Transmitting antenna selection method, device, computer equipment and readable medium

Abstract

The present disclosure provides a transmitting antenna selection method, comprising: performing channel estimation according to a signal transmitted by a user equipment to obtain channel state information at time t; performing channel prediction according to the channel state information at time t by using a channel autocorrelation function to calculate channel state information at time (t+τ d ); determining the optimal transmitting antenna of a millimeter wave MIMO system at a next transmitting time slot according to the channel state information at time (t+τ d ), an original eavesdropping channel matrix and a classification model; the classification model comprises at least two antenna categories of the millimeter wave MIMO system, and the classification model is obtained by training using an SVM algorithm. The channel autocorrelation function is used to represent the time-varying nature of the channel, the CSI error and the out-of-date problem are alleviated by predicting the channel state, the channel state information at time (t+τ d ) obtained by channel prediction is combined with the SVM algorithm to select the optimal transmitting antenna, so that the transmitting antenna can achieve the best performance. The present disclosure also provides a transmitting antenna selection device, a computer device and a readable medium.

Description

Technical Field

[0001] The present disclosure relates to the field of communication technology, and in particular to a method, apparatus, computer device, and readable medium for selecting a transmitting antenna. Background Art

[0002] In MIMO (Multiple Input Multiple Output) systems, antenna selection at the transmitter reduces RF complexity. Selecting the optimal transmit antenna based on known Channel State Information (CSI) maximizes the receiver's signal-to-noise ratio, ultimately effectively preventing eavesdropping. Therefore, optimal Transmit Antenna Selection (TAS) is one of the most important MIMO technologies in physical layer security. The primary advantage of TAS in physical layer security is that it enhances the confidentiality of communications even with partial CSI.

[0003] Currently, research on optimal transmit antenna selection schemes is mostly based on quasi-static fading channel models, where channel element values ​​vary relatively smoothly, representing an idealized channel model. However, in millimeter-wave communications, the channel characteristics differ from those of traditional microwave Rayleigh channels. Millimeter-wave channels are typically line-of-sight channels, with very limited scattering and non-line-of-sight propagation paths, typically with only a few scattering clusters. This leads to the sparse nature of millimeter-wave channels. Currently, research on secure transmission over millimeter-wave channels is still in its infancy, and quasi-static fading models are not suitable for characterizing millimeter-wave channels.

[0004] During the physical layer data transmission process of the millimeter wave communication system, due to the natural time-varying channel and suboptimal channel estimation method, the obtained channel state information contains errors and there is a delay in the transmission processing, resulting in a non-optimal channel state for TAS selection. Summary of the Invention

[0005] The present disclosure provides a transmitting antenna selection method, apparatus, computer equipment, and readable medium.

[0006] In a first aspect, an embodiment of the present disclosure provides a method for selecting a transmit antenna, the method being used to select an optimal transmit antenna for a millimeter-wave multiple-input multiple-output (MIMO) system, the method comprising:

[0007] Perform channel estimation based on the signal sent by the user equipment to obtain the channel state information at time t;

[0008] Using the channel autocorrelation function, the channel state information at the time t is used to perform channel prediction and calculate (t+τ d ) time, τ dis the transmission delay of the signal;

[0009] According to the (t+τ d ) moment, the original eavesdropped channel matrix and the classification model to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein, the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm.

[0010] In some embodiments, the classification model is trained using an SVM-SGD algorithm, and the steps of training the classification model include:

[0011] Constructing learning parameters for each of the antenna categories;

[0012] According to the learned parameters w for all antenna categories l constructing the classification model;

[0013] The SVM-SGD algorithm is used to update the learning parameters corresponding to each category.

[0014] In some embodiments, the learning parameters of each antenna category are constructed in the following manner:

[0015] Extracting a channel eigenvector d and performing normalization processing on the channel eigenvector d to obtain a normalized eigenvector n;

[0016] The learning parameter w is constructed based on at least the Gaussian kernel function f(n) about the normalized eigenvector n and the antenna selection vector l objective function; wherein the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot is the transmitting antenna associated with the maximum value of the selection parameter, and the selection parameter is the transposed matrix of the value of the Gaussian kernel function f(n) and the learning parameter of each antenna category The product of .

[0017] In some embodiments, the method according to (t+τ d ), the original eavesdropped channel matrix G, and the classification model, to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot, including:

[0018] Using the classification model, according to the (t+τ d ) and the original eavesdropping channel matrix to calculate the confidentiality performance index parameters;

[0019] The optimal transmitting antenna of the millimeter wave MIMO system in the next transmitting time slot is determined according to the maximum value of the confidentiality performance indicator parameter.

