A fluid antenna port switching method and device, a terminal device and a storage medium

Abstract

The application discloses a kind of fluid antenna port switching method, device, terminal equipment and storage medium, belong to fluid antenna port switching field, method is: candidate fluid antenna configuration set is input to target fluid antenna performance prediction model, to obtain original predicted performance distribution set;Based on pre-set conformal error coverage and historical prediction deviation set, conformal calibration factor is obtained, to scale original predicted performance distribution set to obtain calibrated predicted performance distribution set;According to candidate fluid antenna configuration set and historical fluid antenna configuration, mechanical switching energy consumption penalty term set is obtained;According to calibrated predicted performance distribution set, obtain potential promotion expectation set, and based on potential promotion expectation set and mechanical switching energy consumption penalty term set, obtain net benefit score set;Based on net benefit score set, target fluid antenna configuration is selected from candidate fluid antenna configuration set, to carry out port switching, to improve the reliability of fluid antenna port switching.

Description

Technical Field

[0001] This invention relates to the field of fluid antenna port switching, and more particularly to a fluid antenna port switching method, apparatus, terminal device, and storage medium. Background Technology

[0002] Integrated Sensing and Communications (ISAC) enables communication links and radar / sensing functions to operate jointly within the same physical layer architecture by sharing wireless spectrum, hardware platforms, and signal waveforms. This has become a key paradigm for achieving extremely high transmission rates and centimeter-level sensing resolution in future 6G wireless communication networks. To improve spatial freedom and environmental adaptability, Fluid Antenna Systems (FAS) have been proposed for dynamically moving or selecting radiation positions within a limited physical aperture, thereby altering the equivalent spatial channel and improving communication and sensing performance. However, existing integrated sensing and communications solutions for fluid antennas typically simplify the antenna port switching process to an instantaneous switch, or treat it as a pure interruption overhead, measurement delay, or energy penalty. This simplification leads to the system over-reliance on the instantaneous performance of the target configuration when selecting ports, neglecting the communication and sensing contributions during the switching process and their impact on time-slot-level performance. This easily results in frequent and low-return port switching, ultimately weakening the sensing gain brought by the spatial flexibility of fluid antennas.

[0003] To address the port switching problem in fluid antenna-assisted sensing integrated systems, three main processing strategies exist, but all suffer from significant application bottlenecks: The first type of approach is based on instantaneous performance-based port selection. This typically selects the port with the best instantaneous performance based on the communication rate, sensing accuracy, or overall sensing performance of each candidate antenna port configuration at the current moment. This method simplifies the port switching process to instantaneous or cost-free switching, focusing only on the performance of the target port configuration in a steady state, while ignoring the actual switching time, movement energy consumption, and changes in sensing performance during the switching process. Therefore, it is prone to overly greedy port switching behavior, where the system frequently moves to port positions with higher instantaneous performance, but the long-term benefits may be offset by the switching overhead. The second type of approach is based on cost-aware methods of switching penalties or movement delays, beginning to consider the impact of fluid antennas or movable antennas on port switching. Physical overhead is a concern, but existing methods mostly treat the handover phase simply as an energy penalty, measurement delay, or interruption overhead of moving before transmitting, without further characterizing the communication and sensing signals that can still be used during port movement. Therefore, although handover costs are considered, the sensing utility inherent in the handover process itself is not fully utilized, resulting in an incomplete characterization of time-slot-level performance. The third type of scheme is based on online configuration search methods using traditional black-box optimization or Bayesian optimization, searching for optimal fluid antenna configurations in a large-scale candidate port space to reduce the training and measurement overhead caused by exhaustive scanning. However, traditional Bayesian optimization methods usually assume that the objective function is relatively stable and mainly rely on the posterior mean and variance of the Gaussian process for decision-making. In dynamic air-to-ground sensing environments, user and target movement, residual self-interference changes, and action-related disturbances introduced by port handover can all lead to mismatch or overconfidence in the surrogate model, resulting in unstable or even erroneous handover decisions by the acquisition function. Therefore, existing technologies for port handover in fluid antenna-assisted sensing integrated systems all have varying degrees of unreliability. Summary of the Invention

[0004] This invention provides a fluid antenna port switching method, apparatus, terminal device, and storage medium, which can solve the above-mentioned problems in the prior art and improve the reliability of fluid antenna port switching.

[0005] This invention provides a method for switching ports of a fluid antenna, comprising: Obtain the set of candidate fluid antenna configurations and historical fluid antenna configurations; The candidate fluid antenna configuration set is input into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model; Obtain the set of historical prediction deviations corresponding to the target fluid antenna performance prediction model, and obtain a conformal calibration factor based on the preset conformal coverage error rate and the set of historical prediction deviations. Then, perform data scaling processing on the original prediction performance distribution set based on the conformal calibration factor to obtain a calibrated prediction performance distribution set. Based on the candidate fluid antenna configuration set and the historical fluid antenna configuration, a set of mechanical switching energy consumption penalty terms is obtained; Based on the calibration prediction performance distribution set, a potential improvement expectation set is obtained, and based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, a net benefit score set is obtained. Based on the net benefit score set, a target fluid antenna configuration for the current time slot is selected from the candidate fluid antenna configuration set, and port switching is performed based on the target fluid antenna configuration.

[0006] In the above scheme, by inputting the candidate fluid antenna configuration set into the target fluid antenna performance prediction model, an original predicted performance distribution set is obtained, and then a historical prediction deviation set corresponding to the target fluid antenna performance prediction model is obtained. A conformal calibration factor is then obtained based on the historical prediction deviation set and a preset conformal error coverage rate, so that the conformal calibration factor can reflect the overall level of the model's recent prediction deviation. The original predicted performance distribution set is then scaled using the conformal calibration factor to obtain a calibrated predicted performance distribution set, making the calibrated predicted performance distribution set correlated with the overall level of the model's recent prediction deviation reflected by the conformal calibration factor, thus more closely reflecting the model's recent actual prediction error level and improving the data reliability of the calibrated predicted performance distribution set. Finally, based on the candidate fluid antenna configuration set and the historical fluid antenna configuration set... The mechanical switching energy consumption penalty term is calculated to represent the penalty of mechanical energy consumption required to switch from the historical fluid antenna configuration to the candidate fluid antenna configuration set. A potential improvement expectation set is obtained based on the calibration prediction performance distribution set to represent the performance improvement potential of the candidate fluid antenna configuration set. Finally, the potential improvement expectation set after deducting the mechanical switching energy consumption penalty term set is obtained as the net benefit score set to quantify the comprehensive net benefit generated when switching between the candidate fluid antenna configuration set and the historical fluid antenna configuration. By selecting the configuration with the highest net benefit score as the target configuration for the current time slot for port switching, unnecessary port switching is avoided. This effectively maintains reliable port switching decision capability in dynamic air-to-ground sensing environments, thereby improving the reliability of fluid antenna port switching.

