Noma assisted drone backscattering sensing integrated beamforming method

Abstract

The embodiment of the application provides a NOMA auxiliary unmanned plane backscattering communication and sensing integrated beam forming method, relates to the technical field of unmanned plane communication and sensing integration, and through a same NOMA downlink signal of a transmitting unmanned plane, multi-user communication, backscattering tag excitation, unmanned plane state back transmission and spatial orientation sensing are simultaneously completed, the use of independent communication waveform, sensing waveform and additional spectrum resources is reduced, and the utilization efficiency of limited communication resources of the unmanned plane is improved. That is, the technical scheme of the application can solve the problem that the traditional unmanned plane communication and sensing integrated beam design technology is difficult to simultaneously guarantee multi-user communication, state back transmission and spatial sensing when the number of antennas, transmitting power, spectrum resources and channel estimation accuracy are limited.

Description

Technical Field

[0001] This application relates to the field of integrated communication and sensing technology for unmanned aerial vehicles (UAVs), and in particular to a NOMA-assisted UAV backscattering sensing integrated beamforming method. Background Technology

[0002] Currently, Integrated Sensing and Communication (ISAC) technology for unmanned aerial vehicles (UAVs) has become a core research direction in fields such as 6G mobile communication, low-altitude economy, and smart IoT. This technology relies on a single hardware platform and spectrum resources to simultaneously achieve target perception and detection as well as data communication transmission. It effectively solves problems such as spectrum conflicts, hardware redundancy, and excessive power consumption caused by the independent deployment of sensing and communication devices in UAV systems. It has significant application value in scenarios such as UAV inspection, low-altitude target detection, swarm collaborative operations, and emergency communication. In ISAC systems, beamforming design is a core technical aspect for coordinating sensing and communication performance and achieving efficient resource reuse; its design accuracy directly determines the overall performance of the system.

[0003] Among related technologies, research on ISAC beam design for unmanned aerial vehicles (UAVs) faces significant technical limitations. On the one hand, most research schemes are based on the assumption of ideal Channel State Information (CSI) for beam optimization, assuming that the base station or UAV can obtain completely accurate channel parameters. However, in actual UAV communication scenarios, due to multiple factors such as rapid channel changes caused by high-speed movement, limited feedback bandwidth, channel estimation errors, and hardware limitations, the obtained CSI inevitably contains errors. The beam scheme designed under the ideal CSI assumption will experience severe performance degradation in actual deployment. On the other hand, existing beam designs mostly optimize only a single communication or single sensing indicator, failing to achieve a joint balance between communication and sensing performance. Moreover, for non-orthogonal multiple access (NOMA) communication scenarios under imperfect CSI conditions, the relevant schemes lack a systematic robust beam design method and have several technical defects, including the inability to effectively suppress the communication interruption risk caused by channel uncertainty, insufficient precision in inter-user interference control in NOMA, difficulty in guaranteeing the main lobe pointing and null depth of the sensing beam, and difficulty in quantifying the trade-off between communication and sensing performance. These problems make it difficult to meet the high reliability and high robustness performance requirements of actual UAV ISAC systems.

[0004] In summary, traditional UAV integrated sensing beam design technology is limited by the ideal CSI assumption and the single-index optimization approach. It lacks an effective robust design scheme for NOMA communication scenarios under imperfect CSI conditions, and suffers from multiple defects such as insufficient communication reliability, weak inter-user interference suppression capability, and difficulty in guaranteeing sensing performance. It cannot meet the high-performance requirements of sensing collaboration in the complex channel environment of actual UAVs, and has become a key technical bottleneck restricting the transition of UAV ISAC technology from theoretical research to engineering implementation. Summary of the Invention

[0005] The main objective of this application is to propose a NOMA-assisted UAV backscattering sensing-integrated beamforming method, which aims to solve the problem that traditional UAV sensing-integrated beamforming technology is difficult to simultaneously guarantee multi-user communication, status backhaul, and spatial awareness when the number of antennas, transmit power, spectrum resources, and channel estimation accuracy are limited.

[0006] To achieve the above objectives, a first aspect of this application proposes a NOMA-assisted unmanned aerial vehicle (UAV) backscattering inductive beamforming method, the method comprising: The system acquires current estimates and channel error statistics for multiple communication channels of the launching UAV; the launching UAV uses a non-orthogonal multiple access method to superimpose the communication signals of multiple communication users in the power domain to form a downlink signal, and multiplexes the downlink signal into radio frequency excitation signals for multiple backscatter tags; the multiple communication channels include user channels from the launching UAV to the multiple communication users, and forward channels from the launching UAV to the multiple backscatter tags; Based on the channel error statistics parameters, channel error samples are extracted and superimposed on the current estimated value to generate the current channel implementation. Based on the current channel, a joint beamforming optimization problem is constructed. The joint beamforming optimization problem takes the weighted sum of the average communication rate of the multiple communication users and the average RF excitation power of the multiple backscattering tags as the optimization objective, and takes the communication rate of each of the multiple communication users as not lower than the corresponding rate threshold, the average power difference of different backscattering tag directions as not exceeding the first threshold, and the average cross-correlation of the tag detection signals as not exceeding the second threshold as random constraints, and the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix as deterministic constraints. The joint beamforming optimization problem is solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition; the beam vector is used by the launching UAV to transmit the downlink signal; The reflected signals generated by the multiple backscattering tags based on the downlink signal are despread in the code domain to obtain the despread signal; Based on the despread signal, the state information of the multiple backscattering tags is recovered and the angle of arrival of each of the multiple backscattering tags is estimated.

[0007] In some embodiments, the method further includes: The communication rate of the target communication user is determined based on the effective rate of the target communication user; the target communication user is any one of the plurality of communication users, the effective rate is: the achievable rate of the target communication user itself, and the minimum achievable rate of the higher-order communication users whose signals of the target communication user need to be decoded in the serial interference cancellation decoding order of the plurality of communication users, and the average communication rate is the average of the effective rate over the channel error samples.

[0008] In some embodiments, solving the joint beamforming optimization problem based on the channel error samples to obtain a beam vector satisfying the rank-one condition includes: In the inner iteration, based on the current channel, a convex surrogate function for the optimization objective and the random constraint is constructed at the current beam variable, forming the current sample surrogate subproblem and solving it to obtain the current sample surrogate solution; The average channel anchor point is updated recursively based on historical channel error samples, and an anchor point proxy subproblem is constructed and solved based on the average channel anchor point to obtain the anchor point proxy solution. The current sample iteration and the anchor iteration are weighted and fused according to the anchor weight that gradually decreases with the number of iterations. The current beam variable is then updated in combination with the variable update step size to obtain the updated beam variable. The updated beam variable is a beam vector that satisfies the rank-one condition, and the updated beam variable is used as the current beam variable in the next inner layer iteration.

[0009] In some embodiments, the step of constructing a convex surrogate function of the optimization objective and the random constraints at the current beam variable based on the current channel, forming a current sample surrogate subproblem and solving it to obtain the current sample surrogate solution includes: Determine whether there is a feasible point in the set of surrogate constraints corresponding to the convex surrogate function; In response to the determination that a feasible point exists, the agent objective function is optimized under the agent constraint set to obtain the agent solution for the current sample; In response to the determination that no feasible point exists, a feasibility recovery subproblem is constructed and solved with the goal of minimizing the maximum surrogate constraint violation, thus obtaining the current sample surrogate solution.

[0010] In some embodiments, the step of solving the joint beamforming optimization problem based on the channel error samples to obtain a beam vector that satisfies the rank-one condition further includes: The beam vector to be optimized is elevated to a beam covariance matrix, and the rank-one violation degree, which characterizes the rank-oneness of the beam covariance matrix, is weighted by a penalty parameter and incorporated into the optimization objective to form an outer penalty objective; the rank-one violation degree is the difference between the trace of the beam covariance matrix and the maximum singular value, and the rank-one violation degree is incorporated into the optimization objective after being processed by differentiability smoothing. In the outer layer iteration, after each inner layer random successive convex approximation converges based on the outer layer penalty target, the penalty parameter is adjusted to enhance the rank-one penalty. In response to the rank-one violation of the beam covariance matrix being less than the outer threshold, the beam vector is extracted from the eigenvector corresponding to the largest eigenvalue of the beam covariance matrix.

[0011] In some embodiments, the method further includes: Based on the channel conditions corresponding to the current estimated value, the multiple communication users are sorted to obtain the serial interference cancellation decoding order of the multiple communication users; the serial interference cancellation decoding order remains fixed within a beam optimization interval and is re-determined after the channel estimation is updated.

[0012] To achieve the above objectives, a second aspect of this application proposes a NOMA-assisted unmanned aerial vehicle backscattering inductive beamforming system, the system comprising: The acquisition module is used to acquire the current estimated values ​​and channel error statistics of multiple communication channels of the launching UAV; the launching UAV superimposes the communication signals of multiple communication users in the power domain to form a downlink signal in a non-orthogonal multiple access manner, and multiplexes the downlink signal into radio frequency excitation signals of multiple backscatter tags; the multiple communication channels include the user channel from the launching UAV to the multiple communication users, and the forward channel from the launching UAV to the multiple backscatter tags; A beam vector generation module is used to extract channel error samples based on the channel error statistical parameters, and superimpose the channel error samples onto the current estimated value to generate the current channel implementation; and to construct a joint beamforming optimization problem based on the current channel implementation; the joint beamforming optimization problem has the following stochastic constraints: the weighted sum of the average communication rate of the multiple communication users and the average RF excitation power of the multiple backscattering tags as the optimization objective; the communication rate of each of the multiple communication users is not lower than the corresponding rate threshold; the average power difference of different backscattering tag directions does not exceed a first threshold; and the average cross-correlation of the tag detection signals does not exceed a second threshold; and the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix are used as deterministic constraints; the joint beamforming optimization problem is solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition; the beam vector is used for the launching UAV to transmit the downlink signal; The integrated inductive beamforming module is used to perform code domain despreading on the reflected signals generated by the multiple backscattering tags based on the downlink signal to obtain the despread signal; and to recover the state information of the multiple backscattering tags based on the despread signal and estimate the angle of arrival of each of the multiple backscattering tags.

[0013] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the NOMA-assisted UAV backscattering inductive beamforming method described in the first aspect.

[0014] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the NOMA-assisted unmanned aerial vehicle backscattering inductive beamforming method described in the first aspect.

[0015] To achieve the above objectives, a fifth aspect of the present application provides a computer program product comprising a computer program that, when executed by a processor, implements the NOMA-assisted unmanned aerial vehicle backscattering inductive beamforming method provided in the first aspect above.

[0016] The NOMA-assisted UAV backscattering sensing integrated beamforming method, system, electronic device, computer-readable storage medium, and computer program product proposed in this application obtain current estimates and channel error statistical parameters of multiple communication channels of the launching UAV; the launching UAV superimposes the communication signals of multiple communication users in a non-orthogonal multiple access manner to form a downlink signal, and multiplexes the downlink signal into radio frequency excitation signals of multiple backscattering tags; the multiple communication channels include user channels from the launching UAV to the multiple communication users, and forward channels from the launching UAV to the multiple backscattering tags; channel error samples are extracted based on the channel error statistical parameters, and the channel error samples are superimposed on the current estimates to generate the current channel implementation; a joint beamforming optimization problem is constructed based on the current channel implementation; the joint beamforming optimization problem uses the multiple The optimization objective is a weighted sum of the average communication rate of the communication users and the average RF excitation power of the multiple backscattering tags. Random constraints include that the communication rate of each of the multiple communication users is not lower than a corresponding rate threshold, the average power difference between different backscattering tag directions does not exceed a first threshold, and the average cross-correlation of the tag detection signals does not exceed a second threshold. Deterministic constraints include the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix. The joint beamforming optimization problem is solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition. This beam vector is used by the launching UAV to transmit the downlink signal. The reflected signals generated by the multiple backscattering tags based on the downlink signal are despread in the code domain to obtain the despread signal. The state information of the multiple backscattering tags is recovered based on the despread signal, and the angle of arrival of each of the multiple backscattering tags is estimated.

