2, then execute Step G9;

[0229] Step G9: Compare the magnitudes of m and M;

[0230] If m < M, then update m to m + 1, and re - execute Steps G2 to G8;

[0231] Otherwise, if m = M, then execute Step H;

[0232] Step H: Construct the transform order of the second time - data block of each path as the time - data - block transform - order vector α, and feedback it to the transmitter:

[0233] α = [α1, α2, … α M

[0234] where α1 is the transform order of the second time - data block of the first path, α2 is the transform order of the second time - data block of the second path, and α M is the transform order of the second time - data block of the M - th path;

[0235] Step ③: Similarly, for the l - th time - data block of each path, where l ≥ 3, the calculation method of its transform order is as follows:

[0236] During the processing of the (l - 1) - th time - data block of each path by the transmitter and the receiver, obtain the ZF equalization matrix, and use the obtained ZF equalization matrix and the method in Step ② to calculate the transform order when processing the l - th time - data block of each path.

[0237] Denote the transform order of the m - th path as α m (-2 < α m ≤2), and the result after WFRFT of the m - th path time - data block at the transmitter is

[0238] The calculation method of the N - point weighted fractional Fourier transform with the transform order α m is as follows:

[0239]

[0240] In the above formula, "×" represents matrix multiplication, the superscript "T" represents matrix transpose, and W(α m ) represents the WFRFT matrix with a transform length of N and a transform order of α m . ​

[0241] The space-time processing block after WFRFT is recorded as:

[0242]

[0243] At the receiving end, step 7 performs a -α operation on the m-th time data block. m The N-point weighted fractional Fourier transform of recovers the initial space-time processing block:

[0244]

[0245] The distribution characteristics of symbol energy in the time-frequency domain are adjusted through linear precoding. The linear precoding methods that can be selected in the present invention also include orthogonal time-frequency modulation (OTFS), wavelet transform, short-time Fourier transform and orthogonal linear frequency modulation multiplexing (OCDM), but are not limited to the above-mentioned linear precoding methods.

[0246] Specific implementation method three: This implementation method is a further limitation of specific implementation method one. The linear equalization in step 5 is MMSE equalization, and the MMSE equalization matrix is:

[0247]

[0248] Among them, G MMSE is the MMSE equalization matrix, σ z 2 is the noise power, P x is the useful signal power, I M×M is the unit matrix of order M, and the superscript “H” indicates the Hermitian transpose of the matrix;

[0249] The linear precoding method in step 4 is weighted fractional Fourier transform (WFRFT); the transformation order of the weighted fractional Fourier transform is:

[0250] α1=α2=…=α M =1

[0251] Among them, α1 is the transformation order of the time data block of the first channel, α2 is the transformation order of the time data block of the second channel, and α M is the transformation order for the Mth time data block.

[0252] When the linear equalization is MMSE equalization, the transformation order of each time data block of each channel is 1.

[0253] Specific implementation method four: Combination Figure 6 and Figure 7This embodiment is described. In this embodiment, a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy is described. In the method, a multi-antenna transmitter at the transmitting end is configured with T (T>1) transmitting antennas, and each transmitting antenna has a dedicated radio frequency chain; a multi-antenna nonlinear receiver at the receiving end is configured with R receiving antennas, and each receiving antenna has a dedicated radio frequency chain;

[0254] The workflow of the transmitter is:

[0255] Step 1: Convert the bit data generated by the digital signal source into serial-to-parallel data to obtain M parallel bit streams, where M≤min(T,R), where min(T,R) represents the smaller value of T and R.

[0256] Step 2: digitally modulate each bit stream in step 1 to obtain M constellation symbols;

[0257] Step 3, grouping each constellation symbol in step 2, that is, for any constellation symbol, grouping the constellation symbols from the 1st moment to the Nth moment in the constellation symbol into the 1st group, using the 1st group of constellation symbols as the 1st time data block, and then grouping the constellation symbols from the N+1th moment to the 2Nth moment into the 2nd group, using the 2nd group of constellation symbols as the 2nd time data block, until all constellation symbols in the constellation symbol are grouped, wherein N≥M, and N is an integer multiple of M;

[0258] Similarly, the other constellation symbols are grouped and processed respectively;

[0259] Step 4: The first time data block of the mth path It is expressed as:

[0260]

[0261] Among them, s m1 ,s m2 ,s m3 ,…,s mN are respectively the first, second, third, ..., Nth moment constellation symbols in the mth constellation symbol, m = 1, 2, ..., M;

[0262] Linear precoding is performed on the first time data block of each channel respectively, and the linear precoding matrix of the first time data block of the mth channel is represented as W (m) , then the precoding result of the first time data block of the mth channel is for:

[0263]

[0264] Among them, "×" represents matrix multiplication, and the superscript "T" represents matrix transposition;

[0265] The precoding results corresponding to the first time data block of each channel together form a space-time processing block X 1 :

[0266]

[0267] Step 5: Process the space-time block X output in step 4 1 The airspace energy distribution is adjusted. The specific process of the airspace energy distribution adjustment is as follows:

[0268] Process block X when empty 1 In the above, the data of the time data block of the mth path at the nth moment is used to replace the data of the time data block of the [(m+n-2)mod M]+1th path at the nth moment (both data refer to the data in the space-time processing block), where mod represents the remainder operation;

[0269] From the perspective of the entire space-time processing block, it is equivalent to converting the space-time processing block X 1 The nth column of cyclically shifts downward [(n-1) mod M] times;

[0270] Then the space-time processing block after the spatial energy distribution is adjusted It is expressed as:

[0271]

[0272] In the above formula, the matrix multiplication symbol is omitted without causing ambiguity. Represents space-time processing block X 1 The nth column of d is an M-order cyclic downshift matrix, and the superscripts 0, 1, ..., N-1 represent the matrix Π d The power of

[0273] The M-order cyclic downshift matrix Π d for:

[0274]

[0275] Step 6: Adjust the spatial energy distribution and then perform space-time processing Each time data block in is processed separately, and the processed data is emitted;

[0276] The space-time processing block after adjusting the spatial energy distribution Each time data block in is processed separately, which is as follows:

[0277] For space-time processing blocks For each time data block in, the operations of adding a cyclic prefix, mapping to the transmit antennas, digital-to-analog conversion, and up-conversion are performed in sequence; the rule for mapping the time data block to the transmit antennas is as follows:

