Ultrasonic transparent metal OFDM (Orthogonal Frequency Division Multiplexing) receiving method and device for resisting multipath fading

Abstract

The invention discloses an ultrasonic transparent metal OFDM receiving method and device for resisting multipath fading, and relates to the technical field of ultrasonic transparent metal communication and digital signal processing. The method comprises the following steps: acquiring a frequency domain signal received in a multipath channel; an ultrasonic transparent metal OFDM system is constructed; the ultrasonic metal-penetrating OFDM system comprises an MCGBNet receiver, a metal-penetrating OFDM (Orthogonal Frequency The MCGBNet receiver comprises a channel estimation sub-network and a signal detection sub-network; a received frequency domain signal and a known sending pilot symbol are input into a channel estimation sub-network to estimate channel frequency responses of all subcarriers; and inputting the channel frequency responses of all the subcarriers and the received data symbols into a signal detection sub-network, and converting soft bits into final binary bits through a decision rule. According to the invention, the original information bit can be effectively extracted and restored from the fading signal.

Description

An ultrasonic transmissive metal OFDM receiver method and device for resisting multipath fading Technical Field

[0001] This invention relates to the fields of ultrasonic through-metal communication and digital signal processing technology, and in particular to an ultrasonic through-metal OFDM receiving method and device for resisting multipath fading. Background Technology

[0002] In industrial monitoring and data exchange applications within sealed containers, Ultrasonic Through-Metal (UTM) communication technology has attracted attention due to its ability to penetrate metal shielding. However, the inhomogeneities within and at the interfaces of metallic media lead to complex multipath propagation effects of ultrasonic waves. This multipath effect manifests in the frequency domain as severe frequency-selective fading, where the channel frequency response exhibits dramatic amplitude fluctuations and phase changes across different subcarriers. This results in deep fading of some subcarriers in OFDM systems, leading to a sharp deterioration in the signal-to-noise ratio.

[0003] Traditional OFDM receivers employ linear least squares or least mean square error methods for channel estimation, combined with zero-forcible or least mean square error equalization for signal detection. However, in the highly frequency-selective channels of UTM, traditional methods struggle to accurately estimate the channel state in deep fading regions, and the linear equalizer amplifies noise, leading to a significant deterioration in the system's bit error rate performance.

[0004] In recent years, deep learning technology has been introduced to improve the performance of communication receivers. However, most existing deep learning-based receivers have not been specifically optimized for the multipath fading characteristics of UTM channels. Purely data-driven black-box models suffer from poor interpretability and insufficient utilization of multipath channel structural features, resulting in limited generalization ability and bit recovery performance in complex multipath environments.

[0005] Therefore, there is an urgent need for an ultrasonic transmissive metal OFDM receiver that can effectively cope with frequency-selective fading caused by multipath propagation, and has strong interference suppression capabilities and high-precision bit recovery performance. Summary of the Invention

[0006] To address the problem of frequency-selective fading caused by multipath propagation in existing technologies, which leads to reduced accuracy in recovering the original bitstream from deeply fading signals, this invention provides an ultrasonic transmissive-metal OFDM receiving method and apparatus for resisting multipath fading. The technical solution is as follows:

[0007] On one hand, an ultrasonic transmissive metal OFDM receiving method for resisting multipath fading is provided. This method is implemented by an ultrasonic transmissive metal OFDM receiving device for resisting multipath fading. The method includes: S1, acquiring the frequency domain signal received in a multipath channel; S2, constructing an ultrasonic transmissive metal OFDM system; the ultrasonic transmissive metal OFDM system includes a multi-scale convolutional gated bidirectional long short-term memory network receiver; the multi-scale convolutional gated bidirectional long short-term memory network receiver includes a channel estimation subnetwork and a signal detection subnetwork; S3, inputting the received frequency domain signal and known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers; S4, inputting the channel frequency response of all subcarriers and the received data symbols into the signal detection subnetwork to convert soft bits into final binary bits using a Softmax function.

[0008] On the other hand, an ultrasonic transmissive metal OFDM receiver for multipath fading is provided. This receiver is applied to an ultrasonic transmissive metal OFDM receiving method for multipath fading. The receiver includes: an acquisition unit for acquiring frequency domain signals received in a multipath channel; a construction unit for constructing an ultrasonic transmissive metal OFDM system; the ultrasonic transmissive metal OFDM system includes a multi-scale convolutionally gated bidirectional long short-term memory network receiver; the multi-scale convolutionally gated bidirectional long short-term memory network receiver includes a channel estimation subnetwork and a signal detection subnetwork; an estimation unit for inputting the received frequency domain signal and known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers; and a recovery unit for inputting the channel frequency response of all subcarriers and received data symbols into the signal detection subnetwork to convert soft bits into final binary bits using a Softmax function.