[0017] Thus, compared to traditional UAV integrated sensing beamforming design technology that designs communication waveforms, sensing waveforms, and state feedback links independently, this application's embodiment acquires current estimates and channel error statistics of multiple communication channels by launching the UAV. Based on the channel error statistics, channel error samples are extracted and the current channel implementation is generated. Based on the current channel implementation, a joint beamforming optimization problem is constructed and solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition for transmitting downlink signals that simultaneously carry communication and excitation functions. Then, the reflected signals from multiple backscattering tags are despread in the code domain and the state information is recovered. Complex angle-of-arrival estimation can integrate multi-user communication, backscatter tag excitation, UAV status return, and spatial orientation awareness onto the same downlink signal, the same spectrum, and the same multi-antenna hardware. It can simultaneously ensure communication and sensing performance under imperfect channel state information conditions, thereby reducing the cost of independent sensing devices, active status return links, and additional spectrum resources, and improving the utilization efficiency of UAV's limited communication resources. In other words, the embodiments of this application can effectively solve the problem that traditional UAV integrated sensing beam design technology is difficult to simultaneously ensure multi-user communication, status return, and spatial awareness when the number of antennas, transmit power, spectrum resources, and channel estimation accuracy are limited. Attached Figure Description

[0018] Figure 1 A schematic diagram of the system structure involved in the NOMA-assisted UAV backscattering sensing integration method provided in the embodiments of this application; Figure 2 A flowchart illustrating the steps of the NOMA-assisted UAV backscattering inductive beamforming method provided in some embodiments of this application. Figure 3 for Figure 2 A detailed flowchart of step S204; Figure 4 for Figure 3 A detailed flowchart of step S301; Figure 5 for Figure 2 A schematic diagram of another detailed step in step S204; Figure 6 This is a block diagram of the NOMA-assisted UAV backscattering inductive beamforming method provided in the embodiments of this application; Figure 7 This is a flowchart of the anchor-assisted rank-one penalty hybrid constraint random successive convex approximation beamforming method involved in the embodiments of this application; Figure 8 This is a schematic diagram of the NOMA-assisted UAV backscattering inductive beamforming system provided in an embodiment of this application. Figure 9 The electronic device provided in the embodiments of this application is shown in some embodiments as a hardware structure diagram. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0020] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0021] 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 belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0022] First, a brief explanation of the relevant technologies involved in the embodiments of this application will be given.

[0023] The integrated sensing and communication IC (ISAC) enables the same transmitted signal to simultaneously perform data communication and target perception functions by sharing spectrum resources, radio frequency links, and multi-antenna arrays. UAVs, with their flexible deployment, high line-of-sight propagation probability, and adjustable coverage, can benefit from having an ISAC deployed on them. This allows them to provide communication services to ground users while simultaneously performing status detection and angle perception on other UAVs, mobile devices, or backscatter tags.

[0024] Non-orthogonal multiple access (NOMA) assisted UAV sensor system: NOMA allows multiple users to transmit power-domain overlays on the same time, frequency, and spatial resources. The receiver uses Successive Interference Cancellation (SIC) to sequentially cancel decoded signals. In related technologies, NOMA-assisted ISAC schemes overlay multiple user data into a single downlink signal and ensure user rate and sensing direction gain through joint power allocation and beamforming. For example: .

[0025] in, For the beamforming vector of the nth user, For corresponding data symbols. When the number of communication users increases and approaches the number of transmitting antennas, inter-user interference will occupy a large number of spatial degrees of freedom; NOMA uses SIC to reduce the required interference suppression degrees of freedom, thereby retaining more beamforming capability for sensing direction.

[0026] Backscattering inductive integrated system: Backscatter tags modulate their own state information onto the incident radio frequency signal and reflect it back to the receiver by changing the antenna load impedance. Backscatter tags do not require actively generating a high-power carrier, making them suitable for deployment on drones, robots, and passive sensing nodes. In related technologies, the backscatter ISAC scheme can reuse communication signals to provide radio frequency excitation for the tag and read the tag's state or estimate the angle of arrival through the reflected signal.

[0027] When multiple tags reflect simultaneously, the receiver receives a superposition of multiple reflected signals. Correlation techniques typically assign Walsh codes, pseudo-random codes, or other spreading sequences to the tags and perform correlation despreading at the receiver. However, finite code lengths, inter-code cross-correlation, channel estimation errors, and receiver combining leakage can still create residual inter-tag interference. Therefore, beam design must simultaneously consider tag excitation power, inter-tag power fairness, and cross-correlation of probe signals.

[0028] Random beamforming under imperfect channel state information: Drones typically employ low-power RF devices and finite pilot training, and are affected by attitude fluctuations and random scattering, making it difficult to obtain completely accurate Channel State Information (CSI). Directly using the estimated channel as the real channel for beam design may result in actual user rates falling below the minimum service requirements, or render perceived fairness and beam cross-correlation constraints ineffective under the real channel. For example: .

[0029] in, This is the current channel estimate. For residual channel estimation error, For path loss, This refers to the Rice factor. In related technologies, robust methods include sample average approximation, stochastic principal minimization, constrained stochastic successive convex approximation, and positive semidefinite relaxation. Sample average approximation requires a large number of channel samples; traditional stochastic algorithms are easily affected by extreme samples with limited sample sizes; and using Gaussian randomization to recover the rank-one beam after positive semidefinite relaxation may also violate the stochastic expectation constraints already satisfied by the original high-rank solution.

[0030] Next, the overall concept of the embodiments of this application will be briefly described.

[0031] Current research on UAV ISAC (Interactive Signal Capability) mainly focuses on beam design for ideal CSI (Communication, Sensing, and Intelligence) or single communication / sensing indicators, resulting in the following shortcomings in related technologies: With limited antenna and power resources, multiple communication users compete with multiple backscatter tags for spatial degrees of freedom, and communication constraints can easily crowd out sensing beam resources.

[0032] Both the user channel and the tag forward channel have estimation errors. User rate, tag excitation power, tag fairness, and probe cross-correlation are all random. Optimization based solely on estimated channels is prone to violating practical constraints.

[0033] Existing sample averaging methods rely on a large number of samples and storage. Traditional random successive convex approximations tend to converge slowly or be affected by abnormal samples under finite sample conditions.

[0034] Existing semidefinite relaxation methods typically use Gaussian randomization to recover the beam vector after optimization. However, the recovery process may reduce the target value and compromise the feasibility of random constraints.

[0035] When multiple tags reflect simultaneously, there is residual interference between tags. Existing methods that only optimize the gain of a single sensing direction cannot simultaneously guarantee state detection, angle sensing, and power fairness among tags.

[0036] To address the aforementioned issues, this application proposes a NOMA-assisted UAV backscattering inductive beamforming method, system, electronic device, computer-readable storage medium, and computer program product. The aim is to solve the problem that traditional UAV inductive beamforming technology struggles to simultaneously guarantee multi-user communication, status feedback, and spatial awareness when limited by antenna quantity, transmit power, spectrum resources, and channel estimation accuracy.

[0037] In this embodiment, the NOMA-assisted UAV backscattering sensing integrated beamforming method is a NOMA-assisted UAV backscattering sensing integrated beamforming method under imperfect channel state information / NOMA-assisted UAV backscattering sensing integrated random beamforming method for imperfect channel state information. This method uses the same NOMA downlink signal to transmit data to multiple communication users and provides radio frequency excitation for backscattering tags on other UAVs or targets. Under imperfect CSI conditions, communication and sensing random constraints are constructed using channel error samples, and beamforming variables are solved using anchor-assisted, rank-one penalty and hybrid constraint random successive convex approximation algorithms.

[0038] The NOMA-assisted UAV backscattering sensing-integrated beamforming method provided in this application addresses the sensing-integrated needs in low-altitude economy, UAV swarms, and 6G mobile communication networks, demonstrating significant engineering implementation value and industrial prospects. As UAVs are increasingly used in emergency communication, logistics transportation, infrastructure inspection, and collaborative operations, they not only need to provide communication services to ground users but also perform formation monitoring, status feedback, and spatial orientation awareness. Traditional systems where communication and sensing are independent require additional spectrum, radio frequency equipment, and energy, making it difficult to meet the UAV platform's requirements for lightweight design, low power consumption, and high spectral efficiency.

[0039] As mentioned earlier, in related technologies, UAV sensing and communication integration methods typically rely on accurate channel state information. However, UAVs are often affected by factors such as low-precision converters, attitude disturbances, limited pilot training, and random scattering, resulting in imperfect channel information. Directly designing beams based on estimated channels may lead to decreased user rates, sensing beam offset, or even violations of communication and sensing constraints.

[0040] The NOMA-assisted UAV backscattering and sensing integrated beamforming method provided in this application combines non-orthogonal multiple access (NOMA), UAV backscattering, and sensing integration in a unified design. The transmitting UAV uses NOMA superimposed signals to serve multiple users and reuses the same signal as the radio frequency excitation source for the backscattering tag. Other UAVs report their own status back via backscattering, while the transmitting UAV uses code domain separation and array processing to complete status identification and orientation awareness, achieving resource sharing in communication, status reporting, and spatial awareness.

[0041] Furthermore, this application proposes an anchor-assisted penalty-type hybrid constraint stochastic successive convex approximation algorithm. This algorithm optimizes the beam using finite channel error samples without deriving a closed-form expression for a complex expectation function. The algorithm utilizes historical samples to construct an average channel anchor point, reducing the impact of extreme samples on the update direction. A feasibility recovery mechanism and rank-one penalty ensure the reliability of the stochastic constraints and the physical feasibility of the beam. Therefore, this application can be applied to scenarios such as UAV swarms, emergency communications, power line inspection, logistics transportation, remote area communications, and low-altitude network infrastructure.

[0042] The technical framework of the NOMA-assisted UAV backscattering induction integrated beamforming method provided in this application mainly includes four parts: NOMA communication, backscattering induction, imperfect channel modeling, and random beam optimization.

[0043] During downlink communication, the launching UAV uses a multi-antenna array to superimpose the power domain of the communication signals from multiple users. Users perform serial interference cancellation according to a preset decoding order, thereby reducing interference between users on the same frequency and improving multi-user access capabilities under limited spectrum and spatial degrees of freedom.

[0044] During backscattering induction, the NOMA downlink signal simultaneously serves as the RF excitation signal for the backscattering modules of other UAVs. Each backscattering UAV modulates its own state symbol and corresponding spreading code onto the reflected signal. The transmitting UAV performs correlation despreading and receive-combining on the superimposed reflected signals from multiple tags, recovers the state information through least-squares estimation, and estimates the tag's angle of arrival using array signal processing methods.

[0045] The NOMA-assisted UAV backscattering sensing integrated beamforming method provided in this application represents the actual channel as the sum of the current channel estimate and the random estimation error, and establishes a communication-sensing weighted stochastic optimization model. The optimization objective simultaneously considers the average user communication rate and the average tag excitation power, and the constraints include the minimum user rate, tag power fairness, cross-correlation of detection signals, antenna transmit power, and the rank-one condition of the beamforming matrix.

[0046] Furthermore, to solve this stochastic nonconvex problem, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application raises the beam vector to a positive semi-definite matrix and constructs a stochastic surrogate function using successive convex approximation. The algorithm obtains candidate solutions based on the current channel sample and the historical average channel anchor point, and then uses attenuation weights for fusion and updating. When the surrogate constraint is infeasible, the algorithm automatically minimizes the maximum constraint violation; simultaneously, through an outer rank-one penalty, the final matrix can be restored to a practically usable beamforming vector.

[0047] The NOMA-assisted UAV backscattering and sensing integrated beamforming method provided in this application first proposes a system architecture for the coordinated operation of NOMA, UAV backscattering, and sensing. The launching UAV utilizes the same NOMA downlink signal to simultaneously complete multi-user communication, backscatter tag excitation, UAV status feedback, and spatial orientation awareness, reducing the use of independent communication waveforms, sensing waveforms, and additional spectrum resources, thus improving the utilization efficiency of the UAV's limited communication resources.

[0048] Furthermore, a communication-sensing stochastic optimization model oriented towards conditional channel uncertainty was established. Unlike deterministic design methods based on estimated channels, the NOMA-assisted UAV backscattering sensing integrated beamforming method provided in this application represents user rate, tag excitation power, inter-tag power difference, and cross-correlation of probe signals as stochastic indices under imperfect channels. It jointly processes stochastic targets, stochastic constraints, and fixed hardware constraints in a unified model, thereby improving the beam design's adaptability to channel estimation errors.

[0049] Furthermore, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application innovatively proposes an anchor-assisted random successive convex approximation mechanism. The algorithm not only constructs the update direction using current channel error samples but also recursively generates average channel anchor points based on historical samples, and solves both the current sample proxy problem and the anchor point proxy problem separately. The two solutions are fused using gradually decaying weights, enabling the algorithm to stabilize the search direction using historical information in the early stages and reduce anchor point bias in the later stages, thereby reducing interference from extreme samples and improving convergence performance under finite sample conditions.