[0278] If M = T, the first time data block is mapped to the first transmit antenna, the second time data block is mapped to the second transmit antenna, …, the Mth time data block is mapped to the Mth transmit antenna; otherwise, if M < T, the first time data block is mapped to the first transmit antenna, the second time data block is mapped to the second transmit antenna, …, the Mth time data block is mapped to the Mth transmit antenna, and the time data blocks mapped to the (M + 1)th, (M + 2)th, …, Tth transmit antennas are all space-time processing blocks For any one of the time data blocks in, that is, for each of the (M + 1)th, (M + 2)th, …, Tth transmit antennas, the time data block mapped to it can be a space-time processing block Any time data block in;

[0279] Step 7: Similarly, perform Steps 4 to 6 on the lth time data block of each path, where l = 2, 3, …, L, and L is the number of time data blocks in each path;

[0280] The working process of the receiving end is as follows:

[0281] Step 1: After the first time data block of each path processed in Step 6 passes through the channel, R receive antennas respectively receive analog signals from free space, and then preprocess the analog signals received by each receive antenna, and R digital sequences are obtained after preprocessing;

[0282] Step 2: Use the R receive antennas to jointly determine the starting sampling point of the received time data block;

[0283] Step 3: Remove the cyclic prefixes of the R digital sequences according to the starting sampling point and the cyclic prefix length respectively;

[0284] Step 4: The receiver performs channel estimation using pilots to obtain an MIMO channel matrix H with R rows and M columns;

[0285] Step 5: Denote the 1 received space-time processing block formed by the R digital sequences after removing the cyclic prefix as Y in , for Y in perform serial-to-parallel conversion to obtain a space data block Y p , Y p = [y p,1 , y p,2 , y p,3 , …, y p,RN T , y p,1 , y p,2 , yp,3 ,…,y p,RN They are spatial data blocks Y p the 1st, 2nd, 3rd, …, RNth element in ;

[0286] The rules of the serial-to-parallel conversion are:

[0287] The empty time processing block Y in The 1st to the Rth row of the nth column is used as the spatial data block Y p From row [(n-1)R+1] to row nR in column 1;

[0288] Step 6: The receiver uses the MIMO channel matrix H and the linear precoding matrix of each time data block to calculate the equivalent channel matrix H for serial interference cancellation (SIC) detection with NR rows and NM columns. eq ;

[0289] The equivalent channel matrix H eq The calculation method for the [(k-1)R+1]th row to the kRth row in the cth column is:

[0290]

[0291] Among them, H eq,k,c Denotes the matrix H eq The submatrix consisting of the elements from row [(k-1)R+1] to row kR in column c, 1≤c≤MN, 1≤k≤N, represents the k-th row and v-th column element of the linear precoding matrix of the u-th time data block. The superscript (k-1) represents the (k-1)th power of the matrix. The subscript “·u” represents all elements of the u-th column of the matrix. “·” represents the multiplication of a scalar and a matrix, where:

[0292]

[0293] In the formula, represents the smallest positive integer not less than c / M;

[0294] Step 7: Use spatial data block Y p and the equivalent channel matrix H eq Perform i (0≤i≤MN) times of serial interference cancellation based on post-detection Euclidean distance sorting and MN-i times of serial interference cancellation based on signal to interference plus noise ratio (SINR) sorting;

[0295] The specific process of step 7 is as follows:

[0296] Step 7.1: Set the detection number record variable iter = 1, and initialize the post-detection vector r p is a vector of all zeros with 1 column and MN rows;

[0297] Step 7.2: Compare iter with MN. If iter>MN, execute step 7.21. Otherwise, continue to compare iter with i. If iter≤i, execute step 7.3; otherwise, execute step 7.12.

[0298] Step 7.3: Use the equivalent channel matrix after iter-1th serial interference cancellation detection Calculate the linear equalization matrix of the iter-th detection, and use the linear equalization matrix to eliminate the spatial data block after the iter-1-th serial interference detection Perform equalization and obtain the iterth linear equalization result in,

[0299] Step 7.4: Take the iterth linear equalization result in step 7.3 Send to buffer;

[0300] Step 7.5: For the iterth linear equalization result in step 7.3 Perform digital demodulation, and then digitally modulate the demodulation result to obtain the constellation symbol vector of the iter-th digital demodulation result

[0301] Step 7.6: Read the iterth linear equalization result in the buffer of step 7.4 Calculate the iterth linear equalization result The constellation symbol vector of the iter-th digital demodulation result The Euclidean distance vector e( iter ):

[0302]

[0303] Where “|·|” means calculating the absolute value of each element of the vector respectively, and “⊙” means multiplying the corresponding elements of the vector;

[0304] Step 7.7: Search for the minimum value of the element in the Euclidean distance vector of step 7.6, and record the minimum value in the Euclidean distance vector e (iter) The index index (1≤index≤MN-iter+1) and the minimum value in the first linear equalization result The corresponding original index ori (1≤ori≤MN);

[0305] Step 7.8: The indexth row element is digitally demodulated, and the demodulation result is used as an element to update the post-detection vector r in step 7.1. p The ori-th row element of ;

[0306] Step 7.9: Keep the value in step 7.5 The index-th row element of Change the other elements of to 0;

[0307] Step 7.10: Calculate the spatial data block after the iterth serial interference elimination detection and the equivalent channel matrix after the iterth serial interference cancellation detection

[0308]

[0309]

[0310] In the formula, Representation Matrix dth column of ;

[0311] Step 7.11, update iter to iter+1, and return to step 7.2;

[0312] Step 7.12: Use the equivalent channel matrix after iter-1th serial interference cancellation detection Calculate the linear equalization matrix G for the iterth serial interference cancellation detection (iter) , The definition of is the same as step 7.3;

[0313] Step 7.13: Use the linear equalization matrix G (iter) The spatial data block after the iter-1th serial interference elimination detection Perform linear equalization and obtain the iterth linear equalization result in The definition of is the same as step 7.3;

[0314] Step 7.14: Use the linear equalization matrix and Calculate the signal to interference and noise ratio of each to-be-demodulated symbol in the iter-th linear equalization result, where the signal to interference and noise ratio SINR of the lth (1≤l≤MN-iter+1)th to-be-demodulated symbol in the iter-th linear equalization result is l The calculation method is:

[0315]

[0316] in, Denote the linear equalization matrix as G (iter) The l-th row of Denote the matrix The l-th column of Denote the matrix The col-th column of, where col ≠ l, and "||·||" represents the 2-norm of a vector;

[0317] Step 7.15: Search for the maximum signal-to-interference-plus-noise ratio (SINR) of each symbol to be demodulated in the iter-th linear equalization result of Step 7.14, record the index of the maximum value as index (1 ≤ index ≤ MN - iter + 1), and record the original index ori (1 ≤ ori ≤ MN) corresponding to the maximum value in the first linear equalization result ;

