Method for optimizing communication power and deployment of multiple listeners based on unmanned aerial vehicle interference assistance
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BIT ZHENGZHOU INTELLIGENT TECH RES INST
- Filing Date
- 2025-11-22
- Publication Date
- 2026-07-10
Smart Images
Figure CN121692280B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wireless communication security technology, specifically to a method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV interference assistance. Background Technology
[0002] The open nature of wireless communication makes its transmitted signals easily detectable, posing a significant challenge to system security. Against this backdrop, covert communication technology, as a high-level security measure to ensure that communication activities themselves remain undetected, has become an important research direction in the field of wireless security. To effectively improve covert performance, existing technologies often introduce friendly jamming nodes, raising the eavesdropping party's decision threshold by emitting artificial noise. Given the high mobility and flexible deployment advantages of unmanned aerial vehicle (UAV) platforms, utilizing UAVs as cooperative jamming platforms has become an important technological trend in this field.
[0003] However, existing technologies still have significant shortcomings in the research and practice of applying drones to covert communication.
[0004] First, existing methods for modeling the spatial distribution of eavesdroppers typically employ simplified settings, failing to effectively reflect the differences in security levels across different geographical areas in real-world adversarial environments. This approach ignores the uneven spatial distribution of eavesdroppers, preventing the system from accurately adjusting jamming resources based on regional security differences and limiting the spatial utilization efficiency of jamming effectiveness.
[0005] Secondly, existing channel modeling capabilities are relatively limited in characterizing heterogeneous propagation environments between air and ground. Many studies only consider idealized line-of-sight propagation conditions, neglecting non-line-of-sight propagation characteristics caused by factors such as building obstruction or terrain reflection. This single channel model cannot accurately reflect the fading patterns of real channels, leading to discrepancies between theoretical analysis results of concealment performance and actual conditions, thereby reducing the reliability of system optimization design.
[0006] More importantly, existing research, in constructing stealth performance measurement systems and optimization frameworks, lacks a matching design for the differences in the certainty of information known to the eavesdropper. Most studies assume a fixed, uniform level of understanding of the eavesdropper's channel state or location information and employ a single optimization model accordingly. However, in real-world systems, the level of certainty of intelligence changes dynamically. If an optimization framework based on a single assumption is still used, the optimization results will be overly aggressive and fail when the intelligence is ambiguous, while the strategy will be too conservative and sacrifice communication performance when the intelligence is accurate. This lack of adaptability to different intelligence qualities makes it difficult to guarantee the robustness of the solution. Furthermore, for scenarios with uncertain information, existing methods typically require solving a high-dimensional, complex joint optimization problem, which has high computational complexity and low solution efficiency, making it difficult to meet the real-time requirements of dynamic communication environments. Summary of the Invention
[0007] The technical problem to be solved by this invention is how to adaptively optimize the transmission power of the covert communication system and the deployment of friendly jamming drones in a complex communication environment with multiple eavesdroppers, based on the different degrees of certainty of the eavesdroppers' information, so as to maximize the covert transmission performance while ensuring the covertness of communication.
[0008] To achieve the above objectives, the present invention provides the following technical solution:
[0009] A method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance includes the following steps:
[0010] Scene identification and threat modeling: The certainty level of the channel state and location information of the multiple eavesdroppers is determined, and the current communication scene is identified into one of two types: a fully known scene or a statistically known scene. In a fully known scene, the location and channel state information of all eavesdroppers are instantaneously known values; in a statistically known scene, only the statistical characteristics of the eavesdropper distribution are known. Simultaneously, the threat of the eavesdroppers is modeled according to the scene type.
[0011] Single-eavesdropper detection performance analysis: An optimal detection model for concealed signals by a single eavesdropper is established. Based on this model, the minimum false detection probability is derived for which, under optimal detection conditions, the eavesdropper still cannot correctly determine the presence of a concealed signal. This probability is the basis for evaluating concealment performance.
[0012] Global stealth performance quantification: Based on the scene recognition results, the minimum error detection probability of multiple eavesdroppers is combined into a global stealth performance metric that reflects the overall stealth level of the system through a specific mathematical method.
