Communication method for suppressing too high peak-to-average ratio in orthogonal time-frequency-space system

Abstract

The invention relates to the technical field of communication, in particular to a communication method for inhibiting an overhigh peak-to-average ratio in an orthogonal time-frequency-space system, which comprises the following steps of: 1, converting a signal from a delay-Doppler domain convenient for channel characterization to a time-frequency domain convenient for modulation implementation; step 2, performing preliminary suppression on the peak-to-average ratio by adopting a T-SLM method, generating a plurality of groups of candidate signals, and selecting one group with the peak-to-average ratio lower than a preset threshold or with the minimum peak-to-average ratio; step 3, carrying out Hisenberg transformation on the selected signal to obtain a time domain signal; carrying out iterative amplitude limiting filtering processing, adding a cyclic prefix, and then sending; step 4, removing the cyclic prefix by a receiving end, and recovering to a time delay-Doppler domain through Wigner transform and Sextile Fourier transform; performing preliminary detection by adopting a message passing algorithm to obtain a hard decision value and a log-likelihood ratio; and selecting a high-reliability observation value based on a log-likelihood ratio, and constructing a reliability selection matrix.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a communication method for suppressing excessively high peak-to-average power ratio in orthogonal time-frequency space systems. Background Technology

[0002] Orthogonal Time-Frequency-Space (OTFS) is a novel multi-carrier modulation technique designed specifically for high-speed mobile communication scenarios. Compared to traditional Orthogonal Frequency Division Multiplexing (OFDM) schemes, OTFS directly maps information symbols into the delay-Doppler domain for transmission. This design can significantly suppress inter-carrier interference caused by the Doppler effect, thus exhibiting superior bit error rate performance and communication reliability in high-dynamic mobile environments such as high-speed rail, vehicle-to-everything (V2X) communication, and UAV communication.

[0003] However, OTFS systems face a significant technical challenge in practical applications: the high peak-to-average power ratio (PAR) of the signal. Theoretical analysis and simulation results show that the upper bound of the PAR of the OTFS signal exhibits an approximately linear relationship with the number of Doppler grid points N. When N is larger to support higher Doppler resolution or more complex mobile environments, the peak power of the signal increases significantly, easily leading to high instantaneous power peaks. When the PAR signal passes through the power amplifier (PA) in the RF front-end, it easily exceeds its linear operating region, causing nonlinear distortion, which in turn leads to out-of-band spectral spread and in-band signal distortion, resulting in a decrease in the system's bit error rate performance. Furthermore, a high PAR also places more stringent requirements on the dynamic range of devices such as digital-to-analog converters (DACs) and analog-to-digital converters (ADCs). This not only increases the complexity of hardware design but also raises the system's implementation cost and energy consumption, becoming a major constraint on the practical deployment of OTFS technology.

[0004] Currently, methods for reducing the peak-to-average power ratio (PAPR) of over-the-counter (OTFS) signals mainly include distortion-based techniques such as μ-law companding and probabilistic techniques such as selective mapping (SLM) and partial transmission sequence (PTS). While distortion-based techniques are simple to implement and have significant suppression effects, nonlinear processing can cause signal distortion, affecting the system's bit error rate (BER). Probabilistic techniques reduce the probability of high peak values ​​by performing linear transformations on the signal without affecting BER performance, but traditional SLM and PTS methods suffer from high computational complexity. Furthermore, amplitude-limiting filtering methods introduce amplitude-limiting noise while suppressing PPR, requiring corresponding compensation processing at the receiver. Therefore, this paper proposes a communication method for suppressing excessively high PPR in orthogonal time-frequency space systems. Summary of the Invention

[0005] Technical problems to be solved: While distortion-based techniques are simple to implement and have a significant suppression effect, nonlinear processing can cause signal distortion, affecting the system's bit error rate. Probabilistic techniques reduce the probability of peak values ​​by performing linear transformations on the signal without affecting bit error rate performance, but traditional SLM and PTS methods suffer from high computational complexity.

[0006] To address the shortcomings of existing technologies, this invention provides an overload protection motor for heavy machinery, thereby solving the technical problems mentioned in the background section.

