Artificial intelligence-based 6g phased array antenna receiving and transmitting group

Abstract

The application discloses to the technical field of communication, specifically 6G phased array antenna group based on artificial intelligence, first through multi-unit channel, environmental data acquisition and fusion processing, output standardization characteristic vector, then undertake intelligent sensing scheduling module local data, rely on block chain identity authentication, complete star earth fusion link data interaction and federal learning computing power scheduling, then according to the global data of cross-domain group cooperation module, through CNN-GPT2 hybrid prediction, extract high-dimensional features of channel, capture long-range dependence, and generate and pre-train fine-tuning with dynamic codebook, output optimal beam parameter, based on this, the application effectively avoids the problem that the cost and power consumption in the traditional phased array antenna group increase sharply with the expansion of array size, greatly improves the near-field beam prediction effect, reduces the pilot overhead, and solves the contradiction between the codebook dimension and the time slot constraint in the traditional beam search.

Description

Technical Field

[0001] This invention relates to the field of communication technology, specifically to a 6G phased array antenna transmitting and receiving group based on artificial intelligence. Background Technology

[0002] As 6G communication evolves towards ubiquitous connectivity and holographic communication, ultra-large-scale MIMO technology has become a core support, and phased array antenna transceiver groups are key components for achieving efficient signal transmission. Current 6G communication demands ultra-high bandwidth, low latency, and wide coverage from antennas, making millimeter wave and terahertz bands inevitable choices due to their abundant spectrum resources. However, traditional phased array antenna transceiver groups face architectural bottlenecks in their technological evolution.

[0003] Existing phased array antennas have evolved from the era of mechanical scanning to the era of fully digital beamforming, but in the 6G ultra-large-scale array scenario, traditional architectures struggle to overcome physical limitations. While new technologies such as holographic metasurfaces and gallium nitride materials offer solutions, they lack deep integration with artificial intelligence, failing to achieve global intelligent optimization of the transmitting and receiving groups. Specifically, the following problems exist:

[0004] 1. Low near-field beam prediction accuracy and high pilot overhead.

[0005] In near-field scenarios, 6G Very Large Scale Antenna Arrays (ELAAs) require a combined angle and distance two-dimensional beam search, leading to codebook dimensionality explosion. Traditional solutions using beam-by-beam scanning require over 1000 pilot symbols, which cannot meet time slot constraints. For example, when a low-altitude drone flies at 110 km / h, even a slight user displacement causes drastic beam index jumps, resulting in a prediction accuracy of less than 70% for traditional RNN models and a signal drop rate exceeding 30%.

[0006] 2. Poor unit consistency and wide range of fault impact.

[0007] Due to limitations in materials and manufacturing processes, antenna elements suffer from inconsistent electromagnetic responses, making real-time compensation difficult with traditional calibration algorithms. For example, the non-uniformity of gallium nitride wafers can cause phase errors of over 15° in phased array elements, resulting in main lobe gain losses exceeding 5dB. Furthermore, element failures can trigger a chain reaction; in one satellite communication system, a single T / R module failure led to the interruption of communication across the entire subarray, requiring over 24 hours of downtime for repair. Summary of the Invention

[0008] To address the aforementioned technical problems of low near-field beam prediction accuracy, high pilot overhead, poor element consistency, and wide fault impact range, this invention provides the following technical solution:

[0009] The AI-based 6G phased array antenna transmitting and receiving group includes:

[0010] The intelligent sensing and scheduling module is used to collect and process channel and environmental data through multiple units, and output a standardized feature vector.

[0011] The cross-domain group collaboration module is used to receive local data from the intelligent sensing and scheduling module. Relying on blockchain identity authentication, it completes data interaction of the space-ground fusion link and federated learning computing power scheduling.

[0012] The AI ​​beamforming calculation module is used to extract high-dimensional channel features and capture long-range dependencies based on the full-domain data of the cross-domain group collaboration module through CNN-GPT2 hybrid prediction, and output the optimal beam parameters in conjunction with dynamic codebook generation and pre-training fine-tuning.

[0013] The multimodal service adaptation module receives general parameters from the AI ​​beamforming calculation module, intelligently identifies and extracts core requirements through service type, customizes beam parameter sets accordingly, and then performs real-time fine-tuning instructions with QoS dynamic protection.

[0014] The self-adaptive unit management module is used to complete gallium nitride intelligent tuning, holographic metasurface driving and dynamic energy allocation according to the customized instructions of the multi-modal service adaptation module.