[0007] Furthermore, the acquisition of the candidate fluid antenna configuration set and historical fluid antenna configurations includes: Obtain the original set of candidate fluid antenna configurations and historical fluid antenna configurations; Based on the port movement distance and port liquid metal movement speed between any original candidate fluid antenna configuration and the set of original candidate fluid antenna configurations, the original candidate switching time for switching from the historical fluid antenna configuration to the original candidate fluid antenna configuration is obtained, and the set of original candidate switching times is obtained based on all the original candidate switching times corresponding to the set of original candidate fluid antenna configurations. Based on the original set of candidate switching times and the preset time slot length, the candidate fluid antenna configuration set is selected from the original set of candidate fluid antenna configurations.

[0008] In the above scheme, by acquiring the original set of candidate fluid antenna configurations and historical fluid antenna configurations, the original candidate switching time for switching from the historical fluid antenna configurations to each original candidate fluid antenna configuration is calculated based on the port movement distance and the port liquid metal movement speed. Based on the preset time slot length, candidate configurations that can complete the switching within the preset time slot length are selected from the original candidate set. This ensures that each candidate configuration participating in the subsequent evaluation and decision-making is a physically achievable switching target. This solves the problem in the prior art of simplifying port switching to instantaneous switching and ignoring the actual switching time constraint. At the same time, by eliminating the original candidate configurations that cannot complete the switching within one time slot, the feasibility of switching is ensured, and the subsequent decision-making is avoided from selecting configurations that cannot be actually executed due to insufficient switching time. This provides a reasonable and executable candidate space for subsequent prediction model evaluation, energy consumption calculation, and net benefit scoring, enabling the entire online decision-making process to operate normally under the actual switching time constraint.

[0009] Further, the step of inputting the candidate fluid antenna configuration set into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model includes: Acquire several switching pairs, wherein each switching pair includes a sample history fluid antenna configuration and a sample target fluid antenna configuration corresponding to the sample history fluid antenna configuration; For any one of the aforementioned switching pairs, obtain the historical observation data corresponding to that switching pair; A set of historical observation data is obtained based on all the historical observation data corresponding to several switching pairs; The target fluid antenna performance prediction model is obtained by training a preset fluid antenna performance prediction model based on several switching pairs and the historical observation dataset.

[0010] In the above scheme, several switching pairs containing sample historical fluid antenna configurations and corresponding sample target fluid antenna configurations are obtained, and corresponding historical observation data are obtained for each switching pair to form a historical observation dataset. Then, a preset fluid antenna performance prediction model is trained based on the switching pairs and the historical observation dataset. This allows the obtained target fluid antenna performance prediction model to learn the mapping relationship between port switching actions and sensing performance, using the switching pairs as input and the historical observation data as output labels. This solves the problem in the prior art that physical information such as communication channels, target responses, and residual self-interference is difficult to obtain accurately in advance in dynamic environments, making it impossible to directly calculate sensing performance. In other words, the model trained with historical data predicts the sensing performance of unknown candidate configurations without relying on a complete closed channel model or exhaustive scanning, providing a usable predictive model foundation for subsequent online decision-making.

[0011] Further, the step of obtaining the historical observation data corresponding to any one of the several handover pairs includes: The sample port movement path is obtained based on the sample history fluid antenna configuration and the sample target fluid antenna configuration in the switching pair; Based on the sample history fluid antenna configuration, a sample history precoding matrix is ​​obtained, and based on the sample history precoding matrix and the sample port movement path, the port handover communication rate and port handover sensing mutual information are obtained. Based on the sample target fluid antenna configuration, a sample target precoding matrix is ​​obtained, and based on the sample target precoding matrix, a stable dwell communication rate and a stable dwell sensing mutual information are obtained. The actual communication performance is obtained based on the port switching communication rate and the stable dwell communication rate. The actual sensing performance is obtained based on the port switching sensing mutual information and the stable dwell sensing mutual information. The historical observation data of the handover pair are obtained based on the actual communication performance and the actual perception performance.

[0012] In the above scheme, sample port movement paths are obtained by using the sample historical fluid antenna configuration and sample target fluid antenna configuration in each handover pair. Port handover communication rate and port handover sensing mutual information are calculated based on the sample historical precoding matrix and the sample port movement paths. Simultaneously, stable dwell communication rate and stable dwell sensing mutual information are calculated based on the sample target precoding matrix. Then, the communication rate and sensing mutual information of the handover phase and the dwell phase are combined to form the true communication performance and true sensing performance, respectively, as the historical observation data for that handover pair. The core of this scheme is that the synesthetic performance within a complete time slot is composed of the handover phase and the dwell phase, weighted by time, rather than only calculating the instantaneous performance of the dwell phase. This solves the problem in existing technologies that simply treat the port handover phase as instantaneous, with interruption overhead or energy penalty, ignoring the fact that communication and sensing signals can still be utilized during the handover process. This allows the historical observation data to truly reflect the actual synesthetic utility of a complete handover process, enabling the trained target fluid antenna performance prediction model to accurately learn the coupling relationship between the handover phase performance and the dwell phase performance, providing a data foundation for subsequent online prediction.

[0013] Further, a set of historical prediction deviations corresponding to the target fluid antenna performance prediction model is obtained, and a conformal calibration factor is obtained based on a preset conformal coverage error rate and the set of historical prediction deviations. Then, based on the conformal calibration factor, the original prediction performance distribution set is subjected to data scaling processing to obtain a calibrated prediction performance distribution set, including: The historical prediction deviation set is processed to obtain a standardized residual score set; The conformal calibration factor is obtained based on the preset conformal error coverage rate and the standardized residual score set; A calibration performance standard deviation set is obtained based on the conformal calibration factor and the predicted performance standard deviation set; The calibration prediction performance distribution set is obtained based on the set of calibration performance standard deviations and the set of predicted performance mean values.

[0014] In the above scheme, a standardized residual score set is obtained based on the historical prediction deviation set and the prediction performance standard deviation set. Then, a conformal calibration factor is calculated based on the preset conformal coverage error rate and the standardized residual score set. The conformal calibration factor is then used to correct the prediction performance standard deviation set to obtain a calibration performance standard deviation set. Finally, this is combined with the prediction performance mean set to obtain a calibration prediction performance distribution set. This dynamically corrects the original prediction uncertainty of the target fluid antenna performance prediction model. It solves the problem that in the dynamic air-to-ground sensing environment, user and target movement, residual self-interference changes, and motion-related disturbances introduced by port switching lead to inaccurate estimation of surrogate model uncertainty, and are prone to overconfidence and wrong decisions. This makes the potential improvement expectation calculated based on the calibrated prediction distribution reflect the true uncertainty level of the model, thereby effectively suppressing suboptimal or high-risk switching decisions caused by overconfidence and improving the reliability of online decision-making in dynamic environments.