[0007] To achieve the above objectives, the present invention provides a communication method for suppressing excessively high peak-to-average power ratios in an orthogonal time-frequency space system, comprising the following steps: Step 1: Convert the signal from the delay-Doppler domain, which is convenient for channel characterization, to the time-frequency domain, which is convenient for modulation implementation; Step 2: Use the T-SLM method to initially suppress the peak-to-average ratio (PAR), generate multiple candidate signals, and select the group with a PAR lower than the preset threshold or the minimum. Step 3: Perform Heisenberg transform on the selected signal to obtain the time-domain signal; perform iterative amplitude limiting filtering, add a cyclic prefix, and then send the signal; Step 4: The receiver removes the cyclic prefix and restores the time-delay-Doppler domain through Wigner transform and Sin-Fourier transform; a message passing algorithm is used for preliminary detection to obtain the hard decision value and log-likelihood ratio; based on the log-likelihood ratio, high-reliability observations are selected to construct a reliability selection matrix; Step 5: Use compressed sensing to recover and remove the amplitude-limiting noise from the high-reliability observations; use the noise-removed signal again to perform final detection using the message passing algorithm to recover the original bit sequence.

[0008] In one possible implementation, the T-SLM method in step two includes: generating a random phase sequence of length M in Q groups, multiplying it with a time-frequency domain signal, performing a Heisenberg transform, calculating the peak-to-average power ratio of each group, and comparing it with a preset threshold. If the current group is lower than the threshold, it is selected immediately; otherwise, the smallest group is selected.

[0009] In one possible implementation, Q is set to 6 and the threshold is set to 9dB.

[0010] In one possible implementation, the iterative amplitude limiting filtering in step three includes: setting an amplitude limiting rate γ, calculating a threshold A, limiting the sampling points whose amplitude exceeds A, switching back to the time delay-Doppler domain for frequency domain filtering, switching back to the time domain to complete one iteration, and repeating the iteration 4 times.

[0011] In one possible implementation, the message passing algorithm described in step four initially detects that the maximum number of iterations is set to 20, approximates the interference and noise as Gaussian variables, updates the mean and variance, obtains the hard decision value after convergence, and calculates the log-likelihood ratio.

[0012] In one possible implementation, step four, selecting highly reliable observations, includes: defining the reliability of an observation node as the minimum log-likelihood ratio among its associated symbols, and selecting the top V most reliable nodes after sorting by reliability to construct a reliability selection matrix.

[0013] In one possible implementation, step five, which involves recovering the clipping noise using compressed sensing, includes: subtracting the initial detection signal from the received signal to obtain the residual, filtering the compressed observations using a reliability selection matrix, and recovering the sparse clipping noise using an orthogonal matching pursuit algorithm.

[0014] In one possible implementation, step five involves using a message-passing algorithm again for final detection, outputting a hard decision symbol, and demapping to recover the binary bit sequence.

[0015] Beneficial effects compared to existing technologies: 1. This scheme employs a two-stage processing architecture combining the T-SLM method and iterative amplitude-limiting filtering at the transmitting end. The T-SLM method avoids traversing all candidate sequences by setting a threshold. When the number of random phase sequence vectors Q is 6, the phase set is {1,-1}, and the threshold is set to 9dB, it can reduce the computational load by more than 70% compared to traditional SLM. It achieves a peak-to-average power ratio (PAPR) suppression gain of approximately 2.1dB when CCDF equals 10^-3. Furthermore, the T-SLM process is a linear transformation and does not affect the system's bit error rate. Subsequent iterative amplitude-limiting filtering further suppresses the PAPR to the target value through multiple amplitude-limiting and filtering operations, while effectively controlling out-of-band radiation.

[0016] 2. This scheme utilizes the sparsity of amplitude-limiting noise at the receiver, selects highly reliable observations based on the log-likelihood ratio provided by the message passing algorithm, and employs compressed sensing technology to recover and remove amplitude-limiting noise. This process effectively compensates for the signal distortion introduced by amplitude-limiting filtering at the transmitter, making the corrected signal approach the ideal received signal when the amplitude is infinite. Through secondary message passing detection after noise removal, the system's bit error rate performance is significantly improved, maintaining the system's communication quality as much as possible while ensuring peak-to-average power ratio suppression. Attached Figure Description

[0017] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0018] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0019] Preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention can also be implemented in various different forms, and therefore the present invention is not limited to the embodiments described below. The technical solution in this application embodiment is to solve the problems mentioned in the background art, and the overall idea is as follows: Orthogonal Time-Frequency Conditioning (OTFS), as a novel multi-carrier modulation technique, carries information symbols in the time-delay-Doppler domain, effectively resisting inter-carrier interference caused by Doppler frequency shift in high-speed mobile scenarios.