[0013] Problem Formulation: An optimization problem is constructed with the objective of maximizing the covert transmission capacity between the covert sender and the covert receiver, while being constrained by two types of constraints. The first type of constraint is that the global covert performance metric must not be lower than a preset covert threshold; the second type of constraint is that the transmit power of each node and the deployment location of the UAV must be within physically permissible limits.
[0014] Optimization problem solving: For the constructed optimization problem, based on the scene recognition results, an optimization algorithm matching the scene type is used to solve it, in order to obtain a set of parameters that optimizes the objective function. This set of parameters includes the optimal transmission power of the covert transmitter, the optimal upper limit of the interference power of the friendly jamming drone, and the optimal deployment location.
[0015] Parameter Execution and Closed-Loop Control: Finally, the solved optimal parameters are applied to the physical control of the concealed transmitter and the friendly jamming drone. Simultaneously, a closed-loop control mechanism is established to continuously monitor changes in environmental conditions. When the change reaches a preset significance threshold, the method is triggered to return to the scene recognition and threat modeling steps for a new round of adaptive optimization.
[0016] In one specific embodiment, when the scenario is identified as a fully known scenario, the threat modeling step specifically involves, for the first... Each eavesdropper is assigned a threat weighting factor. Accordingly, the global stealth performance quantification step specifically involves calculating the global weighted error detection probability as a global stealth performance metric.
[0017] Preferably, the threat weighting factor The threat weighting factor is determined as follows: First, a baseline threat level is calculated for each eavesdropper, which takes into account the eavesdropper's eavesdropping advantages such as channel quality and distance. Then, the baseline threat levels of all eavesdroppers are normalized to obtain the final threat weighting factor.
[0018] In one specific embodiment, when the scene is identified as a statistically known scene, the threat modeling step specifically involves obtaining the spatial probability distribution function of the eavesdropper within a preset unsafe zone.
[0019] Accordingly, the global concealment performance quantification step specifically involves calculating the average minimum false detection probability as a global concealment performance metric.
[0020] Preferably, the calculation of the average minimum false detection probability is based on the probability density function of a pre-derived key intermediate variable. This intermediate variable is the distance ratio between the eavesdropping party and the covert transmitter and friendly jamming drones, and obtaining its probability density function simplifies subsequent integration calculations.
[0021] In one specific embodiment, when the scene is identified as a fully known scene, the optimization problem-solving step employs a particle swarm optimization algorithm. This algorithm simulates the social behavior of a particle swarm and performs global optimization in a multi-dimensional parameter space to solve an optimization problem with complex nonlinear constraints, ultimately obtaining the joint optimal solution for the transmit power, the upper limit of interference power, and the deployment location.
[0022] In one specific embodiment, when the scene is identified as a statistically known scene, the optimization problem solving step decomposes the original joint optimization problem into a location optimization subproblem and a power optimization subproblem, and solves them in stages.
[0023] Preferably, the phased solution first performs a position optimization subproblem. The optimization objective of this subproblem is to maximize the average minimum error detection probability. The optimal deployment position of the UAV can be analytically solved through mathematical derivation.
[0024] Furthermore, after determining the optimal deployment location, a power optimization subproblem is executed. This subproblem is simplified by fixing two conditions: first, the upper limit of the interference power of the friendly jamming UAV is set to the maximum value allowed by its hardware to provide the strongest cover effect; second, the solution is performed under the boundary condition that the average minimum error detection probability is strictly equal to the preset concealment threshold, thereby maximizing the transmission power while satisfying the concealment constraint, and finally obtaining the optimal transmission power.
[0025] Preferably, the closed-loop control mechanism is implemented in the following way: continuously monitoring the statistical characteristics of communication channel conditions or changes in the threat posture of the eavesdropper, wherein changes in the threat posture include, but are not limited to, significant movement of the eavesdropper's location or reassessment of its threat weighting factor. When any monitored change exceeds a preset significance threshold, the system automatically triggers and returns to execute the initial scene identification and threat modeling steps to ensure that the optimization results always match the dynamically changing environment.
[0026] This invention provides a method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance. It has the following beneficial effects:
[0027] 1. This invention establishes a dual-track optimization framework that adaptively switches between two different optimization paths based on the level of certainty regarding the information from the eavesdropping party. When the intelligence is precise, a refined optimization model based on weighted threats is used; when the intelligence is ambiguous, a robust optimization model based on statistical characteristics is used. This method can flexibly cope with various complex adversarial scenarios ranging from precise to ambiguous intelligence, improving the scenario adaptability of the solution and avoiding the failure of a single model in a changing environment.