[0015] Further, the step of obtaining the set of mechanical switching energy consumption penalty terms based on the candidate fluid antenna configuration set and the historical fluid antenna configurations includes: Based on the port movement distance and port liquid metal movement speed between any candidate fluid antenna configuration and the historical fluid antenna configuration set, a candidate switching time for switching from the historical fluid antenna configuration to the candidate fluid antenna configuration is obtained, and a candidate switching time set is obtained based on all the candidate switching times corresponding to the candidate fluid antenna configuration set. Based on the pre-acquired fluid antenna mobile power and the candidate switching time set, the set of mechanical switching energy consumption penalty terms is obtained.

[0016] In the above scheme, the candidate switching time for switching from the historical fluid antenna configuration to the candidate configuration is calculated based on the port movement distance and the port liquid metal movement speed between the historical fluid antenna configuration and the candidate fluid antenna configuration. A set of mechanical switching energy consumption penalty terms is calculated based on the pre-acquired fluid antenna movement power and the set of candidate switching times. This quantifies the physical energy consumed by port movement into a deterministic penalty value proportional to the candidate switching time. This solves the problem in existing technologies where only the sensing performance gain is considered while ignoring port movement delay and energy consumption, leading to frequent low-gain switching that weakens the spatial flexibility gain of the fluid antenna. The net gain score can achieve a quantitative trade-off between improved communication sensing performance and mechanical energy consumption costs, thus providing a basis for determining whether a switch is worthwhile. This effectively suppresses low-gain or unnecessary port movements, achieving a balance between long-term sensing performance and switching costs.

[0017] Further, the step of obtaining a potential improvement expectation set based on the calibration prediction performance distribution set, and obtaining a net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, includes: The set of calibration prediction performance distributions is sampled to obtain a set of sampled prediction performance vectors; Based on the sampled prediction performance vector set, the pre-acquired current approximate Pareto front, and the preset reference point, the potential improvement expectation set is obtained, and based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, the net benefit score set is obtained.

[0018] In the above scheme, a sampled prediction performance vector set is obtained by sampling the calibration prediction performance distribution set. Based on the sampled prediction performance vector set, the pre-acquired current approximate Pareto front, and the preset reference point, a potential improvement expectation set is calculated. Based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, a net benefit score set is obtained. Thus, the potential improvement of the candidate configuration in the two-dimensional targets of communication and sensing and the mechanical switching energy consumption cost are uniformly quantified into a comparable net benefit score. This solves the problems in the prior art of lacking a unified decision criterion for whether the switching is worthwhile and being unable to comprehensively judge communication gain, sensing gain, and switching energy consumption in a dynamic environment. It enables the selection of the target configuration based on the net benefit score set to select the candidate configuration with the highest comprehensive net benefit. While preserving the spatial degrees of freedom of the fluid antenna, it avoids the performance loss caused by excessive movement and maximizes the long-term online cost-sensing communication utility.

[0019] Another embodiment of the present invention also provides a fluid antenna port switching device, comprising: The basic data acquisition module is used to acquire the candidate fluid antenna configuration set and historical fluid antenna configurations; The original predicted performance distribution set acquisition module is used to input the candidate fluid antenna configuration set into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model. The calibration module is used to calibrate the original prediction performance distribution set based on the pre-acquired historical prediction deviation set to obtain a calibrated prediction performance distribution set. The mechanical switching energy consumption penalty term acquisition module is used to obtain the mechanical switching energy consumption penalty term set based on the candidate fluid antenna configuration set and the historical fluid antenna configuration; The net benefit score set acquisition module is used to obtain a potential improvement expectation set based on the calibration prediction performance distribution set, and to obtain a net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set. The port switching module is used to select the target fluid antenna configuration for the current time slot from the candidate fluid antenna configuration set based on the net benefit score set, so as to perform port switching based on the target fluid antenna configuration.

[0020] Another embodiment of the present invention provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the steps of a fluid antenna port switching method as described in the present invention.

[0021] Another embodiment of the present invention also provides a computer-readable storage medium item, including: a stored computer program, which, when the computer program is running, controls the device where the computer-readable storage medium is located to perform steps such as the fluid antenna port switching method of the present invention. Attached Figure Description

[0022] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a method for switching fluid antenna ports according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a fluid antenna port switching device according to an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0026] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0029] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0030] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0031] See Figure 1 To address the aforementioned problems in the prior art and improve the reliability of fluid antenna port switching, an embodiment of the present invention provides a fluid antenna port switching method, comprising: Step S1: Obtain the set of candidate fluid antenna configurations and historical fluid antenna configurations; Step S2: Input the candidate fluid antenna configuration set into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model; Step S3: Obtain the historical prediction deviation set corresponding to the target fluid antenna performance prediction model, obtain the conformal calibration factor based on the preset conformal coverage error rate and the historical prediction deviation set, and then perform data scaling processing on the original prediction performance distribution set based on the conformal calibration factor to obtain the calibrated prediction performance distribution set. Step S4: Based on the candidate fluid antenna configuration set and the historical fluid antenna configuration, obtain the set of mechanical switching energy consumption penalty terms; Step S5: Obtain the potential improvement expectation set based on the calibration prediction performance distribution set, and obtain the net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set; Step S6: Select the target fluid antenna configuration for the current time slot from the candidate fluid antenna configuration set based on the net benefit score set, and perform port switching based on the target fluid antenna configuration.

[0032] In the above scheme, by inputting the candidate fluid antenna configuration set into the target fluid antenna performance prediction model, an original predicted performance distribution set is obtained, and then a historical prediction deviation set corresponding to the target fluid antenna performance prediction model is obtained. A conformal calibration factor is then obtained based on the historical prediction deviation set and a preset conformal error coverage rate, so that the conformal calibration factor can reflect the overall level of the model's recent prediction deviation. The original predicted performance distribution set is then scaled using the conformal calibration factor to obtain a calibrated predicted performance distribution set, making the calibrated predicted performance distribution set correlated with the overall level of the model's recent prediction deviation reflected by the conformal calibration factor, thus more closely reflecting the model's recent actual prediction error level and improving the data reliability of the calibrated predicted performance distribution set. Finally, based on the candidate fluid antenna configuration set and the historical fluid antenna configuration set... The mechanical switching energy consumption penalty term is calculated to represent the penalty of mechanical energy consumption required to switch from the historical fluid antenna configuration to the candidate fluid antenna configuration set. A potential improvement expectation set is obtained based on the calibration prediction performance distribution set to represent the performance improvement potential of the candidate fluid antenna configuration set. Finally, the potential improvement expectation set after deducting the mechanical switching energy consumption penalty term set is obtained as the net benefit score set to quantify the comprehensive net benefit generated when switching between the candidate fluid antenna configuration set and the historical fluid antenna configuration. By selecting the configuration with the highest net benefit score as the target configuration for the current time slot for port switching, unnecessary port switching is avoided. This effectively maintains reliable port switching decision capability in dynamic air-to-ground sensing environments, thereby improving the reliability of fluid antenna port switching.