[0050] Furthermore, the NOMA-assisted UAV backscattering sensing integrated beamforming method provided in this application also incorporates a stochastic feasibility recovery and rank-one enhancement mechanism. When a feasible solution exists for the stochastic surrogate constraint, the algorithm optimizes the communication-sensing target; when the surrogate constraint is infeasible, the algorithm automatically switches to minimizing the maximum constraint violation. Addressing the issue that semidefinite relaxation results are difficult to directly form beam vectors, this application utilizes the difference between the matrix trace and the largest eigenvalue to construct a smooth rank-one penalty. Through inner-layer stochastic optimization and outer-layer penalty enhancement, the output result simultaneously satisfies stochastic constraints and physical feasibility requirements.

[0051] Finally, the NOMA-assisted UAV backscattering sensing integrated beamforming method provided in this application embodiment can achieve integrated processing of tag recognition, state decoding and orientation perception by separating the simultaneous reflection signals of multiple UAVs through spreading codes and combining receive merging, least squares detection and angle estimation.

[0052] In summary, the NOMA-assisted UAV backscattering sensing-integrated beamforming method provided in this application targets UAV communication equipment, low-altitude network infrastructure, cluster control platforms, and sensing-integrated systems. It can generate multi-layered revenue through algorithm licensing, communication module integration, airborne software deployment, industry solutions, and continuous operation and maintenance services. Its core commercial value lies in sharing communication signals, spectrum, and multi-antenna hardware, integrating user communication, UAV status feedback, and spatial perception into a single system, reducing the costs associated with independent sensing devices, active status feedback links, and additional spectrum.

[0053] With the rapid development of the low-altitude economy, drones are increasingly being used in power line inspection, logistics, urban management, emergency rescue, and remote area communications. The increasing number and mission density of drones will continuously generate demands for multi-user access, formation coordination, status monitoring, and spatial awareness. This application's embodiment employs a low-power backscatter method to transmit drone status, which can reduce the energy consumption of onboard active transmission equipment and extend the effective operating time of the drone.

[0054] In terms of industrialization, embodiments of this application can provide embedded algorithm libraries, software development kits, or firmware licenses to UAV manufacturers, flight control platform manufacturers, and communication module manufacturers; it can also integrate multi-antenna communication, backscatter transceiver, code domain detection, and beam control into a unified sensing module. Furthermore, it can provide planning, deployment, and operation and maintenance services for low-altitude communication sensing networks to industries such as power, oil and gas, logistics, and emergency management.

[0055] The embodiments of this application are mainly implemented through software algorithms and backscattering modules, without requiring large-scale modifications to the UAV airframe and power system, and have the conditions for gradual transformation from simulation verification and prototype testing to integration with existing platforms. By establishing a unified technical link between multi-user access, resource reuse, robust beam optimization, and low-power state backhaul, a business model that combines algorithm licensing, module sales, system delivery, and technical services can be formed.

[0056] Next, based on the overall concept of the embodiments of this application described above, specific embodiments of the NOMA-assisted UAV backscattering inductive integrated beamforming method, system, electronic device, computer-readable storage medium, and computer program product proposed in the embodiments of this application are presented. First, each specific embodiment of the NOMA-assisted UAV backscattering inductive integrated beamforming method proposed in the embodiments of this application is described in detail.

[0057] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0058] It should be noted that the NOMA-assisted UAV backscattering sensing integrated beamforming system provided in this application embodiment relates to the field of UAV communication and sensing integration technology. The NOMA-assisted UAV backscattering sensing integrated beamforming system provided in this application embodiment can be applied to a terminal or a server, while the NOMA-assisted UAV backscattering sensing integrated beamforming method provided in this application embodiment can be software running on either the terminal or the server. In some embodiments, the terminal can be a device integrating the NOMA-assisted UAV backscattering sensing integrated beamforming system, such as a UAV, or an electronic device such as a smartphone, tablet, laptop, or desktop computer that controls the device integrating the NOMA-assisted UAV backscattering sensing integrated beamforming system. The server can be the terminal's backend server terminal device, which can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms. The software can be an application, a computer program, and a storage medium carrying the NOMA-assisted UAV backscattering inductive beamforming method, etc. It should be understood that, based on different design needs of practical applications, the terminals, servers, and software involved in the embodiments of this application may, of course, be other forms not listed here, applied in different feasible embodiments. The NOMA-assisted UAV backscattering inductive beamforming method provided in the embodiments of this application does not specifically limit this.

[0059] Furthermore, this application can also be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: drones, personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, personal computers (PCs), minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via communication networks. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0060] Furthermore, the related system applying the NOMA-assisted UAV backscattering inductive beamforming method provided in this application embodiment can include several fixed formations of UAVs and multiple ground communication users. Each UAV is equipped with M transmitting antennas, a backscattering module, and a receiving array. Within a frequency band, one UAV is selected as the transmitting UAV, and the remaining UAVs act as backscattering tags. The transmitting UAV serves N communication users and receives K signals through a co-located or distributed receiving array. u The reflected signals of the backscattering tags can be isolated between different transmitting UAV groups using frequency division. The transmitting UAV includes a channel estimation module, a NOMA signal generation module, a channel error sample generation module, a beam optimization module, and a backscattering signal processing module; the backscattering tag includes a radio frequency energy receiving unit, a reflection coefficient control unit, a state data modulation unit, and a spreading code generation unit.

[0061] For example, the system may include 5 fixed formation drones and 4 ground communication users. Each drone is equipped with 8 transmitting antennas. One drone is selected as the transmitting drone in the 2.4 GHz band, and the other 4 drones are used as backscatter tags. The transmitting drone receives the reflected signals of the 4 backscatter tags through a co-located receiving array.

[0062] In some embodiments, the transmitting antenna array can be a uniform linear array, a planar array, a circular array, a sparse array, a distributed array, or a reconfigurable smart surface-assisted array. The specific form of the transmitting antenna array is not limited in the embodiments of this application.

[0063] In some embodiments, the backscatter tag can be installed on other drones, or on unmanned vehicles, robots, logistics equipment, passive sensors, or fixed infrastructure, as long as the backscatter tag can use the downlink signal as radio frequency excitation and reflect the modulated signal. The embodiments of this application do not limit the mounting carrier of the backscatter tag.

[0064] In some embodiments, such as Figure 1 As shown, the related system of the NOMA-assisted UAV backscattering sensing integrated beamforming method provided in the embodiments of this application can be a NOMA-assisted UAV backscattering sensing integrated system. The overall architecture of this system is based on an urban scenario and consists of four core nodes: the launching UAV, the ground NOMA user group, the backscattering tag node, and other UAVs in the formation.

[0065] The core of the system is a transmitter UAV equipped with a Uniform Linear Array (ULA), which serves as the central transmitting and receiving hub of the system. It transmits signals to ground users, backscatter tags, and other UAVs in the formation simultaneously through multi-antenna beamforming, and receives backscatter echo and sensing echo signals.

[0066] Multiple users are distributed across the ground area, forming a non-orthogonal multiple access (NOMA) user group. The launching UAV transmits data downlink to the ground users via communication signals, including at least beamforming weights Wn. Multiple users achieve non-orthogonal multiplexing and share the same radio resources in the power domain, frequency domain, or time domain.

[0067] The system deploys multiple backscatter tags. These tags do not need to actively transmit radio frequency signals; instead, they transmit information back by reflecting and modulating the carrier wave transmitted by the drone. The drone transmits information through the tag-side sensing channel. After the transmitted signal reaches the tag, the tag modulates the incident signal by backscattering according to its own reflection coefficient, forming backscatter signals that are reflected back to the transmitting drone, thus realizing low-power tag information collection and positioning perception.

[0068] Other UAVs in the formation, together with the launching UAV, constitute the UAV formation. The launching UAV uses sensing signals to detect and perceive other UAVs in the formation, achieving relative position awareness and coordinated control within the formation.

[0069] The launching UAV simultaneously achieves three functions through a unified multi-antenna beamforming waveform: first, NOMA downlink communication for ground users; second, excitation signal transmission and reflected echo reception for backscatter tags; and third, radar sensing and detection for other UAVs in the formation. These three functions share the same transmission waveform and RF link. Through joint design in the signal domain and separate processing at the receiver, communication and sensing are integrated, forming a complete NOMA-assisted UAV backscatter sensing integrated system.

[0070] Please refer to Figure 2 , Figure 2 The flowchart illustrates the steps of the NOMA-assisted UAV backscattering inductive beamforming method provided in some embodiments of this application. It should be understood that, although... Figure 2 The flowcharts illustrating subsequent steps show the execution order of some method steps. However, based on different design needs in practical applications, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application embodiment can, of course, employ a different execution order of method steps than shown in the figures. That is, Figure 2 The order of the method steps shown does not constitute a limitation on the execution logic order of the NOMA-assisted UAV backscattering inductive beamforming method provided in the embodiments of this application. Any other method based on this method is not limited to this method. Figure 2 Reasonable changes to the sequence of steps shown should be included within the protection scope of the NOMA-assisted UAV backscattering inductive beamforming method provided in the embodiments of this application.

[0071] like Figure 2 As shown, in some embodiments, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application may include steps S201 to S205 as shown below.

[0072] Step S201: Obtain the current estimated values ​​and channel error statistics of multiple communication channels of the launching UAV.

[0073] It should be noted that the launching UAV uses a non-orthogonal multiple access method to superimpose the communication signals of multiple communication users in the power domain to form a downlink signal, and multiplexes the downlink signal into radio frequency excitation signals for multiple backscatter tags; the multiple communication channels include the user channel from the launching UAV to the multiple communication users, and the forward channel from the launching UAV to the multiple backscatter tags.

[0074] Among these, Non-Orthogonal Multiple Access (NOMA) is a multi-user access method where the transmitter superimposes the signals of multiple users in the power domain and transmits them on the same time-frequency resource, and the receiver performs Serial Interference Cancellation (SIC) according to a preset decoding order. Multiple users share the same time-frequency resource, which can reduce interference between users on the same frequency and improve the multi-user access capability under limited spectrum and spatial degrees of freedom.

[0075] In addition, backscatter tags can be low-power passive or semi-passive devices without active radio frequency transmission links. The backscatter tag receives the energy of the downlink signal through the radio frequency energy receiving unit, adjusts its own reflection coefficient through the reflection coefficient control unit, and modulates its own status symbol onto the reflected signal of the downlink signal through the status data modulation unit. Thus, it completes the back transmission of its own information by using the downlink signal as radio frequency excitation, without the need for active signal transmission, which can reduce the energy consumption of airborne active transmission equipment.

[0076] Furthermore, the current estimate can be the channel state information estimation result obtained by the launch UAV through channel estimation within a coherent interval and which remains unchanged within that coherent interval; the channel error statistical parameters can be statistics that characterize the distribution characteristics of the channel estimation error, such as the covariance matrix of the channel estimation error, the mean error, or the range of error variation, etc. The channel error statistical parameters are used to characterize the degree of deviation of the current estimate from the real channel.

[0077] The channel estimation module of the launching UAV can estimate the user channels from the launching UAV to each communication user and the forward channels from the launching UAV to each backscattering tag by receiving pilot signals fed back from each communication user and each backscattering tag, thus obtaining the current estimated value of each communication channel within the current coherence interval. Simultaneously, the channel estimation module can statistically analyze the estimation errors of each communication channel based on the residual errors of historical channel estimations or offline calibration results, obtaining channel error statistical parameters. The NOMA signal generation module of the launching UAV superimposes the data symbols of each communication user in the power domain according to a set power allocation ratio to form a downlink signal. This downlink signal carries the communication data of each communication user and serves as the radio frequency excitation signal for each backscattering tag. Therefore, the same downlink signal simultaneously carries three functions: multi-user communication, tag excitation, and spatial sensing, without requiring independent sensing waveforms or additional spectrum resources.

[0078] For example, the launching UAV is equipped with 8 transmitting antennas, serving 4 ground communication users and exciting 3 backscatter tags. The channel estimation module performs channel estimation every 5 milliseconds in the coherent interval and calculates the covariance matrix of the channel estimation error based on the estimation residuals of the most recent 100 coherent intervals, which is used as the channel error statistical parameter. The NOMA signal generation module superimposes the data symbols of the 4 communication users into a downlink signal according to the power allocation coefficient of each communication user and transmits it.

[0079] In some embodiments, the non-orthogonal multiple access (NOMA) method can be replaced by dynamic user packet NOMA, code domain NOMA, multi-carrier NOMA, or other multi-user transmission methods with non-orthogonal superposition and interference cancellation functions. The specific implementation of non-orthogonal superposition transmission is not limited in the embodiments of this application.