[0028] 2. This invention decomposes the complex joint optimization problem into two independent sub-problems: position optimization and power optimization. The optimal UAV deployment position is obtained directly through analytical derivation. This reduces a high-dimensional non-convex optimization problem into a low-dimensional problem that only requires numerical solution. This reduces the computational complexity of the algorithm, improves the solution efficiency, is more suitable for dynamic communication environments with high real-time requirements, and reduces the performance requirements of the computing platform.
[0029] 3. This invention introduces a closed-loop control mechanism that continuously monitors changes in the communication environment and automatically re-optimizes when the threat situation of the eavesdropper or channel conditions change significantly. This ensures the stealth performance and transmission efficiency of the communication system, maintaining a near-optimal state in dynamically changing environments and guaranteeing the long-term effectiveness and reliability of the solution during long-term task execution. Attached Figure Description
[0030] Figure 1 This is a system architecture diagram of the method of the present invention;
[0031] Figure 2 This is a schematic diagram of the method flow of the present invention. Detailed Implementation
[0032] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0033] This invention provides a method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance. This method is applied to an adversarial communication environment with differentiated security levels, which includes a covert sender Alice, a covert receiver Bob, a friendly jamming UAV Carol, and an eavesdropper Willie.
[0034] In this scenario, the friendly jamming drone Carol is deployed in the safe zone. Inside, the eavesdropping party cannot enter this area. The eavesdropping party, Willie, is located in the unsafe zone. Inside.
[0035] Alice and Carol are equipped with multiple transmitting antennas, while Bob and all Willie are equipped with a single receiving antenna.
[0036] Alice employs Maximum Ratio Transmission (MRT) precoding to maximize the received signal power at Bob's location. Her precoding vector... Defined as: ;
[0037] in, Here is the channel matrix from Alice to Bob. This indicates the conjugate transpose. This represents the F-norm. The precoded vector satisfies the normalization constraint. .
[0038] Carol employs interference cancellation precoding to ensure that her interfering signals do not affect the legitimate receiver, Bob. Her precoding vector... The following null space conditions must be met: ;
[0039] in, This is the channel matrix from Carol to Bob. This precoding vector also satisfies the normalization constraint. .
[0040] Carol's interference power At the beginning of each time slot from the interval Randomly selected from a uniform distribution. This represents the preset maximum interference power limit. The probability density function (PDF) for this interference power is also shown. Represented as:
[0041] ;
[0042] To uniformly characterize mixed propagation scenarios that include line-of-sight (LoS) and non-line-of-sight (NLoS) components, this method uses the Nakagami-m channel model to uniformly model channel gain.
[0043] Signal transmitting end ( For Alice or Carol) root transmitting antenna to the first One listening node Channel gain probability density function Represented as:
[0044] ;
[0045] in: For shape parameters (with values ≥ 0.5); For scale parameters; For An exponential function with base 0; by adjusting It can freely characterize the LoS and NLoS features of the channel, and realize unified modeling of all listening links.
[0046] After precoding, the squared equivalent channel gain of the beamformed signal is... It can be approximated as a Gamma distribution using the moment matching method, and its probability density function is... Represented as:
[0047] ;
[0048] in: It is the Gamma function; The shape parameter approximates the Gamma distribution, and its expression is: ; The scaling parameter for approximating the Gamma distribution is expressed as: .
[0049] See attached document Figure 1 The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance can be executed by a communication optimization system, which serves as the execution vehicle for the method and includes:
[0050] Input module 201: Used to receive and preprocess basic parameters required for system operation. Parameters include, but are not limited to: the maximum transmit power limit of the covert transmitter, Alice. Maximum jamming power limit for the friendly jamming drone Carol Number of listening nodes 3D position coordinates of the legitimate communicator and the UAV, and precoded vectors and Specific form, channel fading parameters and And the allowable error for concealment restrictions. , , wait.
[0051] Scene recognition module 202: Used to determine whether the covert communication party is listening to the eavesdropping party. The module measures the certainty level of Channel State Information (CSI) and location information. Based on information availability, it dynamically identifies the current task as either a fully known scenario or a statistically known scenario and outputs the corresponding enable signal.
