Space-time coding and modulation scheme joint identification method for MIMO-OFDM system under unknown noise

Abstract

This invention specifically relates to a method and system for joint identification of space-time coding and modulation schemes in MIMO-OFDM systems under unknown noise conditions. The method includes: designing a feature extraction mechanism based on cyclic moments, calculating cyclic moments under multiple time delay conditions and uniformly projecting cyclic features onto a feature space, and constructing a cyclic peak needle-like feature map by utilizing the differences in peak distribution of different STBCs at the cyclic frequency; introducing a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy, utilizing the cyclic stationarity of modulation symbols, incorporating the multi-kernel concept into cyclic correlation entropy, and combining it with threshold smoothing processing to extract a multi-kernel cyclic smoothing correlation entropy spectral feature map; and designing a joint identification mechanism based on complementary gating fusion, inputting the two types of feature maps into a dual-branch convolutional neural network, adaptively weighting and fusing deep features, and achieving joint identification within a multi-task parallel classification framework. Simulation results show that this invention has good joint identification performance and noise adaptability under unknown noise conditions.

Description

Technical Field

[0001] This invention relates to the field of communication signal recognition technology in cognitive radio, and in particular to a method and system for joint recognition of the space-time coding and modulation scheme of multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) signals under unknown noise. Background Technology

[0002] With the continuous development of fifth-generation (5G) and sixth-generation (6G) mobile communication technologies, the wireless communication environment is becoming increasingly complex, and signal types are becoming more diverse, placing higher demands on the signal recognition capabilities of receivers. Multiple-input multiple-output (MIMO-OFDM) systems combine the advantages of MIMO and OFDM technologies, improving system capacity, spectral efficiency, and anti-fading capabilities without increasing additional spectrum resources, and are therefore widely used in modern wireless communication systems. In MIMO-OFDM systems, space-time coding improves signal transmission reliability by introducing spatial and temporal redundancy, while the modulation scheme directly affects the statistical characteristics and structural form of the signal. In applications such as communication reconnaissance, signal monitoring, and cognitive radio, receivers typically need to identify the space-time coding type and modulation scheme of the incoming signal to provide a basis for subsequent parameter estimation, signal detection, and demodulation. Therefore, research on the joint identification of MIMO-OFDM signal space-time coding and modulation scheme has significant theoretical and practical value.

[0003] Numerous research achievements have been made in STBC-OFDM identification of MIMO-OFDM signals. Existing technologies are mainly based on identification methods using statistical features and decision criteria. These methods extract statistical features of the signal and combine them with peak detection, hypothesis testing, decision trees, or minimum distance decision methods to achieve signal identification. The basic idea is to use theories such as probability statistics, cyclic stationarity, and higher-order moments to analyze the received signal and construct corresponding decision criteria to complete the classification.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a method and system for joint identification of space-time coding and modulation schemes in MIMO-OFDM systems under unknown noise conditions, which addresses the problem that most current space-time coding identification technologies do not fully consider the identification problem under unknown noise environments.

[0006] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to a first aspect of the present invention, a method for joint identification of space-time coding and modulation scheme in a MIMO-OFDM system under unknown noise is provided, the method comprising: Step 1: Design a feature extraction mechanism based on cyclic moments. Calculate cyclic moments under multiple time delay conditions and project cyclic features uniformly onto the feature space. Utilize the differences in peak distribution at cyclic frequencies of different STBCs to construct a needle-like feature map of cyclic peaks. Step 2: Introduce a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. Utilize the cyclic stationarity of modulation symbols to introduce the multi-kernel concept into cyclic correlation entropy. Combine this with threshold smoothing to extract the multi-kernel cyclic smoothing correlation entropy spectral feature map. Step 3: Design a joint recognition mechanism based on complementary gating fusion. Use the two types of feature maps mentioned above as network inputs to construct a dual-branch convolutional neural network that incorporates the complementary gating fusion mechanism. Perform adaptive weighted fusion of deep features and achieve joint recognition under a multi-task parallel classification framework.

[0008] In some exemplary embodiments, the construction of the cyclic peak needle-like feature map in step one specifically includes: The fourth-order time-varying moment under the delay parameter is extracted from the received signal, and the fourth-order cyclic moment is extracted by Fourier transform. The cyclic moments obtained under different time delay conditions are stacked along the time delay dimension to form a three-dimensional feature with time delay as the vertical axis; The three-dimensional features are mapped to their maximum values ​​along the cyclic frequency axis to achieve a centralized representation of multi-delay cyclic peak information, resulting in a cyclic peak needle-shaped feature map.

