Channel estimation method and device for OTFS system, storage medium and equipment
By training a channel sparsity model using a deep neural network and combining it with the OMP algorithm, the problems of selective bias and low accuracy in channel estimation in the OTFS system are solved, achieving high-precision channel estimation in high-speed scenarios and improving system performance and reliability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF POSTS & TELECOMM
- Filing Date
- 2023-08-07
- Publication Date
- 2026-05-29
AI Technical Summary
Existing OTFS systems suffer from selective bias and low accuracy in channel estimation, especially in high-speed scenarios. The selective bias of traditional OMP algorithms leads to inaccurate sparse representation, affecting the accuracy of channel estimation.
A deep neural network is used to train a channel sparsity estimation model. Combined with the OMP algorithm, the channel sparsity is estimated by preprocessing the pilot signal and noise variance. The method of combining offline training and online data-driven approach is used to improve the accuracy of channel estimation.
It overcomes the selective bias of the traditional OMP algorithm, improves the accuracy of channel estimation and system performance, reduces the overhead of pilot signals, and enhances the reliability of the system.
Smart Images
Figure CN116886471B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a channel estimation method, apparatus, storage medium, and device for OTFS systems, and belongs to the field of wireless communication technology. Background Technology
[0002] With the rapid development of high-speed rail and the popularization of autonomous driving and vehicle-to-everything (V2X) technologies, the optimization of wireless communication in high-speed scenarios is considered an indispensable part of future mobile wireless communication systems. To reduce the performance loss caused by rapidly time-varying channels in high-speed scenarios, Orthogonal Time Frequency Space (OTFS) modulation technology has been proposed in recent years. It can be well applied to high-mobility scenarios, meeting the high data rate requirements of future high-mobility environments.
[0003] In recent years, researchers have conducted extensive research and exploration into the channel estimation problem of OTFS systems. Some scholars have utilized methods such as compressed sensing and matrix factorization to improve the performance and efficiency of OTFS system channel estimation, including the traditional OMP (Orthogonal Matching Pursuit) algorithm. However, the OMP algorithm suffers from a selection bias; it tends to select atoms most relevant to the residual for matching, ignoring other potentially useful atoms. This selection bias, when sparsity is unknown, may lead to the selection of inaccurate subsets of atoms, thus affecting the accuracy and stability of the sparse representation and reducing the precision of channel estimation. Summary of the Invention
[0004] The purpose of this invention is to provide a channel estimation method, apparatus, storage medium and device for OTFS system, and to solve the problems of selectivity bias and low accuracy in the prior art.
[0005] To achieve the above objectives, the present invention employs the following technical solution:
[0006] In a first aspect, the present invention provides a channel estimation method for an OTFS system, comprising:
[0007] Acquire the transmitted pilot signal, the received pilot signal, and the noise variance;
[0008] The transmitted pilot signal and the received pilot signal are preprocessed to obtain transmitted pilot data and received pilot data;
[0009] The transmitted pilot data, the received pilot data, and the noise variance are input into a trained channel sparsity estimation model to obtain a channel sparsity estimate. The channel sparsity estimation model is a deep neural network.
[0010] Based on the channel sparsity estimate, the OMP algorithm is used to reconstruct the channel and obtain the channel estimate.
[0011] In conjunction with the first aspect, the transmitted pilot signal, the received pilot signal, and the noise variance are obtained from the OTFS system.
[0012] In conjunction with the first aspect, further, the preprocessing of the transmitted pilot signal and the received pilot signal to obtain transmitted pilot data and received pilot data includes:
[0013] The real and imaginary parts of the transmitted pilot signal are separated using the real and imaginary functions to obtain the transmitted pilot data, which includes the real and imaginary parts of the transmitted pilot signal.
[0014] The real and imaginary parts of the received pilot signal are separated using the real and imaginary functions to obtain the received pilot data, which includes the real and imaginary parts of the received pilot signal.
[0015] In conjunction with the first aspect, the channel sparsity estimation model is further trained using the following method:
[0016] Acquire training and validation sets, which contain transmitted pilot signals, received pilot signals, and noise variance from multiple OTFS systems.
