An adaptive scene pilot design method based on a MIMO-OTFS system

Abstract

An adaptive scenario pilot design method based on a MIMO-OTFS system is proposed. At the transmitting end, information bits undergo preprocessing such as serial-to-parallel conversion, convolutional coding, and interleaving, and are then modulated and mapped into time-delay Doppler domain data symbols. After transmission through the channel, a feedback iteration mechanism is introduced at the receiving end. The initial iteration uses the initial pilot for channel estimation and symbol detection. In subsequent iterations, high-confidence data symbols are dynamically selected to expand the dimension of the observation matrix, thereby improving the accuracy of channel parameter estimation. Subsequently, the high-precision channel estimation results are used to iteratively detect low-confidence symbols multiple times, allowing the system to converge to optimal performance. The method is suitable for complex wireless channel scenarios such as high-speed movement and strong multipath propagation, and can achieve high-precision channel estimation and reliable signal detection.

Description

Technical Field

[0001] This invention relates to an adaptive scene pilot design method based on a MIMO-OTFS system, belonging to the field of wireless communication technology. Background Technology

[0002] In ultra-high-speed mobile scenarios such as unmanned aerial vehicles, high-speed trains, and low-orbit satellites, received signals are often distorted due to high-frequency bias. To recover the bit information from the transmitter, designing a suitable pilot pattern to address issues such as low channel estimation accuracy and high bit error rate during communication is crucial. To solve this problem, existing solutions typically involve inserting pilots in the time-delay Doppler domain. Common methods generally fall into two categories: power-enhanced orthogonal pilot structures and non-orthogonal pilot structures based on complex Gaussian random sequences.

[0003] The time-delay Doppler domain high-power pulse pilot scheme proposed by Raviteja et al. achieves orthogonal isolation between pilot and data symbols by setting up multi-dimensional zero-guard regions, which is simple to implement and widely applicable. However, this scheme has obvious limitations in MIMO-OTFS systems: high pilot overhead, excessively high system PAPR can easily cause nonlinear distortion of the power amplifier, and a large number of zero-guard pilots destroy the full-rank property of the observation matrix, thus limiting the accuracy of channel estimation.

[0004] The non-orthogonal pilot scheme based on complex Gaussian sequences proposed by Shen Wenqian et al. reduces PAPR and improves spectral efficiency by dispersing signal energy through random phase and reducing the zero guard interval. However, the pilot overhead of this scheme increases significantly with the number of antennas and is constrained by hardware quantization errors and storage costs, making its engineering implementation difficult.

[0005] Both existing pilot schemes are fixed designs, lack scene adaptability, and do not make full use of the soft information feedback from symbol detection output, resulting in the channel estimation and symbol detection accuracy failing to meet the reliable communication requirements in high-speed mobile scenarios. Summary of the Invention

[0006] The technical problem solved by this invention is: to address the problems of various requirements being difficult to meet, complex implementation, and insufficient accuracy in the existing technology, an adaptive scene pilot design method based on the MIMO-OTFS system is proposed.

[0007] The present invention solves the above-mentioned technical problem through the following technical solution: An adaptive scene pilot design method based on a MIMO-OTFS system includes: The received bit data is processed by serial-to-parallel conversion, channel coding, and constellation mapping to obtain delay-Doppler domain symbol data. The time-delay-Doppler domain symbol data is transformed into the TF domain by inverse finite symplectic Fourier transform to obtain the TF domain signal matrix; Perform Heisenberg transformation and transmit pulse shaping on the TF domain signal matrix to obtain the time domain signal matrix of the transmitter, which is then transmitted to the wireless channel by the transmitting antenna. The receiving antenna of the wireless channel receives the time-domain signal matrix of the transmitting end, performs signal domain conversion on the time-domain signal matrix of the transmitting end, and obtains the TF domain received signal sample. Perform SFFT transformation on the received signal samples in the TF domain to obtain the corresponding received antenna's DD domain signal; Determine the current wireless channel scenario, determine the initial pilot based on the current wireless channel scenario type, and output the initial pilot pattern; Based on the initial pilot pattern, the adaptive block orthogonal matching pursuit algorithm is used to traverse the antenna array composed of receiving and transmitting antennas, and to perform signal estimation to update the antenna array information to obtain the channel estimation matrix corresponding to the updated antenna array. Based on the channel estimation matrix, the time-domain signal matrix of the transmitter, and the time-domain pilot signal of the receiver after transmission through the wireless channel, the transmission signal is generated by an approximate message passing algorithm. High-confidence symbols are selected from the transmitted signal and merged with the initial pilot to obtain the pilot symbol matrix; The initial pilot is updated according to the current wireless channel scenario type until the pilot symbol matrix meets the convergence condition. The final pilot symbol matrix is ​​used as the joint pilot, and the corresponding joint pilot pattern is output.