[0080] Step S202: Extract channel error samples based on the channel error statistics parameters, and superimpose the channel error samples onto the current estimated value to generate the current channel implementation.

[0081] It should be noted that the channel error sample can be a random sample generated based on the channel error statistical parameters, representing the residual estimation error of the true channel relative to the current estimate; the current channel realization can be a specific channel value obtained by superimposing the channel error sample onto the current estimate, used to represent a possible situation of the true channel under channel uncertainty. Since the true channel is jointly determined by the current estimate and the residual estimation error, the launching UAV directly samples the channel uncertainty by extracting channel error samples, thus eliminating the need to assume that the channel error follows a specific distribution, or to derive a closed-form expectation expression for the random index.

[0082] The channel error sample generation module for launching UAVs can generate random numbers that conform to the channel error distribution characteristics based on channel error statistical parameters, obtain one or more channel error samples, and add the channel error samples to the current estimated value to generate the current channel implementation. This current channel implementation is used for the generation of subsequent beam vectors.

[0083] For example, when the channel estimation error follows a complex Gaussian distribution with a mean of zero and a covariance matrix equal to the given error covariance matrix, the channel error sample generation module can directly sample the channel error samples from this complex Gaussian distribution; the diagonal elements of the error covariance matrix can be 0.05, indicating that the average power of the channel estimation error is 5% of the signal power; after superimposing the sampled channel error samples with the current estimate, the current channel implementation used in this iteration is obtained.

[0084] In some embodiments, the channel error samples are generated based on one or more of the channel estimation error covariance, historical pilot residuals, offline calibration results, or empirical distributions, and solving the joint beamforming optimization problem does not require the optimization objective and the stochastic constraints to have closed-form expectation expressions.

[0085] In some embodiments, channel errors can be modeled using a Gaussian error model, a bounded error model, an ellipsoidal uncertainty set model, a mixed distribution model, an empirical distribution model, or a historical residual resampling model. This application only requires that samples representing channel uncertainty can be extracted, without limiting the specific distribution form of channel errors, thus making it applicable to various channel error distributions and equipment error statistical characteristics.

[0086] Step S203: Construct a joint beamforming optimization problem based on the current channel.

[0087] It should be noted that the joint beamforming optimization problem takes the weighted sum of the average communication rate of the multiple communication users and the average RF excitation power of the multiple backscattering tags as the optimization objective, and takes the communication rate of each of the multiple communication users as not lower than the corresponding rate threshold, the average power difference of different backscattering tag directions as not exceeding the first threshold, and the average cross-correlation of the tag detection signals as not exceeding the second threshold as random constraints, and the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix as deterministic constraints. Among these, the average communication rate can be the rate index obtained by averaging the communication rate of the communication users with the channel uncertainty represented by the channel error samples; the average RF excitation power can be the power index obtained by averaging the RF excitation power obtained by the backscatter tag from the downlink signal with the channel error samples, and the sum of the RF excitation power of all backscatter tags can be used as the perceived utility of the system; the stochastic constraint can be the constraint in which the channel correlation quantity in the constraint condition changes randomly with the channel error samples, and thus is imposed in the statistical form of the channel error samples; the deterministic constraint can be the constraint that does not change with the channel error samples and is directly determined by the hardware conditions; the rank-one condition can be the condition that the rank of the beam covariance matrix is ​​equal to 1, and only the beam covariance matrix that satisfies the rank-one condition can be reduced to a physically implementable beam vector without loss.

[0088] The beam optimization module for launching UAVs can be implemented based on the current channel, constructing a joint beamforming optimization problem that simultaneously considers communication and sensing performance. Specifically, the beam optimization module can use the weighted sum of the normalized conditional average communication rate and the normalized average tag excitation power as the optimization objective. The communication weight and sensing weight can be set according to the service's emphasis on communication and sensing, and the normalization factor is used to eliminate the dimensional differences between communication and sensing indicators. Simultaneously, the beam optimization module imposes the following constraints on the optimization problem: the average rate of each communication user is not lower than the corresponding rate threshold to ensure minimum communication quality for each user; the average power difference between any two backscattering tag directions does not exceed a first threshold to limit the excitation power difference between different backscattering tag directions and ensure fairness in multi-tag sensing; the average cross-correlation of tag detection signals does not exceed a second threshold to limit mutual interference between detection signals from different tag directions and ensure the distinguishability of multi-tag sensing; each antenna meets preset transmit power constraints; and the beam covariance matrix satisfies the positive semi-definite condition and the rank-one condition. Since both the user channel and the tag forward channel have estimation errors, the user rate, tag excitation power, inter-tag power difference, and cross-correlation of the probe signal in the above constraints are all random. Therefore, applying them in the form of random constraints can improve the beam design's adaptability to channel estimation errors and avoid violating actual constraints when optimizing based solely on estimated channels.

[0089] For example, the communication weight can be 0.6, the perception weight can be 0.4, the rate threshold for each communication user can be 1 bit / s / Hz, the first threshold can be 3dB, the second threshold can be the linear value corresponding to -20dB, the upper limit of the total transmit power of the drone can be 20dBm, and each antenna can adopt total power constraint or daily antenna power constraint.

[0090] Step S204: Solve the joint beamforming optimization problem based on the channel error samples to obtain a beam vector that satisfies the rank-one condition; the beam vector is used for the transmitting UAV to transmit the downlink signal.

[0091] It should be noted that the solution based on channel error samples can be a stochastic optimization solution that unifies the optimization objective and stochastic constraints in the joint beamforming optimization problem into the expectation form of a stochastic sampling function, and uses successively extracted channel error samples to approximate the expectation. This solution method only requires that samples representing channel uncertainty can be extracted, and does not require that the stochastic communication and sensing indicators have closed-form expectation expressions, thus it can be applied to various channel error distributions and equipment error statistical characteristics.

[0092] In addition, the beam vector can be a vector composed of the transmission weighting coefficients of each antenna element in the transmitting antenna array of the launching UAV. The launching UAV weights the downlink signal through the beam vector and transmits it through each antenna element, thereby forming a transmission beam in space pointing towards each communication user and each backscatter tag.

[0093] The beam optimization module of the launching UAV can use the anchor-assisted penalty hybrid constraint stochastic successive convex approximation algorithm to solve the constructed joint beamforming optimization problem and obtain the beam vector that satisfies the rank-one condition. The launching UAV transmits downlink signals according to the generated beam vector, which can form a transmission beam that meets the excitation and sensing requirements in the direction of each backscattering tag while ensuring the communication rate of each communication user.

[0094] For example, the beam optimization module for launching UAVs can first initialize the beam covariance matrix, rate auxiliary variable, random surrogate function, average channel anchor point and outer rank-one penalty parameter, and then enter a two-layer iteration consisting of inner random successive convex approximation and outer rank-one penalty enhancement. After the iteration converges, the beam vector that satisfies the rank-one condition is extracted from the beam covariance matrix.

[0095] In some embodiments, the beam optimization module for launching UAVs can extract channel error samples and generate the current channel realization in each iteration, construct a solvable form of the optimization objective and random constraints at the current beam variables based on the current channel realization, and gradually update the beam variables until convergence to obtain a beam vector that satisfies the rank-one condition.

[0096] For example, the beam optimization module extracts one channel error sample in each inner layer iteration, and expresses the optimization objective and random constraints as the expected form of the sampling function of the channel error sample. The beam variables are updated iteratively by successive convex approximation. When the change in beam variables between two adjacent iterations is less than 1e-3, the inner layer is determined to be converged.

[0097] In some embodiments, the optimization problem can be expressed as the weighted sum of the normalized conditional average communication rate and the average tag excitation power, with the objective being: .

[0098] in, and These are communication and sensing weights, respectively. and is the normalization factor. {Wn} is the set of beamforming vectors for each communication user to be optimized, and Wn is the beamforming vector of the nth communication user; P represents the conditional average total communication rate of multiple communication users under channel error samples. s The sum of the radio frequency excitation power of all backscattered tags, i.e., the sensing utility; E represents the current estimate of multiple communication channels. ξ Let the channel estimation error be a random vector. Indicates that given the current estimate Under the condition of channel estimation error E ξ Find the expected value under the given conditions.

[0099] The optimization process simultaneously satisfies the following: the average speed for each user is not lower than the threshold. The average cross-correlation of the tag detection signals does not exceed the threshold. The average power difference between any two tag directions does not exceed the threshold. Each antenna meets the preset power constraints; the beam covariance matrix meets the requirements of positive semi-definiteness and rank one.

[0100] In some embodiments, the solution process can be performed using the interior point method, semidefinite programming, second-order cone programming, alternating direction method of multipliers (ADMM), or an embedded solver. The specific solver is not limited in the embodiments of this application.

[0101] Step S205: Despread the reflected signals generated by the multiple backscattering tags based on the downlink signal in the code domain to obtain the despread signal.

[0102] It should be noted that code domain despreading is a processing method where the receiver uses the spreading sequence corresponding to each backscattering tag to perform correlation operations on the superimposed reflected signals formed by the simultaneous reflections of multiple backscattering tags, thereby distinguishing and separating the reflected signal components of each backscattering tag in the code domain. The spreading code generation unit of each backscattering tag generates its own corresponding spreading sequence, and the state data modulation unit modulates its own state symbol and the corresponding spreading sequence onto the reflected signal of the downlink signal. The spreading sequences corresponding to different backscattering tags are orthogonal to each other or have low cross-correlation, so that multiple backscattering tags can reflect simultaneously on the same time-frequency resource without confusion.

[0103] The backscatter signal processing module of the launching UAV can receive the superimposed reflection signals of multiple backscatter tags through co-located or separate receiving arrays. The superimposed reflection signals are correlated with the spread spectrum sequence corresponding to each backscatter tag, and the correlation results are received and merged to obtain the despread signal corresponding to each backscatter tag.

[0104] For example, three backscattering tags each use three mutually orthogonal Walsh sequences of length 8 as spreading sequences. The backscattering signal processing module performs correlation operations between the superimposed reflected signal received by the receiving array and the Walsh sequence of the first backscattering tag, thereby separating the signal component of the first backscattering tag from the superimposed reflected signal. The signal components of the other backscattering tags are suppressed because the spreading sequences are orthogonal, and so on, to obtain the despread signal corresponding to each backscattering tag.

[0105] In some embodiments, the spreading sequence may be Walsh code, PN code, Gold code or Zadoff-Chu sequence. The specific code type of the spreading sequence is not limited in the embodiments of this application.

[0106] Step S206: Recover the state information of the multiple backscattering tags based on the despread signal and estimate the angle of arrival of each of the multiple backscattering tags.

[0107] It should be noted that the status information can be the status data that the device attached to the backscatter tag needs to transmit back, such as the battery level, location, speed, and mission progress of the drone or logistics equipment; the angle of arrival can be the angle of arrival of the reflected signal of the backscatter tag to the receiving array of the transmitting drone. The angle of arrival is used to characterize the spatial orientation of the backscatter tag relative to the transmitting drone.

[0108] The backscatter signal processing module of the launching UAV can perform state symbol estimation on the despread signal corresponding to each backscatter tag to recover the state information of each backscatter tag. At the same time, the backscatter signal processing module can use the phase difference between the despread signals received by each element of the receiving array to estimate the angle of arrival of each backscatter tag through array signal processing methods, thereby realizing the integrated processing of tag identification, state decoding and orientation perception.

[0109] For example, the backscatter signal processing module can use least squares estimation to recover the state symbols of each backscatter tag from the despread signal, and use the MUSIC algorithm to construct the array covariance matrix based on the despread signal and perform spectral peak search to estimate the angle of arrival of each backscatter tag. The angle estimation accuracy can reach within 1 degree.

[0110] In some embodiments, the state information can be estimated using the least squares (LS) method or the minimum mean square error (MMSE) method, and the angle of arrival can be estimated using the MUSIC, ESPRIT, or sparse reconstruction methods. The specific algorithms for state estimation and angle estimation are not limited in the embodiments of this application.

[0111] In this embodiment, a UAV is launched to acquire current estimates and channel error statistics of multiple communication channels. Channel error samples are extracted based on the channel error statistics to generate the current channel implementation. A beam vector is generated based on the current channel implementation, and a downlink signal carrying both communication and excitation functions is transmitted. Then, code domain despreading, state information recovery, and angle-of-arrival estimation are performed on the reflected signals of multiple backscatter tags. Thus, compared to traditional solutions where communication waveforms, sensing waveforms, and state return links are independent, this embodiment integrates multi-user communication, backscatter tag excitation, UAV state return, and spatial orientation sensing onto the same downlink signal, the same spectrum, and the same multi-antenna hardware. This ensures both communication and sensing performance even under imperfect channel state information conditions, reducing the cost of independent sensing devices, active state return links, and additional spectrum resources, and improving the utilization efficiency of the limited communication resources of the UAV.