[0009] On the other hand, an ultrasonic transmissive metal OFDM receiving device for resisting multipath fading is provided, the ultrasonic transmissive metal OFDM receiving device for resisting multipath fading includes: a processor; a memory, the memory storing computer-readable instructions, which, when executed by the processor, implement any of the methods described above for ultrasonic transmissive metal OFDM receiving for resisting multipath fading.

[0010] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored therein, the at least one instruction being loaded and executed by a processor to implement any of the above-described ultrasonic transmissive metal OFDM receiving methods for resisting multipath fading.

[0011] The beneficial effects of the technical solution provided by the embodiments of the present invention include at least the following: the embodiments of the present invention accurately estimate and compensate for frequency selective fading through the collaborative design of multi-scale convolutional neural networks and bidirectional long short-term memory networks, effectively suppress multipath interference, and achieve low bit error rate bit recovery under severe UTM channels.

[0012] The embodiments of the present invention adopt an end-to-end training objective to directly minimize bit error, so that the joint optimization of the entire receiver is oriented towards the final task. Therefore, compared with the traditional separation optimization method, it can extract and restore the original information bits from the fading signal more effectively.

[0013] The network structure of this invention is designed based on a communication physical model, with clearly defined functions for each module. It combines the nonlinear capabilities of deep learning with the interpretability of model-driven approaches. It also exhibits greater adaptability to channel variations and noise. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 is a flowchart of an ultrasonic transmissive metal OFDM receiving method for resisting multipath fading provided by an embodiment of the present invention; Figure 2 is a structural schematic diagram of a conventional OFDM system receiving method provided by an embodiment of the present invention; Figure 3 is a structural schematic diagram of an ultrasonic transmissive metal OFDM system receiving method provided by an embodiment of the present invention; Figure 4 is a structural schematic diagram of a channel estimation method provided by an embodiment of the present invention; Figure 5 is a structural schematic diagram of a signal detection method provided by an embodiment of the present invention; Figure 6 is a simulation result diagram provided by an embodiment of the present invention; Figure 7 is a block diagram of an ultrasonic transmissive metal OFDM receiving device for resisting multipath fading provided by an embodiment of the present invention; Figure 8 is a structural schematic diagram of an ultrasonic transmissive metal OFDM receiving device for resisting multipath fading provided by an embodiment of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0017] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0018] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0019] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0020] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0021] This invention provides an ultrasonic transmissive OFDM receiving method for resisting multipath fading. This method can be implemented by an ultrasonic transmissive OFDM receiving device for resisting multipath fading, which can be a terminal or a server. As shown in Figure 1, the flowchart of the ultrasonic transmissive OFDM receiving method for resisting multipath fading includes the following steps:

[0022] S1. Obtain the frequency domain signal received in the multipath channel.

[0023] S2. Construct an ultrasonic-to-metal OFDM system; the ultrasonic-to-metal OFDM system includes: a multi-scale convolutional-gated bidirectional network (MCGBNet) receiver; the MCGBNet receiver includes: a channel estimation subnetwork and a signal detection subnetwork.

[0024] The ultrasonic transmissive metal OFDM system also includes a transmitter and an ultrasonic transmissive metal channel; the transmitter is used to generate OFDM signals; and the MCGBNet receiver is responsible for recovering information from distorted signals.

[0025] Figure 2 shows a schematic diagram of a conventional OFDM system receiving method provided in an embodiment of the present invention; Table 1 shows the parameter configuration of the conventional OFDM system.

[0026] Table 1

[0027] The ultrasonic metal-penetrating channel employs a line-of-sight (LOS) time-domain model, generating a large amount of labeled training and testing data within an SNR range of 0–40 dB. The received signal is processed by removing the cyclic prefix and performing an FFT transform to obtain the frequency-domain received signal. The LOS time-domain model is a model well-known to those skilled in the art and will not be described in detail in this embodiment.

[0028] Figure 3 shows a schematic diagram of a receiving method for an ultrasonic translucent metal OFDM system provided in an embodiment of the present invention.

[0029] Optionally, the channel estimation subnetwork includes: a multi-scale feature extraction module, a feature fusion module, and an encoding-decoding module; wherein, the channel estimation subnetwork is used to estimate the channel frequency response of all subcarriers based on the received pilot symbols, and adopts a multi-scale convolutional neural network structure, which can fully extract the feature distribution of the channel at different scales.