[0008] The channel coding process uses convolutional codes for encoding, and the constellation mapping process uses QPSK modulation for data modulation; the method for transforming the time-delay-Doppler domain symbol data to the TF domain is as follows: The inverse finite symplectic Fourier transform is performed on the time-delay-Doppler domain symbol data to obtain the TF domain signal matrix at the transmitting end. The method for calculating the value of each element in the TF domain signal matrix is ​​as follows:

[0009] In the formula, a single element of the TF-domain signal matrix is , For the first Time-domain transmitted signal matrix of the root transmitting antenna , For transmitting antenna index, For time indexing, For frequency index, For delay index, For Doppler indexing.

[0010] The transmitting end time-domain signal matrix The calculation method for each element value is as follows:

[0011] In the formula, It is the pulse shaping function of the transmitting end, and the time-domain signal is sent from the transmitting antenna to the wireless channel.

[0012] The wireless channel selects the receiving antenna based on the transmitting antenna's time-domain signal matrix and performs receiving signal transformation to obtain the receiving time-domain signal matrix. The method for determining the receiving time-domain signal matrix is ​​as follows:

[0013] In the formula, For the first The receiving end time-domain signal matrix of the root receiving antenna. For the receiving antenna index, For DD channel impulse response, For the first The signal transmitted by the root transmitting antenna, It is additive white Gaussian noise.

[0014] The method for obtaining the received signal sample in the TF domain is as follows: For the The time-domain signal matrix of the receiving antenna is subjected to matched filtering to calculate the first... Cross-ambiguity function of root receiving antenna :

[0015] In the formula, This represents the impulse response of the matched filter at the receiving end; Among them, the TF domain received signal samples are processed by the cross-ambiguity function. After sampling the grid points, the method for determining each element in the received signal sample in the TF domain is as follows:

[0016] In the formula, For TF domain received signal samples, It represents any element value in the received signal sample of the TF domain.

[0017] The SFFT transformation method is as follows: TF domain received signal samples Perform two-dimensional transformation, including the received signal samples in the TF domain. Columns M Point IDFT transformation followed by N Point DFT transformation; After conversion Each element in the DD domain signal sample of the root receiving antenna The calculation formula is:

[0018] In the formula, M is the number of subcarriers and N is the number of symbols.

[0019] The wireless channel scenario types include Doppler-dominated single-selection channel scenario and high-speed multipath time-frequency dual-selection channel scenario, wherein: The Doppler-dominated single-selection channel scenario is a high-speed mobile scenario operating at a preset speed, while the high-speed multipath time-frequency dual-selection channel scenario is a data communication transmission scenario operating at a preset speed under high-speed mobile conditions. In the Doppler-dominated single-selection channel scenario, the initial pilot pattern is uniformly distributed along the Doppler axis, with no pilots configured in the time delay dimension, and the pilots are distributed linearly. In the high-speed multipath time-frequency dual-selection channel scenario, the initial pilot pattern is gridded along the time delay and Doppler two-dimensional directions, with pilots configured in both the time delay and Doppler dimensions, and the pilots are distributed in a grid pattern.

[0020] The method for obtaining the channel estimation matrix is ​​as follows: The initial pilot pattern is initialized to obtain the residual and support set, and the maximum contributing atom position of any DD domain channel grid corresponding to the initial pilot pattern is calculated. The residuals and support sets of the initial pilot pattern are updated based on the maximum contributing atomic positions of all DD domain channel grids. Then, using the index positions of the transmit and receive antennas corresponding to the updated pilot pattern and the channel parameter matrix, the channel estimation matrix is ​​reconstructed by traversing each transmit and receive antenna. .