[0033] It should be noted that the operation of this fluid antenna port switching method relies on a fluid antenna-assisted sensing integrated system, which includes... One base station, Single-antenna communication users and There are n targets to be perceived, of which the nth target is... The position and reflectance of each target are expressed as follows: Each base station is equipped with a rotatable fluid antenna surface, on which a set of candidate ports is pre-set, denoted as . ,in Indicates the first One base station. The theoretical principles and model definitions will be introduced below: In each time slot within, no. Each base station selects from its candidate port set Select One transmission port and There are three receiving ports, one for downlink communication transmission and the other for sensing echo reception. To avoid the same port simultaneously handling both transmission and reception functions, the sets of transmitting and receiving ports must not overlap. Each base station in the time slot The selected transmit port coordinate matrix and receive port coordinate matrix are denoted as follows: and tx corresponds to the transmit port within the surface of the fluid antenna, and rx corresponds to the receive port on the surface of the fluid antenna. Considering that the surface of the fluid antenna is rotatable, its spatial attitude is determined by the rotation angle. This indicates that the entire system operates within a time slot. The fluid antenna configuration can be represented as: Based on this fluid antenna configuration Each base station transmits pilot signals and calculates the downlink precoding matrix using zero-forcing precoding based on feedback information. Then, communication data transmission and target perception tasks are completed.

[0034] It should be noted that this invention considers a discrete time-slot operation mechanism, where each time slot has a length of [missing information]. Unlike traditional methods that treat the port switching process as instantaneous, this invention further divides a time slot into two stages: the first stage is the port switching stage, and the second stage is the stable dwell stage. The port switching stage represents the fluid antenna port transitioning from the fluid antenna configuration of the previous time slot, i.e., the historical fluid antenna configuration. Fluid antenna configuration moved to the current time slot The process; the stable dwell phase indicates the target fluid antenna configuration where the port arrives at the current time slot. Then, stable communication and sensing processes are carried out under this configuration.

[0035] Let the velocity of the liquid metal moving in the fluid antenna be... , No. The first base station The activated ports are from the configuration Move to configuration The required path distance is: Then from the configuration Switch to configuration Total switching time required Defined as the maximum value among all relevant port movement times, the handover time formula is: in, and They represent the first The set of transmit port indices and the set of receive port indices selected by each base station. Correspondingly, the duration of the stable dwell phase is: when When, it indicates that the system chooses to keep the fluid antenna configuration of the previous time slot unchanged, i.e., perform a hold operation; when When this occurs, it indicates that a switching operation is being performed.

[0036] Furthermore, the following section elaborates on the communication model portion of the target fluid antenna performance prediction model: This invention employs an orthogonal frequency division multiplexing (OFDM) transmission mechanism, assuming the system subcarrier set is as follows: For the first The first base station, the first The communication user and the first There are 1 subcarrier, and the communication channel vector is denoted as: The spatial response of a fluid antenna is determined by the chosen port location. For a propagation direction... , No. Each base station has its transmission port coordinates The launch steering vector can be expressed as: in, Indicates the carrier wavelength. Indicates the first The first base station Three-dimensional global position of each transmit port. Receive steering vector. The location of the receiving port can be determined in the same way.

[0037] For the The OFDM symbol and the first The subcarrier, the The transmission signals of each base station are: in, Indicates the first Each base station in configuration The following is the first A user-designed downlink precoding vector Indicates sending to the Data signals for each user.

[0038] Therefore, the first The communication user in the first The received signal on each subcarrier can be represented as: in, This represents additive white Gaussian noise with a noise power of . .

[0039] Based on the above definition, the first The user in the first The signal-to-interference-plus-noise ratio (SINR) on each subcarrier is expressed as: Therefore, the downlink communication and rate of the system during the stable dwell phase are defined as follows: This communication rate is used to describe the fluid antenna in the target fluid antenna configuration. The communication performance achievable when maintaining a stable presence.

[0040] Furthermore, the following section elaborates on the sensing model portion of the target fluid antenna performance prediction model: This invention considers a fully cooperative sensing mode, where all base stations simultaneously participate in target detection as both sensing transmitters and receivers. For a given target fluid antenna configuration... , No. The sensed echo signal received by each base station can be represented as: in, express The response vector of each target; Indicates the first The perception dictionary matrix corresponding to each receiving base station consists of the range from each transmitting base station to the target and then to the receiving base station. The cascaded echo responses of each receiving base station constitute the response. Indicates the first Residual self-interference channels of each base station; Indicates the first The base station to the Cross-interference channels of individual base stations; Indicates perceived noise. Let r be the transmitted signal vector of the r-th base station itself. Let be the transmit signal vector of the b-th cooperating base station.

[0041] By stacking the sensing observations from all base stations, a network-level joint sensing model can be obtained: in: Represents the globally aware dictionary matrix; This represents the aggregated interference term, which consists of residual self-interference and cross-interference between base stations. This represents the perceived noise after stacking.

[0042] After interference suppression, the effective perturbation covariance matrix is ​​defined as: in, Represents the perceived noise covariance matrix. This represents the residual disturbance covariance. Assume the target response vector follows a covariance matrix of... The complex Gaussian distribution: Then in The perceptual mutual information is defined as follows: in, for The conjugate transpose of . This is the conjugate transpose notation. This sensing mutual information characterizes the system's ability to acquire target response information under a given fluid antenna configuration. A larger value indicates more comprehensive target-related information obtained from echo observations, and better sensing performance.

[0043] The overall operation of the fluid antenna-assisted sensing system is divided into two stages: the first stage is the characterization of time-slot-level sensing performance and problem derivation, and the second stage is the adaptive switching decision based on online calibration Bayesian optimization.

[0044] Phase one specifically includes: Because the fluid antenna port is from the fluid antenna configuration of the previous time slot, i.e., the historical fluid antenna configuration. Target fluid antenna configuration moved to the current time slot It requires a certain amount of physical time, and the system is not always in a stable port configuration within a time slot. Therefore, this invention divides each time slot... It is divided into two consecutive phases: the port switching phase and the stable dwell phase.

[0045] The duration of the port switching phase is: in, Indicates the first The first base station One activated fluid antenna port from Move to The distance of the path taken, This indicates the speed at which the liquid metal moves within the fluid antenna.

[0046] The duration of the stable residency phase is: Unlike traditional approaches, this invention does not simply treat the port switching phase as an invalid period where communication and sensing are completely interrupted. Instead, it considers it as an unstable but still usable transmission phase. During this phase, the antenna port position changes over time, and the communication channel, sensing response, and interference state also change accordingly. Therefore, the sensing performance during the switching phase should be included in the overall utility evaluation of the time slot, along with the performance during the stable dwell phase.