[0020] However, the peak-to-average power ratio (PAPR) of OTFS signals increases linearly with the number of Doppler grids N. When the signal peak value exceeds the linear operating range of the power amplifier, nonlinear distortion occurs, leading to a decrease in the system's bit error rate performance. To address this issue, a communication method for suppressing excessively high PAPR in orthogonal time-frequency space systems is proposed, comprising the following steps: Step 1: In the signal generation and transformation process at the transmitting end, the source first generates a binary bit sequence to be transmitted. This sequence is mapped into two-dimensional complex information symbols using Quadrature Amplitude Modulation (QAM). These symbols further form a time-delay-Doppler domain signal matrix of dimension N×M. Here, parameter N represents the number of grids in the Doppler dimension, and M represents the number of grids in the time-delay dimension. Each symbol on this two-dimensional grid carries valid information data to be transmitted. Subsequently, an Inverse Symmetric Fourier Transform (ISFFT) is performed on the signal in the time-delay-Doppler domain, and a window function is applied at the transmitting end to convert the signal to a time-frequency domain representation. Specifically, the ISFFT transformation can be decomposed into two steps: performing an N-point Inverse Fast Fourier Transform (IFFT) along the Doppler axis, and performing an M-point Fast Fourier Transform (FFT) along the time-delay axis. This is a two-dimensional orthogonal transformation operation. The main function of this critical processing stage is to convert the signal from the time-delay-Doppler domain, which is more conducive to wireless channel characterization and analysis, to the time-frequency domain, which is more suitable for actual modulation and transmission, thus preparing for subsequent signal transmission.

[0021] Step 2: Initial suppression of peak-to-average ratio using the T-SLM method; Next, the T-SLM method was used to initially suppress the peak-to-average power ratio.

[0022] First, Q groups of random phase sequence vectors of length M are generated. The phase factors in each group are selected from a discrete phase set such as {1, -1}. The Q groups of phase sequences are multiplied by the time-frequency domain signal to obtain Q groups of statistically independent candidate sequences. A Heisenberg transform is performed on each candidate sequence to obtain the time-domain signal; when the transmitted pulse is rectangular, this transform is equivalent to an M-point IFFT. The peak-to-average power ratio (PAPR) of each time-domain signal is calculated sequentially and compared with a preset threshold. The threshold effectively avoids traversing all candidate sequences, thus reducing system complexity. Simulation results show that when the number of random phase sequence vectors Q is 6, the phase set is {1, -1}, and the threshold is set to 9 dB, this method can reduce the computational load by more than 70% compared to the traditional SLM; therefore, the preset threshold is 9 dB. If the peak-to-average power ratio (PAPR) of the current group is lower than the threshold, the signal of that group is immediately selected for transmission and the calculation is terminated; if all Q groups are higher than the threshold, the group with the smallest PAPR is selected for transmission.

[0023] The expression for peak-to-average ratio is: ,in, This represents the instantaneous amplitude value of the signal transmitted at time t.

[0024] Step 3: Perform a Heisenberg transform on the time-frequency domain signal obtained after T-SLM processing to convert the discrete time-frequency domain symbols into a continuous time-domain signal. The Heisenberg transform can be understood as a process of modulating and superimposing time-frequency domain symbols. When a rectangular transmit pulse is used, this transform is equivalent to performing an M-point IFFT on each OFDM symbol, ultimately obtaining a time-domain sampling sequence of length N×M.

[0025] To further reduce the peak-to-average power ratio (PAPR) to the target value, multiple clipping and filtering operations are performed on the time-domain signal. A clipping rate γ is set, and a clipping threshold A is calculated, which is equal to γ ​​multiplied by the square root of the signal's average power.