[0017] The channel sparsity estimation model was initially trained using data from the training set.
[0018] The parameters of the pre-trained channel sparsity estimation model are corrected using data from the validation set to obtain a trained channel sparsity estimation model.
[0019] In conjunction with the first aspect, further, during the training process, the epoch batch size is 100, the initial learning rate is 0.01, the learning rate decay factor is 0.95, the learning rate decay period is 100, ReLU is used as the activation function, stochastic gradient descent is used as the optimizer, and the loss function is... Where L(θ) is the loss function value, s is the number of samples, and L is the sparsity of the true channel. It is a channel sparsity estimate. This represents the square of the L2 norm.
[0020] Secondly, the present invention also provides an OTFS system channel estimation apparatus, comprising:
[0021] The data acquisition module is configured to acquire the transmitted pilot signal, the received pilot signal, and the noise variance.
[0022] The data preprocessing module is configured to preprocess the transmitted pilot signal and the received pilot signal to obtain transmitted pilot data and received pilot data.
[0023] The sparsity estimation module is configured to input the transmitted pilot data, the received pilot data, and the noise variance into a trained channel sparsity estimation model to obtain a channel sparsity estimate, wherein the channel sparsity estimation model is a deep neural network.
[0024] The channel estimation module is configured to: reconstruct the channel using the OMP algorithm based on the channel sparsity estimate to obtain the channel estimate.
[0025] In conjunction with the second aspect, the data preprocessing module is further configured as follows:
[0026] The real and imaginary parts of the transmitted pilot signal are separated using the real and imaginary functions to obtain the transmitted pilot data, which includes the real and imaginary parts of the transmitted pilot signal.
[0027] The real and imaginary parts of the received pilot signal are separated using the real and imaginary functions to obtain the received pilot data, which includes the real and imaginary parts of the received pilot signal.
[0028] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the OTFS system channel estimation method as described in any one of the first aspects.
[0029] Fourthly, the present invention also provides an apparatus comprising:
[0030] Memory, used to store instructions;
[0031] A processor is configured to execute the instructions, causing the device to perform operations implementing the OTFS system channel estimation method as described in any one of the first aspects.
[0032] Compared with the prior art, the beneficial effects achieved by the present invention are:
[0033] The channel estimation method, apparatus, storage medium, and device for the OTFS system provided by this invention estimate the channel sparsity in the OTFS system through a deep neural network, overcoming the selective bias problem in the traditional OMP algorithm. By combining offline training and data-driven methods, the method of this invention can perform accurate channel estimation under real-time changing channel conditions, improving the accuracy of channel estimation, thereby improving system performance and reliability. Moreover, this method can also reduce the overhead of pilot signals. Attached Figure Description
[0034] Figure 1 This is one of the flowcharts of the OTFS system channel estimation method provided in the embodiments of the present invention;
[0035] Figure 2 This is the second flowchart of the OTFS system channel estimation method provided in the embodiments of the present invention;
[0036] Figure 3 This is a comparison chart of experimental results for three OTFS system channel estimation methods provided in the embodiments of the present invention. Detailed Implementation
[0037] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solution of the present invention more clearly, and should not be used to limit the scope of protection of the present invention.
[0038] Example 1
[0039] Taking a single-transmitter, single-receiver OTFS system as an example, appropriate simulation parameters are set under the background of a high-speed train moving at high speed to simulate the signal loss during transmission. When performing channel estimation, a neural network is combined with the OMP algorithm.
[0040] The simulation parameters are set as follows: the high-speed train operating speed is set to 500 km / s, and the carrier frequency f... c 4×10 9 The subcarrier spacing Δf is 15×10 3 The number of subcarriers M = 16, the number of symbols N = 16, the delay spread is 4.7 μs, the transmission bandwidth B = M × Δf, and T × Δf = 1, where T is the symbol period.