[0047] (1) Characterization of time-slot-level communication performance: During the stable residence phase, the system is in configuration. and using the corresponding precoding matrix Communication takes place. The communication rate and speed during this phase are denoted as: During the port switching phase, the fluid antenna from Move to Since the port position changes continuously during this stage, let the continuous configuration during the movement be... ,in During the handover process, the system can reuse the precoding matrix from the previous time slot: The average communication rate during the handover phase can then be expressed as: in, Indicates the instantaneous configuration during the switching process. Next, use the precoding matrix from the previous time slot. The instantaneous communication rate at any given time.

[0048] Under high port density conditions, the channel exhibits strong correlation locally due to the small spatial distance between adjacent ports. Therefore, a discrete port configuration along the port movement path can be used to approximate a continuous movement process. Let's assume... arrive The set of available port configurations corresponding to the movement path is as follows: The communication rate during the handover phase can be approximated as: Therefore, this invention defines the communication utility within a complete time slot as a weighted sum of the communication rate during the handover phase and the communication rate during the stable dwell phase: This formula indicates that the communication benefits of a single port switching action depend not only on the configuration. The stable communication rate also depends on the... Move to The communication contribution that can be obtained during the process, as well as the proportion of time occupied by the handover phase and the dwell phase in the entire time slot.

[0049] (2) Characterization of time-slot level sensing performance: Similarly, during the stable residency phase, the system in The following describes a stable sensing process, where the mutual information of the sensing is denoted as: in: During the port switching phase, the fluid antenna position changes along the moving path, and the system can still receive target echoes and extract sensing information. Let the average sensing mutual information during the switching phase be: Under high port density conditions, the mutual information sensed during this handover phase can also be approximated by the discrete port configuration on the moving path, i.e.: Therefore, the perceived utility within a complete time slot is defined as: This formula indicates that the sensing gains from a single port handover operation come not only from the stable sensing results under the current configuration, but also from the target echo information obtained along the path during port movement. Correspondingly, the paper incorporates both the handover and dwell phases into the time-slot-level communication and sensing objectives.

[0050] (3) Energy consumption characterization of mechanical switching: In addition to communication and sensing performance, port movement also consumes mechanical energy. Let the moving power of the fluid antenna be... Then from the configuration Switch to configuration The energy consumption for mechanical switching is defined as: The summation term covers all active ports that have moved. This energy consumption term is a deterministic physical quantity determined by the port movement distance, the liquid metal movement speed, and the movement power.

[0051] (4) Time slot-level multi-objective reward vector: Based on the aforementioned communication utility, sensing utility, and mechanical switching energy consumption, this invention defines the time-slot-level multi-objective reward vector as follows: Among them, the first item The second term represents the time-slot-level communication utility. The third term represents the time-slot level perception utility. This represents a negative value for mechanical energy consumption. The reason for using a negative sign is that communication and sensing objectives need to be maximized, while the energy consumption of mechanical switching needs to be minimized.

[0052] Therefore, this invention transforms the port configuration selection problem into a long-term multi-objective optimization problem: The constraints include: Among them, the first few constraints ensure the feasibility of port selection and the mutual exclusion of transmit and receive ports; Ensure that port switching can be completed within one time slot; This indicates that the precoding matrix is ​​generated based on the precoding rules of the feedback channel; the last term represents the base station transmit power constraint. The maximum transmit power of the base station is given. The paper also defines the reward vector as communication utility, sensing utility, and negative handover energy consumption, treating it as a multi-objective optimization problem in the Pareto sense.

[0053] Phase Two specifically includes: Phase 1 provides a unified evaluation method for time-slot-level sensing performance and mechanical handover energy consumption. However, in real-world air-to-ground sensing environments, communication channels, target responses, residual self-interference, and inter-base station cross-interference are often difficult to obtain accurately before decision-making. Therefore, time-slot-level communication utility and sensing utility can be represented as unknown implicit physical mappings: in, This is the implicit physical mapping result for time-slot-level communication efficiency; The implicit physical mapping result for perceived utility; and These represent unknown communication performance mapping and unknown perception performance mapping, respectively. Indicates time slot The communication channel status; Represents a perceptual dictionary; This represents residual interference-related terms. Since these physical quantities are difficult to fully resolve in dynamic environments, this invention employs an online Bayesian optimization method to learn the unknown mapping relationship between port switching actions and time-slot-level synergistic performance.

[0054] (1) Online observation feedback modeling: In each time slot System observation and from the candidate fluid antenna configuration set Select a target fluid antenna configuration After executing the hold or switch action, the system receives two types of noisy feedback: communication and sensing. in, Represents communication utility observation, corresponding to ; Represents perceived utility observation, corresponding to ; This indicates measurement noise.

[0055] (2) Cost-aware Gaussian process surrogate model: To learn the unknown synesthetic effect, this invention establishes two Gaussian process surrogate models for the communication effect and the sensing effect, respectively. Since the time-slot level effect depends not only on the selected target fluid antenna configuration... It also depends on the previous configuration. And the switching path between the two, therefore the input to the proxy model is not a single target port configuration, but a configuration switching pair ( , ).

[0056] For the To achieve the following objectives, a Gaussian process model is established: in, Inside, the first parameter It represents the current input switching pair; the symbol ⋅ is a placeholder for the second input variable, representing any other input point of the switching pair; the whole represents the kernel function corresponding to the m-th target. (), whose input is two switching pairs and whose output is the covariance of the two.

[0057] Given a historical observation dataset: For the candidate switching action to candidate fluid antenna configuration a, the Gaussian process model gives the predicted distribution: in, This represents the predicted mean. This represents the standard deviation of the forecast.

[0058] (3) Online conformal calibration mechanism: In dynamic air-to-ground sensing environments, user movement, target movement, changes in residual interference, and motion-related disturbances caused by port switching may lead to inaccurate uncertainty estimations in standard Gaussian process models. To avoid overconfidence in surrogate models, this invention introduces an online conformal calibration mechanism to dynamically correct the prediction variance of Gaussian processes.

[0059] In the time slot Get real feedback Then, define the standardized residuals: in, To prevent extremely small positive numbers with a denominator of zero.

[0060] The system is maintaining the residual cache from the recent period: in, This represents the cache window length. Based on this residual cache, the target coverage rate is calculated. The following conformal calibration factor: in, It is a quantile function.

[0061] Subsequently, the standard deviation of the standard Gaussian process prediction was corrected to: Therefore, candidate actions The predicted distribution is adjusted to a distribution with conformal calibration uncertainty. This mechanism dynamically increases or decreases model uncertainty based on the most recent online observations of the prediction error, thereby reducing the impact of environmental drift and model mismatch on port switching decisions. The paper points out that online conformal calibration adjusts the surrogate model uncertainty through dynamic residual caching to improve the stability of hold / switch decisions in dynamic environments.