[0318] Step 7.16: Demodulate the elements in the index-th row of in Step 7.13, and update the element in the ori-th row of the detected vector r in Step 7.1 with the overall demodulation result as one element p ;

[0319] Step 7.17: Construct the constellation symbol vector of the iter-th digital demodulation result whose number of rows and columns are (MN - iter + 1) and 1 respectively, the element in the index-th row of which is the constellation modulation symbol corresponding to the digital demodulation result in Step 7.16, and the rest of the elements are 0;

[0320] Step 7.18: Calculate the spatial data block after the iter-th successive interference cancellation detection and the equivalent channel matrix after the iter-th successive interference cancellation detection The calculation method is the same as that in Step 7.10;

[0321] Step 7.19: Compare the magnitudes of iter and MN. When iter < MN, execute Step 7.20; otherwise, execute Step 7.21;

[0322] Step 7.20: Update iter to iter + 1 and return to Step 7.12;

[0323] Step 7.21: Output the detected vector r p , and complete i times of successive interference cancellation based on the detected Euclidean distance sorting and (MN - i) times of successive interference cancellation based on the signal-to-interference-plus-noise ratio sorting;

[0324] Step 8: Perform a parallel-to-serial conversion on the detected vector r p output in Step 7.21 to recover the initial space-time processing block r; the parallel-to-serial conversion rule is:

[0325] The detected vector r p The first column of [(n-1)R+1] to the nRth row of the block r is processed as the nth column, the first row to the Rth row of the block r in a space-time manner;

[0326] Step 9, performing parallel-to-serial conversion on the 0 and 1 bit sequences of each channel of the space-time processing block r in step 8, recovering the source information, and completing the nonlinear reception detection of one space-time processing block;

[0327] Step 10: For the data of the lth time data block of each channel after the processing in step 6 at the transmitting end after passing through the channel, the process from step 1 to step 9 is performed, where l=2, 3, ..., L.

[0328] The method of the present invention can achieve full-rate and high-reliability performance under low-complexity nonlinear detection.

[0329] Specific implementation mode 5: This implementation mode is a further limitation of specific implementation mode 4. The linear equalization matrix G in step 7.3 and step 7.12 (iter) is the ZF equilibrium matrix, which is calculated as follows:

[0330]

[0331] Here, the superscript "H" represents the Hermitian transpose of a matrix, and the superscript "-1" represents the inverse of a matrix.

[0332] Specific implementation method 6: This implementation method is a further limitation of specific implementation method 5. The linear precoding method is weighted fractional Fourier transform, and the calculation method of the transform order is:

[0333] Step ①: For the first time data block of each channel, the transformation order is a default order agreed upon by the transmitter and the receiver;

[0334] Step ②: The calculation method for the transformation order of the second time data block of each channel is:

[0335] Step A: The receiver calculates the critical noise power P th ; The specific process is:

[0336] Define the function P e,1 (x) is:

[0337]

[0338] Among them, x is the independent variable of the function, Q(·) represents the right tail function of the standard normal distribution, P erg To test the power;

[0339]

[0340] Where e is the base of natural logarithms;

[0341] Take 1 / 3 as the initial value of the test power, and then gradually increase the test power P by δ1 (δ1>0) erg The value of P erg / M is no longer a function P e,1 When (x) is at its minimum value, the test power P is stopped. erg The update of the final trial power P erg The critical noise power P th ;

[0342] Among them, δ1 is determined comprehensively based on the required accuracy and computational complexity;

[0343] Step B: Calculate the noise power after ZF equalization

[0344]

[0345] Among them, “||·|| F " represents the F-norm of the matrix, G (1) is the ZF equalization matrix obtained in the first iteration in step 7 at the receiving end of the fifth specific implementation mode;

[0346] Step C: Compare the noise power after ZF equalization and critical noise power P th size;

[0347] If the noise power after ZF equalization Not greater than the critical noise power P th , then the weighted fractional Fourier transform order of the second time data block of each channel is: α1=α2=…α M =1, and execute step H;

[0348] Otherwise, if the noise power after ZF equalization is Greater than the critical noise power P th , then execute step D;

[0349] Step D: Define function P e,2 (x) is:

[0350]

[0351] Solving function P e,2 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P low ;

[0352] Step E: Define function P e,3 (x) is:

[0353]

[0354] Solving function P e,3 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P high ;

[0355] Step F: Construct an ideal noise power distribution vector P ideal :

[0356]

[0357] Step G: Use ZF balance matrix G (1) and the ideal noise power distribution vector P ideal Calculate the WFRFT transform order of the second time data block of each channel; the specific process is:

[0358] Step G1, set m=1;

[0359] Step G2, initialize the mean square error value to MSE temp , let the trial order α erg =δ2-2(0<δ2<4), where δ2 is determined based on the required accuracy and computational complexity;

[0360] Step G3, (receiver) calculates the actual noise power distribution vector P of the mth time data block act,m , the calculation formula is:

[0361] P act,m =[P act,m1 ,P act,m2 ,P act,m3 ,…P act,mN ] 1×N

[0362] Among them, P act,m1 is the actual noise power of the first constellation symbol of the mth time data block, P act,m2 is the actual noise power of the second constellation symbol of the mth time data block, P act,m3 is the actual noise power of the third constellation symbol of the mth time data block, P act,mN is the actual noise power of the Nth constellation symbol of the mth time data block;

[0363]

[0364] Where W qM+m,n (α erg ) indicates that the transformation order is α erg The element of the qM+mth row and nth column of the N-point weighted fractional Fourier transform matrix, q=0,1,...,N / M-1; Denote the matrix G (1) for the ath row, where a = [(m - 1) mod M]+1, [m mod M]+1, [(m + 1) mod M]+1, …, [(m + M - 2) mod M]+1;

[0365] Step G4. Sort each element in the actual noise power distribution vector P act,m in descending order. After sorting, obtain the noise power distribution vector P order,m of the mth time data block;

[0366] Step G5. Calculate the mean square error MSE between the noise power distribution vector P order,m and the ideal noise power distribution vector P ideal :

[0367] MSE = ||P order,m - P ideal ||

[0368] Step G6. Compare the MSE calculated in Step G5 with MSE temp ;

[0369] If MSE is less than MSE temp , then execute Step G7; otherwise execute Step G8;

[0370] Step G7. Update MSE temp to MSE, and update the weighted fractional Fourier transform order α m of the mth time data block to α erg , then execute Step G8;

[0371] When initializing MSE temp in Step G2, initialize it to positive infinity to ensure that Step G7 is executed at least once;

[0372] Step G8. If α erg + δ2 is less than or equal to 2, then update α erg to α erg + δ2, and then return to Step G3;