[0041] This invention only considers the integer Doppler case, that is, it assumes a time delay τ. i It is an integer multiple of the time-delay domain resolution 1 / MΔf, and the Doppler frequency shift v i It is an integer multiple of the Doppler domain resolution 1 / NT, that is, it satisfies Maximum path delay Maximum Doppler Shift l max It is the maximum path delay τ max The corresponding index on the time delay axis, k max For the maximum Doppler frequency shift v max The corresponding index on the Doppler axis, k i It is the Doppler frequency shift v i The corresponding index on the Doppler axis.
[0042] Based on the above parameters and the time-delay-Doppler domain spread function Where p is the total number of multipaths, h iτ is the complex channel gain of the i-th path, δ is the Dirac function, and τ is the complex channel gain of the i-th path. i and ν i It represents the time delay and Doppler frequency shift of the i-th path, and the time delay-Doppler domain channel can be sampled to obtain the DD domain channel in the OTFS system.
[0043] like Figure 1 As shown, this invention provides a channel estimation method for an OTFS system, comprising the following steps:
[0044] S1. Acquire the transmitted pilot signal, the received pilot signal, and the noise variance.
[0045] The transmitted pilot signal, the received pilot signal, and the noise variance are obtained from the OTFS system to be estimated.
[0046] S2. Preprocess the transmitted pilot signal and the received pilot signal to obtain transmitted pilot data and received pilot data.
[0047] The real and imaginary parts of the transmitted pilot signal are separated using the real and imaginary functions to obtain the transmitted pilot data, which includes the real and imaginary parts of the transmitted pilot signal.
[0048] The real and imaginary parts of the received pilot signal are separated using the real and imaginary functions to obtain the received pilot data, which includes the real and imaginary parts of the received pilot signal.
[0049] S3. Input the transmitted pilot data, the received pilot data and the noise variance into the trained channel sparsity estimation model to obtain the channel sparsity estimate. The channel sparsity estimation model is a deep neural network.
[0050] like Figure 2 As shown, in the OTFS system, the preprocessing of training data involves preprocessing the data. Since the QAM-modulated transmitted pilot signal X, received pilot signal Y, and the noise variance N generated by the channel are all known, and the neural network discards the imaginary part of the data during training, preprocessing is required to separate the real and imaginary parts to obtain [Y]. R Y1, X R [X1, A], where X R X1 and Y represent the real and imaginary parts of the transmitted pilot signal after QAM modulation. R Y1 and Y2 represent the real and imaginary parts of the received signal, respectively, and A is the noise variance in the channel. The sparsity value of the channel serves as the training label for the neural network. The neural network is trained using a loss function to obtain the optimal set of network model parameters.
[0051] The deep neural network is constructed using the PyTorch framework. The linear mapping network consists of 5 layers, with each hidden layer containing 128, 256, and 512 neurons.
[0052] The channel sparsity estimation model is trained using the following method:
[0053] Acquire training and validation sets, which contain transmitted pilot signals, received pilot signals, and noise variance from multiple OTFS systems.
[0054] The channel sparsity estimation model was initially trained using data from the training set.
[0055] The parameters of the pre-trained channel sparsity estimation model are corrected using data from the validation set to obtain a trained channel sparsity estimation model.
[0056] During training, the training set to validation set ratio was 8:2, the epoch batch size was 100, the initial learning rate was 0.01, the learning rate decay factor was 0.95, the learning rate decay period was 100, ReLU (Rectified LinearUnit) was used as the activation function, stochastic gradient descent was used as the optimizer, and the loss function was... Where L(θ) is the loss function value, s is the number of samples, and L is the sparsity of the true channel. It is the channel sparsity estimate obtained through a linear mapping process. This represents the square of the L2 norm.
[0057] After offline training, a well-trained channel sparsity estimation model can be obtained, which can be used to estimate channel sparsity.
[0058] S4. Based on the channel sparsity estimate, the OMP algorithm is used to reconstruct the channel to obtain the channel estimate.
[0059] During the online computation phase, a data-driven approach is used to feed the preprocessed transmitted pilot signal X, received pilot signal Y, and noise variance A into the trained channel sparsity estimation model, thereby obtaining the channel sparsity estimate through the data-driven method.