[0062] (4) Cost-aware acquisition function and hold / switch decision: To balance communication utility, sensing utility, and switching energy consumption, this invention employs the desired hypervolume improved EHI as the acquisition function for multi-objective Bayesian optimization. Let the current approximate Pareto set be... The reference point is For candidate actions The predicted target vector is obtained by sampling from the conformally calibrated prediction distribution. Its hypervolume improvement is defined as: in, This indicates that the Pareto set is relative to the reference point. The dominated hypervolume. The corresponding desired hypervolume improvement is: Due to energy consumption during mechanical switching This is a deterministic physical quantity determined by the port movement distance and the fluid antenna movement speed. This invention incorporates it as an explicit penalty term into the acquisition function, resulting in a cost-aware acquisition function: in, This represents the energy consumption weighting coefficient, used to adjust the trade-off between improved sensory performance and energy consumption during mechanical switching.

[0063] Finally, in each time slot, the system selects the configuration with the largest acquisition function value from the set of candidate fluid antenna configurations as the target fluid antenna configuration: when When, the system performs a hold operation; when At this time, the system performs a port switching operation. For hold operations, since there is no port movement, the switching time and switching energy consumption are both zero; for switching operations, the system calculates the full time slot revenue based on the switching time and switching energy consumption defined in Phase 1.

[0064] Based on the above theoretical principles and model definitions, the execution flow of the fluid antenna port switching method of the present invention will be described in detail below: In another embodiment, obtaining the candidate fluid antenna configuration set and historical fluid antenna configurations includes: Obtain the original set of candidate fluid antenna configurations and historical fluid antenna configurations; Based on the port movement distance and port liquid metal movement speed between any original candidate fluid antenna configuration and the set of original candidate fluid antenna configurations, the original candidate switching time for switching from the historical fluid antenna configuration to the original candidate fluid antenna configuration is obtained, and the set of original candidate switching times is obtained based on all the original candidate switching times corresponding to the set of original candidate fluid antenna configurations. Based on the original set of candidate switching times and the preset time slot length, the candidate fluid antenna configuration set is selected from the original set of candidate fluid antenna configurations.

[0065] It should be noted that the candidate fluid antenna configuration set is obtained by acquiring the original set of candidate fluid antenna configurations and historical fluid antenna configurations. For any one of the original candidate fluid antenna configurations in the set of original candidate fluid antenna configurations; According to the above switching time formula, the port movement distance between any one of the original candidate fluid antenna configurations in the set of historical fluid antenna configurations and the original candidate fluid antenna configurations is calculated. and port liquid metal movement speed Calculate from Switch to Original candidate switching time The original candidate switching time is compared with the preset time slot length. (i.e., the length of one time slot) are compared, if ≤ Then the original candidate fluid antenna configuration will be included in the candidate fluid antenna configuration set. ,like > If the original candidate fluid antenna configuration is not selected, it will be excluded from the candidate set. This results in the final set of candidate fluid antenna configurations. Candidate fluid antenna configuration .

[0066] In another embodiment, the method for obtaining the target fluid antenna performance prediction model includes inputting the candidate fluid antenna configuration set into a pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model. Acquire several switching pairs, wherein each switching pair includes a sample history fluid antenna configuration and a sample target fluid antenna configuration corresponding to the sample history fluid antenna configuration; For any one of the aforementioned switching pairs, obtain the historical observation data corresponding to that switching pair; A set of historical observation data is obtained based on all the historical observation data corresponding to several switching pairs; The target fluid antenna performance prediction model is obtained by training a preset fluid antenna performance prediction model based on several switching pairs and the historical observation dataset.

[0067] It should be noted that the target fluid antenna performance prediction model is obtained through the following methods: Obtain several switching pairs for each of the switching pairs ( , ), including a sample historical fluid antenna configuration and a sample target fluid antenna configuration corresponding to the sample history fluid antenna configuration. Where n represents the historical time slot number, This is the configuration actually executed in the (n-1)th time slot. This is the target configuration actually executed in the nth time slot.

[0068] For any pair of switching pairs among the plurality of such switching pairs ( , ), obtain the historical observation data corresponding to the switch pair, including When m=1, it represents the actual communication performance and When m=2, it represents the actual perceived performance.

[0069] A set of historical observation data is obtained based on all the historical observation data corresponding to the several switching pairs. .

[0070] A preset fluid antenna performance prediction model is trained based on several switching pairs and the historical observation dataset: the preset fluid antenna performance prediction model is a Gaussian process model, with switching pairs ( , Using historical observation data as input features, the actual communication performance is considered. and real-world performance The output labels are used for training to obtain the performance prediction model of the target fluid antenna. After training, the candidate fluid antenna configuration is assumed to be... The target fluid antenna performance prediction model can predict the performance of any input candidate fluid antenna configuration and historical fluid antenna configurations, forming a candidate switching pair. Output the corresponding average prediction performance and the standard deviation of predicted performance .

[0071] In another embodiment, obtaining the historical observation data corresponding to any one of the plurality of switching pairs includes: The sample port movement path is obtained based on the sample history fluid antenna configuration and the sample target fluid antenna configuration in the switching pair; Based on the sample history fluid antenna configuration, a sample history precoding matrix is ​​obtained, and based on the sample history precoding matrix and the sample port movement path, the port handover communication rate and port handover sensing mutual information are obtained. Based on the sample target fluid antenna configuration, a sample target precoding matrix is ​​obtained, and based on the sample target precoding matrix, a stable dwell communication rate and a stable dwell sensing mutual information are obtained. The actual communication performance is obtained based on the port switching communication rate and the stable dwell communication rate. The actual sensing performance is obtained based on the port switching sensing mutual information and the stable dwell sensing mutual information. The historical observation data of the handover pair are obtained based on the actual communication performance and the actual perception performance.

[0072] It should be noted that, based on the aforementioned switching pair ( , Sample historical fluid antenna configurations The sample target fluid antenna configuration in the switching pair Obtain the sample port movement path The sample port movement path is from liquid metal... arrive The physical path traversed is represented by a set of discrete port configurations along the path.

[0073] Based on the sample history fluid antenna configuration Obtain the sample history precoding matrix And based on the sample history precoding matrix and the sample port movement path Obtain port switching communication rate Port switching awareness mutual information Specifically, the positions of multiple discrete ports along the moving path of the sample port. ∈ The sample history precoding matrix was used respectively. Calculate the instantaneous communication rate at each discrete location. and instantaneous mutual information Then, average the instantaneous values ​​at all discrete locations to obtain... and .

[0074] Based on the sample target fluid antenna configuration Obtain the sample target precoding matrix And based on the sample target precoding matrix Achieve stable dwell communication rate and stable residency sensing mutual information The stable dwell communication rate According to the formula in the above communication model Calculation; the stable resident sensing mutual information According to the formula in the above perception model Calculate, and take one of them As .

[0075] Based on the port switching communication speed With the stable dwell communication rate Obtain the actual communication performance Specifically, the duration of the switching phase is obtained. To obtain the duration of the stable residency phase According to the ratio of the duration of the switching phase to the preset time slot length / Switch the communication rate on the port Weighting is applied based on the ratio of the duration of the stable residence phase to the preset time slot length. / For the stable dwell communication rate Perform a weighted sum, then sum the two weighted terms. = / + / Thus, the actual communication performance is obtained.