[0020] In some embodiments, the confidentiality performance indicator parameter is the channel traversal confidentiality capacity, and the classification model is used according to (t+τ d ) and the original eavesdropping channel matrix to calculate the confidentiality performance index parameters, including:

[0021] At least the (t+τ d ) input into the classification model to calculate a first signal-to-noise ratio of the user equipment;

[0022] At least inputting the original eavesdropping channel matrix into the classification model to calculate a second signal-to-noise ratio of the eavesdropper;

[0023] The channel ergodic confidentiality capacity is calculated according to the first signal-to-noise ratio and the second signal-to-noise ratio.

[0024] In some embodiments, the channel autocorrelation function is used to perform channel prediction based on the channel state information at time t, and (t+τ d ) time, including:

[0025] Calculating a channel correlation coefficient based on a zero-order first-class Bessel function, Doppler spread, and a transmission delay of the signal;

[0026] Utilize the channel autocorrelation function, calculate (t+τ d ) moment.

[0027] In some embodiments, according to the (t+τ d ), the original eavesdropped channel matrix, and the classification model, to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot, the method further comprising:

[0028] An antenna selection accuracy rate is calculated, where the antenna selection accuracy rate is a ratio of the number of the optimal transmitting antennas to the number of antenna categories.

[0029] In another aspect, an embodiment of the present disclosure further provides a transmit antenna selection device, comprising a channel estimation module, a channel prediction module, and an antenna selection module, wherein the channel estimation module is configured to perform channel estimation based on a signal sent by a user equipment to obtain channel state information at time t;

[0030] The channel prediction module is used to use the channel autocorrelation function to perform channel prediction according to the channel state information at time t, and calculate (t+τ d ) time, τ d is the transmission delay of the signal;

[0031] The antenna selection module is used to select the antenna according to the (t+τ d ) moment, the original eavesdropped channel matrix and the classification model to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm.

[0032] On the other hand, an embodiment of the present disclosure also provides a computer device, including: one or more processors; a storage device on which one or more programs are stored; when the one or more programs are executed by the one or more processors, the one or more processors implement the transmitting antenna selection method as described above.

[0033] On the other hand, an embodiment of the present disclosure further provides a computer-readable medium having a computer program stored thereon, wherein when the program is executed, the transmitting antenna selection method as described above is implemented.

[0034] The transmit antenna selection method provided by the embodiment of the present disclosure is used to select the optimal transmit antenna of a millimeter wave MIMO system. The method includes: performing channel estimation based on the signal sent by the user equipment to obtain channel state information at time t; using the channel autocorrelation function to perform channel prediction based on the channel state information at time t, and calculating (t+τ d ) time, τ d is the transmission delay of the signal; according to (t+τ d ) moment, the original eavesdropped channel matrix, and the classification model are used to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein, the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm. The embodiment of the present disclosure uses the channel autocorrelation function to characterize the time-varying nature of the channel, and by predicting the channel state, it alleviates the CSI error and obsolescence problem; according to the channel prediction (t+τ d ) moment, combined with the machine learning SVM algorithm to select the best transmitting antenna, so that the transmitting antenna can achieve the best performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of a process for selecting a transmitting antenna according to an embodiment of the present disclosure;

[0036] Figure 2 A schematic diagram of a transmitting antenna selection process provided in an embodiment of the present disclosure;

[0037] Figure 3 A schematic diagram of a signal preprocessing process provided by an embodiment of the present disclosure;

[0038] Figure 4 A schematic diagram of the process of training the classification model provided in an embodiment of the present disclosure;

[0039] Figure 5 The learning parameter w for constructing the antenna category provided in the embodiment of the present disclosure is l Flowchart of the process;

[0040] Figure 6 Schematic diagram of the process of selecting a transmitting antenna according to a classification model provided in an embodiment of the present disclosure Figure 1 ;

[0041] Figure 7 A schematic diagram of a MIMO eavesdropping channel model provided in an embodiment of the present disclosure;

[0042] Figure 8 Schematic diagram of the process of selecting a transmitting antenna according to a classification model provided in an embodiment of the present disclosure Figure 2 ;

[0043] Figure 9 A schematic diagram of a channel prediction process provided by an embodiment of the present disclosure;

[0044] Figure 10 A schematic diagram comparing channel traversal confidentiality capacities for different antenna selection methods provided in the embodiments of the present disclosure;

[0045] Figure 11 A schematic diagram comparing the antenna selection accuracy of different antenna selection algorithms provided in the embodiments of the present disclosure;

[0046] Figure 12 A schematic diagram comparing the classification model training time of different algorithms provided in the embodiments of the present disclosure;

[0047] Figure 13 A schematic diagram of the structure of the transmitting antenna selection device provided in the embodiment of the present disclosure Figure 1 ;

[0048] Figure 14 A schematic diagram of the structure of the transmitting antenna selection device provided in the embodiment of the present disclosure Figure 2 . DETAILED DESCRIPTION

[0049] Example embodiments will be described more fully hereinafter with reference to the accompanying drawings, but the example embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete and will fully convey the scope of this disclosure to those skilled in the art.