[0009] In some exemplary embodiments, the three-dimensional feature is mapped to its maximum value along the cyclic frequency axis, and the mapping process is expressed as follows:

[0010]

[0011]

[0012] in, The magnitude of the cyclic moment under each time delay. The needle-like characteristics of the cyclic moment peak of various STBC-OFDM signals are represented. It is a fourth-order cyclic moment.

[0013] In some exemplary embodiments, step two, extracting the multi-kernel cyclic smoothing correlation entropy spectrum feature map, specifically includes: The correlation entropy of the received signal is calculated using a Gaussian kernel function, and the cyclic correlation entropy is obtained through Fourier series expansion. We construct multi-kernel correlation entropy by weighted combination of Gaussian kernel functions with different kernel widths, and further obtain multi-kernel cyclic correlation entropy through Fourier series expansion. Threshold smoothing is applied to the multi-kernel cyclic correlation entropy to reduce the interference of unknown noise and irrelevant components on the spectral peak distribution, and the multi-kernel cyclic smoothed correlation entropy spectral feature map is extracted.

[0014] In some exemplary embodiments, the step of weighting and combining Gaussian kernel functions with different kernel widths to construct a multi-kernel correlation entropy specifically involves:

[0015] in, and Indicated by and The Gaussian kernel function with kernel width, For kernel weight coefficients, , These represent the demodulated first and second digits of ofdm. Time and The received signal at any given moment.

[0016] In some exemplary embodiments, the threshold smoothing process applied to the multi-core cyclic correlation entropy specifically involves:

[0017]

[0018] in, This is represented as a parameter for determining the threshold of spectral peaks. The cycle frequency, Indicates the cycle frequency index. Indicates frequency, This represents the period of the correlation entropy function.

[0019] In some exemplary embodiments, the joint identification described in step three specifically includes: The cyclic peak needle-shaped feature map and the multi-kernel cyclic smooth correlation entropy spectrum feature map are respectively input into a dual-branch convolutional neural network to extract high-level feature representations; A complementary gated feature fusion module is introduced to concatenate the high-level features of the dual-branch outputs. A gated weight vector is generated through a "fully connected layer-ReLU-fully connected layer" structure to perform adaptive weighted fusion of the dual-branch features. A multi-task parallel classification module is constructed, with STBC-OFDM classification and modulation scheme classification as auxiliary tasks. Together with the joint classification task, they form a multi-task learning framework. The loss functions of each task are weighted and summed to form a joint optimization objective, thereby achieving joint recognition.

[0020] According to a second aspect of the present invention, a joint identification system for space-time coding and modulation scheme of a MIMO-OFDM system under unknown noise is provided, the system comprising: The space-time coding feature extraction module designs a feature extraction mechanism based on cyclic moments. By calculating cyclic moments under multiple time delay conditions and uniformly projecting cyclic features to the feature space, the module utilizes the differences in peak distribution of different STBCs at the cyclic frequency to construct a cyclic peak needle-like feature map. The modulation mode feature extraction module introduces a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. By utilizing the cyclic stationarity of modulation symbols, the multi-kernel concept is introduced into the cyclic correlation entropy, and combined with threshold smoothing processing, the multi-kernel cyclic smoothing correlation entropy spectral feature map is extracted. The spatiotemporal coding and modulation scheme joint recognition module designs a joint recognition mechanism based on complementary gating fusion. The two types of feature maps mentioned above are used as network inputs to construct a dual-branch convolutional neural network that incorporates the complementary gating fusion mechanism. The deep features are adaptively weighted and fused, and joint recognition is achieved under a multi-task parallel classification framework.

[0021] According to a third aspect of the present invention, a computer program product is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the method for joint identification of space-time coding and modulation scheme of MIMO-OFDM system under unknown noise as described in the first aspect is implemented.

[0022] According to a fourth aspect of the present invention, an electronic device is provided, comprising: Processor; and Memory for storing the executable instructions of the processor; The processor is configured to implement the joint identification method of space-time coding and modulation scheme of MIMO-OFDM system under unknown noise as described in the first aspect when executing the executable instructions.