[0060] Next, the channel is reconstructed using the sparsity estimate through a traditional iterative compressed sensing algorithm. This process yields an estimate of the sparse channel H. This data-driven online training method combines the advantages of offline training and online estimation, improving the reliability of channel estimation.
[0061] For OTFS systems, this invention proposes a novel channel estimation method, NN-OMP. First, a deep neural network is used to train preprocessed training data offline, resulting in a channel sparsity estimation model capable of performing channel sparsity estimation. Using this model, a data-driven method is employed to obtain channel sparsity estimates for the transmitted pilot signal X, the received pilot signal Y, and the noise variance A. Finally, the OMP algorithm is used to reconstruct the signal.
[0062] like Figure 3 As shown in the experimental results, the NN-OMP method proposed in this invention can solve the problems of poor channel reconstruction effect due to too few iterations in the traditional OMP algorithm based on unknown sparsity, and large computational load and long computation time due to too many iterations. At the same time, it reduces the overhead of pilot signals and provides a feasible method for the application of OMP algorithm in channel estimation.
[0063] like Figure 2 As shown, this invention is achieved through the following techniques: First, an OTFS transmission system is built using a high-speed rail communication system as the application background. At the transmitting end, a transmit pilot signal X is generated after QAM (Quadrature Amplitude Modulation) modulation. After X passes through the DD (delay-doppler) domain transmission channel H, a receive pilot signal Y is generated. During the entire signal transmission process, the high-speed movement of the transmitting or receiving end, or the movement of scattering objects (objects that reflect, scatter, or diffract the propagating wave), will cause the Doppler effect. Furthermore, the multipath propagation of the signal will cause the wireless channel to exhibit selectivity in the frequency domain. To better estimate the DD domain transmission channel, it is necessary to estimate the channel parameters in the DD domain, such as P (the number of transmission paths from the transmitting end to the receiving end), h... i (represents the complex channel gain of the i-th path), τ i (delay) and v i (Doppler shift). This invention utilizes data [Y] in the OTFS system. R Y1, X R X1, A] are used as training data and fed into the DNN neural network. Since the input of the neural network is generally real numbers, the above data preprocessing can reduce the error generated during the training process. The sparsity L in the DD domain channel is used as the training label. After offline training, a channel sparsity estimation model is obtained. Then, the preprocessed transmitted pilot signal X, received pilot signal Y, and noise variance A are fed into the model using deep learning. The channel sparsity estimate can be obtained by using a data-driven method. Finally, the DD domain channel is reconstructed using the OMP algorithm to obtain the channel estimate.
[0064] Example 2
[0065] This invention also provides an OTFS system channel estimation apparatus, comprising:
[0066] The data acquisition module is configured to acquire the transmitted pilot signal, the received pilot signal, and the noise variance.
[0067] The data preprocessing module is configured to preprocess the transmitted pilot signal and the received pilot signal to obtain transmitted pilot data and received pilot data.
[0068] The sparsity estimation module is configured to input the transmitted pilot data, the received pilot data, and the noise variance into a trained channel sparsity estimation model to obtain a channel sparsity estimate, wherein the channel sparsity estimation model is a deep neural network.
[0069] The channel estimation module is configured to: reconstruct the channel using the OMP algorithm based on the channel sparsity estimate to obtain the channel estimate.
[0070] Example 3
[0071] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the OTFS system channel estimation method as provided in Embodiment 1:
[0072] Acquire the transmitted pilot signal, the received pilot signal, and the noise variance;
[0073] The transmitted pilot signal and the received pilot signal are preprocessed to obtain transmitted pilot data and received pilot data;
[0074] The transmitted pilot data, the received pilot data, and the noise variance are input into a trained channel sparsity estimation model to obtain a channel sparsity estimate. The channel sparsity estimation model is a deep neural network.
[0075] Based on the channel sparsity estimate, the OMP algorithm is used to reconstruct the channel and obtain the channel estimate.