[0050] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0051] The terms used herein are used only to describe specific embodiments and are not intended to limit the present disclosure. As used herein, the singular forms "a," "an," and "the" are also intended to include the plural forms, unless the context clearly indicates otherwise. It will also be understood that when the terms "comprising" and / or "made of" are used in this specification, the presence of the features, wholes, steps, operations, elements, and / or components is specified, but the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or groups thereof is not excluded.

[0052] The embodiments described herein may be described with reference to plan views and / or cross-sectional views, with the aid of idealized schematic diagrams of the present disclosure. Thus, the example illustrations may be modified based on manufacturing techniques and / or tolerances. Therefore, the embodiments are not limited to the embodiments shown in the accompanying drawings, but include modifications of the configurations formed based on the manufacturing process. Therefore, the regions illustrated in the accompanying drawings are schematic in nature, and the shapes of the regions shown in the drawings illustrate specific shapes of the regions of the elements, but are not intended to be limiting.

[0053] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless expressly defined as such herein.

[0054] Existing transmit antenna selection methods primarily focus on solving optimization problems. Optimization methods search for the optimal solution to an objective function. Among these, exhaustive search algorithms offer the best performance. Exhaustive search involves enumerating multiple candidate antennas to obtain the optimal solution set. Because its search cost and complexity are proportional to the number of candidate solutions, exhaustive search is often used as the optimal criterion for comparison with suboptimal algorithms. Using exhaustive search to select the optimal transmit antennas can select the antenna subset that optimizes system security performance. Transmit antenna selection is typically based on known CSI. With the criterion of maximizing the signal-to-noise ratio, the achievable channel traversal security capacity of different transmit and receive antenna combinations is calculated, and the optimal antenna subset is selected.

[0055] Current research on secure signal transmission typically assumes an ideal channel, but this assumption fails to account for the less-than-ideal factors that exist in real-world situations. Errors in channel estimation and delays in transmission processing can lead to incomplete and outdated estimated CSI. This imperfect CSI used for transmit antenna selection can result in suboptimal performance of the antennas used for signal transmission.

[0056] The present disclosure provides a method for selecting a transmitting antenna, which is used to select the best transmitting antenna for a millimeter wave multiple-input multiple-output (MIMO) system. Figure 1 、 Figure 2 and Figure 3 As shown, the transmitting antenna selection method includes the following steps:

[0057] Step 11: Perform channel estimation based on the signal sent by the user equipment to obtain channel state information at time t.

[0058] The receiver (i.e., user equipment) sends a signal, which can be a pilot signal or other signal. In this embodiment, the pilot signal is used as an example. In this step, after receiving the pilot signal sent by the receiver, the transmitter (i.e., base station equipment) performs channel estimation based on the pilot signal to obtain the channel state information H(t) at time t.

[0059] Millimeter waves have short wavelengths and are greatly affected by path loss during transmission. Multipath clustering ray models are usually used for modeling. t antennas, the receiver is equipped with N r Antennas, transmitting N cl data streams, each containing N ray The narrowband millimeter wave channel in the MIMO communication scenario is represented by a sub-path:

[0060]

[0061] α il is the complex channel gain of the lth ray in the ith cluster and represents the average power in the i-th cluster. il is the arrival angle corresponding to the path gain, a r (θ il ) represents the arrival angle θ corresponding to the receiving end il The normalized array response vector of il is the departure angle corresponding to the path gain, a t (φ il ) represents the departure angle φ corresponding to the transmitter il The normalized array response vector.

[0062] The transmitter sends a pilot signal X(t) at time t. At time t, the signal received by the receiver is expressed as: Among them, P t is the transmit power, H(t) is the channel gain matrix, N~CN(0,σ 2 ) is additive white Gaussian noise.

[0063] With the minimum mean square error (MMSE) channel estimation, the actual obtained CSI is: where H(t) is the estimated CSI, N MMSE (t) is the error generated by the MMSE estimation.

[0064] The embodiments of the present disclosure consider time-varying channel prediction, and select the error value between the real CSI and the prediction result as an index to evaluate the channel prediction effect.

[0065] Step 12, using the channel autocorrelation function, the channel state information at time t is used to perform channel prediction, and the channel state information at time (t+τ d ) is calculated, τ d is the transmission delay of the signal.

[0066] Since there is a transmission delay τ d for the transmission of the pilot signal, in this step, the transmitter uses the channel autocorrelation function to perform channel prediction according to the channel state information H(t) at time t, and obtains the channel state information H(t+τ d ) at time (t+τ d ).