[0015] The intelligent fault diagnosis module is used to monitor the operating status of the self-adaptive unit control module and the preceding modules, complete fault warning and repair, and feed back deviation data to the front-end module.

[0016] As a preferred embodiment of the AI-based 6G phased array antenna receiving and transmitting group described in this invention, the intelligent sensing and scheduling module includes:

[0017] The channel multidimensional detection unit is used to acquire space-frequency samples by combining wide-beam analog precoding and frequency-varying digital precoding.

[0018] Environmental parameter acquisition unit, used for real-time monitoring of environmental indicators;

[0019] The multi-source data fusion unit is used to integrate channel data and environmental data through the Kalman filter algorithm and output a standardized feature vector.

[0020] As a preferred embodiment of the AI-based 6G phased array antenna transmitting and receiving group described in this invention, the cross-domain group coordination module includes:

[0021] The group identity authentication unit is used to assign a unique encrypted identifier to each antenna receiving and transmitting group using blockchain distributed ledger technology, so as to achieve secure identity verification of groups across regions and scenarios.

[0022] The global data interaction unit is used to build a low-latency data transmission channel through the satellite-ground fusion link, and to aggregate channel data and environmental data from multiple local intelligent sensing and scheduling modules to form a multi-dimensional data map covering the entire domain.

[0023] The distributed computing scheduling unit is used to dynamically allocate computing tasks to each group according to the federated learning framework.

[0024] As a preferred embodiment of the AI-based 6G phased array antenna transmitting and receiving group described in this invention, the AI ​​beamforming calculation module includes:

[0025] The CNN-GPT2 hybrid prediction unit is used to extract high-dimensional channel features based on the convolutional neural network and capture long-range dependencies across multiple frames through the GPT2 architecture, directly outputting the optimal near-field codeword index.

[0026] The dynamic codebook generation unit is used to dynamically adjust the codebook dimensions according to the user's movement speed;

[0027] Pre-trained fine-tuning units are used to pre-train the model using channel samples.

[0028] As a preferred embodiment of the AI-based 6G phased array antenna transmitting and receiving group described in this invention, the multi-mode service adaptation module includes:

[0029] The service type identification unit is used to analyze the service characteristics of the access signal in real time based on deep learning algorithms, accurately identify different service types, and extract the core requirement indicators of the service.

[0030] The beam parameter customization unit is used to adjust the basic parameters output by the AI ​​beamforming calculation module in reverse according to business needs.

[0031] The QoS dynamic assurance unit is used to monitor the quality of service transmission in real time, so as to dynamically adjust the control instructions when service requirements change or the network environment fluctuates, and ensure that the quality of service always meets the preset standards.

[0032] As a preferred embodiment of the AI-based 6G phased array antenna transmitting and receiving group described in this invention, the self-adaptive unit management module includes:

[0033] Gallium nitride smart tuning unit, used to achieve phase adjustment using gallium nitride Schottky diode;

[0034] Holographic metasurface driving unit is used to load PIN diodes to achieve real-time reconstruction of electromagnetic response;

[0035] The power consumption dynamic allocation unit is used to allocate power consumption according to beam priority.

[0036] As a preferred embodiment of the AI-based 6G phased array antenna transmitting and receiving group described in this invention, the intelligent fault diagnosis module includes:

[0037] The unit deviation detection unit is used to dynamically compensate for amplitude / phase differences caused by wafer non-uniformity based on the differential evolution algorithm;

[0038] Predictive maintenance units are used to analyze operational data through LSTM networks and provide early warnings of potential faults.

[0039] A redundancy switching unit is used to activate the backup unit in the event of a failure.

[0040] Compared with existing technologies:

[0041] This invention effectively avoids the problem of soaring costs and power consumption with array size expansion in traditional phased array antenna transceiver groups, significantly improves near-field beam prediction performance, reduces pilot overhead, and resolves the contradiction between codebook dimension and time slot constraints in traditional beam search. Simultaneously, it enhances the electromagnetic response consistency of antenna elements through an intelligent dynamic compensation mechanism, avoiding chain reactions caused by single-element failures and significantly reducing the impact of failures on the overall system. Relying on a more adaptable structural design and adaptive parameter adjustment capabilities, it breaks through the limitations of traditional phased arrays in terms of size and weight, enhancing deployment flexibility and conformal adaptation capabilities. It achieves seamless adaptation to different application scenarios such as low-altitude communication, satellite internet, and base stations in densely populated urban areas, comprehensively optimizing the overall operational performance of phased array antenna transceiver groups in terms of cost control, signal transmission, operational reliability, and scenario adaptability. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the overall framework of the present invention;

