A method and system for secure communication of a drone with a movable antenna

Abstract

This invention discloses a secure communication method and system for unmanned aerial vehicles (UAVs) equipped with a movable antenna, relating to the field of wireless communication technology. The method includes: measuring the angle of arrival (Angle of Arrival) of an active eavesdropper's interference signal, estimating its location and the eavesdropping channel gain; inputting the UAV state and this information into a policy network, and outputting a joint action; executing the action to adjust the configuration and regain the channel gain; calculating the security rate and immediate reward; storing the experience in a replay pool and sampling to update network parameters; iterating until convergence, thereby achieving real-time security optimization of UAV communication. This invention utilizes the spatial degrees of freedom of a movable antenna, combined with deep reinforcement learning to achieve real-time decision-making, improving the overall security rate in multi-user scenarios and enhancing the physical layer security of UAV communication.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and specifically to a secure communication method and system for unmanned aerial vehicles (UAVs) equipped with a movable antenna. Background Technology

[0002] With the development of UAV communication and movable antenna technology, deploying movable antennas on UAV platforms to assist secure communication has become a new technological direction. In existing research, fixed antenna arrays can improve signal gain in the direction of legitimate users by optimizing beamforming; UAV-assisted communication can improve channel conditions between the UAV and legitimate users by adjusting the UAV's position, thereby enhancing secure communication performance. However, existing methods are still limited by factors such as fixed array structure, single-position optimization methods, and energy consumption, making it difficult to fully utilize the spatial degrees of freedom of the system.

[0003] Existing research has begun to focus on combining drone mobility with movable antenna technology. By 2025, researchers had already deployed movable antenna arrays on drone platforms to simultaneously utilize the drone's position adjustment capabilities and the local mobility of antenna elements. This is because the macroscopic movement of the drone and the microscopic adjustment of the movable antenna are complementary, which can improve secure communication performance while reducing reliance on frequent drone position adjustments.

[0004] However, existing research mainly focuses on passive eavesdropping scenarios and has not fully considered active eavesdroppers who may actively transmit jamming signals to legitimate users. It also has not adequately addressed the joint optimization problem between the UAV's position, the position of movable antenna elements, and the beamforming strategy under such active eavesdropping conditions. Because these three types of optimization variables are interdependent, and the system state dynamically changes with the UAV's movement, traditional optimization methods are difficult to solve efficiently. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a secure communication method and system for unmanned aerial vehicles (UAVs) equipped with movable antennas. By jointly optimizing the UAV position, movable antenna array element position, and transmit beamforming matrix using a deep reinforcement learning framework, it achieves physical layer secure communication that resists interference from active eavesdroppers and improves the overall security rate of multi-user systems.

[0006] On the one hand, a secure communication method for unmanned aerial vehicles (UAVs) equipped with a movable antenna includes:

[0007] S1, acquire the interference signal transmitted by the active eavesdropper to the legitimate user, measure the angle of arrival of the interference signal, obtain the azimuth and elevation coordinates of the interference signal, and determine the position coordinates of the active eavesdropper and the channel gain of the eavesdropping link based on the azimuth and elevation coordinates.

[0008] S2, obtain the current status information of the drone, and combine the current status information of the drone, the location coordinates of the active eavesdropper, and the channel gain of the eavesdropping link into the current time slot status;

[0009] S3, input the state of the current time slot into the policy network, and map out the joint action by performing forward propagation, feature extraction and nonlinear transformation on the state of the current time slot; the joint action includes the UAV target position, the target position of the movable antenna array element and the transmit beamforming matrix;

[0010] S4, adjust the positions of the UAV and antenna elements based on the UAV target position and antenna element target position in the joint operation, and configure the transmit beamforming matrix; reacquire the channel gain of each legitimate user and the channel gain of the eavesdropping link based on the adjusted UAV and antenna element positions to construct the state of the next time slot;

[0011] S5. Based on the channel gain of each legitimate user, the channel gain of the eavesdropping link, and the transmit beamforming matrix, calculate the signal-to-interference-plus-noise ratio (SIR) of each legitimate user and the eavesdropping SIR of the active eavesdropper. Based on the SIR of each legitimate user and the eavesdropping SIR of the active eavesdropper, calculate the security rate of each legitimate user and the total security rate.

[0012] S6. Calculate the instant reward based on the confidentiality rate of each legitimate user and the total confidentiality rate. Assemble the current time slot's state, joint action, instant reward and the next time slot's state into an experience sample and store it in a preset experience replay pool. Sample experience samples from the experience replay pool to update the policy network parameters.