[0021] The method for generating the transmission signal using an approximate message passing algorithm is as follows: Based on the time-domain signal matrix transmitted from the transmitter through the wireless channel, the time-domain pilot signal at the receiver is... Reconstructing the channel estimation matrix The channel estimation matrix is ​​derived to obtain the channel matrix in the row and column dimensions and the corresponding set of observation nodes; To obtain the prior probability, the mean of the Gaussian random variable is calculated and assigned to the variable node. The prior probability is iterated in a loop. The prior probability is updated using the mean of the Gaussian random variable and then passed to each observation node to obtain the updated set of confidence symbols. The estimated value of the data sign is determined based on the prior probability of reaching a preset number of iterations, and the estimated value of the data sign is used as a single element to derive the following. ,Will The set of credibility symbols is used as the output signal.

[0022] When the pilot symbol matrix satisfies the convergence condition For the first j The joint pilot symbols generated in the next iteration, when , This represents the initial pilot frequency.

[0023] The advantages of this invention compared to the prior art are: (1) The present invention provides an adaptive scene pilot design method based on MIMO-OTFS system, which adapts to different channel scenarios through scene adaptive initial pilot design and iterative joint optimization, and achieves high-precision channel estimation and reliable signal detection with low pilot overhead; (2) This invention addresses the pain points of high PAPR, high pilot overhead, and insufficient channel estimation accuracy in MIMO-OTFS systems under high-speed mobile scenarios by proposing a scenario-adaptive iterative pilot pattern design scheme. By distinguishing between Doppler-dominated scenarios and dual-selection channel scenarios, linear and grid-type pilot layouts are adopted respectively, avoiding the redundant overhead of fixed pilot design, while reducing the power amplifier nonlinear distortion problem caused by peak-to-average power ratio, and significantly improving the system's spectral efficiency and transmission efficiency; (3) This invention proposes a joint channel estimation and symbol detection mechanism based on the ABOMP and AMP algorithms. By dynamically selecting high-confidence data symbols through an iterative process to expand the observation matrix, it fully utilizes the soft information feedback of symbol detection, achieving a synergistic improvement in channel estimation accuracy and symbol detection performance. Compared with existing fixed pilot schemes, this invention can effectively improve the time / frequency selective fading effect in high-speed mobile scenarios without adding additional pilot overhead, enabling the system to maintain excellent bit error rate performance under low pilot overhead and low PAPR conditions. Attached Figure Description

[0024] Figure 1 Schematic diagrams of other existing pilot pattern design methods provided for comparison with the present invention; Figure 2 A diagram illustrating the pilot pattern design method for joint detection and estimation provided by this invention; Figure 3 A model diagram of the MIMO-OTFS joint detection and estimation system provided by the present invention; Figure 4 The JDCE algorithm flowchart is provided for an embodiment of the present invention. Detailed Implementation

[0025] An adaptive scenario pilot design method based on a MIMO-OTFS system is proposed. At the transmitting end, information bits undergo preprocessing such as serial-to-parallel conversion, convolutional coding, and interleaving, and are then mapped to time-delay Doppler domain data symbols via QPSK modulation. After transmission through the channel, a feedback iteration mechanism is introduced at the receiving end. The initial iteration uses the initial pilot for channel estimation and symbol detection. In subsequent iterations, high-confidence data symbols are dynamically selected to expand the dimension of the observation matrix, thereby improving the accuracy of channel parameter estimation. Subsequently, the high-precision channel estimation results are used to iteratively detect low-confidence symbols multiple times, allowing the system to converge to optimal performance. The method is suitable for complex wireless channel scenarios such as high-speed movement and strong multipath propagation, and can achieve high-precision channel estimation and reliable signal detection.