[0023] The embodiment of this invention provides a method for joint identification of spatiotemporal coding and modulation schemes in MIMO-OFDM systems under unknown noise conditions. It utilizes the spatiotemporal redundancy characteristics of STBC-OFDM signals to calculate cyclic moments of the received MIMO-OFDM signals under multiple time delays, and projects the cyclic features uniformly onto the feature space. By leveraging the differences in peak distribution at the cyclic frequency of different STBCs, a cyclic peak needle-like feature map is constructed, thereby achieving effective characterization of the spatiotemporal coding structure features. Simultaneously, utilizing the cyclic stationarity of modulation symbols, a multi-kernel approach is introduced into the cyclic correlation entropy, combined with threshold smoothing processing, to extract a multi-kernel cyclic smoothing correlation entropy spectrum feature map, thereby enhancing the robustness and adaptability of modulation features under unknown noise environments. Furthermore, using the above two types of feature maps as network input, a dual-branch convolutional neural network incorporating a complementary gating fusion mechanism is constructed to adaptively weight and fuse deep features, achieving joint identification of spatiotemporal coding and modulation schemes within a multi-task parallel classification framework. Therefore, this invention can effectively improve the accuracy, robustness, and noise adaptability of joint identification under unknown noise environments, possessing significant engineering application value.

[0024] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0026] Figure 1 This is a flowchart of the research on the joint identification method of space-time coding and modulation mode of MIMO-OFDM system under unknown noise provided in the embodiments of the present invention.

[0027] Figure 2 This is a block diagram of the joint identification system structure of space-time coding and modulation mode of MIMO-OFDM system under unknown noise provided in the embodiments of the present invention.

[0028] In the figure: 1. Space-time coding feature extraction module; 2. Modulation mode feature extraction module; 3. Space-time coding and modulation mode joint identification module.

[0029] Figure 3 This is a schematic diagram illustrating the accuracy of joint type identification of various spatiotemporal coding and modulation methods under unknown noise, provided by an embodiment of the present invention.

[0030] Figure 4This is a schematic diagram illustrating the accuracy of joint type identification of OFDM block number variation space-time coding and modulation method under unknown noise, provided by an embodiment of the present invention. Detailed Implementation

[0031] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0032] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0033] Among related technologies, STBC-OFDM signal recognition based on generalized cross-entropy function (Zhang J, Liu M, Chen Y, et al. Automatic identification of space-time block coding for MIMO-OFDM systems in the presence of impulsive interference[J]. IEEE Transactions on Communications, 2024, 72(8): 4816-4828.), STBC-OFDM signal recognition based on second-order cyclostationarity (Karami E, Dobre OA. Identification of SM-OFDM and AL-OFDM signals based on their second-order cyclostationarity[J]. IEEE Transactions on Vehicular Technology, 2015, 64(3): 942-953.), and STBC-OFDM signal recognition based on cross-correlation (Eldemerdash YA, Dobre OA, Liao BJ. Blind identification of SM and Alamouti STBC-OFDM signals[J]. IEEE Transactions on Wireless Communications, 2015, 14(2): 972-982.), Time-domain STBC-OFDM signal recognition based on feature sequences (Yu Keyuan, Zhang Limin, Yan Wenjun, et al. Blind recognition algorithm for time-domain STBC-OFDM based on feature sequences [J]. Journal of Beijing University of Aeronautics and Astronautics, 2021, 47(8): 1524-1532.), STBC-OFDM signal recognition based on fourth-order cyclostationary characteristics (Huang Bo, Pan Shuang, Li Xue. Blind recognition of STBC-OFDM signals based on fourth-order cyclostationary characteristics [J]. Signal Processing, 2017, 33(9): 1221-1229.), STBC-OFDM signal recognition based on fourth-order hysteresis product (Yan Wenjun, Zhang Limin, Ling Qing. Blind recognition method of STBC-OFDM signals based on FOLP [J]. Acta Electronica Sinica, 2017, 45(9): 2233-2240.), STBC-OFDM Signal Recognition Based on Improved KS Detection (Ling Qing, Zhang Limin, Yan Wenjun. Research on Blind Recognition Algorithm of STBC-OFDM Signal Based on Improved KS Detection [J].Journal of Communications, 2017, 38(4): 46-54.) and STBC-OFDM signal recognition based on fourth-order moments (Ling Qing, Zhang Limin, Yan Wenjun, et al. Blind recognition of STBC-OFDM signals in frequency-selective channels [J]. Systems Engineering & Electronics, 2017, 39(5): 1141-1147.).