[0013] S7. Repeat S2 to S6 until the change in the average value of the obtained instant reward is less than the preset threshold. The algorithm is then considered to have converged, and the final policy network parameters are obtained. Based on the final policy network parameters, the position of the UAV, the position of the movable antenna array elements, and the transmit beamforming matrix are controlled.

[0014] Furthermore, in S1, the position coordinates of the active eavesdropper are determined based on the azimuth and elevation coordinates, using the following calculation formula:

[0015] ;

[0016] Indicates the azimuth coordinate value; This represents the estimated pitch angle. Indicates the time slot for communication; Indicates the location of the active eavesdropper; This indicates the drone's flight altitude.

[0017] Furthermore, in S2, the current status information of the drone includes:

[0018] The spatial coordinates of the UAV are used to indicate the current spatial location of the UAV, serving as the input for the policy network to determine the target location of the UAV.

[0019] The guide rail position coordinates of each movable antenna element are used to indicate the distribution position of each antenna element on the current one-dimensional guide rail, and serve as the input basis for the strategy network to decide the target position of the antenna elements.

[0020] The ground position coordinates of each legitimate user are used to indicate the fixed or real-time position of each legitimate user on the ground, and serve as a reference position for calculating the channel gain and beamforming between the UAV and the legitimate users.

[0021] The channel gain between the drone and each legitimate user is used to indicate the signal propagation quality of each legitimate communication link in the current time slot, and to calculate the signal-to-interference-plus-noise ratio and security rate of each legitimate user.

[0022] Furthermore, S3 also includes constraint processing on the joint actions output by the policy network, specifically including:

[0023] The target position of the UAV is restricted to a preset UAV movable area, the target position of each movable antenna array element is restricted to the corresponding one-dimensional guide rail movement range, and the preset minimum spacing constraint is met between each movable antenna array element.

[0024] The transmit beamforming matrix is ​​subjected to power constraint processing based on the preset minimum spacing constraint so that the total transmit power of the UAV does not exceed the preset maximum transmit power.

[0025] The constrained UAV target position, the movable antenna array element target position, and the transmitted beamforming matrix are used as a combined action for UAV secure communication.

[0026] Furthermore, in S5, the confidentiality rate of each legitimate user... and total security rate The calculation formula is as follows:

[0027] ;

[0028] ;

[0029] in, Indicates taking a non-negative value; This represents the signal-to-interference-plus-noise ratio (SIR) for the i-th user in the k-th time slot; This represents the signal-to-interference-plus-noise ratio (SIR) of the active eavesdropper in the k-th time slot. Indicates the time slot for reinforcement learning; Indicates the number of legitimate users; This represents the i-th legitimate user.

[0030] Furthermore, in S6, instant rewards The calculation formula is as follows:

[0031] ;

[0032] in, This is the penalty coefficient; Penalties for violating minimum security rate constraints; Represents the reward function; Represents the state space of the k-th time slot; This represents the action space of the k-th time slot; This represents a part of the reward function, and is expressed as the total system security rate in the k-th time slot; This represents the time slot of the reinforcement learning algorithm.

[0033] On the other hand, a secure communication system for unmanned aerial vehicles (UAVs) equipped with a movable antenna includes:

[0034] The location coordinate acquisition module is used to acquire the interference signal transmitted by the active eavesdropper to the legitimate user, measure the angle of arrival of the interference signal, obtain the azimuth and elevation coordinate values ​​of the interference signal, and determine the location coordinates of the active eavesdropper and the channel gain of the eavesdropping link based on the azimuth and elevation coordinate values.

[0035] The time slot status acquisition module is used to acquire the current status information of the UAV and combine the current status information of the UAV, the position coordinates of the active eavesdropper, and the channel gain of the eavesdropping link into the status of the current time slot.

[0036] The joint action generation module is used to input the state of the current time slot into the policy network, and map out the joint action by performing forward propagation, feature extraction and nonlinear transformation on the state of the current time slot; the joint action includes the UAV target position, the target position of the movable antenna array element and the transmit beamforming matrix.