[0026] The adaptive scene pilot design method includes the following specific steps: The received bit data is processed by serial-to-parallel conversion, channel coding, and constellation mapping to obtain delay-Doppler domain symbol data. Transform the time-delay-Doppler domain symbol data to the TF domain to obtain the TF domain signal matrix; Perform Heisenberg transformation and transmit pulse shaping on the TF domain signal matrix to obtain the time domain signal matrix of the transmitter, which is then transmitted to the wireless channel by the transmitting antenna. The receiving antenna of the wireless channel receives the time-domain signal matrix of the transmitting end, performs signal domain conversion on the time-domain signal matrix of the transmitting end, and obtains the TF domain received signal sample. Perform SFFT transformation on the received signal samples in the TF domain to obtain the corresponding received antenna's DD domain signal; Determine the current wireless channel scenario, determine the initial pilot based on the current wireless channel scenario type, and output the initial pilot pattern; Based on the initial pilot pattern, the adaptive block orthogonal matching pursuit algorithm is used to traverse the antenna array composed of receiving and transmitting antennas, and to perform signal estimation to update the antenna array information to obtain the channel estimation matrix corresponding to the updated antenna array. Based on the channel estimation matrix, the transmitter's time-domain signal matrix, and the receiver's time-domain pilot signal transmitted via the wireless channel, a transmission signal is generated using an approximate message passing algorithm. Here, the receiver's time-domain pilot signal refers to the transmitter's time-domain signal matrix. After reaching the receiving end via the wireless channel, the signal becomes... ; High-confidence symbols are selected from the transmitted signal and merged with the initial pilot to obtain the pilot symbol matrix; The initial pilot is updated according to the current wireless channel scenario type until the pilot symbol matrix meets the convergence condition. The final pilot symbol matrix is ​​used as the joint pilot, and the corresponding joint pilot pattern is output.

[0027] Channel coding uses convolutional codes for encoding, and constellation mapping uses QPSK modulation for data modulation; the method for transforming the time-delay-Doppler domain symbol data to the TF domain is as follows: The inverse finite symplectic Fourier transform is performed on the time-delay-Doppler domain symbol data to obtain the TF domain signal matrix at the transmitting end. The method for calculating the value of each element in the TF domain signal matrix is ​​as follows:

[0028] In the formula, a single element of the TF-domain signal matrix is , For the first Time-domain transmitted signal matrix of the root transmitting antenna , For transmitting antenna index, For time indexing, For frequency index, For delay index, For Doppler indexing.

[0029] Transmitter time-domain signal matrix The calculation method for each element value is as follows:

[0030] In the formula, It is the pulse shaping function of the transmitting end, and the time-domain signal is sent from the transmitting antenna to the wireless channel.

[0031] The wireless channel selects the receiving antenna based on the transmitting antenna's time-domain signal matrix and performs receiving signal transformation to obtain the receiving time-domain signal matrix. The method for determining the receiving time-domain signal matrix is ​​as follows:

[0032] In the formula, For the first The receiving end time-domain signal matrix of the root receiving antenna. For the receiving antenna index, For DD channel impulse response, For the first The signal transmitted by the root transmitting antenna, It is additive white Gaussian noise.

[0033] The method for obtaining TF domain received signal samples is as follows: For the The time-domain signal matrix of the receiving antenna is subjected to matched filtering to calculate the first... Cross-ambiguity function of root receiving antenna :

[0034] In the formula, This represents the impulse response of the matched filter at the receiving end; Among them, the TF domain received signal samples are processed by the cross-ambiguity function. After sampling the grid points, the method for determining each element in the received signal sample in the TF domain is as follows:

[0035] In the formula, For TF domain received signal samples, It represents any element value in the received signal sample of the TF domain.

[0036] The SFFT transformation method is as follows: TF domain received signal samples Perform two-dimensional transformation, including the received signal samples in the TF domain. Columns M Point IDFT transformation followed by N Point DFT transformation; After conversion Each element in the DD domain signal sample of the root receiving antenna The calculation formula is:

[0037] In the formula, M is the number of subcarriers and N is the number of symbols.

[0038] Wireless channel scenarios include Doppler-dominated single-selection channel scenarios and high-speed multipath time-frequency dual-selection channel scenarios, among which: The Doppler-dominated single-selection channel scenario is a high-speed mobile scenario operating at a preset speed, while the high-speed multipath time-frequency dual-selection channel scenario is a data communication transmission scenario operating at a preset speed under high-speed mobile conditions. In the Doppler-dominated single-selection channel scenario, the initial pilot pattern is uniformly distributed along the Doppler axis, with no pilots configured in the time delay dimension, and the pilots are distributed linearly. In the high-speed multipath time-frequency dual-selection channel scenario, the initial pilot pattern is gridded along the time delay and Doppler two-dimensional directions, with pilots configured in both the time delay and Doppler dimensions, and the pilots are distributed in a grid pattern.