[0034] In the context of wireless communication research, existing STBC-OFDM identification techniques mostly focus on identification under known noise environments, and many independently identify the spatiotemporal coding type itself, rarely considering modulation scheme information in conjunction with it. This makes it difficult to fully utilize the correlation between the two during signal generation and transmission. Meanwhile, existing modulation identification methods based on cyclic correlation entropy typically use a single Gaussian kernel function for modeling, with fixed kernel parameters. This results in insufficient adaptability to changes in noise intensity and statistical characteristics under unknown noise environments, limiting feature stability and robustness. Furthermore, the cyclic moment feature forms extracted by existing STBC-OFDM identification methods often differ under different time delay conditions, lacking a unified feature representation method, which is detrimental to subsequent feature fusion and joint modeling. Therefore, in complex unknown noise environments, existing methods still need further improvement in terms of joint discrimination capability, noise adaptability, and feature representation capability.

[0035] Based on the above analysis, the problems and shortcomings of the existing technology are as follows: Most current space-time coding recognition technologies do not fully consider the recognition problem under unknown noise environments, but mainly focus on known noise interference conditions. In addition, existing methods usually focus on the independent recognition of space-time coding types, and rarely consider modulation mode information together. It is difficult to make full use of the inherent relationship between the two in the signal generation and transmission process. As a result, the decision error in the previous stage in the serial recognition process is easy to propagate to the next stage and accumulate, thus limiting the overall recognition performance.

[0036] The challenge in addressing these issues and shortcomings lies in the following: To achieve efficient joint identification of STBC-OFDM and modulation schemes in unknown noise environments, it is necessary to extract features that are insensitive to noise disturbances and possess strong discriminative capabilities without relying on prior noise distribution models. Simultaneously, it is also necessary to consider the differences and complementarities between STBC-OFDM structural features and modulation scheme features, construct a unified and noise-resistant feature representation, and on this basis, achieve effective fusion and joint decision-making of multi-source features. This not only places higher demands on the robustness, stability, and uniformity of feature extraction methods but also increases the complexity of joint modeling and classification.

[0037] The significance of addressing the above problems and shortcomings is that accurate joint identification of STBC-OFDM and modulation schemes under unknown noise environments is a crucial prerequisite for improving signal sensing, parameter estimation, and subsequent demodulation processing capabilities in complex wireless communication environments. This invention designs a joint identification method for STBC-OFDM and modulation schemes in unknown noise environments. The cyclic peak needle-like feature map utilizes the stable differences in the cyclic frequency peak distribution of different STBC-OFDM signals under multiple time delay conditions to achieve a unified expression of cyclic moment features, exhibiting strong structural characterization capabilities and noise resistance. The multi-kernel cyclic smoothing correlation entropy spectrum feature, through the introduction of multi-scale kernel functions for joint modeling, can adaptively adjust the contribution of statistical information at each order under varying conditions of unknown noise intensity and statistical characteristics, demonstrating good noise suppression capabilities and robustness. Furthermore, by combining a dual-branch convolutional neural network with a complementary gating fusion mechanism, the two types of noise-resistant features are effectively fused, and joint identification of STBC-OFDM and modulation schemes is achieved within a multi-task parallel classification framework. This significantly improves the overall identification performance under complex unknown noise environments, providing technical support for wireless communication signal monitoring and intelligent sensing.

[0038] To address the shortcomings and deficiencies of existing technologies, this invention provides a joint identification mechanism for the space-time coding and modulation schemes of MIMO-OFDM systems under unknown noise conditions. This mechanism primarily involves a space-time coding feature extraction mechanism based on cyclic moments, a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy, and a joint identification mechanism using a dual-branch convolutional neural network based on complementary gating fusion. This achieves joint identification of the signal's space-time coding and modulation schemes. Specifically, it includes: designing a feature extraction mechanism based on cyclic moments, calculating cyclic moments under multiple time delay conditions and uniformly projecting cyclic features onto a feature space, and constructing a cyclic peak needle-like feature map by utilizing the differences in peak distribution at the cyclic frequency of different STBCs; introducing a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy, utilizing the cyclic stationarity of modulation symbols, incorporating multi-kernel concepts into cyclic correlation entropy, and combining threshold smoothing processing to extract a multi-kernel cyclic smoothing correlation entropy spectral feature map; and designing a joint identification mechanism based on complementary gating fusion, using the above two types of feature maps as network input, constructing a dual-branch convolutional neural network incorporating the complementary gating fusion mechanism, adaptively weighting and fusing deep features, and achieving joint identification within a multi-task parallel classification framework.