[0026] In each iteration, sampling points with amplitudes exceeding A are clipped, maintaining the phase while reducing the amplitude to A. The clipped signal is then converted back to the time-delay-Doppler domain via Wigner transform and SFFT, and frequency domain filtering is performed to remove out-of-band radiation. The filtered signal is then converted back to the time domain via ISFFT and Heisenberg transform to complete one iteration. Typically, four iterations are sufficient to achieve the target peak-to-average power ratio, effectively suppressing peak values ​​and controlling out-of-band radiation. Finally, a cyclic prefix with a length greater than or equal to the maximum channel delay spread is added before the time-domain signal to avoid inter-symbol interference, and then it is transmitted to the wireless channel via antenna.

[0027] Step 4: Preliminary testing and reliability selection at the receiving end; After receiving the signal, the receiver first removes the cyclic prefix to obtain the received signal in the time domain. The received signal is then subjected to a Wigner transform to map it back to the time-frequency domain, and then subjected to a SFFT transform with a receiving window function added to restore it to the time-delay-Doppler domain to obtain the received symbol vector.

[0028] A message-passing algorithm is employed to perform preliminary detection of received symbols, leveraging the sparsity of the equivalent channel matrix. This algorithm calculates the posterior probability of each transmitted symbol through iterative message passing between observation and variable nodes in the factor graph. In each iteration, interference and additive noise from other symbols are approximated as Gaussian random variables, with their mean and variance calculated from the current probability estimate. A maximum of 20 iterations is set. After convergence, the hard decision value for each symbol is obtained based on the posterior probability, and the log-likelihood ratio of each decision is calculated for subsequent reliability selection.

[0029] Amplitude-limited noise exhibits sparse characteristics in the time domain and can be recovered using compressed sensing techniques. To obtain high-quality noise observations, highly reliable observations must be selected. Each observation node is associated with a set of symbol indices, the size of which is equal to the number of channel paths. The reliability of an observation node is defined as the minimum log-likelihood ratio among all its associated symbols. All observation nodes are sorted in descending order of reliability, and the top V most reliable observation nodes are selected to form a reliability set. A reliability selection matrix is ​​constructed based on this set, with each row corresponding to a selected observation node.

[0030] Step 5: Compressed sensing to recover amplitude-limiting noise and secondary detection; The sparsity of amplitude-limiting noise in the time domain stems from the fact that only a few signal sampling points exceed the amplitude-limiting threshold. The residual obtained by subtracting the initial detection signal from the received signal contains amplitude-limiting noise, decision error, and additive noise.

[0031] Compressed observations are obtained by filtering the residuals using a reliability selection matrix. The observation matrix is ​​then defined as the product of the selection matrix, the time-frequency transformation matrix, and the channel matrix. An orthogonal matched pursuit algorithm is employed to recover the sparse amplitude-limiting noise signal from the compressed observations.

[0032] After successfully recovering the estimated value of the clipping noise, this estimated noise component is subtracted from the original received signal to obtain the corrected communication signal. This compensation operation mainly targets the nonlinear distortion caused by the clipping filtering process at the transmitting end, and can significantly suppress the signal distortion caused by clipping, making the corrected received signal closer to the ideal received signal without clipping processing in terms of waveform and statistical characteristics. Subsequently, this corrected signal is used as input, and the message passing algorithm is used again to perform the final symbol detection step. Since the clipping noise has been effectively estimated and eliminated, the equivalent additive noise power is significantly reduced, the signal quality is significantly improved, and the message passing algorithm can perform symbol-level decision more accurately under better signal-to-noise ratio conditions. Through multiple iterations until the algorithm converges, the final hard decision symbol is output based on the iterative posterior probability calculation result. Then, the symbol sequence is converted back to the original binary bit sequence through a demapping operation, thus completing the signal detection and recovery process of the entire communication link.

[0033] In summary, by combining transceiver-side T-SLM preprocessing with iterative amplitude-limiting filtering and receiver-side compressed sensing noise recovery, the peak-to-average power ratio (PAPR) and bit error rate (BER) of the OTFS system are optimized. The transceiver-side T-SLM method reduces the probability of peak values ​​at the source without affecting the BER, while iterative amplitude-limiting filtering further finely controls the peak values ​​to the target value. At the receiver, the sparsity of amplitude-limiting noise and the reliability selection provided by the message passing algorithm effectively compensate for signal distortion caused by amplitude-limiting filtering, thus effectively reducing the BER.