[0039] The method for obtaining the channel estimation matrix is ​​as follows: The initial pilot pattern is initialized to obtain the residual and support set, and the maximum contributing atom position of any DD domain channel grid corresponding to the initial pilot pattern is calculated. The residuals and support sets of the initial pilot pattern are updated based on the maximum contributing atomic positions of all DD domain channel grids. Then, using the index positions of the transmit and receive antennas corresponding to the updated pilot pattern and the channel parameter matrix, the channel estimation matrix is ​​reconstructed by traversing each transmit and receive antenna. .

[0040] The method for generating the transmission signal using an approximate message passing algorithm is as follows: Based on the time-domain signal matrix transmitted from the transmitter through the wireless channel, the time-domain pilot signal at the receiver is... Reconstructing the channel estimation matrix The channel estimation matrix is ​​derived to obtain the channel matrix in the row and column dimensions and the corresponding set of observation nodes; To obtain the prior probability, the mean of the Gaussian random variable is calculated and assigned to the variable node. The prior probability is iterated in a loop. The prior probability is updated using the mean of the Gaussian random variable and then passed to each observation node to obtain the updated set of confidence symbols. The estimated value of the data sign is determined based on the prior probability of reaching a preset number of iterations, and the estimated value of the data sign is used as a single element to derive the following. ,Will The set of credibility symbols is used as the output signal.

[0041] in, yes One element, with This is equivalent to knowing the matrix. The value at each index position in the text, therefore directly through Write it down directly

[0042] When the pilot symbol matrix satisfies the convergence condition, For the first j The joint pilot symbols generated in the next iteration, when , This represents the initial pilot frequency.

[0043] The following description, in conjunction with the accompanying drawings and preferred embodiments, provides further details: Traditional solutions such as Figure 1 As shown, including as Figure 1 (a) Figure 1 As shown in (b), both existing pilot schemes are fixed designs, lack scene adaptability, and do not make full use of the soft information feedback of symbol detection output, resulting in the channel estimation and symbol detection accuracy being difficult to meet the reliable communication requirements in high-speed mobile scenarios.

[0044] In the current embodiment, the scenario-adaptive pilot design method of the MIMO-OTFS system adapts to different channel scenarios through scenario-adaptive initial pilot design and iterative joint optimization, achieving high-precision channel estimation and reliable signal detection with low pilot overhead. Specific steps are as follows: Figure 2 As shown: (1) Transmitter signal preprocessing: After serial-to-parallel conversion, channel coding, and constellation mapping of the input bits, each bit stream is converted into a symbol in the delay-Doppler (DD) domain; where convolutional coding is used for channel coding and QPSK modulation is used for constellation mapping; let the first bit be... The time-domain transmitted signal matrix of the root transmitting antenna is as follows , For transmitting antenna index; (2) Transformation of the transmitter signal to the TF domain: The DD domain symbols obtained in step (1) are subjected to inverse finite symplectic Fourier transform (ISFFT) to obtain the time domain (TF) signal of the transmitter. Specifically, each element of the TF domain signal matrix The calculation formula is: , in, For time indexing, For frequency index, For delay index, For Doppler indexing; (3) Transformation of the signal in the time domain and transmission at the transmitting end: Through step (2), the TF domain signal matrix is ​​transformed. Perform Heisenberg transform and transmit pulse shaping operations to generate a time-domain signal matrix. Each element of the time-domain signal matrix can be calculated using the following formula: , in, This is the pulse shaping function at the transmitting end. This time-domain signal is ultimately transmitted to the wireless channel via the transmitting antenna. (4) Receiver signal transformation: The receiver's first... Time-domain transmitted signal of the root receiving antenna It can be calculated using the following formula:

[0045] in, For the receiving antenna index, Indicates the impulse response of the DD channel. It is the first The signal transmitted by the root transmitting antenna, This represents additive white Gaussian noise.