[0076] Based on the port switching sensing mutual information With the stable resident sensing mutual information Achieve true perceived performance Specifically, based on the ratio of the duration of the switching phase to the preset time slot length. / Mutual information for port switching Weighting is applied based on the ratio of the duration of the stable residence phase to the preset time slot length. / The stable resident sensing mutual information Perform a weighted sum, then sum the two weighted terms. = / + / Thus, the actual perception performance is obtained.

[0077] Based on the actual communication performance and the aforementioned real-world perception performance The switching pair is obtained ( , Historical observation data: = , = .

[0078] In another embodiment, the step of obtaining the historical prediction deviation set corresponding to the target fluid antenna performance prediction model, obtaining a conformal calibration factor based on a preset conformal coverage error rate and the historical prediction deviation set, and then performing data scaling processing on the original prediction performance distribution set based on the conformal calibration factor to obtain a calibrated prediction performance distribution set includes: The historical prediction deviation set is processed to obtain a standardized residual score set; The conformal calibration factor is obtained based on the preset conformal error coverage rate and the standardized residual score set; A calibration performance standard deviation set is obtained based on the conformal calibration factor and the predicted performance standard deviation set; The calibration prediction performance distribution set is obtained based on the set of calibration performance standard deviations and the set of predicted performance mean values.

[0079] It should be noted that the original set of predicted performance distributions includes the set of predicted performance standard deviations. and the set of predictive performance mean .

[0080] The historical prediction bias set is processed to obtain a standardized residual score set. Each standardized residual score According to the formula calculate.

[0081] Based on the preset conformal error coverage rate (The preset conformal error coverage rate is a probability threshold preset by the system for allowing prediction deviations to exceed the calibrated prediction range, preferably between 0.05 and 0.2) and the standardized residual score set. The conformal calibration factor was obtained. .

[0082] Based on the conformal calibration factor and the set of standard deviations of the predicted performance The standard deviation set of calibration performance is obtained. .

[0083] Based on the set of standard deviations of the calibration performance and the set of predicted performance averages The calibration prediction performance distribution set is obtained. .

[0084] In another embodiment, obtaining the set of mechanical switching energy consumption penalty terms based on the candidate fluid antenna configuration set and the historical fluid antenna configurations includes: Based on the port movement distance and port liquid metal movement speed between any candidate fluid antenna configuration and the historical fluid antenna configuration set, a candidate switching time for switching from the historical fluid antenna configuration to the candidate fluid antenna configuration is obtained, and a candidate switching time set is obtained based on all the candidate switching times corresponding to the candidate fluid antenna configuration set. Based on the pre-acquired fluid antenna mobile power and the candidate switching time set, the set of mechanical switching energy consumption penalty terms is obtained.

[0085] It should be noted that, based on the aforementioned historical fluid antenna configuration With the aforementioned candidate fluid antenna configuration set Port movement distance between any candidate fluid antenna configuration a and port liquid metal movement speed According to the formula Calculations from the historical fluid antenna configuration The candidate switching time for switching to the candidate fluid antenna configuration a is described above. And based on all the candidate switching times corresponding to the candidate fluid antenna configuration set, a candidate switching time set is obtained.

[0086] Based on pre-acquired fluid antenna mobile power and the candidate switching time set The fluid antenna's moving power is the mechanical power required to drive the liquid metal, a preset parameter of the system determined by the hardware's physical characteristics, according to the formula... Computational machine switching energy consumption Then multiply the mechanical switching energy consumption by a preset energy consumption weighting coefficient. The energy consumption weighting coefficient mentioned above is a system preset parameter used to adjust the weight of mechanical switching energy consumption in the net benefit score, thereby obtaining a mechanical switching energy consumption penalty item. For each candidate configuration in the candidate fluid antenna configuration set, the set of mechanical switching energy consumption penalty terms is calculated separately.

[0087] In another embodiment, the step of obtaining a potential improvement expectation set based on the calibration predicted performance distribution set, and obtaining a net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, includes: The set of calibration prediction performance distributions is sampled to obtain a set of sampled prediction performance vectors; Based on the sampled prediction performance vector set, the pre-acquired current approximate Pareto front, and the preset reference point, the potential improvement expectation set is obtained, and based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, the net benefit score set is obtained.

[0088] It should be noted that the set of calibration prediction performance distributions... Sampling is performed to obtain a set of sampled prediction performance vectors. Specifically, the joint distribution of communication performance and perception performance corresponding to each candidate configuration is randomly sampled multiple times. Each sampling yields a sampled prediction performance vector, which includes communication performance sample values ​​and perception performance sample values.

[0089] Based on the sampling prediction performance vector set Pre-acquired current approximate Pareto front And a preset reference point r, the current approximate Pareto front is a non-dominated solution set front constructed based on all historical true observations in the historical observation dataset, representing all optimal trade-offs discovered by the system so far. The preset reference point is a fixed coordinate point pre-defined in the two-dimensional space of communication performance and sensing performance, usually selected as the theoretical lower bound of communication performance and sensing performance, used to determine the lower bound of hypervolume calculation, and obtain the potential improvement expectation set. Specifically, for each sampled prediction performance vector Calculate its relative to the current approximate Pareto front. The overvolume improvement amount of the preset reference point r ,in Let r represent the hypervolume dominated by the Pareto set relative to the reference point r; the average of multiple hypervolume improvements obtained from multiple samplings for each candidate configuration is taken to obtain the expected hypervolume improvement of the candidate configuration, which is used as the potential improvement expectation of the candidate configuration. .

[0090] Based on the set of potential improvement expectations and the set of energy consumption penalty terms for mechanical switching To obtain the net income score set The net benefit score comprehensively reflects the net gain between the improvement in communication sensing performance and the mechanical energy consumption cost when switching to the corresponding candidate configuration. Ultimately, the system selects the candidate configuration with the highest net benefit score as the target fluid antenna configuration for the current time slot. .

[0091] like Figure 2 As shown, based on the above method embodiments, corresponding apparatus embodiments are provided; An embodiment of the present invention provides a fluid antenna port switching device, comprising: The basic data acquisition module is used to acquire the candidate fluid antenna configuration set and historical fluid antenna configurations; The original predicted performance distribution set acquisition module is used to input the candidate fluid antenna configuration set into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model. The calibration module is used to calibrate the original prediction performance distribution set based on the pre-acquired historical prediction deviation set to obtain a calibrated prediction performance distribution set. The mechanical switching energy consumption penalty term acquisition module is used to obtain the mechanical switching energy consumption penalty term set based on the candidate fluid antenna configuration set and the historical fluid antenna configuration; The net benefit score set acquisition module is used to obtain a potential improvement expectation set based on the calibration prediction performance distribution set, and to obtain a net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set. The port switching module is used to select the target fluid antenna configuration for the current time slot from the candidate fluid antenna configuration set based on the net benefit score set, so as to perform port switching based on the target fluid antenna configuration.