[0067] The best current transmit antenna selection scheme is mostly based on a quasi-static fading channel model, that is, the channel remains unchanged within the coherence time, thereby simplifying the performance analysis process. However, in the millimeter wave scenario, due to the mobility of the receiver, the Doppler spread gradually changes in the propagation environment, and the corresponding channel impulse response also gradually changes over time. Therefore, the gap between the CSI estimation value obtained by the transmitter and the real CSI gradually increases, that is, there is a transmission delay between the transmission of the pilot signal to the transmitter for signal estimation and the actual start of information transmission. The transmission delay is defined as τ d , and the embodiments of the present disclosure consider using the channel autocorrelation function to represent its time-varying characteristics.

[0068] The output of the network is H(t+τ d ), that is, the predicted value of the channel matrix at time (t+τ d ). The channel parameters obtained by sampling are used to train and predict the channel coefficients at time (t+τ d ), and the generated channel matrix H(t+τ d ) has time series characteristics, that is, time domain correlation.

[0069] Step 13, according to the channel state information at time (t+τ d ), the original eavesdropping channel matrix and the classification model, determine the best transmit antenna of the millimeter wave MIMO system in the next transmission time slot; wherein the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained by using the SVM algorithm.

[0070] In this step, the antennas of the millimeter wave MIMO system are divided into multiple categories, and the SVM (Support Vector Machine) algorithm is used to train the classification model including the categories, and (t+τ d ) time channel state information H(t+τ d ) and the original eavesdropping channel matrix G are input into the trained classification model, and the classification model is used to select the best transmitting antenna for the next transmission time slot, where the classification model outputs the best antenna index.

[0071] The disclosed embodiment combines the machine learning algorithm SVM to propose an intelligent antenna selection algorithm, which can select the best transmitting antenna for secure signal transmission of the transmitter, effectively improve the channel traversal confidentiality capacity, and effectively reduce the computational complexity.

[0072] Figure 3 This is the signal preprocessing process of the embodiment of the present disclosure, such as Figure 3 As shown, taking the TDD (Time Division Duplexing) system as an example, the signal preprocessing process includes: channel estimation, channel prediction, and antenna selection, which correspond to the above steps 11, 12, and 13 respectively.

[0073] The transmit antenna selection method provided by the embodiment of the present disclosure is used to select the optimal transmit antenna of a millimeter wave MIMO system. The method includes: performing channel estimation based on the signal sent by the user equipment to obtain the channel state information H(t) at time t; using the channel autocorrelation function, performing channel prediction based on the channel state information H(t) at time t, and calculating (t+τ d ) time channel state information H(t+τ d ), τ d is the transmission delay of the signal; according to (t+τ d ) time channel state information H(t+τ d ), the original eavesdropping channel matrix G and the classification model, determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm. The embodiment of the present disclosure uses the channel autocorrelation function to characterize the time-varying nature of the channel, and predicts the channel state to alleviate the CSI error and obsolescence problem; according to the channel prediction (t+τ d ) time channel state information H(t+τ d ), combined with the machine learning SVM algorithm to select the best transmitting antenna, so that the transmitting antenna can achieve the best performance.

[0074] Current methods for selecting transmit antennas using machine learning require training on a large number of CSI samples. The SVM algorithm's parameter update strategy, called Grid Search, loops through all possible parameters and selects the best-performing parameters, resulting in a high time consumption. To address this issue, in some embodiments, a classification model is trained using the SVM-SGD (Stochastic Gradient Descent) algorithm. The disclosed embodiments utilize the principles of the SGD algorithm to optimize the SVM algorithm's parameter update strategy, implementing a transmit antenna selection solution. This solution models transmit antenna selection as a multi-class classification problem, trains a multi-class model using the SVM-SGD algorithm, and uses this classification model to select the optimal transmit antenna.

[0075] Because H(t) can generate estimation errors during the channel estimation process, and transmission delays in the signal transmission process can cause actual CSI to become outdated, the CSI obtained by traditional channel estimation is imperfect. The disclosed embodiments employ a machine learning-based approach, using the generated CSI to train a network. This trained network is then used for CSI prediction to mitigate the negative impacts of estimation errors and delays. The predicted CSI is then used to train a classification model, and the SGD approach is employed to improve the SVM parameter update method for greater efficiency.

[0076] like Figure 4 As shown, the steps for training a classification model include:

[0077] Step 41: construct learning parameters for each antenna category.

[0078] In the embodiment of the present disclosure, the transmit antenna selection is modeled as a multi-class classification problem, and the classification labels of the transmit antennas are defined as l∈{1,...,N t}, N t is the total number of transmitting antennas in the millimeter wave MIMO system. In this step, for each antenna category l, the corresponding learning parameter w is constructed. l .

[0079] Step 42: Building a classification model based on the learned parameters of all antenna categories.

[0080] Step 43: Use the SVM-SGD algorithm to update the learning parameters corresponding to each category.

[0081] The traditional SVM algorithm uses grid search to update parameters and compare all possible parameter combinations, which has the problem of low training efficiency. The embodiment of the present disclosure uses the SVM-SGD algorithm to update the learning parameters w corresponding to each category. l, the convergence speed is fast, and therefore, the parameter updating strategy of the antenna selection scheme based on the SVM-SGD algorithm can improve the training efficiency of the classification model.