[0112] Furthermore, this embodiment constructs a joint beamforming optimization problem based on the current channel. The optimization objective is a weighted sum of average communication rate and average RF excitation power. Stochastic constraints include minimum user rate, tag directional power fairness, and cross-correlation of detection signals. Deterministic constraints include antenna power and rank-one conditions. This problem is then solved based on channel error samples. Thus, this embodiment can unify multi-user communication, backscatter tag excitation, inter-tag power fairness, and detection distinguishability into a single optimization model under imperfect channel state information conditions. This improves the beamforming design's adaptability to channel estimation errors and avoids the problem of violating practical constraints when performing deterministic design based solely on estimated channels.

[0113] In some embodiments, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application may further include the following steps: Based on the channel conditions corresponding to the current estimated value, the multiple communication users are sorted to obtain the serial interference cancellation decoding order of the multiple communication users; the serial interference cancellation decoding order remains fixed within a beam optimization interval and is re-determined after the channel estimation is updated.

[0114] It should be noted that the serial interference cancellation decoding order can be the order in which each communication user decodes other users' signals when performing serial interference cancellation. Communication users with different channel conditions decode and reconstruct the signal components of the already decoded users in this order, thereby reducing co-channel interference between users.

[0115] The launching UAV can sort multiple communication users according to the channel gain corresponding to the current estimate, for example, determine the serial interference cancellation decoding order according to the channel gain from large to small; the serial interference cancellation decoding order remains fixed within a beam optimization interval. When the channel estimate is updated, the launching UAV redetermines the serial interference cancellation decoding order according to the updated current estimate, so that the decoding order adapts to the channel change.

[0116] In some embodiments, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application may further include the following steps: The communication rate of the target communication user is determined based on the effective rate of the target communication user.

[0117] It should be noted that the target communication user is any one of the plurality of communication users, the effective rate is: the achievable rate of the target communication user itself, and the minimum achievable rate of the higher-order communication users whose signals of the target communication user need to be decoded in the serial interference cancellation decoding order of the plurality of communication users, and the average communication rate is the average of the effective rate over the channel error samples.

[0118] Furthermore, the effective rate can be the upper limit of the rate at which the target communication user can simultaneously ensure correct decoding by itself and correct elimination of the signal by higher-order communication users that need to decode the target communication user's signal in the decoding sequence under the serial interference cancellation mechanism. That is, the minimum of the achievable rate of the target communication user and the achievable rate of each higher-order communication user when decoding the target communication user's signal. The higher-order communication users can be the communication users whose decoding order is before the target communication user in the serial interference cancellation decoding sequence and who need to decode and eliminate the target communication user's signal first.

[0119] In some embodiments, the effective rate R of user n n The achievable rate R for the user itself and all higher-order users who need to decode the user's signal is taken as an example. n，l The minimum value in the range is used to obtain the total communication rate R of the NOMA user group. c : .

[0120] For example, for the second-ranked communication user in the serial interference cancellation decoding order, the first-ranked high-order communication user needs to decode the signal of the second-ranked communication user first and then cancel it. The effective rate of the second-ranked communication user is the smaller of its own achievable rate and the achievable rate of the first-ranked communication user when decoding its signal. The average communication rate of the communication user is obtained by averaging the effective rate over multiple sampled channel error samples.

[0121] Please refer to Figure 3 , Figure 3 for Figure 2 A detailed flowchart of step S204.

[0122] like Figure 3 As shown, in some embodiments, step S204 above: solving the joint beamforming optimization problem based on the channel error samples to obtain a beam vector that satisfies the rank-one condition may include steps S301 to S303 as shown below.

[0123] Step S301: In the inner iteration, based on the current channel, construct the convex surrogate function of the optimization objective and the random constraint at the current beam variable, form the current sample surrogate subproblem and solve it to obtain the current sample surrogate solution.

[0124] It should be noted that the convex surrogate function can be a convex function obtained by first-order linearizing the non-convex part of the random sampling function at the current beam variable and superimposing a proximal regularization term to limit the iteration step size. The convex surrogate function is used to approximate the original non-convex random optimization objective and random constraints in a convex form in each inner iteration. The current sample surrogate subproblem can be a convex optimization subproblem formed by replacing the original objective and original random constraints with the convex surrogate function. The current sample surrogate solution can be the solution obtained by solving the current sample surrogate subproblem based on the current channel realization corresponding to the channel error sample extracted in this iteration.

[0125] The beam optimization module for launching UAVs can extract channel error samples and generate the current channel implementation in the t-th inner iteration. It performs first-order linearization on the non-convex part of the sampling function at the current beam variable and adds a near-end regularization term to obtain a convex surrogate function for the optimization objective and random constraints, forming a surrogate subproblem for the current sample. This surrogate subproblem is then solved to obtain the current sample surrogate solution. Furthermore, the beam optimization module can update the surrogate function of the conditional expectation recursively by recursively fusing the sampling surrogate function corresponding to the current channel error sample with the historical surrogate function. The recursive update only needs to save the coefficients or sufficient statistics of the previous surrogate function, without needing to save all historical channel samples, thus enabling execution under limited storage conditions.

[0126] For example, the beam optimization module can use a first-order Taylor expansion to linearize the non-convex terms in the sampling function at the current beam variable. The coefficient of the near-end regularization term can be 0.1, and the weight coefficient during recursive fusion can be 1 / t, where t is the number of inner iterations. The t-th iteration only needs to add the coefficient of the current sampling proxy function to the coefficient of the previous proxy function according to the weight to complete the proxy function update.

[0127] In some embodiments, the convex surrogate function can be constructed using the first-order Taylor approximation, the difference convex approximation, the principalization function, or the trust region method. The embodiments of this application do not limit the specific construction method of the convex surrogate function.

[0128] Step S302: Recursively update the average channel anchor point based on historical channel error samples, and construct and solve the anchor point surrogate subproblem based on the average channel anchor point to obtain the anchor point surrogate solution.

[0129] It should be noted that the average channel anchor point can be the iterative reference point corresponding to the average channel constructed recursively using historical channel error samples. The average channel anchor point integrates the statistical information of historical samples, which can provide a stable update direction in a finite iteration stage and reduce the impact of extreme samples on the update direction. The anchor point surrogate subproblem can be a convex optimization subproblem with the same form as the current sample surrogate subproblem constructed based on the average channel anchor point. The anchor point surrogate solution can be the solution obtained by solving the anchor point surrogate subproblem. Since the average channel anchor point is updated recursively, the beam optimization module only needs to save the current anchor point value and does not need to save all historical channel error samples.

[0130] The beam optimization module for launching UAVs can fuse the extracted channel error samples with the existing average channel anchor points in each inner iteration according to the recursive formula to obtain the updated average channel anchor points. Based on the average channel anchor points, it constructs and solves the anchor point proxy subproblem in the same way as the current sample proxy subproblem to obtain the anchor point proxy solution.

[0131] For example, the average channel anchor point can be updated recursively using an arithmetic average method. That is, the average channel anchor point of the t-th iteration is equal to the difference between the average channel anchor point of the (t-1)-th iteration and the channel realization corresponding to the current channel error sample, divided by t, and then added to the average channel anchor point of the (t-1)-th iteration. The beam optimization module constructs an anchor point surrogate subproblem based on this average channel anchor point and calls the interior point method to solve it, thus obtaining the anchor point surrogate solution.

[0132] In some embodiments, the average channel anchor point can be recursively updated using arithmetic average, exponential moving average, finite window average, truncated average, confidence-weighted average, multi-anchor point, or historical generation average. The specific construction method of the average channel anchor point is not limited in the embodiments of this application.

[0133] Step S303: According to the anchor point weight that gradually decreases with the number of iterations, the current sample algebra and the anchor point algebra are weighted and fused, and the current beam variable is updated in combination with the variable update step size to obtain the updated beam variable.

[0134] It should be noted that the updated beam variable is a beam vector that satisfies the rank-one condition, and the updated beam variable serves as the current beam variable in the next inner layer iteration.

[0135] In addition, the anchor weight can be the weight of the anchor substitution in the weighted fusion. The anchor weight gradually decreases with the number of iterations, so that the algorithm can use historical information to stabilize the search direction in the early stage of the iteration and reduce the impact of anchor deviation on convergence in the later stage of the iteration. The variable update step size can be a step size parameter that controls the update magnitude of each iteration.

[0136] The beam optimization module for launching UAVs can perform a weighted summation of the current sample iteration and the anchor iteration according to the anchor weights that gradually decrease with the number of iterations to obtain a fused solution. Then, it updates the current beam variables in the direction of the fused solution according to the variable update step size to obtain the updated beam variables. The updated beam variables are then used as the current beam variables for the next inner layer iteration. This process is repeated until convergence.

[0137] For example, the anchor weight can be initially set to 0.5, and then multiplied by 0.99 for decay in each inner iteration. The variable update step size can be set to 0.3. In the t-th iteration, the beam optimization module multiplies the anchor weight by the anchor substitution solution and subtracts the anchor weight from the current sample substitution solution by 1. The two are added together to obtain the fused solution. Then, the difference between the current beam variable and the fused solution is multiplied by the variable update step size and superimposed on the current beam variable to obtain the updated beam variable.

[0138] In this embodiment, a surrogate subproblem is constructed by recursively updating the average channel anchor point based on the current channel error sample and historical samples in each inner iteration. This yields the current sample surrogate solution and the anchor point surrogate solution, which are then fused and updated according to progressively decaying anchor point weights. Thus, this embodiment can stabilize the search direction in the early stages of iteration using historical information provided by the average channel anchor point, and reduce anchor point bias in the later stages of iteration through the decay of anchor point weights. This reduces the interference of extreme samples on the update direction without storing a large number of historical samples, improving convergence stability and performance under finite sample conditions.

[0139] Please refer to Figure 4 , Figure 4 for Figure 3 A detailed flowchart of step S301.

[0140] like Figure 4 As shown, in some embodiments, the step S301 above, "constructing a convex surrogate function of the optimization objective and the random constraint at the current beam variable based on the current channel, forming a current sample surrogate subproblem and solving it to obtain the current sample surrogate solution", may include steps S401 to S403 as shown below.

[0141] Step S401: Determine whether there is a feasible point in the surrogate constraint set corresponding to the convex surrogate function.

[0142] It should be noted that the surrogate constraint set can be the set of all constraints formed by replacing the original random constraints with convex surrogate functions in the current sample surrogate subproblem; the feasible point can be the beam variable value that simultaneously satisfies all constraints in the surrogate constraint set. Since the random constraints are applied in the statistical form of channel error samples, the surrogate constraint set may not have feasible points in some iterations due to sample fluctuations. Therefore, a feasibility assessment needs to be performed before solving the problem.

[0143] Before solving the current sample surrogate subproblem, the beam optimization module for launching UAVs can determine whether there are feasible points in the surrogate constraint set by solving a feasibility check problem with constraint violation as the objective. If the optimal value of the feasibility check problem is zero, it is determined that there are feasible points in the surrogate constraint set; otherwise, it is determined that there are no feasible points.

[0144] For example, the beam optimization module can introduce non-negative relaxation variables for each constraint in the surrogate constraint set, and solve the feasibility test problem with the goal of minimizing the sum of all relaxation variables. When the optimal objective value is zero, it is determined that there is a feasible point, and when the optimal objective value is greater than zero, it is determined that there is no feasible point.

[0145] Step S402: In response to the determination that a feasible point exists, optimize the agent objective function under the agent constraint set to obtain the current sample agent solution.

[0146] It should be noted that the surrogate objective function can be the convex objective function obtained by substituting the optimization objective of the joint beamforming optimization problem with a convex surrogate function, that is, the convex approximation form of the communication and sensing weighted sum objective.

[0147] When the beam optimization module determines that there are feasible points in the surrogate constraint set, it means that the random constraints corresponding to the current sample can be satisfied. Under the constraints of the surrogate constraint set, the beam optimization module optimizes the surrogate objective function, solves the surrogate subproblem of the current sample, and obtains the surrogate solution of the current sample, thereby improving the weighted performance of communication and sensing while ensuring the feasibility of random constraints.

[0148] For example, the beam optimization module can call the interior point method to minimize the negative value of the surrogate objective function within the surrogate constraint set, and obtain the beam variable that satisfies all surrogate constraints and has the optimal objective value, which can be used as the current sample surrogate.