[0030] The multi-scale feature extraction module includes: multiple parallel one-dimensional convolutional layers with different kernel sizes; the multi-scale feature extraction module is used to extract multi-scale features from the pilot channel response, capture the correlation of different frequency domain components, and enhance the channel representation capability; the feature fusion module is used to fuse the multi-scale features and output the fused features; the encoding-decoding module consists of three convolutional layers and three deconvolutional layers cascaded together, used to perform nonlinear mapping and channel reconstruction on the fused features; the convolutional layers are used for feature compression and encoding, and the deconvolutional layers are used for feature upsampling and reconstruction.

[0031] Optionally, the signal detection subnetwork includes: a differentiable equalization layer, a gated fusion module, a bidirectional long short-term memory network module, and a decision output layer; wherein, the signal detection subnetwork is connected to the channel estimation subnetwork and is used to perform signal detection based on the estimated channel frequency response and the received data symbols to achieve transmitted bit recovery. This subnetwork combines the traditional zero-approach equalization algorithm with the time-series modeling capability of deep learning.

[0032] The system comprises the following components: a differentiable equalization layer, which uses the channel frequency response of the subcarriers output by the channel estimation subnetwork to perform preliminary equalization on the received data symbols, obtaining an initial equalization signal; a gated fusion module, which concatenates the received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal, and generates adaptive weight coefficients through a fully connected layer and a Softmax function to perform weighted fusion on different input components, outputting a fused sequence signal; a bidirectional long short-term memory (LSTM) network module, which consists of three layers of bidirectional LSM networks, used for bidirectional time-series modeling of the fused sequence signal, extracting time-domain correlations through forward and backward information interaction, suppressing residual interference and noise, and outputting high-dimensional features; and a decision output layer, which consists of a fully connected layer and a sigmoid activation function, used to map the high-dimensional features into soft-bit information and complete binary bit decisions through a preset threshold.

[0033] S3. Input the received frequency domain signal and the known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers.

[0034] Figure 4 shows a schematic diagram of a channel estimation method provided in an embodiment of the present invention.

[0035] Optionally, the specific implementation process of S3 includes S31-S34: S31, inputting the received frequency domain signal and the known transmitted pilot symbol into the channel estimation sub-network to calculate the initial least squares channel estimate; concatenating the real and imaginary parts of the least squares estimate to construct the input tensor; in one feasible implementation, the process of calculating the initial least squares channel estimate is represented by the following formula (1): (1) Among them, This represents the initial least-squares channel estimate at pilot position k; This represents the received frequency domain signal; This indicates a known transmitted pilot symbol.

[0036] The input tensor has dimensions (Batch, 512, 2).

[0037] S32. Input the input tensor into the multi-scale feature extraction module, and extract multi-scale features through three parallel one-dimensional convolutional layers; wherein each convolutional layer is followed by a ReLU activation function. In one feasible implementation, the process of extracting multi-scale features is represented by the following formulas (2)-(4):

[0038] (2) (3) (4) Among them, This represents the scale features extracted by the first convolutional layer; Indicates the weights of the first convolutional layer; This represents the initial least-squares channel estimate at pilot position k; Indicates the bias of the first convolutional layer; This represents the scale features extracted by the second convolutional layer; Indicates the weights of the second convolutional layer; Indicates the bias of the second convolutional layer; This represents the scale features extracted by the third convolutional layer; Indicates the weights of the third convolutional layer; Indicates the bias of the third convolutional layer; This indicates a convolution operation.

[0039] The three convolutional kernels are set to sizes of 4, 8, and 16, respectively, to capture frequency correlations of varying widths.

[0040] S33. Input the multi-scale features into the feature fusion module for splicing and dimensionality reduction to obtain the fused features; in one feasible implementation, the extracted multi-scale feature dimensions are spliced, and dimensionality reduction and feature fusion are performed using one 1×1 convolution after another to reduce the number of channels. The specific implementation process is represented by the following formulas (5)-(6):

[0041] (5) (6) Among them, Indicates the features after splicing; Represents the weights of a 1×1 convolution; This represents the bias of a 1×1 convolution; This indicates the characteristics after fusion.

[0042] S34. Input the fused features into the encoder-decoder module, and use a skip connection structure to directly pass the output features of the convolutional layer to the corresponding deconvolutional layer for splicing, and output the channel frequency response of all subcarriers.