[0039] like Figure 1 As shown, the method for joint identification of space-time coding and modulation scheme of MIMO-OFDM system under unknown noise provided in this embodiment of the invention includes the following steps: S101, Design a feature extraction mechanism based on cyclic moments. By calculating cyclic moments under multiple time delay conditions and uniformly projecting cyclic features to the feature space, the difference in peak distribution of different STBCs at the cyclic frequency is utilized to construct a cyclic peak needle-like feature map. S102 introduces a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. By utilizing the cyclic stationarity of modulation symbols, the multi-kernel concept is introduced into cyclic correlation entropy, and combined with threshold smoothing processing, the multi-kernel cyclic smoothing correlation entropy spectral feature map is extracted. S103, and designed a joint recognition mechanism based on complementary gating fusion, taking the above two types of feature maps as network input, constructing a dual-branch convolutional neural network with complementary gating fusion mechanism, adaptively weighting and fusing deep features, and realizing joint recognition under a multi-task parallel classification framework.

[0040] like Figure 2 As shown in the embodiment of the present invention, the joint identification system for space-time coding and modulation scheme of MIMO-OFDM system under unknown noise includes: The space-time coding feature extraction module is used to design a feature extraction mechanism based on cyclic moments for the received MIMO-OFDM signal. By calculating cyclic moments under multiple time delay conditions and uniformly projecting cyclic features to the feature space, the module utilizes the differences in peak distribution of different STBCs at the cyclic frequency to construct a cyclic peak needle-shaped feature map, providing space-time coding features for the classification network. The modulation mode feature extraction module introduces a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. By utilizing the cyclic stationarity of modulation symbols, the multi-kernel concept is introduced into cyclic correlation entropy, and combined with threshold smoothing processing, multi-kernel cyclic smoothing correlation entropy spectral feature map is extracted to provide modulation features for the classification network. The spatiotemporal coding and modulation scheme joint recognition module is used to design a joint recognition mechanism based on complementary gating fusion. The two types of feature maps mentioned above are used as network inputs to construct a dual-branch convolutional neural network that introduces the complementary gating fusion mechanism. The deep features are adaptively weighted and fused, and joint recognition is achieved under a multi-task parallel classification framework.

[0041] The present invention will be further described below with reference to embodiments.

[0042] Example 1 The present invention provides a method for joint identification of space-time coding and modulation scheme in a MIMO-OFDM system under unknown noise conditions, comprising the following steps: The first step involves designing a feature extraction mechanism based on cyclic moments for the received MIMO-OFDM signal. This involves calculating cyclic moments under multiple time delay conditions and uniformly projecting the cyclic features onto the feature space. Utilizing the differences in peak distribution at the cyclic frequency among different STBCs, a needle-like feature map of cyclic peaks is constructed. The main purpose of this step is to extract the spatiotemporal coding features of the signal, including: First, regarding the received signal Extracting different delays and Fourth-order time-varying moment under typical time delay parameters And perform Fourier transform on it to extract the fourth-order cyclic moments. .

[0043] Then, and The cyclic moments obtained under time delay conditions are stacked along the time delay dimension to form a three-dimensional feature with time delay as the vertical axis. This three-dimensional feature is then mapped to its maximum value along the cyclic frequency axis to achieve a centralized representation of multi-time-delay cyclic peak information. The mapping process can be expressed as follows:

[0044]

[0045]

[0046] in, The magnitude of the cyclic moment under each time delay. The needle-like characteristics of the cyclic moment peak of various STBC-OFDM signals are represented.