[0082] In some embodiments, as shown in Figure 5 , the step of constructing the learning parameter w l of each antenna category includes:

[0083] Step 141, extract the channel feature vector d and normalize the channel feature vector d to obtain the normalized feature vector n.

[0084] In this step, for each antenna category, the channel feature vector d is extracted and normalized to obtain the normalized feature vector n, n = (d - E[d]) / (max(d) - min(d)), where E[d] represents the average value of d. By normalizing the channel feature vector d, the accuracy of the classification model can be improved.

[0085] Step 142, at least according to the Gaussian kernel function f(n) of the normalized feature vector n and the antenna selection vector, construct the objective function of the learning parameter w l ; wherein the optimal transmission antenna of the millimeter wave MIMO system in the next transmission time slot is the transmission antenna associated with the maximum value of the selection parameter, and the selection parameter is the product of the value of the Gaussian kernel function f(n) and the transpose matrix of the learning parameter of each antenna category .

[0086] The SVM-SGD algorithm is used for classification decision, and a "one versus the rest" classification model objective function is constructed by solving an alternative logistic regression problem. The formula of the objective function of the learning parameter is formula (1):

[0087]

[0088] where w l is the learning parameter, l ∈ {1, 2,..., N t}; c is a non-negative scalar representing the trade-off between bias and overfitting; b l is the antenna selection vector, b l = 1 indicates that the antenna is selected, b l = 0 indicates that the antenna is not selected; f(n) represents a kernel function based on Gaussian radial basis, which is used to increase the distribution probability of the original sample features.

[0089] When all the parameters w l are obtained, i.e., loss minimization, the classification model can be used for TAS. The antenna label l * to be selected is the antenna label that achieves the maximum in all categories.

[0090] The signal received by the receiving antenna can be expressed as: Where B(t) is the antenna selection vector. It should be noted that in order to highlight the role of antenna selection, after antenna selection, the channel matrix only retains the selected column vector elements and sets the remaining elements to 0. d ) represents the actual channel matrix, not the estimated CSI matrix H(t).

[0091] In some embodiments, as Figure 6 As shown, the basis (t+τ d ) time channel state information H(t+τ d ), the original eavesdropping channel matrix G and the classification model, determining the transmitting antenna of the millimeter wave MIMO system in the next transmission time slot (i.e., step 13), including the following steps:

[0092] Step 131, using the classification model, according to (t+τ d ) and the original eavesdropping channel matrix to calculate the confidentiality performance index parameters.

[0093] In this step, (t+τ d ) time channel state information H(t+τ d ) and the original eavesdropping channel matrix G are input into the trained classification model, and the classification model is used to calculate the confidentiality performance index parameters.

[0094] Step 132: Determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmitting time slot according to the maximum value of the confidentiality performance indicator parameter.

[0095] The optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot is determined according to the maximum value of the confidentiality performance index parameter. The antenna that maximizes the channel traversal confidentiality capacity can be selected for signal transmission, thereby improving the security of the physical layer of wireless communication.

[0096] Figure 7 A schematic diagram of a MIMO eavesdropping channel model provided in an embodiment of the present disclosure is shown in FIG. Figure 7 As shown in Figure 1, the signal sent by the transmitter can be received by a normal user device (i.e., the receiver) or by an illegal user device (i.e., the eavesdropper). The channel between the transmitter and the receiver is the primary channel, and its channel state information is H(t). The channel between the transmitter and the eavesdropper is the eavesdropping channel, and its channel state is the original eavesdropping channel matrix G.

[0097] In some embodiments, the confidentiality performance indicator parameter is the channel traversal confidentiality capacity. Figure 8 As shown, the classification model is used according to (t+τ d) and the original eavesdropping channel matrix to calculate the security performance index parameter (i.e., step 131), including the following steps:

[0098] Step 1311, at least input the channel state information at time (t+τ d ) into the classification model to calculate the first signal-to-noise ratio of the user equipment.

[0099] In this step, the SNR (i.e., the first signal-to-noise ratio) of the receiver at time t is calculated according to the channel state information at time (t+τ d ) H(t+τ d ): γ B = γ B,1 , γ B,2 ,..., γ B,Nt , where

[0100] Step 1312, at least input the original eavesdropping channel matrix into the classification model to calculate the second signal-to-noise ratio of the eavesdropper.

[0101] In this step, the SNR (i.e., the second signal-to-noise ratio) of the eavesdropper at time t is calculated according to the original eavesdropping channel matrix G: γ E = γ E,1 , γ E,2 ,..., γ E,Nt , where

[0102] Step 1313, calculate the channel traversal secrecy capacity according to the first signal-to-noise ratio and the second signal-to-noise ratio.