[0043] Figure 2 This is a schematic diagram of the intelligent sensing and scheduling module framework of the present invention;

[0044] Figure 3 This is a schematic diagram of the cross-domain group collaboration module framework of the present invention;

[0045] Figure 4 This is a schematic diagram of the AI ​​beamforming calculation module framework of the present invention;

[0046] Figure 5 This is a schematic diagram of the multimodal service adaptation module framework of the present invention;

[0047] Figure 6 This is a schematic diagram of the self-adaptive unit control module framework of the present invention;

[0048] Figure 7 This is a schematic diagram of the fault intelligent diagnosis module framework of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0050] This invention provides an artificial intelligence-based 6G phased array antenna transceiver group. Please refer to [link / reference]. Figure 1 ,include:

[0051] The intelligent sensing and scheduling module is used to collect and process channel and environmental data through multiple units, and output a standardized feature vector.

[0052] The cross-domain group collaboration module is used to receive local data from the intelligent sensing and scheduling module. Relying on blockchain identity authentication, it completes data interaction of the space-ground fusion link and federated learning computing power scheduling.

[0053] The AI ​​beamforming calculation module is used to extract high-dimensional channel features and capture long-range dependencies based on the full-domain data of the cross-domain group collaboration module through CNN-GPT2 hybrid prediction, and output the optimal beam parameters in conjunction with dynamic codebook generation and pre-training fine-tuning.

[0054] The multimodal service adaptation module receives general parameters from the AI ​​beamforming calculation module, intelligently identifies and extracts core requirements through service type, customizes beam parameter sets accordingly, and then performs real-time fine-tuning instructions with QoS dynamic protection.

[0055] The self-adaptive unit management module is used to complete gallium nitride intelligent tuning, holographic metasurface driving and dynamic energy allocation according to the customized instructions of the multi-modal service adaptation module.

[0056] The intelligent fault diagnosis module is used to monitor the operating status of the self-adaptive unit control module and the preceding modules, complete fault warning and repair, and feed back deviation data to the front-end module.

[0057] Please see Figure 2 The intelligent sensing and scheduling module includes:

[0058] The channel multidimensional detection unit is used to acquire space-frequency samples by combining wide-beam analog precoding and frequency-varying digital precoding, thereby reducing pilot overhead by an order of magnitude.

[0059] The environmental parameter acquisition unit is used to monitor environmental indicators such as electromagnetic interference intensity, temperature, and wind speed in real time, with a sampling frequency of up to 200MHz.

[0060] The multi-source data fusion unit is used to integrate channel data and environmental data through the Kalman filter algorithm and output a standardized feature vector.

[0061] Please see Figure 3The cross-domain group collaboration module includes:

[0062] The group identity authentication unit uses blockchain distributed ledger technology to assign a unique encrypted identifier to each antenna receiving and transmitting group, enabling secure identity verification of groups across regions and scenarios, and preventing signal interference and data leakage caused by unauthorized group access.

[0063] The global data interaction unit is used to build a low-latency data transmission channel through the satellite-ground fusion link, and to aggregate channel data and environmental data from multiple local intelligent sensing and scheduling modules to form a multi-dimensional data map covering the entire domain, supporting global beam optimization calculation.

[0064] The distributed computing power scheduling unit is used to dynamically allocate computing tasks to each group according to the federated learning framework, splitting complex beam computing tasks into groups with redundant computing power to avoid overload of a single group, while protecting the data privacy of each group without uploading original sensitive data.

[0065] Please see Figure 4 The AI ​​beamforming calculation module includes:

[0066] The CNN-GPT2 hybrid prediction unit is used to extract high-dimensional channel features based on the convolutional neural network and capture long-range dependencies across multiple frames through the GPT2 architecture, directly outputting the optimal near-field codeword index.

[0067] The dynamic codebook generation unit is used to dynamically adjust the codebook dimensions according to the user's movement speed, maintaining codebook adaptability in the speed range of 30-110km / h.

[0068] The pre-trained fine-tuning unit is used to pre-train the model with 80k standard channel samples and fine-tune it with 20k scene samples. It can achieve a prediction accuracy of over 85% in 5 epochs.

[0069] Please see Figure 5 The multimodal service adaptation module includes:

[0070] The service type identification unit is used to analyze the service characteristics of the access signal in real time based on deep learning algorithms, accurately identify different service types such as holographic communication, high-definition video, and IoT sensing, and extract core requirement indicators such as latency, bandwidth, and reliability of the service.