[0047] The second step introduces a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. By utilizing the cyclic stationarity of modulation symbols, the multi-kernel concept is incorporated into cyclic correlation entropy, and combined with threshold smoothing processing, a multi-kernel cyclic smoothed correlation entropy spectral feature map is extracted, including: First, the correlation entropy of the received signal is calculated, and then the cyclic correlation entropy is obtained by Fourier series expansion. The correlation entropy is constructed using a Gaussian kernel function, and the kernel width parameter is used to adjust the contribution of different order statistical information in the feature representation. Next, the concept of multi-kernel is introduced, and Gaussian kernel functions with different kernel widths are weighted and combined to construct multi-kernel correlation entropy:

[0048] in, and Indicated by and The Gaussian kernel function with kernel width, For kernel weight coefficients, , These represent the demodulated first and second digits of ofdm. Time and The received signal at any given moment.

[0049] Then, by performing a Fourier series expansion, the expression for the multi-kernel cyclic correlation entropy is obtained:

[0050] in, The cycle frequency, Indicates the cycle frequency index. Indicates frequency, This represents the period of the correlation entropy function.

[0051] Subsequently, the multi-kernel cyclic correlation entropy is threshold-smoothed to reduce the interference of unknown noise and irrelevant components on the spectral peak distribution. The zero-cycle frequency axis and the zero-frequency axis are set to zero, retaining only the maximum peak value at the origin. This reduces the feature dimension and complexity while preserving effective information such as peak amplitude and position, allowing for the extraction of the multi-kernel cyclic smoothed correlation entropy spectrum.

[0052] in, This is represented as the parameter for determining the threshold of the spectral peak.

[0053] The third step involves designing a joint recognition mechanism based on complementary gating fusion. This mechanism uses the two types of feature maps mentioned above as network input to construct a dual-branch convolutional neural network incorporating the complementary gating fusion mechanism. This network adaptively weights and fuses deep features, and then implements joint recognition within a multi-task parallel classification framework. The mechanism includes: First, the cyclic peak needle-like feature map and the multi-kernel cyclic smooth correlation entropy spectrum feature map are input to two LeNet single-branch networks. The high-level features extracted by the two single-branch networks are then processed. and The two types of features are concatenated as input to form a joint feature vector. Subsequently, a nonlinear dependency relationship between channels is established using a "fully connected layer-ReLU-fully connected layer" structure. Finally, the sigmoid function is used to generate normalized channel weights, thereby achieving adaptive selection of gating weights and generating gating weight vectors corresponding to the two branches. The expression is:

[0054] in, These represent the gating weights for the corresponding branch features. and These are two fully connected layers of the complementary gating fusion module.

[0055] After calculating the gate control weights, the high-level features of the two branches are then weighted and fused, as expressed in the following expression:

[0056] Gated weights are used to model the correlation between different feature channels. This fusion mechanism can adaptively adjust the contribution of each branch feature in the joint modeling, effectively suppress the interference of unreliable features on the joint classification results under low signal-to-noise ratio conditions, reduce the impact of noise and channel uncertainties on feature representation, and thus improve the robustness and stability of the model in complex environments.

[0057] Subsequently, STBC-OFDM classification and modulation scheme classification are used as auxiliary tasks, forming a multi-task learning framework together with the joint classification task. A joint optimization objective is constructed by weighted summation of the loss functions of each task, thereby ensuring the stability of network training. The overall loss function is defined as:

[0058] in, , , Let represent the loss functions for STBC-OFDM, modulation scheme, and joint classification task, respectively. , , These represent the corresponding loss weights.

[0059] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0060] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0061] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is defined only by the appended claims.

Claims

1. A method for joint identification of space-time coding and modulation scheme in a MIMO-OFDM system under unknown noise, characterized in that, The method includes: Step 1: Design a feature extraction mechanism based on cyclic moments. Calculate cyclic moments under multiple time delay conditions and project cyclic features uniformly onto the feature space. Utilize the differences in peak distribution at cyclic frequencies of different STBCs to construct a needle-like feature map of cyclic peaks. Step 2: Introduce a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. Utilize the cyclic stationarity of modulation symbols to introduce the multi-kernel concept into cyclic correlation entropy. Combine this with threshold smoothing to extract the multi-kernel cyclic smoothing correlation entropy spectral feature map. Step 3: Design a joint recognition mechanism based on complementary gating fusion. Use the two types of feature maps mentioned above as network inputs to construct a dual-branch convolutional neural network that incorporates the complementary gating fusion mechanism. Perform adaptive weighted fusion of deep features and achieve joint recognition under a multi-task parallel classification framework.