[0043] Among them, the skip connection structure directly passes the output features of the convolutional layer to the corresponding deconvolutional layer for fusion in the encoder-decoder module, which alleviates the gradient vanishing problem and preserves high-frequency detail features, thereby improving the channel reconstruction accuracy and training stability.

[0044] In one feasible implementation, the fused features are input into an encoder-decoder module consisting of three convolutional layers and three deconvolutional layers. The encoder downsamples the features using convolutions with a stride greater than 1 to compress them, while the decoder upsamples them using deconvolutions to restore resolution. To avoid gradient vanishing and preserve details, skip connections are introduced into the network, such as the first skip connection. The output of the convolutional layer is concatenated with the output of the deconvolutional layer j, and used as the input of the next deconvolutional layer, as expressed by the following formula (7):

[0045] (7) Among them, Indicates the first The output of a convolutional layer; This represents the output of the j-th deconvolution layer; This represents the input to the next deconvolution layer; This represents the concatenation function.

[0046] In one feasible implementation, the entire encoding-decoding process can be represented by the following formula (8): (8) Where o represents the composition of functions; For the first Activation function of the layer; This represents the learnable bias parameters of the nth layer; This represents the learnable weight parameters of the nth layer, specifically the kernel weights of the convolutional or deconvolutional layer. This represents the output feature tensor of the encoder-decoder module, which is the channel frequency response corresponding to all subcarriers after reconstruction.

[0047] Figure 5 shows a schematic diagram of a signal detection method provided in an embodiment of the present invention.

[0048] S4. The channel frequency response of all subcarriers and the received data symbols are converted into final binary bits in the input signal detection subnetwork using the Softmax function.

[0049] Optionally, the specific implementation process of S4 includes S41-S44: S41, taking the channel frequency response of all subcarriers and the received data symbols, and inputting them into the differentiable equalization layer of the signal detection subnetwork, the received data symbols are subjected to differentiable zero-approach equalization to obtain the initial equalization signal; wherein, the process of the initial equalization signal is represented by the following formula (9): (9) Among them, This represents the initial signal estimate; Indicates the received data symbols; This represents the channel frequency response of all subcarriers.

[0050] S42. The received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal are concatenated by feature dimensions and input into the gating fusion module to adaptively learn the importance weights of each component; based on the importance weights of each component, the fused sequence signal is obtained; optionally, the process of obtaining the fused sequence signal is represented by the following formulas (10)-(12): (10) (11) (12) Among them, Indicates the received data symbols; This represents the channel frequency response of all subcarriers; This represents the initial equalization signal; This represents the fused feature vector formed by splicing the received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal. Indicates the first weighting coefficient; This represents the second weighting coefficient; This represents the third weighting coefficient; This represents the fused sequence signal; This represents the concatenation function.

[0051] Among them, the adaptive learning mechanism in the gated fusion module can dynamically adjust the robustness of imperfect channel estimation and noise interference.

[0052] S43. Input the fused sequence signal into the bidirectional long short-term memory network module, and process the forward sequence and the backward sequence simultaneously at each time step to obtain the forward hidden state and the backward hidden state; the forward hidden state and the backward hidden state are passed through the output gate to obtain high-dimensional features; the specific process of obtaining high-dimensional features is represented by the following formulas (13)-(15): (13) (14) (15) Among them, Indicates a forward-hidden state; Indicates the first weighting coefficient; Indicates the current LSTM input; This represents the second weighting coefficient; This represents the output of a forward LSTM; Indicates a backward hidden state; This represents the third weighting coefficient; This represents the fourth weighting coefficient; This represents the output of the inverse LSTM; This represents the high-dimensional features output from the BiLSTM module; This represents the fifth weighting coefficient; Indicates the output gate; This represents the sixth weighting coefficient; Indicates the cell state.

[0053] S44. Input the high-dimensional features into the decision output layer, process them through a fully connected layer and a Sigmoid activation function, and output the likelihood probability of each bit; based on the likelihood probability of each bit, convert it into the final binary bit through the decision rule.

[0054] In one feasible implementation, the calculation process of the likelihood probability of each bit is expressed by the following formula (16): (16) Among them, This represents the likelihood probability of each bit; This represents the weight matrix of the decision output layer, i.e., the fully connected layer. This represents a fixed-length feature vector after global average pooling. This represents the bias vector of the decision output layer, i.e., the fully connected layer.

[0055] In one feasible implementation, the final binary bits are represented by the following formula (17): (17) Optionally, both the channel estimation subnetwork and the signal detection subnetwork are trained using an end-to-end joint training method; wherein, during the training process, both the channel estimation subnetwork and the signal detection subnetwork use the mean square error function as the loss function.