[0037] The time slot state construction module is used to adjust the positions of the UAV and antenna array elements based on the UAV target position and antenna array element target position in the joint operation, and configure the transmit beamforming matrix; based on the adjusted UAV and antenna array element positions, it reacquires the channel gain of each legitimate user and the channel gain of the eavesdropping link to construct the state of the next time slot;

[0038] The security rate calculation module is used to calculate the signal-to-interference-plus-noise ratio (SIR) of each legitimate user and the SIR of the active eavesdropper based on the channel gain of each legitimate user, the channel gain of the eavesdropping link, and the transmit beamforming matrix. Based on the SIR of each legitimate user and the SIR of the active eavesdropper, the security rate of each legitimate user and the total security rate are calculated.

[0039] The instant reward calculation module is used to calculate the instant reward based on the confidentiality rate of each legitimate user and the total confidentiality rate. It assembles the current time slot state, joint action, instant reward and the next time slot state into an experience sample and stores it in a preset experience replay pool. It samples experience samples from the experience replay pool to update the policy network parameters.

[0040] The strategy network parameter acquisition module is used to repeatedly execute the time slot state acquisition module to the instant reward calculation module until the change in the average value of the obtained instant rewards is less than a preset threshold. The algorithm is then considered to have converged, and the final strategy network parameters are obtained. Based on the final strategy network parameters, the position of the UAV, the position of the movable antenna array elements, and the transmit beamforming matrix are controlled.

[0041] The present invention adopts the above technical solution and has beneficial effects:

[0042] (1) This invention measures the angle of arrival of the interference signal of the active eavesdropper, accurately estimates its location and the channel gain of the eavesdropping link, so that the system can perceive the eavesdropping threat in real time. Based on this, in the subsequent beamforming design, the signal energy is accurately aimed at the legitimate user while the null trap is aimed at the eavesdropper, thereby effectively resisting malicious interference and eavesdropping, and significantly improving the physical layer security of UAV communication.

[0043] (2) This invention uses a deep reinforcement learning strategy network to fuse the current state of the UAV with the eavesdropper information, and then outputs the target position of the UAV, the target position of the movable antenna array element and the transmit beamforming matrix. Based on the channel gain feedback after execution, the security rate of each user is calculated. By making full use of the spatial degree of freedom of the movable antenna, the antenna array element layout is flexibly adjusted, thereby effectively improving the security rate of each user in the scenario where multiple legitimate users coexist, and finally maximizing the total security rate of the system.

[0044] (3) The present invention uses the state, action, instant reward and next state of each time slot as experience samples and randomly samples to update the policy network parameters. It iterates until the reward converges, so that the system does not need to obtain the precise instantaneous channel state information of the active eavesdropper. It has good real-time decision-making ability and environmental adaptability, and provides an efficient and robust engineering solution for actual UAV communication deployment. Attached Figure Description

[0045] Figure 1 This is a flowchart of a secure communication method for a drone equipped with a movable antenna, according to an embodiment of the present invention.

[0046] Figure 2 This is a diagram of a secure communication system for a drone equipped with a movable antenna, according to an embodiment of the present invention. Detailed Implementation

[0047] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0048] like Figure 1 As shown, the present invention provides a secure communication method for unmanned aerial vehicles (UAVs) equipped with a movable antenna, comprising:

[0049] S1. Acquire the interference signal transmitted by the active eavesdropper to the legitimate user, measure the angle of arrival of the interference signal, obtain the azimuth and elevation coordinates of the interference signal, and determine the position coordinates of the active eavesdropper and the channel gain of the eavesdropping link based on the azimuth and elevation coordinates (the equivalent channel gain of the eavesdropping link needs to be estimated by combining the preset statistical channel model).

[0050] Specifically, the location coordinates of the active eavesdropper are determined based on the azimuth and elevation coordinates, using the following calculation formula:

[0051] ;

[0052] Indicates the azimuth coordinate value; This represents the estimated pitch angle. Indicates the time slot for communication; Indicates the location of the active eavesdropper; This indicates the drone's flight altitude.

[0053] Specifically, in this embodiment, a scenario is simulated where a drone equipped with a one-dimensional movable antenna array sends confidential information to a legitimate user via a downlink at a fixed altitude. Flight, in The time slot, the drone's coordinates are There is on the ground A number of legitimate users, whose locations are represented as follows: Meanwhile, there is also an active eavesdropper on the ground. The one-dimensional movable antenna array is equipped with... A movable antenna, whose coordinates on a one-dimensional array are represented as follows: ,in, Represents the three-dimensional coordinates of the UAV; This represents the position coordinates of the nth time slot UAV on the horizontal x-axis; This represents the position coordinates of the nth time slot UAV on the horizontal y-axis; This represents the position coordinates of the i-th legitimate user in the n-th time slot along the x-axis. This represents the position coordinates of the i-th legitimate user in the n-th time slot along the y-axis.