[0103] The channel traversal secrecy capacity is:

[0104] When the first signal-to-noise ratio of the main channel is greater than the second signal-to-noise ratio of the eavesdropping channel, the main channel capacity is higher than the eavesdropping channel capacity. By selecting the optimal transmit antenna, the antenna with the maximum channel traversal secrecy capacity is selected for signal transmission, thereby improving the physical layer security of wireless communication.

[0105] In some embodiments, as shown in Figure 9 , the channel state information at time (t+τ d ) is calculated by using the channel autocorrelation function to perform channel prediction according to the channel state information at time t (i.e., step 12), including the following steps:

[0106] Step 121, calculate the channel correlation coefficient according to the zero-order first Bessel function, the Doppler spread, and the transmission delay of the signal.

[0107] In some embodiments, the channel correlation coefficient can be calculated according to the following formula (2):

[0108] ρ d =J0(2πf d τ d ) (2)

[0109] Where J0 is a zero-order first-class Bessel function, f d is the Doppler spread, ρ d is the channel correlation coefficient. d The larger the transmission delay τ d The smaller the impact on the channel state information, the d =1 means within the coherence time τ d The impact on channel state information is negligible.

[0110] Step 122: Calculate (t+τ d ) moment.

[0111] In some embodiments, the channel autocorrelation function can be expressed as formula (3):

[0112]

[0113] Where H(t) is the channel state information at time t estimated in step 11, and ρ is the outdated CSI matrix; d is the channel correlation coefficient calculated in step 121; R(t+τ d ) is a random interference matrix independent of H(t); H(t+τ d ) is (t+τ d ) time, H(t+τ d ) is the real-time channel status.

[0114] In some embodiments, according to (t+τ d After determining the optimal transmit antenna for the millimeter-wave MIMO system in the next transmit time slot (i.e., step 13) based on the channel state information at the time instant, the original eavesdropped channel matrix, and the classification model, the transmit antenna selection method further includes the following steps: calculating the antenna selection accuracy, where the antenna selection accuracy is the ratio of the number of optimal transmit antennas to the number of antenna categories. In this step, the classification model is used to calculate and output the antenna selection accuracy Pc. It should be noted that the classification model can also be used to calculate and output the training time of the classification model. The antenna selection accuracy Pc and the training time of the classification model are used to evaluate the quality of the classification model.

[0115] After training the classification model, input (t+τ d ) moment, the classification model can output the best antenna index l* , the channel can achieve channel traversal confidentiality capacity C, antenna selection accuracy Pc and model training time.

[0116] In the embodiment of the present disclosure, in a MIMO communication scenario, it is assumed that the transmitter is equipped with 8 antennas (Nt=8), the receiver is equipped with 2 antennas (Nr=2), and the eavesdropper is equipped with 2 antennas; the antenna array is a uniform linear array (ULA); the CSI is obtained by performing channel estimation on the transmitter, and H(t+τ d ) to select the optimal transmit antenna and use the SVM-SGD algorithm to train the classification model. The channel model is a slowly varying channel, which means that the arrival angle, departure angle, and channel gain of all paths in the channel only change slightly within a short period of time.

[0117] The embodiment of the present disclosure uses a time series prediction channel matrix H(t+τ d ). The sequence of channel gain coefficient matrices is used as the input of the classification model in chronological order for its training and selection. The performance results achieved by this scheme are as follows Figure 10-12 shown.

[0118] Figure 10 A schematic diagram comparing the channel traversal confidentiality capacity of different antenna selection methods provided in the embodiments of the present disclosure is provided. Figure 10 The security performance achieved by antenna selection using traditional exhaustive search, the SVM algorithm, and the SGD algorithm is shown. Three schemes are considered, using the modulus of the complex-valued channel elements as the eigenvector. It can be seen that the channel traversal security capacity achievable by the three schemes increases with increasing SNR. The SGD algorithm achieves a higher average security capacity than the SVM algorithm, and the SGD scheme has lower computational complexity.

[0119] Figure 11 A schematic diagram comparing the antenna selection accuracy of different antenna selection algorithms provided in the embodiment of the present disclosure, from Figure 11 It can be seen that the SGD algorithm has higher accuracy than the SVM algorithm, which shows that the SGD algorithm has better classification effect when used for training the current data set.

[0120] Figure 12 A schematic diagram comparing the training time of classification models of different algorithms provided in the embodiment of the present disclosure is shown in FIG. Figure 12 As shown in Figure 2, the SGD algorithm takes about 1.02 seconds to perform a model training, while the SVM algorithm takes about 137.8 seconds to perform a model training. Therefore, the training speed of the SGD algorithm is much faster than that of the SVM algorithm, especially when the data set reaches 10 5The SGD algorithm uses gradient descent to solve parameters, making it simple and effective to fit linear classification models under convex loss functions, reducing time complexity and improving the training efficiency of classification models.