[0056] In this embodiment of the invention, the mean square error function is used as the loss function. By minimizing the error between the predicted bits and the actual bits, the global optimal optimization of the receiver parameters is achieved.

[0057] In one feasible implementation, this embodiment of the invention uses the Adam optimizer during training, with an initial learning rate of 0.0001. If the verification loss does not decrease for several consecutive epochs, the learning rate is multiplied by 0.01 to decay. The loss function uses mean squared error, expressed by the following formula (18):

[0058] (18) Among them, This represents the loss value during model training. This represents the actual bit value at the k-th position; This represents the receiver network's prediction (soft bit) for the k-th bit. This represents the total number of bits in a training sample.

[0059] The training batch size is 128, and both the training and test sets are generated based on the UTM channel model. During training, the training and validation loss and bit error rate (BER) curves are monitored, and an early stopping mechanism is used to prevent overfitting.

[0060] Figure 6 shows a simulation result provided by an embodiment of the present invention. The MCGBNet receiver achieves optimal BER performance across the entire SNR range, particularly outperforming Communication Network (ComNet) and Fully Connected Deep Neural Network (FC-DNN) when SNR > 20 dB. Under a 25 dB SNR condition, the BER is reduced by approximately 22%, 34%, 64%, and 82% compared to ComNet, FC-DNN, MMSE-ZF, and LS-ZF, respectively.

[0061] In one feasible implementation, Table 2 shows the comparison results of the complexity of the MCGBNet model, ComNet model, and FC-DNN model in the embodiments of the present invention.

[0062] Table 2

[0063] Among them, the MCGBNet receiver proposed in the embodiments of the present invention has better BER performance, lower computational complexity and faster convergence speed in ultrasonic metal OFDM systems, and is suitable for high reliability and low complexity internal and external communication scenarios of sealed metal containers.

[0064] The embodiments of the present invention accurately estimate and compensate for frequency-selective fading through the collaborative design of multi-scale convolutional neural networks and bidirectional long short-term memory networks, effectively suppress multipath interference, and achieve low bit error rate bit recovery under severe UTM channels.

[0065] The embodiments of the present invention adopt an end-to-end training objective to directly minimize bit error, so that the joint optimization of the entire receiver is oriented towards the final task. Therefore, compared with the traditional separation optimization method, it can extract and restore the original information bits from the fading signal more effectively.

[0066] The network structure of this invention is designed based on a communication physical model, with clearly defined functions for each module. It combines the nonlinear capabilities of deep learning with the interpretability of model-driven approaches. It also exhibits greater adaptability to channel variations and noise.

[0067] Figure 7 is a block diagram of an ultrasonic transmissive metal OFDM receiver for resisting multipath fading according to an embodiment of the present invention. This device is used in an ultrasonic transmissive metal OFDM receiving method for resisting multipath fading. Referring to Figure 7, the device includes an acquisition unit 710, a construction unit 720, an estimation unit 730, and a recovery unit 740. Wherein:

[0068] Acquisition unit 710 is used to acquire the frequency domain signal received in the multipath channel; construction unit 720 is used to construct an ultrasonic transmissive metal OFDM system; the ultrasonic transmissive metal OFDM system includes: a multi-scale convolutional gated bidirectional long short-term memory network receiver; the multi-scale convolutional gated bidirectional long short-term memory network receiver includes: a channel estimation subnetwork and a signal detection subnetwork; estimation unit 730 is used to input the received frequency domain signal and the known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers; recovery unit 740 is used to input the channel frequency response of all subcarriers and the received data symbols into the signal detection subnetwork to convert the soft bits into the final binary bits through the Softmax function.

[0069] Optionally, the channel estimation subnetwork includes: a multi-scale feature extraction module, a feature fusion module, and an encoder-decoder module; wherein, the multi-scale feature extraction module includes: multiple parallel one-dimensional convolutional layers with different kernel sizes; the multi-scale feature extraction module is used to extract multi-scale features from the pilot channel response, capture the correlation of different frequency domain components, and enhance the channel representation capability; wherein, the feature fusion module is used to fuse the multi-scale features and output the fused features; wherein, the encoder-decoder module consists of three convolutional layers and three deconvolutional layers cascaded together, used to perform nonlinear mapping and channel reconstruction on the fused features; wherein, the convolutional layers are used for feature compression and encoding, and the deconvolutional layers are used for feature upsampling and reconstruction.