[0149] Step S403: In response to the determination that no feasible point exists, construct and solve the feasibility recovery subproblem with the goal of minimizing the maximum surrogate constraint violation, and obtain the current sample surrogate solution.

[0150] It should be noted that the feasibility recovery subproblem can be a convex optimization subproblem with the optimization objective of minimizing the maximum surrogate constraint violation, used to make the beam variable regress to the feasible region of the random constraints first when the set of surrogate constraints is infeasible; the maximum surrogate constraint violation can be the violation amount of the constraint with the greatest degree of violation among all surrogate constraints.

[0151] When the beam optimization module determines that there are no feasible points in the surrogate constraint set, it means that the random constraints corresponding to the current sample cannot be satisfied at the same time. The beam optimization module automatically switches to the feasibility recovery mode, constructs and solves the feasibility recovery subproblem with the goal of minimizing the maximum surrogate constraint violation, and uses the solution obtained as the surrogate solution for the current sample, so that the beam variables return to the feasible region of the random constraints first. After the constraints are restored to feasibility, the communication sensing target is optimized again.

[0152] For example, the beam optimization module can introduce an auxiliary variable to represent the maximum violation of all surrogate constraints. With the goal of minimizing this auxiliary variable and the constraint that the violation of each surrogate constraint does not exceed this auxiliary variable, a feasibility recovery subproblem is constructed and solved by calling a semidefinite programming solver. The resulting optimal beam variable is then used as the current sample surrogate.

[0153] In this embodiment, before solving the current sample surrogate subproblem, it is determined whether there is a feasible point in the surrogate constraint set. If a feasible point exists, the communication sensing objective is optimized; if no feasible point exists, it automatically switches to the feasibility recovery subproblem of minimizing the maximum surrogate constraint violation. Thus, this embodiment enables stochastic constraints such as minimum user rate, tag power fairness, and probe cross-correlation to gradually return to the feasible region during the iteration process, avoiding update divergence caused by infeasible iterations and ensuring the reliability of stochastic constraints.

[0154] Please refer to Figure 5 , Figure 5 for Figure 2 A schematic diagram of another detailed step in step S204.

[0155] like Figure 5 As shown, in some embodiments, step S204 above, which involves solving the joint beamforming optimization problem based on the channel error samples to obtain a beam vector that satisfies the rank-one condition, may also include steps S501 to S503 as shown below.

[0156] Step S501: The beam vector to be optimized is improved into a beam covariance matrix, and the rank-one violation degree, which characterizes the rank-one degree of the beam covariance matrix, is weighted by the penalty parameter and incorporated into the optimization objective to form an outer penalty objective.

[0157] It should be noted that the rank-one violation is the difference between the trace and the maximum singular value of the beam covariance matrix, and the rank-one violation is incorporated into the optimization objective after differentiable smoothing. In the inner-layer random successive convex approximation, the convex surrogate function is updated by recursively fusing the sampling surrogate function corresponding to the current channel error sample with the historical surrogate function, and only the coefficients or sufficient statistics of the previous surrogate function are saved during the update process.

[0158] It should be noted that variable lifting can be a transformation that replaces the beam vector to be optimized with the beam covariance matrix formed by the outer product of the beam vector and its conjugate transpose. Variable lifting transforms some quadratic constraints into affine constraints with respect to the beam covariance matrix, thereby giving the original non-convex problem a convex approximation form. At the same time as variable lifting, rate auxiliary variables can be introduced to transform rate constraints into a tractable form with respect to the beam covariance matrix.

[0159] Furthermore, the rank-one violation can be used as a measure of the degree to which the beam covariance matrix deviates from the rank-one matrix. For a non-zero positive semi-definite matrix, when the rank-one violation is 0, it is equivalent to the beam covariance matrix being a rank-one matrix. The outer penalty objective can be the objective function formed by weighting the rank-one violations of each beam covariance matrix with penalty parameters and incorporating them into the optimization objective.

[0160] Furthermore, differentiable smoothing can be used to smooth the non-differentiable terms in the rank-one violation, making the rank-one violation continuously differentiable. For example, Moreau envelope can be used to smooth the singular value summation term, and Log-Sum-Exp function can be used to smooth the maximum singular value term, thereby obtaining a smoothed rank-one penalty function, so that the outer penalty objective can be optimized by gradient-based methods.

[0161] The beam optimization module for launching UAVs can elevate the beam vector to be optimized into a beam covariance matrix and introduce a rate auxiliary variable. After differentiable smoothing of the rank-one violation of each beam covariance matrix, it is multiplied by a penalty parameter and incorporated into the optimization objective to form the outer penalty objective. In the inner random successive convex approximation, the beam optimization module recursively fuses the sampling surrogate function corresponding to the current channel error sample with the historical surrogate function to update the convex surrogate function. During the update process, only the coefficients or sufficient statistics of the previous surrogate function are saved, and it is not necessary to save all historical channel samples, so that the algorithm can be executed under limited sample and limited storage conditions.

[0162] For example, the rank-one violation can be taken as the difference between the sum of all singular values ​​of the beam covariance matrix and the largest singular value, i.e., the difference between the trace of the matrix and the largest singular value; the smoothing parameter of the Moreau envelope and the Log-Sum-Exp function can be taken as 0.01, and the initial value of the penalty parameter can be taken as 0.1; during recursive fusion, the sampling surrogate function of the t-th iteration can be fused with the previous surrogate function using a weight coefficient of 1 / t, and the update can be completed by simply saving the quadratic coefficient matrix, the linear coefficient vector and the constant term of the previous surrogate function.

[0163] In some embodiments, the rank-one violation can be constructed by the difference between the nuclear norm and the spectral norm, the difference between the trace and the largest eigenvalue, the sum of non-principal eigenvalues, eigenvalue entropy, or low-rank factorization; the differentiable smoothing process can be implemented using Moreau envelope, Log-Sum-Exp, Huber function, or other differentiable approximations. The specific form of the rank-one violation and its smoothing method is not limited in the embodiments of this application.

[0164] In some embodiments, variable boosting and rank-one penalty can be achieved by boosting the beam vector to a beam covariance matrix and introducing a rate auxiliary variable: .

[0165] Where Wn is the beam covariance matrix obtained by beam vector lifting of the nth communication user, and wn is the beamforming vector of the nth communication user. H This represents the conjugate transpose; q is the introduced rate auxiliary variable vector, qN is the rate auxiliary variable corresponding to the nth communication user, used to transform the rate constraint of this communication user into a tractable form with respect to the beam covariance matrix, and N is the number of communication users. T This indicates transpose.

[0166] Variable lifting transforms some quadratic constraints into affine constraints on the beam covariance matrix, but it requires ensuring that the rank of each beam covariance matrix is ​​1. Furthermore, instead of using post-optimized Gaussian randomization, the rank-one violation is directly added to the objective function: .

[0167] in, This represents the singular value of the corresponding beam covariance matrix.

[0168] For a non-zero positive semi-definite matrix, a rank-one violation of 0 is equivalent to the matrix being rank-one. To make the penalty function continuously differentiable, the Moreau envelope can be used to smooth the singular sum terms, and the Log-Sum-Exp can be used to smooth the maximum singular term, resulting in a smoothed rank-one penalty function.

[0169] In the In the next outer iteration, the smoothing penalty of each beam matrix is ​​added to the objective function, and the penalty parameter is gradually reduced. This led to a gradual increase in the severity of the rank-based punishment.

[0170] Step S502: In the outer layer iteration, after each inner layer random successive convex approximation converges based on the outer layer penalty target, the penalty parameter is adjusted to enhance the rank-one penalty.

[0171] It should be noted that the outer iteration can be an outer loop process that starts with the convergence result of the inner random successive convex approximation and gradually enhances the rank-one penalty by adjusting the penalty parameters. As the penalty parameters are adjusted, the penalty weight of the rank-one violation in the outer penalty target gradually increases, and the beam covariance matrix is ​​gradually driven to rank-one.

[0172] The beam optimization module for launching UAVs can determine whether the change in the inner layer variable is less than the inner layer threshold after each inner layer random successive convex approximation converges. If it is less than the inner layer threshold, the inner layer is considered to have converged, and the penalty parameter is reduced to gradually enhance the rank-one penalty. Then, the enhanced outer layer penalty target is used to enter the next inner layer random successive convex approximation. If the inner layer has not converged, the inner layer iteration continues.

[0173] For example, the initial value of the penalty parameter can be 0.1, and the penalty parameter is halved in each outer iteration, so that the penalty weight of the rank-one violation is doubled accordingly; the inner convergence criterion can be that the change in beam variable between two adjacent inner iterations is less than 1e-3.

[0174] Step S503: In response to the rank-one violation of the beam covariance matrix being less than the outer threshold, extract the beam vector from the eigenvector corresponding to the largest eigenvalue of the beam covariance matrix.

[0175] It should be noted that the outer threshold can be used as a threshold value to determine whether the beam covariance matrix has reached approximately rank one. When the rank one violation is less than the outer threshold, the beam covariance matrix can be regarded as a rank one matrix, and the eigenvector corresponding to its largest eigenvalue can be used to reconstruct the physically feasible beam vector without loss.

[0176] The beam optimization module for launching UAVs can determine whether the rank-one violation of all beam covariance matrices is less than the outer threshold after each outer iteration. If it is, the module performs eigenvalue decomposition on each beam covariance matrix, taking the product of the square root of the largest eigenvalue and the corresponding eigenvector as the beam vector. If there exists a beam covariance matrix with a rank-one violation not less than the outer threshold, the outer iteration continues. Because the rank-one penalty is directly embedded in the optimization process, the rank-one approximation and the satisfaction of random constraints are jointly completed during the optimization process, thus avoiding the target performance loss and random constraint failure caused by the existing positive semidefinite relaxation schemes that use Gaussian randomization to recover the beam vector after optimization.

[0177] For example, the outer threshold can be 1e-4. When the rank-one violation of all beam covariance matrices is less than 1e-4, the beam optimization module performs eigenvalue decomposition on the beam covariance matrix, extracts the eigenvector corresponding to the largest eigenvalue, and multiplies it by the square root of the largest eigenvalue to obtain the final beam vector used to transmit downlink signals.

[0178] In this embodiment, the beam vector to be optimized is elevated to a beam covariance matrix. The rank-one violation, after being processed by differentiable smoothing, is weighted by a penalty parameter and incorporated into the optimization objective to form an outer penalty objective. During the outer iteration, the penalty parameter is progressively adjusted to enhance the rank-one penalty. When the rank-one violation is less than the outer threshold, the beam vector is directly extracted from the eigenvector corresponding to the largest eigenvalue. Thus, this embodiment jointly achieves rank-one approximation and the satisfaction of stochastic constraints, ensuring that the final output beam simultaneously satisfies both stochastic constraints and physical feasibility requirements, avoiding the damage to target performance and the feasibility of stochastic constraints caused by the Gaussian randomization recovery process.

[0179] Next, a complete embodiment of the NOMA-assisted UAV backscattering inductive beamforming method provided in this application is presented.

[0180] The basic flow of the NOMA-assisted UAV backscattering inductive integrated beamforming method provided in this application embodiment includes the following steps S1 to S8: Step S1: Obtain the current estimates, error statistics parameters, communication sensing weights and constraint thresholds of the user channel and the tag forward channel, and determine the NOMA user order and SIC decoding order.

[0181] The power domain superposition of user data symbols can be performed by superimposing them according to the following formula when launching a drone: .

[0182] Where, x DL [ For the launch of the drone in the The downlink superimposed signal vector transmitted at each symbol time. is the symbol time index; Wn is the beamforming vector of the nth communication user; s n DL [ For the nth communication user in the nth communication period The downlink data symbols at each symbol time; N is the number of communication users.

[0183] Furthermore, the SIC decoding order is determined based on the current CSI estimate.

[0184] Step S2: Initialize the beam covariance matrix, rate auxiliary variable, random surrogate function, average channel anchor point, and outer rank-one penalty parameter.

[0185] Step S3: Extract channel error samples, generate the current channel implementation, and construct a convex surrogate function for the target and random constraints at the current variable.

[0186] Here, the objective function and random constraints are uniformly expressed as the expectation form of a random sampling function: .

[0187] Among them, F i (X) represents the expected form of the i-th objective function or stochastic constraint function, where i is the function index; i=0 corresponds to the objective function, and i≥1 corresponds to each stochastic constraint function; X is the set of variables to be optimized; h is the stochastic channel vector; f i (X,h) represents the sampled function value of the i-th function under a given channel h; E[.] represents the expectation of the random channel.