[0054] Specifically, due to the interference signals emitted by active eavesdroppers ,in For the interference power of the active eavesdropper, therefore the first The signal received by a legitimate user is represented as follows: ,in, This represents the interference channel coefficient between legitimate users and active eavesdroppers. This represents the normalized jamming signal symbol sent by the active eavesdropper in the nth time slot.

[0055] Specifically, in this embodiment, the drone is set to be in the first... The downlink signal sent to the user in each time slot is ,in Send to users via drone terminal The beamforming vector is defined. Further legitimate users In the The signal received in each time slot is ,in For noise; set the signal table received by the active eavesdropper as follows: ,in Noise at the location of an active eavesdropper; This represents the confidential information symbol sent by the drone to the j-th legitimate user in the nth time slot. Represents a set consisting of M×I dimensional complex matrices; This represents the conjugate transpose of the channel vector between the UAV and the i-th legitimate user; This represents the interference channel coefficient between the active eavesdropper and the i-th legitimate user; This represents the conjugate transpose of the eavesdropping channel vector between the drone and the active eavesdropper.

[0056] Specifically, in this embodiment, the location of the eavesdropper is not precisely obtained, but is estimated by using the angle of arrival information of the interference signal emitted by the active eavesdropper. Therefore, the subsequent channel gain of the eavesdropping link is not based on the real CSI, but is equivalently modeled by combining the estimated location parameters with the statistical channel model.

[0057] S2, acquire the current status information of the drone, and combine the current status information of the drone, the location coordinates of the active eavesdropper, and the channel gain of the eavesdropping link into the current time slot status.

[0058] Specifically, the current status information of the drone includes:

[0059] The spatial coordinates of the UAV are used to indicate the current spatial location of the UAV, serving as the input for the policy network to determine the target location of the UAV.

[0060] The guide rail position coordinates of each movable antenna element are used to indicate the distribution position of each antenna element on the current one-dimensional guide rail, and serve as the input basis for the strategy network to decide the target position of the antenna elements.

[0061] The ground position coordinates of each legitimate user are used to indicate the fixed or real-time position of each legitimate user on the ground, and serve as a reference position for calculating the channel gain and beamforming between the UAV and the legitimate users.

[0062] The channel gain between the drone and each legitimate user is used to indicate the signal propagation quality of each legitimate communication link in the current time slot, and to calculate the signal-to-interference-plus-noise ratio and security rate of each legitimate user.

[0063] Specifically, in this scheme, the active eavesdropper is a non-cooperative node, therefore its real instantaneous channel state information is not directly obtained. First, the angle of arrival is measured by the interference signal emitted by the active eavesdropper, and the spatial position of the active eavesdropper is estimated based on information such as azimuth and elevation angles. Then, based on the estimated eavesdropper position and the current position of the UAV, the geometric parameters of the eavesdropping link are determined, and the eavesdropping link is modeled in conjunction with a preset statistical channel model, thereby obtaining the equivalent channel gain or statistical channel gain of the eavesdropping link.

[0064] Among them, the statistical channel model can adopt the corresponding path loss model, line-of-sight / non-line-of-sight propagation model, and small-scale fading statistical model according to the actual communication scenario. This application does not limit the specific statistical channel model.

[0065] The obtained eavesdropping link channel gain is used to characterize the channel quality of the eavesdropping link between the UAV and the active eavesdropper. On the one hand, it is input into the reinforcement learning policy network as part of the current time slot system state information, enabling the policy network to perceive the current communication conditions of the eavesdropping link and adjust the UAV position, movable antenna position, and transmit beamforming matrix accordingly. On the other hand, after performing the joint action, the received signal-to-interference-plus-noise ratio and corresponding eavesdropping rate of the active eavesdropper are calculated based on the updated eavesdropping link channel gain, and the confidentiality rate is further calculated together with the communication rate of the legitimate user, serving as the basis for optimizing the reinforcement learning policy.

[0066] S3, input the state of the current time slot into the policy network, and map out the joint action by performing forward propagation, feature extraction and nonlinear transformation on the state of the current time slot; the joint action includes the UAV target position, the target position of the movable antenna array element and the transmit beamforming matrix.

[0067] Specifically, in this embodiment, the joint actions output by the policy network are also subject to constraint processing, including:

[0068] The target position of the UAV is restricted to a preset UAV movable area, the target position of each movable antenna array element is restricted to the corresponding one-dimensional guide rail movement range, and the preset minimum spacing constraint is met between each movable antenna array element.