[0373] Otherwise α erg + δ2 is greater than 2, then execute Step G9;

[0374] Step G9. Compare the magnitudes of m and M;

[0375] If m < M, then update m to m + 1 and re - execute Steps G2 to G8;

[0376] Otherwise m = M, then execute Step H;

[0377] Step H: construct the transformation order of the second time data block of each channel into a time data block transformation order vector α, and feed it back to the transmitter:

[0378] α=[α1,α2,…α M ]

[0379] Among them, α1 is the transformation order of the second time data block of the first path, α2 is the transformation order of the second time data block of the second path, and α M is the transformation order of the second time data block of the Mth path;

[0380] Step ③: For the lth time data block of each channel, l≥3, the calculation method of its transformation order is:

[0381] When the transmitter and receiver process the l-1th time data block of each channel, the ZF equalization matrix obtained in the first iteration is obtained, and the transformation order when processing the lth time data block of each channel is calculated using the obtained ZF equalization matrix and the method of step ②.

[0382] Specific implementation method 7: This implementation method is a further limitation of specific implementation method 6. When the weighted fractional Fourier transform order of each time data block is α1=α2=…α M =1, the value of the number of serial interference elimination times i based on the post-detection Euclidean distance sorting is 1; otherwise, the value of the number of serial interference elimination times i based on the post-detection Euclidean distance sorting is 0.

[0383] Specific implementation eight: This implementation is a further limitation of specific implementation four. The linear equalization matrix G in steps 7.3 and 7.12 (iter) is the MMSE equalization matrix, and its calculation method is:

[0384]

[0385] Among them, I (MN-iter+1)×(MN-iter+1) is the (MN-iter+1)-order unit matrix, the superscript “H” represents the Hermitian transpose of the matrix, the superscript “-1” represents the inverse of the matrix, and σ z 2 is the noise power, P x is the useful signal power.

[0386] Specific implementation method 9: This implementation method is a further limitation of specific implementation method 8. The linear precoding method is weighted fractional Fourier transform, and the transformation order of each time data block is:

[0387] α1=α2=…=α M =1

[0388] Among them, α1 is the transformation order of the time data block of the first channel, α2 is the transformation order of the time data block of the second channel, and α M is the transformation order for the Mth time data block.

[0389] When the linear equalization is MMSE equalization, the transformation order of each time data block of each channel is 1.

[0390] Specific implementation method ten: This implementation method is a further limitation of specific implementation method nine, wherein the number of serial interference elimination based on post-detection Euclidean distance sorting is i=1.

[0391] Figure 8 This is a bit error rate performance simulation diagram of the second and third embodiments of the multi-antenna system transmission method with time-frequency-space dispersion of symbol energy of the present invention under quasi-static flat fading channel and QPSK modulation; for comparison, the performance simulation of the vertical layered space-time code system (V-BLAST) and the golden space-time code system are also given. During the simulation process, the transmission rates of the three are limited to be consistent, which are all twice the rate of the single-antenna transmission system under the same modulation mode. It can be seen that in the 2-transmit 3-receive linear equalization scenario, when using ZF equalization, the second embodiment of the present invention has the lowest bit error rate among the three; when using MMSE equalization, the third embodiment of the present invention has the lowest bit error rate among the three, thereby verifying the full-rate and high-reliability performance of the present invention when using a low-complexity receiver with linear equalization.

[0392] Fig. 9 This is a bit error rate performance simulation diagram of the seventh and tenth embodiments of a multi-antenna system transmission method with time-frequency-space dispersion of symbol energy of the present invention under quasi-static flat fading channel and QPSK modulation; for comparison, the performance simulation of the vertical layered space-time code system and the golden space-time code system are also given. During the simulation process, the transmission rates of the three are also limited to be consistent, which are all twice the rate of the single-antenna transmission system under the same modulation mode, and in the nonlinear detection process, the vertical layered space-time code system and the golden space-time code system use the classic serial interference elimination method based on SINR sorting. It can be seen that in the 2-transmit 2-receive nonlinear detection scenario, if the iterative linear equalization uses ZF equalization, then the specific embodiment of the present invention has the lowest bit error rate among the three; if the iterative linear equalization uses MMSE equalization, then in the bit signal-to-noise ratio E b When / N0 is greater than 10dB, the tenth embodiment of the present invention has the best bit error rate performance among the three, and the performance advantage continues to expand with the increase of the bit signal-to-noise ratio, thereby verifying the full-rate and high-reliability performance of the present invention in low-complexity nonlinear detection.