[0071] The beam parameter customization unit is used to adjust the basic parameters output by the AI ​​beamforming calculation module in reverse according to business needs, such as optimizing beam bandwidth and gain for holographic communication services, optimizing beam coverage and energy consumption for IoT services, and generating a dedicated beam control parameter set.

[0072] The QoS dynamic assurance unit is used to monitor the quality of service transmission in real time. When service requirements change (such as holographic communication switching from static to dynamic) or the network environment fluctuates, it dynamically fine-tunes the control instructions to ensure that the quality of service always meets the preset standards.

[0073] Please see Figure 6 The self-adaptive unit management module includes:

[0074] Gallium nitride intelligent tuning unit, using gallium nitride Schottky diode, to achieve continuous phase adjustment from 0-210° with a phase error as low as 1.8°;

[0075] A holographic metasurface driving unit is used to load PIN diodes to achieve real-time reconstruction of electromagnetic response, compressing the antenna thickness to less than 0.5mm and supporting conformal deployment;

[0076] The dynamic power allocation unit is used to allocate power consumption according to beam priority, reducing power consumption of non-core area units by 60%.

[0077] Please see Figure 7 The intelligent fault diagnosis module includes:

[0078] The unit deviation detection unit is used to dynamically compensate for amplitude / phase differences caused by wafer non-uniformity according to the differential evolution algorithm, and the main lobe gain is improved by 4.2dB.

[0079] Predictive maintenance unit is used to analyze operational data through LSTM network and provide early warning of potential faults up to 2 hours in advance;

[0080] The redundancy switching unit is used to activate the backup unit within 1ms when a fault occurs, so as to control the impact of the fault to a single unit.

[0081] In practical use, the specific process is as follows:

[0082] 1. Data Acquisition Phase (Intelligent Sensing and Scheduling Module)

[0083] Unit Actions: The channel multi-dimensional detection unit collects spatial-frequency samples through wide beam + frequency-variable digital precoding; the environmental parameter acquisition unit simultaneously monitors 12 indicators such as electromagnetic interference and temperature; and the multi-source data fusion unit integrates the two types of data using the Kalman filter algorithm.

[0084] Output: Standardized feature vectors (providing basic data for global collaboration).

[0085] 2. Global Integration Phase (Cross-Domain Group Collaboration Module)

[0086] Unit actions: The group identity authentication unit completes multi-group security verification through blockchain technology; the global data interaction unit aggregates the perception data of each group through the space-ground fusion link; and the distributed computing power scheduling unit splits computing tasks based on federated learning.

[0087] Output: Global collaborative data + balanced computing power allocation scheme (to expand the global perspective of beam computing).

[0088] 3. Decision-making calculation stage (AI beamforming calculation module)

[0089] Unit actions: The CNN-GPT2 hybrid prediction unit extracts high-dimensional features from the global data and captures long-range dependencies; the dynamic codebook generation unit adjusts the codebook dimension according to the user's movement state; and the pre-training fine-tuning unit optimizes the model parameters through scene samples.

[0090] Output: Generalized optimal beam parameters (providing core decision-making basis for service adaptation).

[0091] 4. Business Adaptation Phase (Multimodal Business Adaptation Module)

[0092] Unit Actions: The service type identification unit analyzes signal characteristics, accurately identifies service types such as holographic communication and Internet of Things, and extracts latency and bandwidth requirements; the beam parameter customization unit adjusts general beam parameters in a targeted manner; and the QoS dynamic guarantee unit monitors transmission quality in real time.

[0093] Output: Customized business control instructions (bridging the gap between general decision-making and hardware execution).

[0094] 5. Execution and Control Phase (Self-Adaptive Unit Control Module)

[0095] Unit Actions: The gallium nitride intelligent tuning unit achieves continuous phase adjustment from 0 to 210°; the holographic metasurface driving unit reconstructs the electromagnetic response through PIN diodes to adapt to the deployment form; and the power consumption dynamic allocation unit allocates power consumption according to beam priority.

[0096] Output: Precise operating status of the antenna unit (translating upper-layer commands into actual transmission actions).

[0097] 6. Stable Feedback Phase (Intelligent Fault Diagnosis Module)

[0098] Unit Actions: The unit deviation detection unit compensates for electromagnetic response consistency deviations through differential evolution algorithm; the predictive maintenance unit analyzes operational data based on LSTM network to warn of potential faults; and the redundancy switching unit activates the backup unit within 1ms in the event of a fault.