[0188] In the t-th inner layer iteration, channel error samples are extracted and the current channel implementation is generated. At the current variable, the non-convex part of the sampling function is linearized to order one, and a near-end regularization term is added. .

[0189] in, Let X be the convex surrogate function constructed for the i-th sampling function at the t-th iteration; X is the set of variables to be optimized. t h is the current value of the variable at the t-th iteration. t f represents the channel implementation sample extracted in the t-th iteration; i C (.) represents the convex portion retained in the i-th sampling function, f i NC (.) represents the non-convex part of the i-th sampling function. f i NC (X t ,ht ) represents the non-convex part at the current variable value X t The gradient at τ is the first-order linearization applied to the non-convex part; i A positive proximal regularization coefficient is used to ensure the strong convexity of the surrogate function; ||.||2 2 Represents the Euclidean norm; (.) T This indicates transpose.

[0190] The surrogate function that updates the conditional expectation using a recursive approach: .

[0191] in, Let i be the conditional expectation surrogate function after the t-th iteration update. For the tth The proxy function obtained in one iteration; Let X be the current variable value based on the t-th iteration. t and channel implementation of sample h t Constructed convex surrogate function; ρ t Let be the recursive average step size of the t-th iteration, satisfying ρ t The weights are ∈(0,1] and gradually decrease with the number of iterations, used to balance the weights of historical agent information and current sample information.

[0192] Recursive updates only need to save the coefficients or sufficient statistics of the previous surrogate function, and do not need to save all historical channel samples. If there is a feasible point in the current surrogate constraint set, the objective is optimized under the surrogate constraints; if there is no feasible point, the feasibility recovery mode is entered to minimize the maximum surrogate constraint violation, so that the variables return to the feasible region of random constraints first.

[0193] Step S4: Determine whether the current sample proxy constraints are feasible; if feasible, solve the proxy objective optimization subproblem; if not feasible, solve the recovery subproblem with the minimum maximum constraint violation to obtain the current sample proxy solution.

[0194] Step S5: Update the average channel anchor point, obtain the anchor point arithmetic in the same way, and update the beam variables by fusing the anchor point weights and variable step sizes.

[0195] To reduce update oscillations caused by limited and outlier samples, historical channel samples are used to construct an average channel anchor point: .

[0196] in, The average channel anchor point after the t-th iteration update is the arithmetic mean of the channel implementation samples extracted in the previous t iterations. For the tth Average channel anchor point after one iteration; ht This represents the current channel implementation sample extracted in the t-th iteration; t is the iteration number index. =0 indicates that the initial value of the average channel anchor point is a zero vector.

[0197] Based on the current sample and average channel anchor point Solve the surrogate subproblem to obtain the current sample surrogate solution. and anchor point representation Then, perform the fusion update according to the following formula: .

[0198] in, Update the step size for the variable. Anchor weights that gradually decrease in weight.

[0199] Anchor points provide a stable direction in the finite iteration phase, while their influence gradually weakens in the asymptotic phase. The average channel anchor point can also be replaced by an exponential moving average, a finite window average, a confidence-weighted average, or a historical iteration average.

[0200] Step S6: Determine whether the change in the inner variable is less than the threshold; if not, return to step S3; if so, reduce the outer penalty parameter.

[0201] Step S7: Determine whether the rank-one violation of all beam matrices is less than the outer threshold; if not, continue the outer iteration; if so, extract the beam vector from the maximum eigenvalue and corresponding eigenvector of each matrix.

[0202] Step S8: Send NOMA superimposed signal according to the obtained beam; after receiving the tag reflection signal, perform code domain despreading, state estimation and angle of arrival estimation.

[0203] In this process, the launching drone reuses the NOMA communication signal as the tag excitation signal. The radio frequency excitation power obtained by the k-th tag... for: .

[0204] in, For the forward channel to launch the drone to tag k.

[0205] The label will be a status symbol With spread spectrum sequence It is modulated onto the reflected signal.

[0206] Furthermore, code-domain despreading is performed on the superimposed reflected signals: .

[0207] Among them, y kUL [q] is the uplink signal vector at the q-th symbol time obtained after despreading the k-th backscatter tag, where q is the index of the despread symbol time; To receive drones in the The superimposed reflected signal vector received at each chip time; c k [ ] is the spreading code of the k-th backscatter tag in the th... The value taken at each chip moment, (.) represents conjugate; Lc is the spreading code length.

[0208] The despread signal is used for tag state estimation and angle of arrival estimation. The state symbol can be estimated using least squares; the angle can be processed using MUSIC, ESPRIT, or other array processing methods. By using the sum of all tag excitation power as the sensing utility and limiting the power difference between different tag directions and the cross-correlation of the detection signal, fairness and distinguishability of multi-tag sensing are ensured.

[0209] Please refer to Figure 6 , Figure 6 This is a block diagram of the NOMA-assisted UAV backscattering inductive beamforming method provided in an embodiment of this application.

[0210] like Figure 6 As shown, in some embodiments, the NOMA-assisted UAV backscattering inductive beamforming method provided in this application can be derived from, for example... Figure 1 The system shown includes several fixed formation UAVs and multiple ground communication users. It is implemented according to the complete implementation process shown below to achieve NOMA-assisted UAV backscattering inductive integrated random beamforming for imperfect channel state information.

[0211] I. Input Parameter Preprocessing Stage: The system receives four types of input parameters, which are then fed into corresponding preprocessing modules: the channel error statistics parameter input condition error sample generation module generates error samples under imperfect channel conditions; the channel user parameter input NOMA signal generation module generates non-orthogonal multiple access signals; and the user channel information and tag orientation channel estimation value are jointly input into the beamforming calculation module as the input basis for beamforming calculation.

[0212] II. Random Beamforming and Multi-Antenna Transmission Stage: The beamforming calculation module performs random beamforming calculations based on user channel information, tag orientation channel estimates, and conditional error samples, taking into account imperfect channel state information. This generates a beamforming weight vector, which is then output to the launch UAV multi-day array. The launch UAV multi-day array then performs weighted signal transmission based on the beamforming weights, forming a directional beam that simultaneously covers both the communication user and the backscattering tag.

[0213] III. Communication signal multiplexing and backscatter tag modulation stage: After the transmitted signal reaches multiple backscatter tag nodes, each tag node executes the same processing procedure in parallel: first, the communication signal is multiplexed into a tag modulation signal, which is then input into the backscatter tag modulation module to complete tag modulation; subsequently, the status symbol and the tag are superimposed and modulated to generate a backscatter superimposed signal that is reflected outward. The figure shows multiple parallel tag modulation links, enabling simultaneous backscatter communication between multiple tags.

[0214] IV. User-side reception and decoding stage: After the NOMA superimposed downlink communication signal arrives at the communication user group, each user performs serial interference cancellation decoding process to eliminate interference from other user signals step by step, and finally decodes to obtain the corresponding user data output, thus completing the downlink communication function.

[0215] V. Radar Sensing and State Estimation Phase: After the multiple backscattered superimposed signals are received, tag code correlation despreading is first performed to separate the reflected signals of different tags, thereby completing the angle of arrival estimation; then the multiple signals are received and merged, and least squares state estimation is performed to obtain tag direction and formation control information, realizing the radar sensing function in the integrated sensing and communication system.

[0216] VI. Closed-loop feedback control stage: The estimated tag orientation and formation control information are transmitted back to the beamforming calculation module through the dashed feedback link, serving as the update input for the next round of beamforming calculation. This forms a complete closed loop of "beamforming-transmission-backscattering-sensing estimation-feedback adjustment," enabling adaptive random beamforming and UAV formation coordinated control for imperfect channel state information.

[0217] In this embodiment, the launching UAV superimposes the communication signals of multiple communication users in the power domain to form a downlink signal in a non-orthogonal multiple access manner, and multiplexes the downlink signal into radio frequency excitation signals of multiple backscattering tags; each backscattering tag modulates its own state symbol and corresponding spreading sequence onto the reflected signal of the downlink signal; the launching UAV receives the superimposed reflected signals of multiple backscattering tags, and completes multi-tag simultaneous reflection sensing through code domain despreading, state estimation and angle of arrival estimation.

[0218] Please refer to Figure 7 , Figure 7 This is a flowchart of the anchor-assisted rank-one penalty hybrid constraint random successive convex approximation beamforming method involved in the embodiments of this application.

[0219] like Figure 7 As shown, in some embodiments, such as Figure 1 When the system shown executes the NOMA-assisted UAV backscattering inductive integrated beamforming method provided in the embodiments of this application, it can realize anchor-assisted rank-one penalty hybrid constraint random successive convex approximation beamforming according to the following complete implementation process.

[0220] I. Algorithm Input and Initialization Phase: First, the current channel state information (CSI) estimate, channel error statistics parameters, communication and sensing weight coefficients, and various constraint thresholds are input as the basic parameters for algorithm operation. Then, initialization operations are performed: the beamforming matrix, auxiliary optimization variables, surrogate functions for the objective and constraints, and penalty parameters for the outer iteration are initialized, establishing initial conditions for subsequent iterative optimization.

[0221] II. Outer layer iteration startup: After initialization, the k-th outer layer iteration begins. At the start of each outer layer iteration, the inner layer iteration counter and the historical average anchor point are initialized first. The anchor point is used to record the average state of the beam variables in the historical iterations, providing a reference benchmark for subsequent anchor-assisted optimization.

[0222] III. Inner-layer random sampling and surrogate function construction: After entering the inner iterative loop, the i-th channel error sample is first extracted. Based on this error sample and the current CSI estimate, the current channel realization is generated. Then, the historical average anchor point is updated using the current channel realization, and the new sample information is incorporated into the anchor point statistics. Based on the updated channel realization and anchor point, a sampling function for the objective function and stochastic constraints is constructed. The non-convex part is linearized and a proximal regularization term is added. Then, the surrogate function for the objective function and constraints is recursively updated, transforming the original non-convex, non-stochastic problem into a solvable convex surrogate problem.

[0223] IV. Solving the current sample proxy problem: After the surrogate function is constructed, it is first determined whether the surrogate constraints corresponding to the current sample are feasible. If feasible, the surrogate objective optimization subproblem for the current sample is directly solved. If not feasible, the recovery subproblem with the minimum maximum constraint violation is solved, and the feasibility of the problem is restored by minimizing the degree of constraint violation. After solving, the surrogate solution corresponding to the current sample is obtained.

[0224] V. Solving the Anchor Point Proxy Problem and Variable Fusion: After obtaining the current sample algebraic solution, the feasibility of the surrogate constraints corresponding to the anchor point is further determined: if feasible, the anchor point surrogate objective optimization subproblem is solved; if not feasible, the recovery subproblem with the minimum maximum constraint violation at the anchor point is solved. The anchor point algebraic solution is obtained after solving this subproblem. Subsequently, according to the preset anchor point weights and variable update step size, the current sample algebraic solution and the anchor point algebraic solution are weighted and fused to update the beamforming variables, completing one inner-layer iteration.

[0225] VI. Inner Iteration Convergence Judgment: After the variable update is completed, it is determined whether the change in the variable during the inner iteration is less than a preset threshold. If it is not less than the threshold, the process returns to the beginning of the inner iteration and continues to extract the next channel error sample to perform a new round of inner optimization. If it is less than the threshold, the inner iteration is considered to have converged, and the outer penalty parameter β is reduced. k It also enhances the operation of the rank-one penalty term, strengthening the constraint on the rank-oneness of the beam matrix.

[0226] VII. Outer Layer Iteration Convergence and Result Output: After the inner layer converges, it is determined whether the rank-one violation of the current beam matrix is ​​less than the preset threshold of the outer layer. If it is not less than the threshold, the outer layer iteration starts and the (k+1)th outer layer iteration begins to continue optimization. If it is less than the threshold, the algorithm is determined to have converged as a whole. Eigenvalue decomposition is performed on the final beam matrix to extract the maximum eigenvalue and the corresponding eigenvector. Finally, the near-rank-one NOMA communication and backscatter sensing joint beamforming result is output, completing the entire beamforming optimization process.

[0227] Please see Figure 8 This application provides a NOMA-assisted unmanned aerial vehicle backscattering inductive integrated beamforming system.