2. The method according to claim 1, characterized in that, The construction of the cyclic peak needle-like feature map in step one specifically includes: The fourth-order time-varying moment under the delay parameter is extracted from the received signal, and the fourth-order cyclic moment is extracted by Fourier transform. The cyclic moments obtained under different time delay conditions are stacked along the time delay dimension to form a three-dimensional feature with time delay as the vertical axis; The three-dimensional features are mapped to their maximum values ​​along the cyclic frequency axis to achieve a centralized representation of multi-delay cyclic peak information, resulting in a cyclic peak needle-shaped feature map.

3. The method according to claim 2, characterized in that, The maximum value mapping of the three-dimensional feature along the cyclic frequency axis is expressed as follows: in, The magnitude of the cyclic moment under each time delay. The needle-like characteristics of the cyclic moment peak of various STBC-OFDM signals are represented. It is a fourth-order cyclic moment.

4. The method according to claim 1, characterized in that, Step two, specifically, involves extracting the multi-kernel cyclic smoothing correlation entropy spectrum feature map, which includes: The correlation entropy of the received signal is calculated using a Gaussian kernel function, and the cyclic correlation entropy is obtained through Fourier series expansion. We construct multi-kernel correlation entropy by weighted combination of Gaussian kernel functions with different kernel widths, and further obtain multi-kernel cyclic correlation entropy through Fourier series expansion. Threshold smoothing is applied to the multi-kernel cyclic correlation entropy to reduce the interference of unknown noise and irrelevant components on the spectral peak distribution, and the multi-kernel cyclic smoothed correlation entropy spectral feature map is extracted.

5. The method according to claim 4, characterized in that, The process of weighting and combining Gaussian kernel functions with different kernel widths to construct a multi-kernel correlation entropy is as follows: in, and Indicated by and The Gaussian kernel function with kernel width, For kernel weight coefficients, , These represent the demodulated first and second digits of ofdm. Time and The received signal at any given moment.

6. The method according to claim 5, characterized in that, The threshold smoothing process for the multi-core cyclic correlation entropy is specifically as follows: in, This is represented as a parameter for determining the threshold of spectral peaks. The cycle frequency, Indicates the cycle frequency index. Indicates frequency, This represents the period of the correlation entropy function.

7. The method according to claim 1, characterized in that, Step three, which describes the implementation of joint identification, specifically includes: The cyclic peak needle-shaped feature map and the multi-kernel cyclic smooth correlation entropy spectrum feature map are respectively input into a dual-branch convolutional neural network to extract high-level feature representations; A complementary gated feature fusion module is introduced to concatenate the high-level features of the dual-branch outputs. A gated weight vector is generated through a "fully connected layer-ReLU-fully connected layer" structure to perform adaptive weighted fusion of the dual-branch features. A multi-task parallel classification module is constructed, with STBC-OFDM classification and modulation scheme classification as auxiliary tasks. Together with the joint classification task, they form a multi-task learning framework. The loss functions of each task are weighted and summed to form a joint optimization objective, thereby achieving joint recognition.

8. A joint identification system for space-time coding and modulation scheme of a MIMO-OFDM system under unknown noise, characterized in that, The system includes: The space-time coding feature extraction module designs a feature extraction mechanism based on cyclic moments. By calculating cyclic moments under multiple time delay conditions and uniformly projecting cyclic features to the feature space, the module utilizes the differences in peak distribution of different STBCs at the cyclic frequency to construct a cyclic peak needle-like feature map. The modulation mode feature extraction module introduces a modulation feature extraction mechanism based on multi-kernel cyclic correlation entropy. By utilizing the cyclic stationarity of modulation symbols, the multi-kernel concept is introduced into the cyclic correlation entropy, and combined with threshold smoothing processing, the multi-kernel cyclic smoothing correlation entropy spectral feature map is extracted. The spatiotemporal coding and modulation scheme joint recognition module designs a joint recognition mechanism based on complementary gating fusion. The two types of feature maps mentioned above are used as network inputs to construct a dual-branch convolutional neural network that incorporates the complementary gating fusion mechanism. The deep features are adaptively weighted and fused, and joint recognition is achieved under a multi-task parallel classification framework.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for joint identification of space-time coding and modulation scheme of MIMO-OFDM system under unknown noise as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the joint identification method of space-time coding and modulation scheme of MIMO-OFDM system under unknown noise as described in any one of claims 1 to 7 by executing the executable instructions.

Images