[0092] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the fluid antenna port switching method provided by any of the above-described method embodiments of the present invention.

[0093] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0094] Based on the above embodiment of a fluid antenna port switching method, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a fluid antenna port switching method according to any embodiment of the present invention.

[0095] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0096] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0097] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0098] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute a fluid antenna port switching method as described in any of the above-described method embodiments of the present invention.

[0099] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0100] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for switching ports of a fluid antenna, characterized in that, include: Obtain the set of candidate fluid antenna configurations and historical fluid antenna configurations; The candidate fluid antenna configuration set is input into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model; Obtain the set of historical prediction deviations corresponding to the target fluid antenna performance prediction model, and obtain a conformal calibration factor based on the preset conformal coverage error rate and the set of historical prediction deviations. Then, perform data scaling processing on the original prediction performance distribution set based on the conformal calibration factor to obtain a calibrated prediction performance distribution set. Based on the candidate fluid antenna configuration set and the historical fluid antenna configuration, a set of mechanical switching energy consumption penalty terms is obtained; Based on the calibration prediction performance distribution set, a potential improvement expectation set is obtained, and based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, a net benefit score set is obtained. Based on the net benefit score set, a target fluid antenna configuration for the current time slot is selected from the candidate fluid antenna configuration set, and port switching is performed based on the target fluid antenna configuration.

2. The fluid antenna port switching method according to claim 1, characterized in that, The acquisition of the candidate fluid antenna configuration set and historical fluid antenna configurations includes: Obtain the original set of candidate fluid antenna configurations and historical fluid antenna configurations; Based on the port movement distance and port liquid metal movement speed between any original candidate fluid antenna configuration and the set of original candidate fluid antenna configurations, the original candidate switching time for switching from the historical fluid antenna configuration to the original candidate fluid antenna configuration is obtained, and the set of original candidate switching times is obtained based on all the original candidate switching times corresponding to the set of original candidate fluid antenna configurations. Based on the original set of candidate switching times and the preset time slot length, the candidate fluid antenna configuration set is selected from the original set of candidate fluid antenna configurations.

3. The fluid antenna port switching method according to claim 1, characterized in that, The step of inputting the candidate fluid antenna configuration set into a pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model includes: Acquire several switching pairs, wherein each switching pair includes a sample history fluid antenna configuration and a sample target fluid antenna configuration corresponding to the sample history fluid antenna configuration; For any one of the several switching pairs, obtain the historical observation data corresponding to that switching pair; A set of historical observation data is obtained based on all the historical observation data corresponding to several switching pairs; The target fluid antenna performance prediction model is obtained by training a preset fluid antenna performance prediction model based on several switching pairs and the historical observation dataset.

4. The fluid antenna port switching method according to claim 3, characterized in that, The step of obtaining the historical observation data corresponding to any one of the several switching pairs includes: The sample port movement path is obtained based on the sample history fluid antenna configuration and the sample target fluid antenna configuration in the switching pair; Based on the sample history fluid antenna configuration, a sample history precoding matrix is ​​obtained, and based on the sample history precoding matrix and the sample port movement path, the port handover communication rate and port handover sensing mutual information are obtained. Based on the sample target fluid antenna configuration, a sample target precoding matrix is ​​obtained, and based on the sample target precoding matrix, a stable dwell communication rate and a stable dwell sensing mutual information are obtained. The actual communication performance is obtained based on the port switching communication rate and the stable dwell communication rate. The actual sensing performance is obtained based on the port switching sensing mutual information and the stable dwell sensing mutual information. The historical observation data of the handover pair are obtained based on the actual communication performance and the actual perception performance.

5. The fluid antenna port switching method according to claim 1, characterized in that, The process of obtaining the historical prediction deviation set corresponding to the target fluid antenna performance prediction model, obtaining a conformal calibration factor based on a preset conformal coverage error rate and the historical prediction deviation set, and then performing data scaling processing on the original prediction performance distribution set based on the conformal calibration factor to obtain a calibrated prediction performance distribution set includes: The historical prediction deviation set is processed to obtain a standardized residual score set; The conformal calibration factor is obtained based on the preset conformal error coverage rate and the standardized residual score set; A calibration performance standard deviation set is obtained based on the conformal calibration factor and the predicted performance standard deviation set; The calibration prediction performance distribution set is obtained based on the set of calibration performance standard deviations and the set of predicted performance mean values.

6. The fluid antenna port switching method according to claim 1, characterized in that, The step of obtaining a set of mechanical switching energy consumption penalty terms based on the candidate fluid antenna configuration set and the historical fluid antenna configurations includes: Based on the port movement distance and port liquid metal movement speed between any candidate fluid antenna configuration and the historical fluid antenna configuration set, a candidate switching time for switching from the historical fluid antenna configuration to the candidate fluid antenna configuration is obtained, and a candidate switching time set is obtained based on all the candidate switching times corresponding to the candidate fluid antenna configuration set. Based on the pre-acquired fluid antenna mobile power and the candidate switching time set, the set of mechanical switching energy consumption penalty terms is obtained.

7. The fluid antenna port switching method according to claim 1, characterized in that, The step of obtaining a potential improvement expectation set based on the calibration prediction performance distribution set, and obtaining a net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, includes: The set of calibration prediction performance distributions is sampled to obtain a set of sampled prediction performance vectors; Based on the sampled prediction performance vector set, the pre-acquired current approximate Pareto front, and the preset reference point, the potential improvement expectation set is obtained, and based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set, the net benefit score set is obtained.

8. A fluid antenna port switching device, characterized in that, include: The basic data acquisition module is used to acquire the candidate fluid antenna configuration set and historical fluid antenna configurations; The original predicted performance distribution set acquisition module is used to input the candidate fluid antenna configuration set into the pre-trained target fluid antenna performance prediction model to obtain the original predicted performance distribution set output by the target fluid antenna performance prediction model. The calibration module is used to calibrate the original prediction performance distribution set based on the pre-acquired historical prediction deviation set to obtain a calibrated prediction performance distribution set. The mechanical switching energy consumption penalty term acquisition module is used to obtain the mechanical switching energy consumption penalty term set based on the candidate fluid antenna configuration set and the historical fluid antenna configuration; The net benefit score set acquisition module is used to obtain a potential improvement expectation set based on the calibration prediction performance distribution set, and to obtain a net benefit score set based on the potential improvement expectation set and the mechanical switching energy consumption penalty term set. The port switching module is used to select the target fluid antenna configuration for the current time slot from the candidate fluid antenna configuration set based on the net benefit score set, so as to perform port switching based on the target fluid antenna configuration.

9. A terminal device, characterized in that, The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements a fluid antenna port switching method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, include: A stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform a fluid antenna port switching method as described in any one of claims 1-7.

Images