[0099] Two-way feedback: On the one hand, it outputs fault repair instructions to ensure the stability of the current process; on the other hand, it pushes deviation data and fault information back to the intelligent perception and scheduling module (to optimize data acquisition strategy) and the AI ​​beamforming calculation module (to adjust beam parameters), forming a closed loop.

[0100] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A 6G phased array antenna transmitting and receiving group based on artificial intelligence, characterized in that, include: The intelligent sensing and scheduling module is used to collect and process channel and environmental data through multiple units, and output a standardized feature vector. The cross-domain group collaboration module is used to receive local data from the intelligent sensing and scheduling module. Relying on blockchain identity authentication, it completes data interaction of the space-ground fusion link and federated learning computing power scheduling. The AI ​​beamforming calculation module is used to extract high-dimensional channel features and capture long-range dependencies based on the full-domain data of the cross-domain group collaboration module through CNN-GPT2 hybrid prediction, and output the optimal beam parameters in conjunction with dynamic codebook generation and pre-training fine-tuning. The multimodal service adaptation module receives general parameters from the AI ​​beamforming calculation module, intelligently identifies and extracts core requirements through service type, customizes beam parameter sets accordingly, and then performs real-time fine-tuning instructions with QoS dynamic protection. The self-adaptive unit management module is used to complete gallium nitride intelligent tuning, holographic metasurface driving and dynamic energy allocation according to the customized instructions of the multi-modal service adaptation module. The intelligent fault diagnosis module is used to monitor the operating status of the self-adaptive unit control module and the preceding modules, complete fault warning and repair, and feed back deviation data to the front-end module.

2. The 6G phased array antenna transmitting and receiving group based on artificial intelligence according to claim 1, characterized in that, The intelligent sensing and scheduling module includes: The channel multidimensional detection unit is used to acquire space-frequency samples by combining wide-beam analog precoding and frequency-varying digital precoding. Environmental parameter acquisition unit, used for real-time monitoring of environmental indicators; The multi-source data fusion unit is used to integrate channel data and environmental data through the Kalman filter algorithm and output a standardized feature vector.

3. The 6G phased array antenna transmitting and receiving group based on artificial intelligence according to claim 1, characterized in that, The cross-domain group collaboration module includes: The group identity authentication unit is used to assign a unique encrypted identifier to each antenna receiving and transmitting group using blockchain distributed ledger technology. The global data interaction unit is used to establish a low-latency data transmission channel through the space-ground fusion link and to aggregate channel data and environmental data from multiple local intelligent sensing and scheduling modules. The distributed computing scheduling unit is used to dynamically allocate computing tasks to each group according to the federated learning framework.

4. The 6G phased array antenna transmitting and receiving group based on artificial intelligence according to claim 1, characterized in that, The AI ​​beamforming calculation module includes: The CNN-GPT2 hybrid prediction unit is used to extract high-dimensional channel features based on the convolutional neural network and capture long-range dependencies across multiple frames through the GPT2 architecture, directly outputting the optimal near-field codeword index. The dynamic codebook generation unit is used to dynamically adjust the codebook dimensions according to the user's movement speed; Pre-trained fine-tuning units are used to pre-train the model using channel samples.

5. The 6G phased array antenna transmitting and receiving group based on artificial intelligence according to claim 1, characterized in that, The multimodal service adaptation module includes: The service type identification unit is used to analyze the service characteristics of the access signal in real time based on deep learning algorithms, identify different service types, and extract the core requirement indicators of the service. The beam parameter customization unit is used to adjust the basic parameters output by the AI ​​beamforming calculation module in reverse according to business needs. The QoS dynamic protection unit is used to monitor the quality of service transmission in real time, so as to dynamically fine-tune the control instructions when service requirements change or the network environment fluctuates.

6. The 6G phased array antenna transmitting and receiving group based on artificial intelligence according to claim 1, characterized in that, The self-adaptive unit management module includes: Gallium nitride smart tuning unit, used to achieve phase adjustment using gallium nitride Schottky diode; Holographic metasurface driving unit is used to load PIN diodes to achieve real-time reconstruction of electromagnetic response; The power consumption dynamic allocation unit is used to allocate power consumption according to beam priority.

7. The 6G phased array antenna transmitting and receiving group based on artificial intelligence according to claim 1, characterized in that, The intelligent fault diagnosis module includes: The unit deviation detection unit is used to dynamically compensate for amplitude / phase differences caused by wafer non-uniformity based on the differential evolution algorithm; Predictive maintenance units are used to analyze operational data through LSTM networks and provide early warnings of potential faults. A redundancy switching unit is used to activate the backup unit in the event of a failure.

Images