[0393] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. A multi-antenna system transmission method with symbol energy time-frequency-space dispersion, characterized in that: In the method, a multi-antenna transmitter at a transmitting end is configured with T transmitting antennas, and each transmitting antenna exclusively uses a radio frequency chain; a multi-antenna linear receiver at a receiving end is configured with R receiving antennas, and each receiving antenna exclusively uses a radio frequency chain; The workflow of the transmitter is: Step 1: Convert the bit data generated by the digital signal source into serial-to-parallel data to obtain M parallel bit streams, where M≤min(T,R), where min(T,R) represents the smaller value of T and R. Step 2: digitally modulate each bit stream in step 1 to obtain M constellation symbols; Step 3, grouping each constellation symbol in step 2, that is, for any constellation symbol, grouping the constellation symbols from the 1st moment to the Nth moment in the constellation symbol into the 1st group, using the 1st group of constellation symbols as the 1st time data block, and then grouping the constellation symbols from the N+1th moment to the 2Nth moment into the 2nd group, using the 2nd group of constellation symbols as the 2nd time data block, until all constellation symbols in the constellation symbol are grouped, wherein N≥M, and N is an integer multiple of M; Similarly, the other constellation symbols are grouped and processed respectively; Step 4: The first time data block of the mth path It is expressed as: Among them, s m1 ,s m2 ,s m3 ,…,s mN are respectively the constellation symbols at the 1st, 2nd, 3rd, …, Nth time in the mth constellation symbol, m = 1, 2, …, M; Linear precoding is performed on the first time data block of each channel respectively, and the linear precoding matrix of the first time data block of the mth channel is represented as W (m) , then the precoding result of the first time data block of the mth channel is for: Among them, "×" represents matrix multiplication, and the superscript "T" represents matrix transposition; The precoding results corresponding to the first time data block of each channel together form a space-time processing block X 1 : Step 5: Process the space-time block X output in step 4 1 The airspace energy distribution is adjusted. The specific process of the airspace energy distribution adjustment is as follows: Process block X when empty 1 , the data of the time data block of the mth path at the nth moment is used to replace the data of the time data block of the [(m+n-2)modM]+1th path at the nth moment, where mod represents a remainder operation; Then the space-time processing block after the spatial energy distribution is adjusted It is expressed as: in Represents space-time processing block X 1 The nth column of d is an M-order cyclic downshift matrix, and the superscripts 0, 1, ..., N-1 represent the matrix Π d The power of The M-order cyclic downshift matrix Π d for: Step 6: Adjust the spatial energy distribution and then perform space-time processing Each time data block in is processed separately, and the processed data is emitted; The space-time processing block after adjusting the spatial energy distribution Each time data block in is processed separately, which is as follows: For space-time processing blocks Each time data block in is sequentially processed by adding a cyclic prefix, mapping to a transmitting antenna, digital-to-analog conversion, and up-conversion; the rule for mapping a time data block to a transmitting antenna is: If M = T, the first time data block is mapped to the first transmitting antenna, the second time data block is mapped to the second transmitting antenna, …, the M-th time data block is mapped to the M-th transmitting antenna; otherwise, if M < T, the first time data block is mapped to the first transmitting antenna, the second time data block is mapped to the second transmitting antenna, …, the M-th time data block is mapped to the M-th transmitting antenna, and the time data blocks mapped to the (M + 1)-th, (M + 2)-th, …, T-th transmitting antennas are all space-time processing blocks Any one of the time data blocks in; that is, for each of the (M + 1)-th, (M + 2)-th, …, T-th transmitting antennas, the time data block mapped to it is a space-time processing block Any time data block in; Step 7: Similarly, for the l-th time data block of each channel, execute steps 4 to 6, where l=2, 3, ..., L, and L is the number of time data blocks of each channel; The workflow on the receiving end is: After the first time data block of each channel processed in step 1 and step 6 passes through the channel, R receiving antennas respectively receive analog signals from free space, and then pre-process the analog signals received by each receiving antenna respectively, and obtain R digital sequences after pre-processing; The analog signal received by each receiving antenna is preprocessed respectively, which is specifically as follows: The analog signal received by each receiving antenna passes through a low noise amplifier, a down converter, a filter, and an analog-to-digital converter in sequence; Step 2: Use R receiving antennas to jointly determine the starting sampling point of the receiving time data block; Step 3, removing the cyclic prefix of the R-way digital sequence respectively according to the starting sampling point and the cyclic prefix length; Step 4: The receiver uses the pilot to perform channel estimation and obtains a MIMO channel matrix H with R rows and M columns. Step 5: The receiving space-time processing block consisting of the R-way digital sequence after removing the cyclic prefix is ​​recorded as Y in , and then to Y in Perform linear equalization to obtain the space-time processing block Y after linear equalization r ; Step 6: Process the equalized space-time block Y output from step 5 r Perform spatial energy distribution recovery to obtain a space-time processing block Y after the spatial energy distribution is restored. The spatial energy distribution recovery is specifically: After equalization, block Y is processed in time r In the above example, the data of the time data block of the mth path at time n is used to replace the data of the time data block of the [(mn)modM]+1th path at time n. The space-time processing block Y after the spatial energy distribution is restored is expressed as: Among them, y r,·n Denotes the space-time processing block Y after equalization r The nth column of u is an M-order cyclic up-shift matrix, and the superscripts 0, 1, ... N-1 represent the matrix Π u The power of The M-order cyclic up-shift matrix Π u for: Step 7, linearly decode each time data block in the space-time processing block Y output in step 6 to restore the initial space-time processing block r; Among them, y m＝ Represents the row vector composed of N symbols of the m-th time data block in the space-time processing block Y, r m＝ W represents the row vector composed of N symbols of the m-th time data block in the initial space-time processing block r. (-m) Denotes the precoding matrix W (m) The inverse matrix of Step 8, digitally demodulate each time data block in the initial space-time processing block r of step 7 to recover the 0 and 1 bit sequences; Step 9: Perform parallel-to-serial conversion on the 0 and 1 bit sequences obtained in step 8 to recover the source information and complete the linear reception detection of a space-time processing block; Step 10: For the data of the lth time data block of each channel after the processing in step 6 at the transmitting end after passing through the channel, the process from step 1 to step 9 is performed, where l=2, 3, ..., L.

2. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 1, characterized in that: The linear balance in step 5 is ZF balance, and the ZF balance matrix is: G ZF =(H H H) -1 H H Among them, G ZF is the ZF balanced matrix, the superscript "H" indicates the Hermitian transpose of the matrix, and the superscript "-1" indicates the inverse of the matrix; The linear precoding method in step 4 is weighted fractional Fourier transform, and the calculation method of the transform order is: Step ①: For the first time data block of each path, the transformation order is a default order agreed upon by both the transmitter and the receiver. Step ②: The calculation method for the transformation order of the second time data block of each path is as follows: Step A: The receiver calculates the critical noise power P th ; The specific process is: Define the function P e,1 (x) is: Among them, x is the independent variable of the function, Q(·) represents the right tail function of the standard normal distribution, P erg To test the power; where \(e\) is the base of the natural logarithm. Take 1 / 3 as the initial value of the test power, and then gradually increase the test power P by δ1. erg The value of P erg / M is no longer a function P e,1 When (x) is at its minimum value, stop testing the power P erg The update of the final trial power P erg The critical noise power P th ; Step B: Calculate the noise power after ZF equalization Among them, "||·|| F " represents the F-norm of the matrix, σ z 2 is the noise power before ZF equalization; Step C: Compare the noise power after ZF equalization and critical noise power P th size; If the noise power after ZF equalization Not greater than the critical noise power P th , then the weighted fractional Fourier transform order of the second time data block of each channel is: α1=α2=…α M =1, and execute step H; Otherwise, if the noise power after ZF equalization is Greater than the critical noise power P th , then execute step D; Step D: Define function P e,2 (x) is: Solving function P e,2 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P low ; Step E: Define function P e,3 (x) is: Solving function P e,3 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P high ; Step F: Construct an ideal noise power distribution vector P ideal : Step G: Use ZF balance matrix G ZF and the ideal noise power distribution vector P ideal Calculate the transformation order of the second time data block of each channel; the specific process is: Step G1: Let \(m = 1\). Step G2, initialize the mean square error value to MSE temp , let the trial order α erg is δ2-2; Step G3: Calculate the actual noise power distribution vector P of the mth time data block act,m , the calculation formula is: P act,m =[P act,m1 ,P act,m2 ,P act,m3 ,…P act,mN ] 1×N Among them, P act,m1 is the actual noise power of the first constellation symbol of the mth time data block, P act,m2 is the actual noise power of the second constellation symbol of the mth time data block, P act,m3 is the actual noise power of the third constellation symbol of the mth time data block, P act,mN is the actual noise power of the Nth constellation symbol of the mth time data block; Where W qM+m,n (α erg ) indicates that the transformation order is α erg The element of the qM+mth row and nth column of the N-point weighted fractional Fourier transform matrix, q=0,1,…,N / M-1, "|·|" represents the absolute value of the scalar, G ZF,a· Denotes the matrix G ZF The ath row of , a=[(m-1)modM]+1,[mmodM]+1,[(m+1)modM]+1,…,[(m+M-2)modM]+1, "||·||" represents the vector 2-norm; Step G4: The actual noise power distribution vector P act,m The elements in are sorted from large to small, and the noise power distribution vector P of the m-th time data block is obtained after sorting. order,m ; Step G5: Calculate the noise power distribution vector P order,m With the ideal noise power distribution vector P ideal Mean square error MSE: MSE=||P order,m -P ideal || Step G6: Compare the MSE calculated in step G5 with the MSE temp size; If MSE is less than MSE temp , then execute step G7; otherwise, execute step G8; Step G7: MSE temp Update to MSE, and change the weighted fractional Fourier transform order α of the m-th time data block m Updated to alpha erg , and then execute step G8; Step G8: If α erg +δ2 is less than or equal to 2, then α erg Updated to alpha erg +δ2, then return to step G3; Otherwise α erg +δ2 is greater than 2, then execute step G9; Step G9: Compare the magnitudes of \(m\) and \(M\). If \(m < M\), then update \(m\) to \(m + 1\) and re - execute Steps G2 to G8. Otherwise, if \(m = M\), then execute Step H. Step H: Construct the transformation order of the second time data block of each path into a time data block transformation order vector \(\alpha\) and feedback it to the transmitter: α＝[α1,α2,…α M ] Among them, α1 is the transformation order of the second time data block of the first path, α2 is the transformation order of the second time data block of the second path, and α M is the transformation order of the second time data block of the Mth path; Step ③: Similarly, for the \(l\) - th time data block of each path, where \(l\geq3\), the calculation method for its transformation order is as follows: During the processing of the \((l - 1)\) - th time data block of each path by the transmitter and the receiver, obtain the ZF equalization matrix, and use the obtained ZF equalization matrix and the method of Step ② to calculate the transformation order when processing the \(l\) - th time data block of each path.

3. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 2, characterized in that: The linear equalization in Step 5 is MMSE equalization, and the MMSE equalization matrix is: Among them, G MMSE is the MMSE equalization matrix, σ z 2 is the noise power, P x is the useful signal power, I M×M is the M-order unit matrix, and the superscript "H" indicates the Hermitian transpose of the matrix; The linear precoding method in Step Four is weighted fractional Fourier transform; the transformation order of the weighted fractional Fourier transform is: α1＝α2＝…＝α M ＝1 Among them, α1 is the transformation order of the time data block of the first channel, α2 is the transformation order of the time data block of the second channel, and α M is the transformation order for the Mth time data block.

4. A multi-antenna system transmission method with symbol energy time-frequency-space dispersion, characterized in that: In the multi - antenna transmitter at the transmitting end, there are \(T\) transmitting antennas, and each transmitting antenna has a dedicated radio frequency chain; in the multi - antenna non - linear receiver at the receiving end, there are \(R\) receiving antennas, and each receiving antenna has a dedicated radio frequency chain. The working process at the transmitting end is as follows: Step One: Perform serial - to - parallel conversion on the bit data generated by the digital data source to obtain \(M\) parallel bit streams, where \(M\leq\min(T,R)\), and \(\min(T,R)\) represents taking the smaller value of \(T\) and \(R\). Step Two: Perform digital modulation on each of the bit streams in Step One to obtain \(M\) constellation symbols. Step Three: Perform grouping processing on each constellation symbol in Step Two. That is, for any constellation symbol, divide the constellation symbols from the 1st moment to the \(N\) - th moment of this constellation symbol into the 1st group, and use the 1st group of constellation symbols as the 1st time data block. Then divide the constellation symbols from the \((N + 1)\) - th moment to the \(2N\) - th moment into the 2nd group, and use the 2nd group of constellation symbols as the 2nd time data block until the grouping of all constellation symbols in this constellation symbol is completed, where \(N\geq M\) and \(N\) is an integer multiple of \(M\). Similarly, perform grouping processing on the other constellation symbols respectively. Step 4: The first time data block of the mth path It is expressed as: Among them, s m1 ,s m2 ,s m3 ,…,s mN are respectively the first, second, third, ..., Nth moment constellation symbols in the mth constellation symbol, m = 1, 2, ..., M; Linear precoding is performed on the first time data block of each channel respectively, and the linear precoding matrix of the first time data block of the mth channel is represented as W (m) , then the precoding result of the first time data block of the mth channel is for: where "\(\times\)” represents matrix multiplication, and the superscript "\(T\)” represents matrix transpose. The precoding results corresponding to the first time data block of each channel together form a space-time processing block X 1 : Step 5: Process the space-time block X output in step 4 1 The airspace energy distribution is adjusted. The specific process of the airspace energy distribution adjustment is as follows: Process block X when empty 1 , the data of the time data block of the mth path at the nth moment is used to replace the data of the time data block of the [(m+n-2)modM]+1th path at the nth moment, where mod represents a remainder operation; Then the space-time processing block after the spatial energy distribution is adjusted It is expressed as: in Represents space-time processing block X 1 The nth column of d is an M-order cyclic downshift matrix, and the superscripts 0, 1, ..., N-1 represent the matrix Π d The power of The M-order cyclic downshift matrix Π d for: Step 6: Adjust the spatial energy distribution and then perform space-time processing Each time data block in is processed separately, and the processed data is emitted; The space-time processing block after adjusting the spatial energy distribution Each time data block in is processed separately, which is as follows: For space-time processing blocks Each time data block in is sequentially processed by adding a cyclic prefix, mapping to a transmitting antenna, digital-to-analog conversion, and up-conversion; the rule for mapping a time data block to a transmitting antenna is: If M = T, the first time data block is mapped to the first transmitting antenna, the second time data block is mapped to the second transmitting antenna, …, and the M-th time data block is mapped to the M-th transmitting antenna; otherwise, if M < T, the first time data block is mapped to the first transmitting antenna, the second time data block is mapped to the second transmitting antenna, …, and the M-th time data block is mapped to the M-th transmitting antenna, and the time data blocks mapped to the (M + 1)-th, (M + 2)-th, …, T-th transmitting antennas are all space-time processing blocks Any one of the time data blocks in , that is, for each of the (M + 1)-th, (M + 2)-th, …, T-th transmitting antennas, the time data block mapped to it is a space-time processing block Any time data block in ; Step Seven: Similarly, for the \(l\) - th time data block of each path, execute Steps Four to Six, where \(l = 2,3,\cdots,L\), and \(L\) is the number of time data blocks of each path. The working process at the receiving end is as follows: Step 1: After the first time data block of each path processed in Step Six passes through the channel, \(R\) receiving antennas respectively receive analog signals from free space, and then perform pre - processing on the analog signals received by each receiving antenna. After pre - processing, \(R\) digital sequences are obtained. Step 2: Use the \(R\) receiving antennas together to determine the starting sampling point of the received time data block. Step 3: Remove the cyclic prefix of the R-channel digital sequence according to the starting sampling point and the cyclic prefix length respectively; Step 4: The receiver performs channel estimation using pilots to obtain the MIMO channel matrix H with R rows and M columns; Step 5: The receiving space-time processing block consisting of the R-way digital sequence after removing the cyclic prefix is ​​recorded as Y in , for Y in Perform serial-to-parallel conversion to obtain spatial data block Y p , Y p =[y p,1 ,y p,2 ,y p,3 ,…,y p,RN ] T ,y p,1 ,y p,2 ,y p,3 ,…,y p,RN They are spatial data blocks Y p the 1st, 2nd, 3rd, …, RNth element in ; The rule of the serial-to-parallel conversion is as follows: The empty time processing block Y in The 1st to the Rth row of the nth column is used as the spatial data block Y p From row [(n-1)R+1] to