[0228] like Figure 8 As shown in the embodiments of this application, the NOMA-assisted UAV backscattering inductive integrated beamforming system includes: The acquisition module is used to acquire the current estimated values ​​and channel error statistics of multiple communication channels of the launching UAV; the launching UAV superimposes the communication signals of multiple communication users in the power domain to form a downlink signal in a non-orthogonal multiple access manner, and multiplexes the downlink signal into radio frequency excitation signals of multiple backscatter tags; the multiple communication channels include the user channel from the launching UAV to the multiple communication users, and the forward channel from the launching UAV to the multiple backscatter tags; A beam vector generation module is used to extract channel error samples based on the channel error statistical parameters, and superimpose the channel error samples onto the current estimated value to generate the current channel implementation; and to construct a joint beamforming optimization problem based on the current channel implementation; the joint beamforming optimization problem has the following stochastic constraints: the weighted sum of the average communication rate of the multiple communication users and the average RF excitation power of the multiple backscattering tags as the optimization objective; the communication rate of each of the multiple communication users is not lower than the corresponding rate threshold; the average power difference of different backscattering tag directions does not exceed a first threshold; and the average cross-correlation of the tag detection signals does not exceed a second threshold; and the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix are used as deterministic constraints; the joint beamforming optimization problem is solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition; the beam vector is used for the launching UAV to transmit the downlink signal; The integrated inductive beamforming module is used to perform code domain despreading on the reflected signals generated by the multiple backscattering tags based on the downlink signal to obtain the despread signal; and to recover the state information of the multiple backscattering tags based on the despread signal and estimate the angle of arrival of each of the multiple backscattering tags.

[0229] In some embodiments, the beam vector generation module is further configured to determine the communication rate of the target communication user based on the effective rate of the target communication user; the target communication user is any one of the plurality of communication users, the effective rate is: the achievable rate of the target communication user itself, and the minimum value among the achievable rates of the higher-order communication users whose signals of the target communication user need to be decoded in the serial interference cancellation decoding sequence, and the average communication rate is the average of the effective rate over the channel error samples.

[0230] In some embodiments, the beam vector generation module is further configured to, in the inner iteration, construct a convex surrogate function of the optimization objective and the random constraint at the current beam variable based on the current channel, form a current sample surrogate subproblem and solve it to obtain the current sample surrogate solution; recursively update the average channel anchor point based on historical channel error samples, and construct and solve an anchor point surrogate subproblem based on the average channel anchor point to obtain the anchor point surrogate solution; perform weighted fusion of the current sample surrogate solution and the anchor point surrogate solution according to the anchor point weight that gradually decays with the number of iterations, and update the current beam variable in combination with the variable update step size to obtain the updated beam variable; the updated beam variable is a beam vector that satisfies the rank-one condition, and the updated beam variable is used as the current beam variable in the next inner iteration.

[0231] In some embodiments, the beam vector generation module is further configured to determine whether there is a feasible point in the surrogate constraint set corresponding to the convex surrogate function; in response to determining that there is a feasible point, optimize the surrogate objective function under the surrogate constraint set to obtain the current sample surrogate solution; in response to determining that there is no feasible point, construct and solve a feasibility recovery subproblem with the objective of minimizing the maximum surrogate constraint violation to obtain the current sample surrogate solution.

[0232] In some embodiments, the beam vector generation module is further configured to elevate the beam vector to be optimized into a beam covariance matrix, and to incorporate the rank-one violation degree, which characterizes the rank-oneness of the beam covariance matrix, into the optimization objective after being weighted by a penalty parameter, forming an outer penalty objective; the rank-one violation degree is the difference between the trace and the maximum singular value of the beam covariance matrix, and the rank-one violation degree is incorporated into the optimization objective after being processed by differentiability smoothing; in the inner random successive convex approximation, the convex surrogate function is updated by recursively fusing the sampling surrogate function corresponding to the current channel error sample with the historical surrogate function, and only the coefficients or sufficient statistics of the previous surrogate function are saved during the update process; in the outer iteration, after each inner random successive convex approximation based on the outer penalty objective converges, the penalty parameter is adjusted to enhance the rank-one penalty; in response to the rank-one violation degree of the beam covariance matrix being less than the outer threshold, the beam vector is extracted from the eigenvector corresponding to the maximum eigenvalue of the beam covariance matrix.

[0233] In some embodiments, the beam vector generation module is further configured to sort the plurality of communication users based on the channel conditions corresponding to the current estimate, thereby obtaining the serial interference cancellation decoding order of the plurality of communication users; the serial interference cancellation decoding order remains fixed within a beam optimization interval and is re-determined after the channel estimate is updated.

[0234] It should be noted that the specific implementation of the NOMA-assisted UAV backscattering inductive beamforming system provided in this application is basically the same as the specific implementation of the NOMA-assisted UAV backscattering inductive beamforming method for launching UAVs described above, and will not be repeated here.

[0235] Please see Figure 9 This application also provides an electronic device that can be applied to the aforementioned NOMA-assisted UAV backscattering inductive beamforming system. The electronic device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned NOMA-assisted UAV backscattering inductive beamforming method.

[0236] In some embodiments, the electronic device may be a device that integrates the NOMA-assisted UAV backscattering sensing integrated beamforming system, or it may be an electronic device such as a smartphone, tablet, laptop, or desktop computer that controls the device that integrates the NOMA-assisted UAV backscattering sensing integrated beamforming system.

[0237] like Figure 9 As shown, the electronic device provided in this application embodiment may include: The processor 901 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 to execute the NOMA-assisted UAV backscattering inductive beamforming method of the embodiments of this application. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0238] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned NOMA-assisted UAV backscattering inductive beamforming method.

[0239] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0240] This application also provides a computer program product, including a computer program. The steps implemented by the computer program when executed by a processor are basically the same as those in the specific embodiments of the NOMA-assisted UAV backscattering inductive beamforming method described above, and will not be repeated here.

[0241] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of this application, and do not constitute a limitation on the technical solutions provided in this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in this application are also applicable to similar technical problems.

[0242] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0243] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0244] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0245] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0246] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding factor is divided by the following factor, or that the related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0247] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0248] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0249] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0250] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0251] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A NOMA-assisted UAV backscattering inductive beamforming method, characterized in that, The method includes: The system acquires current estimates and channel error statistics for multiple communication channels of the launching UAV; the launching UAV uses a non-orthogonal multiple access method to superimpose the communication signals of multiple communication users in the power domain to form a downlink signal, and multiplexes the downlink signal into radio frequency excitation signals for multiple backscatter tags; the multiple communication channels include user channels from the launching UAV to the multiple communication users, and forward channels from the launching UAV to the multiple backscatter tags; Based on the channel error statistics parameters, channel error samples are extracted and superimposed on the current estimated value to generate the current channel implementation. Based on the current channel, a joint beamforming optimization problem is constructed. The joint beamforming optimization problem takes the weighted sum of the average communication rate of the multiple communication users and the average RF excitation power of the multiple backscattering tags as the optimization objective, and takes the communication rate of each of the multiple communication users as not lower than the corresponding rate threshold, the average power difference of different backscattering tag directions as not exceeding the first threshold, and the average cross-correlation of the tag detection signals as not exceeding the second threshold as random constraints, and the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix as deterministic constraints. The joint beamforming optimization problem is solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition; the beam vector is used by the launching UAV to transmit the downlink signal; The reflected signals generated by the multiple backscattering tags based on the downlink signal are despread in the code domain to obtain the despread signal; Based on the despread signal, the state information of the multiple backscattering tags is recovered and the angle of arrival of each of the multiple backscattering tags is estimated.

2. The NOMA-assisted UAV backscattering inductive beamforming method according to claim 1, characterized in that, The method further includes: The communication rate of the target communication user is determined based on the effective rate of the target communication user; the target communication user is any one of the plurality of communication users, the effective rate is: the achievable rate of the target communication user itself, and the minimum achievable rate of the higher-order communication users whose signals of the target communication user need to be decoded in the serial interference cancellation decoding order of the plurality of communication users, and the average communication rate is the average of the effective rate over the channel error samples.

3. The NOMA-assisted UAV backscattering inductive beamforming method according to claim 1, characterized in that, The step of solving the joint beamforming optimization problem based on the channel error samples to obtain a beam vector that satisfies the rank-one condition includes: In the inner iteration, based on the current channel, a convex surrogate function for the optimization objective and the random constraint is constructed at the current beam variable, forming the current sample surrogate subproblem and solving it to obtain the current sample surrogate solution; The average channel anchor point is updated recursively based on historical channel error samples, and an anchor point proxy subproblem is constructed and solved based on the average channel anchor point to obtain the anchor point proxy solution. The current sample iteration and the anchor iteration are weighted and fused according to the anchor weight that gradually decreases with the number of iterations. The current beam variable is then updated in combination with the variable update step size to obtain the updated beam variable. The updated beam variable is a beam vector that satisfies the rank-one condition, and the updated beam variable is used as the current beam variable in the next inner layer iteration.

4. The NOMA-assisted UAV backscattering inductive beamforming method according to claim 3, characterized in that, The process of constructing a convex surrogate function for the optimization objective and the random constraints at the current beam variable based on the current channel, forming a surrogate subproblem for the current sample and solving it to obtain the surrogate solution for the current sample includes: Determine whether there is a feasible point in the set of surrogate constraints corresponding to the convex surrogate function; In response to the determination that a feasible point exists, the agent objective function is optimized under the agent constraint set to obtain the agent solution for the current sample; In response to the determination that no feasible point exists, a feasibility recovery subproblem is constructed and solved with the goal of minimizing the maximum surrogate constraint violation, thus obtaining the current sample surrogate solution.

5. The NOMA-assisted UAV backscattering inductive beamforming method according to claim 1, characterized in that, The step of solving the joint beamforming optimization problem based on the channel error samples to obtain a beam vector that satisfies the rank-one condition includes: The beam vector to be optimized is elevated to a beam covariance matrix, and the rank-one violation degree, which characterizes the rank-oneness of the beam covariance matrix, is weighted by a penalty parameter and incorporated into the optimization objective to form an outer penalty objective; the rank-one violation degree is the difference between the trace of the beam covariance matrix and the maximum singular value, and the rank-one violation degree is incorporated into the optimization objective after being processed by differentiability smoothing. In the outer layer iteration, after each inner layer random successive convex approximation converges based on the outer layer penalty target, the penalty parameter is adjusted to enhance the rank-one penalty. In response to the rank-one violation of the beam covariance matrix being less than the outer threshold, the beam vector is extracted from the eigenvector corresponding to the largest eigenvalue of the beam covariance matrix.

6. The NOMA-assisted UAV backscattering inductive beamforming method according to any one of claims 1 to 5, characterized in that, The method further includes: Based on the channel conditions corresponding to the current estimated value, the multiple communication users are sorted to obtain the serial interference cancellation decoding order of the multiple communication users; the serial interference cancellation decoding order remains fixed within a beam optimization interval and is re-determined after the channel estimation is updated.

7. A NOMA-assisted UAV backscattering inductive beamforming system, characterized in that, The system includes: The acquisition module is used to acquire the current estimated values ​​and channel error statistics of multiple communication channels of the launching UAV; the launching UAV superimposes the communication signals of multiple communication users in the power domain to form a downlink signal in a non-orthogonal multiple access manner, and multiplexes the downlink signal into radio frequency excitation signals of multiple backscatter tags; the multiple communication channels include the user channel from the launching UAV to the multiple communication users, and the forward channel from the launching UAV to the multiple backscatter tags; A beam vector generation module is used to extract channel error samples based on the channel error statistical parameters, and superimpose the channel error samples onto the current estimated value to generate the current channel implementation; and to construct a joint beamforming optimization problem based on the current channel implementation; the joint beamforming optimization problem has the following stochastic constraints: the weighted sum of the average communication rate of the multiple communication users and the average RF excitation power of the multiple backscattering tags as the optimization objective; the communication rate of each of the multiple communication users is not lower than the corresponding rate threshold; the average power difference of different backscattering tag directions does not exceed a first threshold; and the average cross-correlation of the tag detection signals does not exceed a second threshold; and the antenna power constraint of the launching UAV and the rank-one condition of the beam covariance matrix are used as deterministic constraints; the joint beamforming optimization problem is solved based on the channel error samples to obtain a beam vector that satisfies the rank-one condition; the beam vector is used for the launching UAV to transmit the downlink signal; The integrated inductive beamforming module is used to perform code domain despreading on the reflected signals generated by the multiple backscattering tags based on the downlink signal to obtain the despread signal; and to recover the state information of the multiple backscattering tags based on the despread signal and estimate the angle of arrival of each of the multiple backscattering tags.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the NOMA-assisted UAV backscattering inductive beamforming method according to claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the NOMA-assisted unmanned aerial vehicle backscattering inductive beamforming method as described in claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the NOMA-assisted unmanned aerial vehicle backscattering inductive beamforming method as described in claims 1 to 6.

Images