[0046] (5) Receiver signal domain conversion: For the signal domain conversion in step (4) Perform matched filtering and calculate The cross-ambiguity function on the root receiving antenna is defined as follows:

[0047] in, This represents the impulse response of the matched filter at the receiving end. Subsequently, the signal... Grid point sampling is performed to obtain a TF-domain sample matrix, the elements of which are defined as:

[0048] (6) Receiver signal domain conversion: Convert the received signal in the TF domain. Perform SFFT to obtain the first DD domain signal matrix of the root receiving antenna SFFT corresponds to a two-dimensional transform, firstly... Columns M Click IDFT, and then perform it. N Point DFT. Each element The calculation formula is:

[0049] (7) Scene-adaptive initial pilot selection: Determine the current wireless channel scene and select the corresponding initial pilot pattern according to the scene type. .

[0050] (8) Channel estimation: Input the pilot and received pilot matrices in the DD domain, and combine them with the Adaptive Block Orthogonal Matching Pursuit (ABOMP) algorithm to perform signal estimation and obtain the channel estimation matrix. ; (9) Signal detection: The channel estimation matrix obtained in step (7) is used to detect the signal. Receiver time-domain pilot signal In the Common Input Approximate Message Passing (AMP) algorithm, the estimated transmitted signal is obtained. ; (10) Pilot update: The transmitted signal estimated in step (8) Selecting high-confidence symbols This is combined with the initial pilot signal to form a new pilot symbol matrix. ; (11) Iterative optimization: Repeat steps (5) to (7) until the convergence condition is met. The corresponding pilot symbol matrix is ​​then the joint pilot pattern for the MIMO-OTFS system; where, For the first j The joint pilot symbols generated in the next iteration, when ,Right now , representing the initial pilot frequency.

[0051] Furthermore, the scenarios in step (7) are divided into two types: Scenario 1: Doppler-dominated single-selection channel scenarios mainly include high-speed movement scenarios such as maglev trains, high-speed maneuvering drones, and high-speed weapon platforms such as missiles. In these scenarios, the communication transceivers have extremely high relative motion speeds, but few multipath components, causing the channel response to change rapidly over time, exhibiting significant time-selective fading characteristics. Therefore, the initial pilots are uniformly distributed only along the Doppler axis, with no pilots configured along the delay dimension, resulting in an overall linear pilot distribution.

[0052] Scenario 2: High-speed multipath time-frequency dual-selection channel scenarios mainly include complex transmission scenarios such as broadband communication in high-speed mobile environments, high-speed aircraft communication with significant multipath delay spread, high-speed satellite data transmission, and broadband data links for high-speed maneuvering UAVs. In these scenarios, the communication transceivers not only have extremely high relative motion speeds but also rich multipath propagation components, causing the channel response to change rapidly with both time and frequency, exhibiting significant time-selective and frequency-selective fading characteristics. Therefore, the initial pilots are arranged in a grid pattern in both the time delay and Doppler dimensions, with pilots configured in both the time delay and Doppler dimensions, resulting in an overall grid-like pilot distribution.

[0053] Furthermore, the specific steps of the ABOMP algorithm described in step (8) are as follows: (8a) Initialization: Residual Support set and number of iterations Fractional Doppler effect approximation parameters Define a block sparse channel model and the search range of the channel support set; (8b) Calculate the position of the maximum contributing atom: For each DD channel grid, calculate the correlation between the residual and each atom in the dictionary matrix, using the following formula: ; (8c) Solve for the index interval, update the channel parameter matrix and residuals: using atomic positions And the index range derived from the sparsity of signal blocks, ,pass Update support set Using support sets Iterative update of least squares solution and residual ; (8d) Reconstructing the channel matrix Traverse all transmit and receive antennas, and reconstruct the system signal matrix using the index positions and channel parameter matrix. .