[0069] The transmit beamforming matrix is ​​subjected to power constraint processing based on the preset minimum spacing constraint so that the total transmit power of the UAV does not exceed the preset maximum transmit power.

[0070] The constrained UAV target position, the movable antenna array element target position, and the transmitted beamforming matrix are used as a combined action for UAV secure communication.

[0071] Specifically, this scheme employs reinforcement learning to jointly optimize the UAV position, movable antenna position, and beamforming matrix. The policy network input is not a single state variable, but a state vector composed of current time-slot system state information, including the UAV's current position, movable antenna element position, legitimate user channel state, and current communication parameters.

[0072] Based on the aforementioned state information, the policy network outputs joint control actions through feature extraction and nonlinear mapping, including UAV position adjustment, movable antenna position adjustment, and transmit beamforming matrix.

[0073] After the action is executed, the communication performance of legitimate users and eavesdropping users is recalculated based on the updated drone position, antenna position, and beam parameters. The confidentiality rate is then fed back to the reinforcement learning network as a reward. Through a multi-round interactive optimization strategy, the joint optimization of drone position, movable antenna position, and beamforming matrix is ​​achieved.

[0074] S4, adjust the positions of the UAV and antenna elements based on the UAV target position and antenna element target position in the joint operation, and configure the transmit beamforming matrix; based on the adjusted UAV and antenna element positions, reacquire the channel gain of each legitimate user and the channel gain of the eavesdropping link to construct the state of the next time slot.

[0075] Specifically, in this embodiment, a training environment for the reinforcement learning algorithm and a policy network are constructed. Generate actions based on the current state; two evaluation networks. Estimate the soft-state-action value under the maximum entropy objective. Two target evaluation networks are used to construct a stable target Q-value. The network parameters are initialized, and the experience replay pool is emptied. Then initialize the drone position. One-dimensional coordinate vector of movable antenna array element and beamforming matrix Based on the initialized drone position and the position of the movable antenna, the location of the active eavesdropper is estimated. Then, the channel of the eavesdropping link is estimated to be in the initial state. Channel between drone and legitimate user In the first Each time slot, based on the drone's current location Position of movable antenna array elements Beamforming matrix , legitimate user location Channel gain between drones and legal users and the location information of active eavesdroppers and the channel gain of the eavesdropping link This constitutes the current state. .

[0076] Specifically, after the joint action in the current time slot is executed, the position of the UAV and the position of each movable antenna unit are first adjusted according to the UAV target position and the movable antenna unit target position in the joint action, and the corresponding transmit beamforming matrix is ​​configured.

[0077] After the position adjustment is completed, the relative geometric relationship between the UAV, legitimate users, and active eavesdroppers changes, and therefore the channel state of the corresponding communication links also changes. For legitimate users, since they are cooperative communication nodes, their channel state information can be obtained through conventional channel measurement methods such as pilot signals and training sequences. For active eavesdroppers, since their true instantaneous CSI cannot be directly obtained, the equivalent channel gain of the eavesdropping link is recalculated based on the eavesdropper position estimation results obtained in the current time slot, the adjusted positions of the UAV and movable antennas, and a statistical channel model.

[0078] Finally, the adjusted UAV state information, the adjusted movable antenna unit position information, the reacquired legitimate user channel state information, the active eavesdropper's position estimation information, and the recalculated eavesdropping link equivalent channel gain are combined to form the next time slot state, which serves as the input for the next reinforcement learning decision.

[0079] Specifically, legitimate users are cooperative communication nodes in the system, capable of cooperating with drones or transmitters to complete channel measurements. Therefore, the channel state information of legitimate users can be obtained using conventional channel estimation methods in existing wireless communication systems. For example, a legitimate user can transmit a known pilot signal, and the transmitter can perform channel estimation based on the received pilot signal; alternatively, the channel state information of legitimate users can be obtained through channel sounding reference signals, training sequences, uplink and downlink channel reciprocity, or channel state information feedback.

[0080] S5. Based on the channel gain of each legitimate user, the channel gain of the eavesdropping link, and the transmit beamforming matrix, calculate the signal-to-interference-plus-noise ratio (SIR) of each legitimate user and the eavesdropping SIR of the active eavesdropper. Based on the SIR of each legitimate user and the eavesdropping SIR of the active eavesdropper, calculate the security rate of each legitimate user and the total security rate.