[0034] This scheme employs a two-stage processing architecture at the transmitter, combining the T-SLM method with iterative amplitude-limiting filtering. The T-SLM method avoids traversing all candidate sequences by setting a threshold. When the number of random phase sequence vectors Q is 6, the phase set is {1,-1}, and the threshold is set to 9dB, it can reduce the computational load by more than 70% compared to traditional SLM. It achieves a peak-to-average power ratio (PAPR) suppression gain of approximately 2.1dB when CCDF equals 10^-3. Furthermore, the T-SLM process is a linear transformation and does not affect the system's bit error rate. Subsequent iterative amplitude-limiting filtering further suppresses the PAPR to the target value through multiple amplitude-limiting and filtering operations, while effectively controlling out-of-band radiation.

[0035] This scheme utilizes the sparsity of amplitude-limiting noise at the receiver, selects highly reliable observations based on the log-likelihood ratio provided by the message passing algorithm, and employs compressed sensing technology to recover and remove the amplitude-limiting noise. This process effectively compensates for the signal distortion introduced by the amplitude-limiting filter at the transmitter, making the corrected signal approach the ideal received signal when the amplitude is infinite. Through secondary message passing detection after noise removal, the system's bit error rate performance is significantly improved, maintaining the communication quality of the system as much as possible while ensuring the peak-to-average power ratio suppression effect.

[0036] Finally, it should be noted that the above embodiments are merely examples for clearly illustrating the present invention and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A communication method for suppressing excessively high peak-to-average power ratio in an orthogonal time-frequency-space system, characterized in that, Includes the following steps: Step 1: Convert the signal from the delay-Doppler domain, which is convenient for channel characterization, to the time-frequency domain, which is convenient for modulation implementation; Step 2: Use the T-SLM method to initially suppress the peak-to-average ratio (PAR), generate multiple candidate signals, and select the group with a PAR lower than the preset threshold or the minimum. Step 3: Perform Heisenberg transform on the selected signal to obtain the time-domain signal; perform iterative amplitude limiting filtering, add a cyclic prefix, and then send the signal; Step 4: The receiver removes the cyclic prefix and restores the time-delay-Doppler domain through Wigner transform and Sin-Fourier transform; A message passing algorithm is used for preliminary detection to obtain hard decision values ​​and log-likelihood ratios; based on the log-likelihood ratios, high-reliability observations are selected to construct a reliability selection matrix. Step 5: Use compressed sensing to recover and remove clipping noise from high-reliability observations; The signal after noise removal is then used again with a message passing algorithm for final detection to recover the original bit sequence.

2. The method according to claim 1, characterized in that, The T-SLM method described in step two includes: generating a random phase sequence of length M in Q groups, multiplying it with the time-frequency domain signal, performing a Heisenberg transform, calculating the peak-to-average power ratio of each group and comparing it with a preset threshold; if the current group is lower than the threshold, it is selected immediately; otherwise, the smallest group is selected.

3. The method according to claim 2, characterized in that, The value of Q is 6, and the threshold value is set to 9dB.

4. The method according to claim 1, characterized in that, Step 3, the iterative amplitude limiting filtering, includes: setting the amplitude limiting rate γ, calculating the threshold A, limiting the sampling points whose amplitude exceeds A, switching back to the time delay-Doppler domain for frequency domain filtering, switching back to the time domain to complete one iteration, and repeating the iteration 4 times.

5. The method according to claim 1, characterized in that, Step four describes a message passing algorithm with a preliminary detection setting of a maximum of 20 iterations. It approximates interference and noise as Gaussian variables, updates the mean and variance, and obtains a hard decision value after convergence, then calculates the log-likelihood ratio.

6. The method according to claim 1, characterized in that, Step four, which involves selecting highly reliable observations, includes: defining the reliability of an observation node as the minimum log-likelihood ratio among its associated symbols, and selecting the top V most reliable nodes after sorting by reliability to construct a reliability selection matrix.

7. The method according to claim 1, characterized in that, Step 5, which describes the recovery of clipping noise using compressed sensing, includes: subtracting the initial detection signal from the received signal to obtain the residual, filtering the compressed observations using a reliability selection matrix, and using an orthogonal matching pursuit algorithm to recover the sparse clipping noise.

8. The method according to claim 1, characterized in that, Step five describes the use of a message passing algorithm for final detection, outputting a hard decision symbol and demapping to recover the binary bit sequence.

Images