[0070] Optionally, the estimation unit 730 is configured to: input the received frequency domain signal and the known transmitted pilot symbols into the channel estimation subnetwork to calculate the initial least squares channel estimate; concatenate the real and imaginary parts of the least squares estimate to construct an input tensor; input the input tensor into the multi-scale feature extraction module to extract multi-scale features through three parallel one-dimensional convolutional layers; input the multi-scale features into the feature fusion module for concatenation and dimensionality reduction processing to obtain fused features; and input the fused features into the encoder-decoder module, where the output features of the convolutional layers are directly passed to the corresponding deconvolutional layers for concatenation through a skip connection structure to output the channel frequency response of all subcarriers.

[0071] Optionally, the signal detection subnetwork includes: a differentiable equalization layer, a gated fusion module, a bidirectional long short-term memory (LSTM) network module, and a decision output layer; wherein, the differentiable equalization layer is used to perform preliminary equalization on the received data symbols using the channel frequency response of the subcarriers output by the channel estimation subnetwork to obtain an initial equalized signal; wherein, the gated fusion module is used to concatenate the received data symbols, the channel frequency response of all subcarriers, and the initial equalized signal, and generate adaptive weight coefficients through a fully connected layer and a Softmax function to perform weighted fusion on different input components and output the fused sequence signal; wherein, the bidirectional LSM network module consists of three layers of bidirectional LSM networks; the bidirectional LSM network module is used to perform bidirectional time-series modeling on the fused sequence signal, extract time-domain correlation through forward and backward information interaction, suppress residual interference and noise, and output high-dimensional features; wherein, the decision output layer consists of a fully connected layer and a Sigmoid activation function; the decision output layer is used to map the high-dimensional features into soft bit information and complete binary bit decision through a preset threshold.

[0072] Optionally, the recovery unit 740 is configured to: input the channel frequency response of all subcarriers and the received data symbols into the differentiable equalization layer in the signal detection subnetwork to perform differentiable zero-approach equalization on the received data symbols to obtain an initial equalization signal; concatenate the received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal by feature dimensions, and input them into the gated fusion module to adaptively learn the importance weights of each component; obtain the fused sequence signal based on the importance weights of each component; input the fused sequence signal into the bidirectional long short-term memory network module to process the forward sequence and the backward sequence simultaneously at each time step to obtain the forward hidden state and the backward hidden state; pass the forward hidden state and the backward hidden state through the output gate to obtain high-dimensional features; input the high-dimensional features into the decision output layer, process them through a fully connected layer and a sigmoid activation function, and output the likelihood probability of each bit; and convert the likelihood probability of each bit into the final binary bits through the Softmax function.

[0073] Optionally, the process of obtaining the fused sequence signal is represented by the following formulas (1)-(3): (1) (2) (3) Among them, Indicates the received data symbols; This represents the channel frequency response of all subcarriers; This represents the initial equalization signal; This represents the fused feature vector formed by splicing the received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal. Indicates the first weighting coefficient; This represents the second weighting coefficient; This represents the third weighting coefficient; This represents the fused sequence signal; This represents the concatenation function.

[0074] Optionally, both the channel estimation subnetwork and the signal detection subnetwork are trained using an end-to-end joint training method; wherein, during the training process, both the channel estimation subnetwork and the signal detection subnetwork use the mean square error function as the loss function.

[0075] The embodiments of the present invention accurately estimate and compensate for frequency-selective fading through the collaborative design of multi-scale convolutional neural networks and bidirectional long short-term memory networks, effectively suppress multipath interference, and achieve low bit error rate bit recovery under severe UTM channels.

[0076] The embodiments of the present invention adopt an end-to-end training objective to directly minimize bit error, so that the joint optimization of the entire receiver is oriented towards the final task. Therefore, compared with the traditional separation optimization method, it can extract and restore the original information bits from the fading signal more effectively.

[0077] The network structure of this invention is designed based on a communication physical model, with clearly defined functions for each module. It combines the nonlinear capabilities of deep learning with the interpretability of model-driven approaches. It also exhibits greater adaptability to channel variations and noise.

[0078] Figure 8 is a schematic diagram of an ultrasonic transmissive metal OFDM receiver for multipath fading according to an embodiment of the present invention. As shown in Figure 8, the ultrasonic transmissive metal OFDM receiver for multipath fading may include the ultrasonic transmissive metal OFDM receiver shown in Figure 7 above. Optionally, the ultrasonic transmissive metal OFDM receiver 810 for multipath fading may include a first processor 2001.