[0081] Specifically, the confidentiality rate for each legitimate user and total security rate The calculation formula is as follows:

[0082] ;

[0083] ;

[0084] in, Indicates taking a non-negative value; This represents the signal-to-interference-plus-noise ratio (SIR) for the i-th user in the k-th time slot; This represents the signal-to-interference-plus-noise ratio (SIR) of the active eavesdropper in the k-th time slot. Indicates the time slot for reinforcement learning; Indicates the number of legitimate users; This represents the i-th legitimate user.

[0085] Specifically, in this embodiment, the above process is as follows: [The process involves setting the state...] Input to policy network Get action The drone's position, the position of movable antenna elements, and the beamforming vector. The drone's position is based on action. Calculate the signal-to-interference-plus-noise ratio (SIR) for legitimate users and active eavesdroppers. , Furthermore, legitimate users report the received interference power to the drone. Then calculate the confidentiality rate for each user. ,in The final system's total security rate is obtained. .

[0086] S6 calculates the instant reward based on the confidentiality rate of each legitimate user and the total confidentiality rate, assembles the state of the current time slot, joint action, instant reward and the state of the next time slot into an experience sample and stores it in a preset experience replay pool, and samples experience samples from the experience replay pool to update the policy network parameters.

[0087] Specifically, instant rewards The calculation formula is as follows:

[0088] ;

[0089] in, This is the penalty coefficient; Penalties for violating minimum security rate constraints; Represents the reward function; Represents the state space of the k-th time slot; This represents the action space of the k-th time slot; This represents a part of the reward function, and is expressed as the total system security rate in the k-th time slot; This represents the time slot of the reinforcement learning algorithm.

[0090] S7. Repeat S2 to S6 until the change in the average value of the obtained instant reward is less than the preset threshold. The algorithm is then considered to have converged, and the final policy network parameters are obtained. Based on the final policy network parameters, the position of the UAV, the position of the movable antenna array elements, and the transmit beamforming matrix are controlled.

[0091] Specifically, this embodiment will also include empirical samples. Store in the experience replay pool and from A batch of randomly sampled experience is used to update the policy network and evaluation network. The training target value of the Q-network is calculated as follows: ,in Subsequently, the dual-value network parameters were updated. : Then, the policy update network is estimated based on the evaluation network. Parameters enable the strategy output to gradually increase the security rate; based on the target entropy Adaptive adjustment of entropy temperature parameters To balance exploration and utilization; finally, a soft update is performed on the target value network: Repeat this process until the average reward change is less than a threshold, at which point the algorithm converges, and the trained policy network is obtained. This is used for joint control of UAV trajectory, movable antenna position, and beamforming strategy; among which, This represents the immediate reward obtained after performing the joint action in the k-th time slot; Indicates the reward discount factor; The parameter is The soft state-action value function corresponding to the j-th evaluation network; This represents the system state of the (k+1)th time slot obtained after executing the joint action of the kth time slot; The entropy-temperature parameter represents the SAC. Represents a random policy network with parameter φ; Indicates the state The next joint action is obtained based on the policy network sampling. This represents the network parameters of the j-th evaluation network; The learning rate is used to evaluate the updates of network parameters. This represents the target soft state corresponding to the k-th empirical sample; This represents the network parameters of the evaluation network for the j-th objective.

[0092] like Figure 2 As shown, this embodiment also discloses a drone security communication system equipped with a movable antenna, including:

[0093] The position coordinate acquisition module 21 is used to acquire the interference signal transmitted by the active eavesdropper to the legitimate user, measure the angle of arrival of the interference signal, obtain the azimuth coordinate value and elevation coordinate value of the interference signal, and determine the position coordinate of the active eavesdropper and the channel gain of the eavesdropping link based on the azimuth coordinate value and elevation coordinate value.

[0094] The time slot status acquisition module 22 is used to acquire the current status information of the UAV and combine the current status information of the UAV, the position coordinates of the active eavesdropper, and the channel gain of the eavesdropping link into the current time slot status.

[0095] The joint action generation module 23 is used to input the state of the current time slot into the policy network, and to map and output the joint action by performing forward propagation, feature extraction and nonlinear transformation on the state of the current time slot; the joint action includes the UAV target position, the target position of the movable antenna array element and the transmit beamforming matrix.