row nR in column 1; Step 6: The receiver uses the MIMO channel matrix H and the linear precoding matrix of each time data block to calculate the equivalent channel matrix H for serial interference cancellation detection with NR rows and NM columns. eq ; The equivalent channel matrix H eq The calculation method for the [(k-1)R+1]th row to the kRth row in the cth column is: Among them, H eq,k,c Denotes the matrix H eq The submatrix consisting of the elements from row [(k-1)R+1] to row kR in column c, 1≤c≤MN, 1≤k≤N, represents the k-th row and v-th column element of the linear precoding matrix of the u-th time data block. The superscript (k-1) represents the (k-1)th power of the matrix. The subscript "·u" represents all elements of the u-th column of the matrix. "·" represents the multiplication of a scalar and a matrix, where: In the formula, represents the smallest positive integer not less than c / M; Step 7: Use spatial data block Y p and the equivalent channel matrix H eq Perform i-times of serial interference elimination based on post-detection Euclidean distance sorting and MN-i-times of serial interference elimination based on signal-to-interference-noise ratio sorting; The specific process of Step 7 is as follows: Step 7.1: Set the detection number record variable iter = 1, and initialize the post-detection vector r p is a vector of all zeros with 1 column and MN rows; Step 7.2: Compare the magnitudes of iter and MN. If iter > MN, then execute Step 7.21; otherwise, continue to compare the magnitudes of iter and i. If iter ≤ i, then execute Step 7.3; otherwise, execute Step 7.12; Step 7.3: Use the equivalent channel matrix after iter-1th serial interference cancellation detection Calculate the linear equalization matrix of the iter-th detection, and use the linear equalization matrix to eliminate the spatial data block after the iter-1-th serial interference detection Perform equalization and obtain the iterth linear equalization result in, Step 7.4: Take the iterth linear equalization result in step 7.3 Send to buffer; Step 7.5: For the iterth linear equalization result in step 7.3 Perform digital demodulation, and then digitally modulate the demodulation result to obtain the constellation symbol vector of the iter-th digital demodulation result Step 7.6: Read the iterth linear equalization result in the buffer of step 7.4 Calculate the iterth linear equalization result The constellation symbol vector of the iter-th digital demodulation result The Euclidean distance vector e between (iter) : Where "|·|" represents calculating the absolute value of each element of the vector respectively, and "⊙" represents multiplying the corresponding elements of the vectors; Step 7.7: Search for the minimum value of the element in the Euclidean distance vector of step 7.6, and record the minimum value in the Euclidean distance vector e (iter) The index index and the minimum value in the first linear equalization result The corresponding original index ori; Step 7.8: The indexth row element is digitally demodulated, and the demodulation result is used as an element to update the post-detection vector r in step 7.

1. p The ori-th row element of ; Step 7.9: Keep the value in step 7.5 The index-th row element of Change the other elements of to 0; Step 7.10: Calculate the spatial data block after the iterth serial interference elimination detection and the equivalent channel matrix after the iterth serial interference cancellation detection In the formula, Representation Matrix dth column of ; Step 7.11: Update iter to iter + 1, and return to Step 7.2; Step 7.12: Use the equivalent channel matrix after iter-1th serial interference cancellation detection Calculate the linear equalization matrix G for the iterth serial interference cancellation detection (iter) ; Step 7.13: Use the linear equalization matrix G (iter) The spatial data block after the iter-1th serial interference elimination detection Perform linear equalization and obtain the iterth linear equalization result Step 7.14: Use the linear equalization matrix and Calculate the signal to interference noise ratio of each to-be-demodulated symbol in the iter-th linear equalization result, where the signal to interference noise ratio SINR of the lth to-be-demodulated symbol in the iter-th linear equalization result is l The calculation method is: in, Denotes the linear equalization matrix G (iter) The lth row of Representation Matrix The lth column of Representation Matrix The col-th column of , col≠l, "||·||" represents the 2-norm of the vector; Step 7.15: Search for the maximum value of the signal-to-interference-noise ratio of each symbol to be demodulated in the iterth linear equalization result of step 7.14, record the index of the maximum value as index, and record the maximum value in the first linear equalization result. The corresponding original index ori; Step 7.16: The indexth row element is digitally demodulated, and the demodulation result is used as an element to update the post-detection vector r in step 7.

1. p The ori-th row element of ; Step 7.17: Construct the constellation symbol vector of the iter-th digital demodulation result The number of rows and columns are (MN-iter+1) and 1 respectively. The indexth row element of is the constellation modulation symbol corresponding to the digital demodulation result in step 7.16, and the remaining elements are 0; Step 7.18: Calculate the spatial data block after the iterth serial interference elimination detection and the equivalent channel matrix after the iterth serial interference cancellation detection Step 7.19: Compare the magnitudes of iter and MN. When iter < MN, execute Step 7.20; otherwise, execute Step 7.21; Step 7.20: Update iter to iter + 1, and return to Step 7.12; Step 7.21: Output the detected vector r p , complete i times of serial interference elimination based on post-detection Euclidean distance sorting and (MN-i) times of serial interference elimination based on signal-to-interference-noise ratio sorting; Step 8: The post-detection vector r output from step 7.21 p Perform parallel-to-serial conversion to restore the initial space-time processing block r; the parallel-to-serial conversion rule is: The detected vector r p The first column of [(n-1)R+1] to the nRth row of the first column of is used as the space-time processing block r's first column of n to the Rth row; Step 9: Perform serial-to-parallel conversion on the 0 / 1 bit sequences of each path of the space-time processing block r in Step 8 to recover the source information, and complete the non-linear receiving detection of one space-time processing block; Step 10: For the data of the l-th time data block of each path after being processed in Step 6 at the transmitting end through the channel, the processes of Steps 1 to 9 are all executed, where l = 2, 3,..., L.

5. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 4, characterized in that: The linear equalization matrix G in step 7.3 and step 7.12 (iter) is the ZF equilibrium matrix, which is calculated as follows: Among them, the superscript "H" represents the Hermitian transpose of the matrix, and the superscript "-1" represents the inverse of the matrix.

6. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 5, characterized in that: The method of the linear precoding is the weighted fractional Fourier transform, and the calculation method of the transform order is as follows: Step ①: For the first time data block of each path, the transform order is a default order agreed upon by both the transmitter and the receiver; Step ②: The calculation method of the transform order for the second time data block of each path is as follows: Step A: The receiver calculates the critical noise power P th ; The specific process is: Define the function P e,1 (x) is: Among them, x is the independent variable of the function, Q(·) represents the right tail function of the standard normal distribution, P erg To test the power; Where e is the base of the natural logarithm; Take 1 / 3 as the initial value of the test power, and then gradually increase the test power P by δ1. erg The value of P erg / M is no longer a function P e,1 When (x) is at its minimum value, stop testing the power P erg The update of the final trial power P erg The critical noise power P th ; Step B: Calculate the noise power after ZF equalization Among them, "||·|| F " represents the F-norm of the matrix, G (1) is the ZF equilibrium matrix obtained in the first iteration; Step C: Compare the noise power after ZF equalization and critical noise power P th size; If the noise power after ZF equalization Not greater than the critical noise power P th , then the weighted fractional Fourier transform order of the second time data block of each channel is: α1=α2=…α M =1, and execute step H; Otherwise, if the noise power after ZF equalization is Greater than the critical noise power P th , then execute step D; Step D: Define function P e,2 (x) is: Solving function P e,2 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P low ; Step E: Define function P e,3 (x) is: Solving function P e,3 (x) is the value of the independent variable when it takes the minimum value, and the solved independent variable value is recorded as P high ; Step F: Construct an ideal noise power distribution vector P ideal : Step G: Use ZF balance matrix G (1) and the ideal noise power distribution vector P ideal Calculate the transformation order of the second time data block of each channel; the specific process is: Step G1: Let m = 1; Step G2, initialize the mean square error value to MSE temp , let the trial order α erg is δ2-2; Step G3: Calculate the actual noise power distribution vector P of the mth time data block act,m , the calculation formula is: Among them, P act,m1 is the actual noise power of the first constellation symbol of the mth time data block, P act,m2 is the actual noise power of the second constellation symbol of the mth time data block, P act,m3 is the actual noise power of the third constellation symbol of the mth time data block, P act,mN is the actual noise power of the Nth constellation symbol of the mth time data block; Where W qM+m,n (α erg ) indicates that the transformation order is α erg The element of the qM+mth row and nth column of the N-point weighted fractional Fourier transform matrix, q=0,1,…,N / M-1; Denotes the matrix G (1) In the ath row, a=[(m-1)mod M]+1,[m mod M]+1,[(m+1)mod M]+1,…,[(m+M-2)mod M]+1; Step G4: The actual noise power distribution vector P act,m The elements in are sorted from large to small, and the noise power distribution vector P of the m-th time data block is obtained after sorting. order,m ; Step G5: Calculate the noise power distribution vector P order,m With the ideal noise power distribution vector P ideal Mean square error MSE: MSE=||P order,m -P ideal || Step G6: Compare the MSE calculated in step G5 with the MSE temp size; If MSE is less than MSE temp , then execute step G7; otherwise, execute step G8; Step G7: MSE temp Update to MSE, and change the weighted fractional Fourier transform order α of the m-th time data block m Updated to alpha erg , and then execute step G8; Step G8: If α erg +δ2 is less than or equal to 2, then α erg Updated to alpha erg +δ2, then return to step G3; Otherwise α erg +δ2 is greater than 2, then execute step G9; Step G9: Compare the magnitudes of m and M; If m < M, then update m to m + 1, and re-execute Steps G2 to G8; Otherwise, if m = M, then execute Step H; Step H: Construct the transform order of the second time data block of each path as the time data block transform order vector α, and feedback it to the transmitter: α＝[α1,α2,…α M ] Among them, α1 is the transformation order of the second time data block of the first path, α2 is the transformation order of the second time data block of the second path, and α M is the transformation order of the second time data block of the Mth path; Step ③: For the l-th time data block of each path, where l ≥ 3, the calculation method of its transform order is as follows: During the processing of the l - 1-th time data block of each path by the transmitter and the receiver, obtain the ZF equalization matrix obtained in the first iteration, and use the obtained ZF equalization matrix and the method in Step ② to calculate the transform order for processing the l-th time data block of each path.

7. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 6, characterized in that: When the weighted fractional Fourier transform order of each time data block is α1=α2=…α M =1, the value of the number of serial interference elimination times i based on the post-detection Euclidean distance sorting is 1; otherwise, the value of the number of serial interference elimination times i based on the post-detection Euclidean distance sorting is 0.

8. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 4, characterized in that: The linear equalization matrix G in steps 7.3 and 7.12 (iter) is the MMSE equalization matrix, and its calculation method is: Among them, I (MN-iter+1)×(MN-iter+1) is the (MN-iter+1)-order unit matrix, the superscript "H" represents the Hermitian transpose of the matrix, the superscript "-1" represents the inverse of the matrix, σ z 2 is the noise power, P x is the useful signal power.

9. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 8, characterized in that: The method of the linear precoding is the weighted fractional Fourier transform, and the transform order of each time data block is as follows: α1＝α2＝…＝α M ＝1 Among them, α1 is the transformation order of the time data block of the first channel, α2 is the transformation order of the time data block of the second channel, and α M is the transformation order for the Mth time data block.

10. A multi-antenna system transmission method for symbol energy time-frequency-space dispersion according to claim 9, characterized in that: The number of serial interference cancellations i based on the sorted Euclidean distance after detection is i = 1.

Images