[0079] Optionally, the ultrasonic transmissive metal OFDM receiver 810 for resisting multipath fading may also include a memory 2002 and a transceiver 2003.

[0080] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.

[0081] The following describes the various components of the ultrasonic transmissive metal OFDM receiver 810 for multipath fading resistance in detail with reference to Figure 8: The first processor 2001 is the control center of the ultrasonic transmissive metal OFDM receiver 810 for multipath fading resistance. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0082] Optionally, the first processor 2001 can perform various functions of the ultrasonic transmissive metal OFDM receiver 810 for anti-multipath fading by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0083] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, such as CPU0 and CPU1 shown in FIG8.

[0084] In a specific implementation, as one embodiment, the ultrasonic transmissive metal OFDM receiver 810 for resisting multipath fading may also include multiple processors, such as the first processor 2001 and the second processor 2004 shown in FIG. 8. Each of these processors may be a single-core processor or a multi-core processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0085] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0086] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently and be coupled to the first processor 2001 through an interface circuit (not shown in FIG. 8) for the ultrasonic transmissive metal OFDM receiver 810 for anti-multipath fading. This embodiment of the invention does not specifically limit this.

[0087] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0088] Optionally, transceiver 2003 may include a receiver and a transmitter (not shown separately in Figure 8). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0089] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or exist independently, and is coupled to the first processor 2001 through the interface circuit (not shown in FIG8) of the ultrasonic transmissive metal OFDM receiver 810 for anti-multipath fading. This embodiment of the invention does not specifically limit this.

[0090] It should be noted that the structure of the ultrasonic transmissive metal OFDM receiver 810 for anti-multipath fading shown in Figure 8 does not constitute a limitation on the router. Actual ultrasonic transmissive metal OFDM receivers for anti-multipath fading may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0091] Furthermore, the technical effects of the ultrasonic translucent metal OFDM receiver 810 for resisting multipath fading can be referred to the technical effects of the ultrasonic translucent metal OFDM receiver method for resisting multipath fading described in the above method embodiments, and will not be repeated here.

[0092] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or it may be any conventional processor, etc.

[0093] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0094] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0095] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0096] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0097] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0098] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0099] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0100] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0101] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0103] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0104] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for ultrasonic transmissive metal OFDM receivers to resist multipath fading, characterized in that, The method includes: S1, acquiring the frequency domain signal received in a multipath channel; S2, constructing an ultrasonic transmissive metal OFDM system; the ultrasonic transmissive metal OFDM system includes a multi-scale convolutional gated bidirectional long short-term memory network receiver; the multi-scale convolutional gated bidirectional long short-term memory network receiver includes a channel estimation subnetwork and a signal detection subnetwork; S3, inputting the received frequency domain signal and known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers; S4, inputting the channel frequency response of all subcarriers and the received data symbols into the signal detection subnetwork to convert the soft bits into final binary bits through the Softmax function.

2. The ultrasonic transmissive metal OFDM receiving method for resisting multipath fading according to claim 1, characterized in that, The channel estimation subnetwork includes a multi-scale feature extraction module, a feature fusion module, and an encoder-decoder module. The multi-scale feature extraction module comprises multiple parallel one-dimensional convolutional layers with different kernel sizes. This module extracts multi-scale features from the pilot channel response, captures the correlation between different frequency domain components, and enhances channel representation capabilities. The feature fusion module fuses the multi-scale features and outputs the fused features. The encoder-decoder module consists of three cascaded convolutional layers and three deconvolutional layers, used for nonlinear mapping and channel reconstruction of the fused features. The convolutional layers are used for feature compression and encoding, while the deconvolutional layers are used for feature upsampling and reconstruction.

3. The ultrasonic transmissive metal OFDM receiving method for resisting multipath fading according to claim 2, characterized in that, The step S3 involves inputting the received frequency domain signal and the known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers. This includes: S31, inputting the received frequency domain signal and the known transmitted pilot symbols into the channel estimation subnetwork to calculate the initial least-squares channel estimate; concatenating the real and imaginary parts of the least-squares estimate to construct an input tensor; S32, inputting the input tensor into the multi-scale feature extraction module, where multi-scale features are extracted through three parallel one-dimensional convolutional layers; S33, inputting the multi-scale features into the feature fusion module for concatenation and dimensionality reduction processing to obtain the fused features; and S34, inputting the fused features into the encoder-decoder module, where the output features of the convolutional layers are directly passed to the corresponding deconvolutional layers for concatenation through a skip connection structure to output the channel frequency response of all subcarriers.