[0096] The time slot state construction module 24 is used to adjust the positions of the UAV and antenna array elements based on the UAV target position and antenna array element target position in the joint action, and configure the transmit beamforming matrix; based on the adjusted UAV and antenna array element positions, it reacquires the channel gain of each legitimate user and the channel gain of the eavesdropping link to construct the state of the next time slot;

[0097] The security rate calculation module 25 is used to calculate the signal-to-interference-plus-noise ratio (SIR) of each legitimate user and the SIR of the active eavesdropper based on the channel gain of each legitimate user, the channel gain of the eavesdropping link, and the transmit beamforming matrix, and to calculate the security rate of each legitimate user and the total security rate based on the SIR of each legitimate user and the SIR of the active eavesdropper.

[0098] The instant reward calculation module 26 is used to calculate the instant reward based on the confidentiality rate of each legitimate user and the total confidentiality rate. It assembles the state of the current time slot, joint actions, instant reward and the state of the next time slot into an experience sample and stores it in a preset experience replay pool. It samples experience samples from the experience replay pool to update the policy network parameters.

[0099] The strategy network parameter acquisition module 27 repeatedly executes the time slot state acquisition module to the instant reward calculation module until the change in the average value of the obtained instant rewards is less than a preset threshold. The algorithm is then determined to have converged, and the final strategy network parameters are obtained. Based on the final strategy network parameters, the position of the UAV, the position of the movable antenna array element, and the transmit beamforming matrix are controlled.

[0100] A specific implementation of a UAV secure communication system equipped with a movable antenna is described in this embodiment, which is the same as the UAV secure communication method equipped with a movable antenna.

[0101] Although the invention has been specifically shown and described in conjunction with preferred embodiments, those skilled in the art should understand that various changes in form and detail may be made to the invention without departing from the spirit and scope of the invention as defined in the appended claims, all of which shall be within the scope of protection of the invention.

Claims

1. A secure communication method for unmanned aerial vehicles (UAVs) equipped with a movable antenna, characterized in that, Includes the following steps: S1, acquire the interference signal transmitted by the active eavesdropper to the legitimate user, measure the angle of arrival of the interference signal, obtain the azimuth and elevation coordinates of the interference signal, and determine the position coordinates of the active eavesdropper and the channel gain of the eavesdropping link based on the azimuth and elevation coordinates. S2, obtain the current status information of the drone, and combine the current status information of the drone, the location coordinates of the active eavesdropper, and the channel gain of the eavesdropping link into the current time slot status; S3, input the state of the current time slot into the policy network, and map out the joint action by performing forward propagation, feature extraction and nonlinear transformation on the state of the current time slot; the joint action includes the UAV target position, the target position of the movable antenna array element and the transmit beamforming matrix; S4, adjust the positions of the UAV and antenna array elements based on the UAV target position and antenna array element target position in the joint operation, and configure the transmit beamforming matrix; Based on the adjusted positions of the drone and antenna array elements, the channel gain of each legitimate user and the channel gain of the eavesdropping link are reacquired to construct the state of the next time slot; S5. Based on the channel gain of each legitimate user, the channel gain of the eavesdropping link, and the transmit beamforming matrix, calculate the signal-to-interference-plus-noise ratio (SIR) of each legitimate user and the eavesdropping SIR of the active eavesdropper. Based on the SIR of each legitimate user and the eavesdropping SIR of the active eavesdropper, calculate the security rate of each legitimate user and the total security rate. S6. Calculate the instant reward based on the confidentiality rate of each legitimate user and the total confidentiality rate. Assemble the current time slot's state, joint action, instant reward and the next time slot's state into an experience sample and store it in a preset experience replay pool. Sample experience samples from the experience replay pool to update the policy network parameters. S7. Repeat S2 to S6 until the change in the average value of the obtained instant reward is less than the preset threshold. The algorithm is then considered to have converged, and the final policy network parameters are obtained. Based on the final policy network parameters, the position of the UAV, the position of the movable antenna array elements, and the transmit beamforming matrix are controlled.

2. The secure communication method for a UAV equipped with a movable antenna according to claim 1, characterized in that, In S1, the position coordinates of the active eavesdropper are determined based on the azimuth and elevation coordinates, using the following formula: ； Indicates the azimuth coordinate value; This represents the estimated pitch angle. Indicates the time slot for communication; Indicates the location of the active eavesdropper; This indicates the drone's flight altitude.