[0076] Example 4
[0077] This invention also provides an apparatus, including:
[0078] Memory, used to store instructions;
[0079] A processor is configured to execute the instructions, causing the device to perform the OTFS system channel estimation method as provided in Example 1:
[0080] Acquire the transmitted pilot signal, the received pilot signal, and the noise variance;
[0081] The transmitted pilot signal and the received pilot signal are preprocessed to obtain transmitted pilot data and received pilot data;
[0082] The transmitted pilot data, the received pilot data, and the noise variance are input into a trained channel sparsity estimation model to obtain a channel sparsity estimate. The channel sparsity estimation model is a deep neural network.
[0083] Based on the channel sparsity estimate, the OMP algorithm is used to reconstruct the channel and obtain the channel estimate.
[0084] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0085] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0086] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0087] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0088] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A channel estimation method for an OTFS system, characterized in that, include: Acquire the transmitted pilot signal, the received pilot signal, and the noise variance; The transmitted pilot signal and the received pilot signal are preprocessed to obtain transmitted pilot data and received pilot data; The transmitted pilot data, the received pilot data, and the noise variance are input into a trained channel sparsity estimation model to obtain a channel sparsity estimate. The channel sparsity estimation model is a deep neural network. Based on the channel sparsity estimate, the OMP algorithm is used to reconstruct the channel to obtain the channel estimate. The transmitted pilot signal, received pilot signal, and noise variance are obtained from the OTFS system; The preprocessing of the transmitted pilot signal and the received pilot signal to obtain transmitted pilot data and received pilot data includes: The real and imaginary parts of the transmitted pilot signal are separated using the real and imaginary functions to obtain the transmitted pilot data, which includes the real and imaginary parts of the transmitted pilot signal. The real and imaginary parts of the received pilot signal are separated using the real and imaginary functions to obtain the received pilot data, which includes the real and imaginary parts of the received pilot signal.
2. The channel estimation method for the OTFS system according to claim 1, characterized in that, The channel sparsity estimation model is trained using the following method: Acquire training and validation sets, which contain transmitted pilot signals, received pilot signals, and noise variance from multiple OTFS systems. The channel sparsity estimation model was initially trained using data from the training set. The parameters of the pre-trained channel sparsity estimation model are corrected using data from the validation set to obtain a trained channel sparsity estimation model.
3. The channel estimation method for the OTFS system according to claim 2, characterized in that, During the training process, the epoch batch size is 100, the initial learning rate is 0.01, the learning rate decay factor is 0.95, the learning rate decay period is 100, ReLU is used as the activation function, stochastic gradient descent is used as the optimizer, and the loss function is... Where L(θ) is the loss function value, s is the number of samples, and L is the sparsity of the true channel. It is a channel sparsity estimate. This represents the square of the L2 norm.
4. An OTFS system channel estimation device, characterized in that, include: The data acquisition module is configured to acquire the transmitted pilot signal, the received pilot signal, and the noise variance. The data preprocessing module is configured to preprocess the transmitted pilot signal and the received pilot signal to obtain transmitted pilot data and received pilot data. The sparsity estimation module is configured to input the transmitted pilot data, the received pilot data, and the noise variance into a trained channel sparsity estimation model to obtain a channel sparsity estimate, wherein the channel sparsity estimation model is a deep neural network. The channel estimation module is configured to: reconstruct the channel using the OMP algorithm based on the channel sparsity estimate to obtain the channel estimate; The data preprocessing module is specifically configured as follows: The real and imaginary parts of the transmitted pilot signal are separated using the real and imaginary functions to obtain the transmitted pilot data, which includes the real and imaginary parts of the transmitted pilot signal. The real and imaginary parts of the received pilot signal are separated using the real and imaginary functions to obtain the received pilot data, which includes the real and imaginary parts of the received pilot signal. The transmitted pilot signal, received pilot signal, and noise variance are obtained from the OTFS system.
5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the OTFS system channel estimation method as described in any one of claims 1-3.
6. An OTFS system channel estimation device, characterized in that, include: Memory, used to store instructions; A processor is configured to execute the instructions, causing the device to perform operations implementing the OTFS system channel estimation method as described in any one of claims 1-3.