4. The ultrasonic transmissive metal OFDM receiving method for resisting multipath fading according to claim 1, characterized in that, The signal detection subnetwork includes: a differentiable equalization layer, a gated fusion module, a bidirectional long short-term memory (LSTM) network module, and a decision output layer. The differentiable equalization layer uses the channel frequency response of the subcarriers output by the channel estimation subnetwork to perform preliminary equalization on the received data symbols, obtaining an initial equalized signal. The gated fusion module concatenates the received data symbols, the channel frequency response of all subcarriers, and the initial equalized signal, and generates adaptive weight coefficients through a fully connected layer and a Softmax function to perform weighted fusion on different input components, outputting a fused sequence signal. The bidirectional LSM network module consists of three layers and is used for bidirectional time-series modeling of the fused sequence signal. It extracts temporal correlation through forward and backward information interaction, suppresses residual interference and noise, and outputs high-dimensional features. The decision output layer consists of a fully connected layer and a sigmoid activation function. It maps the high-dimensional features to soft-bit information and completes binary bit decision through a preset threshold.

5. The ultrasonic transmissive metal OFDM receiving method for resisting multipath fading according to claim 4, characterized in that, The step S4, which involves inputting the channel frequency response of all subcarriers and the received data symbols into the signal detection subnetwork and converting soft bits into final binary bits using decision rules, includes: S41, inputting the channel frequency response of all subcarriers and the received data symbols into the differentiable equalization layer of the signal detection subnetwork, and performing differentiable zero-approximation equalization on the received data symbols to obtain an initial equalization signal; S42, concatenating the received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal by feature dimensions, and inputting this concatenation into the gated fusion module to adaptively learn the importance of each component. Weights; Based on the importance weights of each component, the fused sequence signal is obtained; S43, The fused sequence signal is input into the bidirectional long short-term memory network module, and the forward and backward sequences are processed simultaneously at each time step to obtain the forward hidden state and the backward hidden state; The forward hidden state and the backward hidden state are passed through the output gate to obtain high-dimensional features; S44, The high-dimensional features are input into the decision output layer, processed through the fully connected layer and the sigmoid activation function, and the likelihood probability of each bit is output; Based on the likelihood probability of each bit, it is converted into the final binary bits through the Softmax function.

6. The ultrasonic transmissive metal OFDM receiving method for resisting multipath fading according to claim 5, characterized in that, The process of obtaining the fused sequence signal is represented by the following formulas (1)-(3): （1） （2） (3) Among them, Indicates the received data symbols; This represents the channel frequency response of all subcarriers; This represents the initial equalization signal; This represents the fused feature vector formed by splicing the received data symbols, the channel frequency response of all subcarriers, and the initial equalization signal. Indicates the first weighting coefficient; This represents the second weighting coefficient; This represents the third weighting coefficient; This represents the fused sequence signal; This represents the concatenation function.

7. The ultrasonic transmissive metal OFDM receiving method for resisting multipath fading according to claim 1, characterized in that, Both the channel estimation subnetwork and the signal detection subnetwork are trained using an end-to-end joint training method; during the training process, both the channel estimation subnetwork and the signal detection subnetwork use the mean square error function as the loss function.

8. An ultrasonic transmissive metal OFDM receiver for resisting multipath fading, wherein the ultrasonic transmissive metal OFDM receiver is used to implement the ultrasonic transmissive metal OFDM receiving method for resisting multipath fading as described in any one of claims 1-7, characterized in that, The apparatus includes: an acquisition unit for acquiring frequency domain signals received in a multipath channel; a construction unit for constructing an ultrasonic transmissive metal OFDM system; the ultrasonic transmissive metal OFDM system includes: a multi-scale convolutional gated bidirectional long short-term memory network receiver; the multi-scale convolutional gated bidirectional long short-term memory network receiver includes: a channel estimation subnetwork and a signal detection subnetwork; an estimation unit for inputting the received frequency domain signal and known transmitted pilot symbols into the channel estimation subnetwork to estimate the channel frequency response of all subcarriers; and a recovery unit for inputting the channel frequency response of all subcarriers and the received data symbols into the signal detection subnetwork to convert soft bits into final binary bits using a Softmax function.

9. An ultrasonic transmissive metal OFDM receiver for resisting multipath fading, characterized in that, The ultrasonic transmissive metal OFDM receiver for resisting multipath fading includes: a processor; and a memory storing computer-readable instructions, which, when executed by the processor, implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 7.

Images