3. The secure communication method for a UAV equipped with a movable antenna according to claim 1, characterized in that, In S2, the current status information of the drone includes: The spatial coordinates of the drone are used to indicate the drone's current spatial location, serving as input for the policy network to determine the drone's target location; The guide rail position coordinates of each movable antenna element are used to indicate the distribution position of each antenna element on the current one-dimensional guide rail, and serve as the input basis for the strategy network to decide the target position of the antenna elements. The ground position coordinates of each legitimate user are used to indicate the fixed or real-time position of each legitimate user on the ground, and serve as a reference position for calculating the channel gain and beamforming between the UAV and the legitimate users. The channel gain between the drone and each legitimate user is used to indicate the signal propagation quality of each legitimate communication link in the current time slot, and to calculate the signal-to-interference-plus-noise ratio and security rate of each legitimate user.

4. The secure communication method for a UAV equipped with a movable antenna according to claim 1, characterized in that, S3 also includes constraint processing on the joint actions output by the policy network, specifically including: The target position of the UAV is restricted to a preset UAV movable area, the target position of each movable antenna array element is restricted to the corresponding one-dimensional guide rail movement range, and the preset minimum spacing constraint is met between each movable antenna array element. The transmit beamforming matrix is ​​subjected to power constraint processing based on the preset minimum spacing constraint so that the total transmit power of the UAV does not exceed the preset maximum transmit power. The constrained UAV target position, the movable antenna array element target position, and the transmitted beamforming matrix are used as a combined action for UAV secure communication.

5. The secure communication method for a UAV equipped with a movable antenna according to claim 1, characterized in that, In S5, the confidentiality rate of each legitimate user and total security rate The calculation formula is as follows: ； ； in, Indicates taking a non-negative value; This represents the signal-to-interference-plus-noise ratio (SIR) for the i-th user in the k-th time slot; This represents the signal-to-interference-plus-noise ratio (SIR) of the active eavesdropper in the k-th time slot; Indicates the time slot for reinforcement learning; Indicates the number of legitimate users; This represents the i-th legitimate user.

6. The secure communication method for a UAV equipped with a movable antenna according to claim 1, characterized in that, In Season 6, instant rewards The calculation formula is as follows: ； in, This is the penalty coefficient; Penalties for violating minimum security rate constraints; Represents the reward function; Represents the state space of the k-th time slot; This represents the action space of the k-th time slot; This represents the total system security rate in the k-th time slot.

7. A secure communication system for unmanned aerial vehicles (UAVs) equipped with a movable antenna, characterized in that, include: The location coordinate acquisition module is used to acquire the interference signal transmitted by the active eavesdropper to the legitimate user, measure the angle of arrival of the interference signal, obtain the azimuth and elevation coordinate values ​​of the interference signal, and determine the location coordinates of the active eavesdropper and the channel gain of the eavesdropping link based on the azimuth and elevation coordinate values. The time slot status acquisition module is used to acquire the current status information of the UAV and combine the current status information of the UAV, the position coordinates of the active eavesdropper, and the channel gain of the eavesdropping link into the status of the current time slot. The joint action generation module is used to input the state of the current time slot into the policy network, and map out the joint action by performing forward propagation, feature extraction and nonlinear transformation on the state of the current time slot; the joint action includes the UAV target position, the target position of the movable antenna array element and the transmit beamforming matrix. The time slot state construction module is used to adjust the positions of the UAV and antenna array elements based on the UAV target position and antenna array element target position in the joint operation, and to configure the transmit beamforming matrix. Based on the adjusted positions of the drone and antenna array elements, the channel gain of each legitimate user and the channel gain of the eavesdropping link are reacquired to construct the state of the next time slot; The security rate calculation module is used to calculate the signal-to-interference-plus-noise ratio (SIR) of each legitimate user and the SIR of the active eavesdropper based on the channel gain of each legitimate user, the channel gain of the eavesdropping link, and the transmit beamforming matrix. Based on the SIR of each legitimate user and the SIR of the active eavesdropper, the security rate of each legitimate user and the total security rate are calculated. The instant reward calculation module is used to calculate the instant reward based on the confidentiality rate of each legitimate user and the total confidentiality rate. It assembles the current time slot state, joint action, instant reward and the next time slot state into an experience sample and stores it in a preset experience replay pool. It samples experience samples from the experience replay pool to update the policy network parameters. The strategy network parameter acquisition module is used to repeatedly execute the time slot state acquisition module to the instant reward calculation module until the change in the average value of the obtained instant rewards is less than a preset threshold. The algorithm is then considered to have converged, and the final strategy network parameters are obtained. Based on the final strategy network parameters, the position of the UAV, the position of the movable antenna array elements, and the transmit beamforming matrix are controlled.

Images