[0054] Furthermore, the specific steps of the AMP algorithm described in step (9) are as follows: (9a) Initialization: Receive signal Channel matrix Initialization parameters Maximum number of iterations High-reliability index set High confidence probability threshold ; (9b) Derive the relevant parameters of the channel matrix: from Derive the channel matrix in row and column dimensions , and the corresponding set of observation nodes , ; (9c) Iterative loop (when) or (Execution at the time): Utilize the prior probability from the previous round. Calculate Gaussian random variables mean and Passed to the variable node ;use , and renew And transmit it to the observation node. ; Update the set of high-confidence symbols ; (9d) Based on the final sign probability Determine the estimated value of the data sign. Output With high-confidence symbol index set .

[0055] The scene-adaptive iterative pilot pattern designed in this embodiment requires the use of the ABOMP and AMP algorithms to construct a joint channel estimation and detection mechanism, as follows: Figure 4 As shown. In the iterative update of the pilot, high-confidence data symbols need to be selected and combined with the current pilot to form the pilot structure for the next iteration stage. This process is implemented by the AMP algorithm. The AMP algorithm is based on the channel matrix estimated by the ABOMP algorithm. The two work together to achieve adaptive pilot optimization.

[0056] Process design such as Figure 2As shown, the information bits at the transmitting end undergo preprocessing such as serial-to-parallel conversion, convolutional coding, and interleaving, and are then mapped into time-delay Doppler domain data symbols via QPSK modulation. After transmission through the channel, a feedback iteration mechanism is introduced at the receiving end. The first iteration uses initial pilots for channel estimation and symbol detection. In subsequent iterations, high-confidence data symbols are dynamically selected to expand the dimension of the observation matrix, thereby improving the accuracy of channel parameter estimation. Subsequently, the high-precision channel estimation results are used to perform multiple iterations to detect low-confidence symbols, enabling the system to converge to optimal performance. The scene-adaptive pilot patterns for each scenario are shown below. Figure 3 As shown.

[0057] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

[0058] The contents not described in detail in this specification are common knowledge to those skilled in the art.

Claims

1. A method for adaptive scenario pilot design based on MIMO-OTFS system, characterized in that include: The received bit data is processed by serial-to-parallel conversion, channel coding, and constellation mapping to obtain delay-Doppler domain symbol data. The time-delay-Doppler domain symbol data is transformed into the TF domain by inverse finite symplectic Fourier transform to obtain the TF domain signal matrix; Perform Heisenberg transformation and transmit pulse shaping on the TF domain signal matrix to obtain the time domain signal matrix of the transmitter, which is then transmitted to the wireless channel by the transmitting antenna. The receiving antenna of the wireless channel receives the time-domain signal matrix of the transmitting end, performs signal domain conversion on the time-domain signal matrix of the transmitting end, and obtains the TF domain received signal sample. Perform SFFT transformation on the received signal samples in the TF domain to obtain the corresponding received antenna's DD domain signal; Determine the current wireless channel scenario, determine the initial pilot based on the current wireless channel scenario type, and output the initial pilot pattern; Based on the initial pilot pattern, the adaptive block orthogonal matching pursuit algorithm is used to traverse the antenna array composed of receiving and transmitting antennas, and to perform signal estimation to update the antenna array information to obtain the channel estimation matrix corresponding to the updated antenna array. Based on the channel estimation matrix, the time-domain signal matrix of the transmitter, and the time-domain pilot signal of the receiver after transmission through the wireless channel, the transmission signal is generated by an approximate message passing algorithm. High-confidence symbols are selected from the transmitted signal and merged with the initial pilot to obtain the pilot symbol matrix; The initial pilot is updated according to the current wireless channel scenario type until the pilot symbol matrix meets the convergence condition. The final pilot symbol matrix is ​​used as the joint pilot, and the corresponding joint pilot pattern is output.

2. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 1, characterized in that: The channel coding process uses convolutional codes for encoding, and the constellation mapping process uses QPSK modulation for data modulation; the method for transforming the time-delay-Doppler domain symbol data to the TF domain is as follows: The inverse finite symplectic Fourier transform is performed on the time-delay-Doppler domain symbol data to obtain the TF domain signal matrix at the transmitting end. The method for calculating the value of each element in the TF domain signal matrix is ​​as follows: where a single element of the T-F domain signal matrix is , is the time domain transmit signal matrix for the th transmit antenna , is the transmit antenna index, is the time index, is the frequency index, is the delay index, is the Doppler index.

3. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 2, characterized in that: The transmitting end time-domain signal matrix The calculation method for each element value is as follows: In the formula, It is the pulse shaping function of the transmitting end, and the time-domain signal is sent from the transmitting antenna to the wireless channel.

4. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 3, characterized in that: The wireless channel selects the receiving antenna based on the transmitting antenna's time-domain signal matrix and performs receiving signal transformation to obtain the receiving time-domain signal matrix. The method for determining the receiving time-domain signal matrix is ​​as follows: In the formula, For the first The receiving end time-domain signal matrix of the root receiving antenna. For the receiving antenna index, For DD channel impulse response, For the first The signal transmitted by the root transmitting antenna, It is additive white Gaussian noise.

5. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 4, characterized in that: The method for obtaining the received signal sample in the TF domain is as follows: For the The time-domain signal matrix of the receiving antenna is subjected to matched filtering to calculate the first... Cross-ambiguity function of root receiving antenna : In the formula, This represents the impulse response of the matched filter at the receiving end; Among them, the TF domain received signal samples are processed by the cross-ambiguity function. After sampling the grid points, the method for determining each element in the received signal sample in the TF domain is as follows: In the formula, For TF domain received signal samples, It represents any element value in the received signal sample of the TF domain.

6. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 5, characterized in that: The SFFT transformation method is as follows: TF domain received signal samples Perform two-dimensional transformation, including the received signal samples in the TF domain. Columns M Point IDFT transformation followed by N Point DFT transformation; After conversion Each element in the DD domain signal sample of the root receiving antenna The calculation formula is: In the formula, M is the number of subcarriers and N is the number of symbols.

7. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 1, characterized in that: The wireless channel scenario types include Doppler-dominated single-selection channel scenario and high-speed multipath time-frequency dual-selection channel scenario, wherein: The Doppler-dominated single-selection channel scenario is a high-speed mobile scenario operating at a preset speed, while the high-speed multipath time-frequency dual-selection channel scenario is a data communication transmission scenario operating at a preset speed under high-speed mobile conditions. In the Doppler-dominated single-selection channel scenario, the initial pilot pattern is uniformly distributed along the Doppler axis, with no pilots configured in the time delay dimension, and the pilots are distributed linearly. In the high-speed multipath time-frequency dual-selection channel scenario, the initial pilot pattern is gridded along the time delay and Doppler two-dimensional directions, with pilots configured in both the time delay and Doppler dimensions, and the pilots are distributed in a grid pattern.

8. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 1, characterized in that: The method for obtaining the channel estimation matrix is ​​as follows: The initial pilot pattern is initialized to obtain the residual and support set, and the maximum contributing atom position of any DD domain channel grid corresponding to the initial pilot pattern is calculated. The residuals and support sets of the initial pilot pattern are updated based on the maximum contributing atomic positions of all DD domain channel grids. Then, using the index positions of the transmit and receive antennas corresponding to the updated pilot pattern and the channel parameter matrix, the channel estimation matrix is ​​reconstructed by traversing each transmit and receive antenna. .

9. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 8, characterized in that: The method for generating the transmission signal using an approximate message passing algorithm is as follows: Based on the time-domain signal matrix transmitted from the transmitter through the wireless channel, the time-domain pilot signal at the receiver is... Reconstructing the channel estimation matrix The channel estimation matrix is ​​derived to obtain the channel matrix in the row and column dimensions and the corresponding set of observation nodes; To obtain the prior probability, the mean of the Gaussian random variable is calculated and assigned to the variable node. The prior probability is iterated in a loop. The prior probability is updated using the mean of the Gaussian random variable and then passed to each observation node to obtain the updated set of confidence symbols. The estimated value of the data sign is determined based on the prior probability of reaching a preset number of iterations, and the estimated value of the data sign is used as a single element to derive the following. ,Will The set of credibility symbols is used as the output signal.

10. The adaptive scene pilot design method based on a MIMO-OTFS system according to claim 9, characterized in that: When the pilot symbol matrix satisfies the convergence condition For the first j The joint pilot symbols generated in the next iteration, when , This represents the initial pilot frequency.

Images