[0121] The disclosed embodiments propose a machine learning-based transmit antenna selection scheme for millimeter-wave MIMO communication systems. This scheme uses channel prediction to mitigate the issue of transmit antenna selection obsolescence caused by channel estimation errors and transmission delays. By leveraging the time-varying nature of wireless channels and building a classification model based on effective channel prediction, the scheme employs a machine learning SVM algorithm. Furthermore, the SVM algorithm's parameter update process is improved using stochastic gradient descent, enhancing the physical layer security of transmit antenna selection and improving model training efficiency.

[0122] The disclosed embodiments are based on innovations in millimeter wave MIMO systems. Millimeter wave frequencies are high and spectrum resources are abundant, providing high-speed transmission for 5G wireless communications. However, due to the short wavelength of millimeter waves and their significant impact on path loss, they must be combined with large-scale MIMO technology to leverage the benefits of antenna arrays to compensate for their inherent shortcomings. In a multi-antenna system, keeping antennas open simultaneously results in reduced energy efficiency and waste of resources. By utilizing transmit antenna selection technology, a subset of antennas that maximize the receiver's signal-to-noise ratio is selected for signal transmission, effectively reducing resource waste. Furthermore, when some antennas are operating, the channel quality gap between legitimate users and eavesdroppers can be widened, thereby improving transmission confidentiality.

[0123] Based on the same technical concept, the embodiment of the present disclosure also provides a transmitting antenna selection device, such as Figure 13 As shown, the transmitting antenna selection device includes a channel estimation module 101, a channel prediction module 102 and an antenna selection module 103. The channel estimation module 101 is used to perform channel estimation based on the signal sent by the user equipment to obtain channel state information at time t.

[0124] The channel prediction module 102 is used to perform channel prediction based on the channel state information at time t using the channel autocorrelation function, and calculate (t+τ d ) time, τ d is the transmission delay of the signal.

[0125] The antenna selection module 103 is used to select the antenna according to the (t+τ d ) moment, the original eavesdropped channel matrix and the classification model to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm.

[0126] In some embodiments, as Figure 14 As shown, the transmitting antenna selection device further includes a model training module 104, which is used to train the classification model using the SVM-SGD algorithm, wherein the learning parameters of each antenna category are constructed for each antenna category; according to the learning parameters w of all antenna categories l Construct the classification model; and use the SVM-SGD algorithm to update the learning parameters corresponding to each category.

[0127] In some embodiments, the model training module 104 is used to construct the learning parameters of each antenna category by extracting the channel feature vector d and normalizing the channel feature vector d to obtain a normalized feature vector n; constructing the learning parameter w based on at least the Gaussian kernel function f(n) about the normalized feature vector n and the antenna selection vector l objective function; wherein the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot is the transmitting antenna associated with the maximum value of the selection parameter, and the selection parameter is the transposed matrix of the value of the Gaussian kernel function f(n) and the learning parameter of each antenna category The product of .

[0128] In some embodiments, the antenna selection module 103 is configured to use the classification model according to the (t+τ d ) moment and the original eavesdropped channel matrix to calculate a confidentiality performance index parameter; and determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmitting time slot according to the maximum value of the confidentiality performance index parameter.

[0129] In some embodiments, the antenna selection module 103 is configured to at least set the (t+τ d ) moment into the classification model to calculate a first signal-to-noise ratio of the user equipment; at least the original eavesdropping channel matrix is ​​input into the classification model to calculate a second signal-to-noise ratio of the eavesdropper; and the channel ergodic confidentiality capacity is calculated based on the first signal-to-noise ratio and the second signal-to-noise ratio.

[0130] In some embodiments, the channel prediction module 102 is used to calculate the channel correlation coefficient based on the zero-order first-class Bessel function, Doppler spread and the transmission delay of the signal; and calculate (t+τ) using the channel autocorrelation function based on the channel correlation coefficient, the channel state information at time t and the random interference matrix. d ) moment.

[0131] In some embodiments, the antenna selection module 103 is further configured to: dThe channel state information at the time, the original eavesdropping channel matrix, and the classification model are used to determine the optimal transmit antennas of the millimeter wave MIMO system in the next transmit time slot, and then the antenna selection accuracy is calculated, where the antenna selection accuracy is a ratio of the number of the optimal transmit antennas to the number of the antenna categories.

[0132] The embodiments of the present disclosure further provide a computer device, which comprises one or more processors and a storage device; wherein the storage device stores one or more programs; and when the one or more programs are executed by the one or more processors, the one or more processors implement the transmit antenna selection method provided in the foregoing embodiments.

[0133] The embodiments of the present disclosure further provide a computer readable medium, which stores a computer program; and when the computer program is executed, the transmit antenna selection method provided in the foregoing embodiments is implemented.

[0134] Those skilled in the art can understand that all or some of the steps in the above disclosed method and the function modules / units in the device can be implemented as software, firmware, hardware or a combination thereof. In the hardware implementation, the division between the function modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component can have multiple functions, or one function or step can be performed by several physical components in cooperation. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software can be distributed on a computer readable medium, which can include computer storage media (or non-transitory media) and communication media (or transitory media). As known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. In addition, as known to those skilled in the art, communication media typically includes computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism, and can include any information delivery medium.

[0135] Example embodiments have been disclosed herein and, although specific terms are employed, they are used in a generic and descriptive sense only and not for purposes of limitation. In some instances, it will be apparent to those skilled in the art that features, characteristics or / and elements described in connection with a particular embodiment can be used in conjunction with other embodiments unless otherwise explicitly stated. Accordingly, it will be understood that various changes in form and details can be made without departing from the scope of the present application as set forth in the appended claims.

Claims

1. A method for selecting a transmitting antenna, characterized in that: The method is used to select an optimal transmitting antenna for a millimeter wave multiple-input multiple-output (MIMO) system, and the method includes: Perform channel estimation based on the signal sent by the user equipment to obtain the channel state information at time t; Using the channel autocorrelation function, the channel state information at the time t is used to perform channel prediction and calculate (t+τ d ) time, τ d is the transmission delay of the signal; According to the (t+τ d ) moment, the original eavesdropped channel matrix and the classification model to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein, the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm.

2. The method according to claim 1, wherein The classification model is trained using the SVM-SGD algorithm, and the steps of training the classification model include: Constructing learning parameters for each of the antenna categories; According to the learned parameters w for all antenna categories l constructing the classification model; The SVM-SGD algorithm is used to update the learning parameters corresponding to each category.

3. The method according to claim 2, wherein The learning parameters for each of the antenna classes are constructed as follows: Extracting a channel eigenvector d and performing normalization processing on the channel eigenvector d to obtain a normalized eigenvector n; The learning parameter w is constructed based on at least the Gaussian kernel function f(n) about the normalized eigenvector n and the antenna selection vector l objective function; wherein the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot is the transmitting antenna associated with the maximum value of the selection parameter, and the selection parameter is the transposed matrix of the value of the Gaussian kernel function f(n) and the learning parameter of each antenna category The product of .

4. The method according to claim 1, wherein According to the (t+τ d ), the original eavesdropped channel matrix G, and the classification model, to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot, including: Using the classification model, according to the (t+τ d ) and the original eavesdropping channel matrix to calculate the confidentiality performance index parameters; The optimal transmitting antenna of the millimeter wave MIMO system in the next transmitting time slot is determined according to the maximum value of the confidentiality performance indicator parameter.

5. The method according to claim 4, wherein The confidentiality performance indicator parameter is the channel ergodic confidentiality capacity, and the classification model is used according to (t+τ d ) and the original eavesdropping channel matrix to calculate the confidentiality performance index parameters, including: At least the (t+τ d ) input into the classification model to calculate a first signal-to-noise ratio of the user equipment; At least inputting the original eavesdropping channel matrix into the classification model to calculate a second signal-to-noise ratio of the eavesdropper; The channel ergodic confidentiality capacity is calculated according to the first signal-to-noise ratio and the second signal-to-noise ratio.

6. The method according to claim 1, wherein The channel autocorrelation function is used to perform channel prediction according to the channel state information at time t, and (t+τ d ) time, including: Calculating a channel correlation coefficient based on a zero-order first-class Bessel function, Doppler spread, and a transmission delay of the signal; Utilize the channel autocorrelation function, calculate (t+τ d ) moment.

7. The method according to any one of claims 1 to 6, wherein: According to the (t+τ d ), the original eavesdropped channel matrix, and the classification model, to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot, the method further comprising: An antenna selection accuracy rate is calculated, where the antenna selection accuracy rate is a ratio of the number of the optimal transmitting antennas to the number of antenna categories.

8. A transmitting antenna selection device, characterized in that: It includes a channel estimation module, a channel prediction module and an antenna selection module. The channel estimation module is used to perform channel estimation based on the signal sent by the user equipment to obtain the channel state information at time t; The channel prediction module is used to use the channel autocorrelation function to perform channel prediction according to the channel state information at time t, and calculate (t+τ d ) time, τ d is the transmission delay of the signal; The antenna selection module is used to select the antenna according to the (t+τ d ) moment, the original eavesdropped channel matrix and the classification model to determine the optimal transmitting antenna of the millimeter wave MIMO system in the next transmission time slot; wherein the classification model includes at least two antenna categories of the millimeter wave MIMO system, and the classification model is trained using the SVM algorithm.

9. A computer device comprising: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the transmit antenna selection method according to any one of claims 1 to 7.

10. A computer-readable medium having a computer program stored thereon, wherein: When the program is executed, the transmitting antenna selection method according to any one of claims 1 to 7 is implemented.

Images