[0052] Area security assessment module 203: Controlled activation in statistically known scenarios. This module is based on a preset security zone. and unsafe areas The division of (hidden areas) leads to the deduction of the location of the listening nodes. Spatial probability distribution within, for example, deriving the key ratio probability density function .
[0053] The eavesdropper channel evaluation module 204 is used to characterize the channel fading characteristics of the eavesdropper link. This module uses the Nakagami-m channel model to uniformly model the channel gain, and approximates the squared magnitude of the equivalent channel gain after beamforming as a Gamma distribution using the moment matching method. The parameters for the output channel gain distribution.
[0054] Threat assessment module 205 for the eavesdropping party: Controlled activation in fully known scenarios. This module is based on the eavesdropping party's... Based on accurate CSI and location information, as well as the threat level, calculate and output the threat weighting factor. The threat weighting factor satisfies the normalization constraint.
[0055] Covert performance analysis module 206: Used to calculate the detection performance indicators of the covert communication system. This module includes a single eavesdropper detection performance analysis unit and an eavesdropper global detection performance analysis unit.
[0056] Single-eavesdropper detection performance analysis unit: Based on binary hypothesis testing, derive each false detection probability and the probability of missed detection The expression for the total false detection probability is derived, and the result is obtained. Minimum optimal decision threshold Under this optimal threshold, The minimum error detection probability is denoted as .
[0057] Global detection performance analysis unit of the eavesdropping party: Calculates global concealment performance metrics based on scene recognition results.
[0058] In a fully known scenario, calculate the globally weighted error detection probability. ;
[0059] In statistically known scenarios, calculate the average minimum false detection probability. ;
[0060] Parameter optimization module 207: Used to maximize covert transmission capacity under covert constraints. With the goal of optimizing variables Solve the problem.
[0061] Execution module 208: Used to receive the optimal parameters output by parameter optimization module 207. , and optimal deployment location and generate corresponding control commands; for transmission power, upper limit of interference power and The flight attitude is adjusted in real time.
[0062] See attached document Figure 2 This invention provides a method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance. The method includes the following steps:
[0063] S301, Scene Recognition and Threat Modeling: In step S301 of this method, the scene recognition module 202 first performs scene determination. Based on the system's understanding of the eavesdropper's channel state information (CSI) and location information, the scene recognition module 202 determines the current task into one of the following two scenarios:
[0064] Completely Known Scenario: In this scenario, the system assumes that all... One eavesdropping party The precise CSI and instantaneous location coordinates of each listening device are known. Specifically, the system holds or can acquire real-time information about each listening device. Precise Cartesian coordinates Furthermore, it possesses instantaneous channel state information within the current communication time slot, i.e., the information from the covert sender Alice to... Channel matrix And the friendly interference drone Carol to Channel matrix .
[0065] Statistically known scenario: In this scenario, the system assumes that it only knows the eavesdropping party. Located in an unsafe zone The system obtains the spatial probability distribution within the listening area, but its precise CSI and instantaneous location are unknown. Specifically, the system cannot acquire the location of each listening party. Instead of specific coordinates, it is about mastering Geographic boundaries (e.g., defined by radius) The system defines the annular region and the spatial distribution model of the listeners within that region (e.g., following a spatial Poisson process or a uniform distribution). Simultaneously, the system does not possess instantaneous CSI, only knowing the statistical properties of the channel, such as the shape parameters of the Nakagami-m distribution. and scale parameters and path loss factor .
[0066] The scene recognition module 202 determines the completeness of the above two types of information by checking the internal data bus of the system or querying external sensor input:
[0067] If both instantaneous CSI and precise location information are available, the scenario is determined to be fully known, and a first enable signal is output to enable the subsequent eavesdropper threat assessment module 205.
[0068] If only statistical information and area boundaries are available, it is determined to be a statistically known scenario, and a second enable signal is output to enable the subsequent area security assessment module 203.
[0069] If the scenario is determined to be completely known, the eavesdropping threat assessment module 205 is activated. This module is based on known... The CSI, distance, and potential threat level to the system are used for each Assign a threat weighting factor The weighting factor Used to quantify different listening methods Differentiated threat levels to covert communications.
[0070] In one specific implementation, the eavesdropping threat assessment module 205 first assesses each... Calculate a baseline threat level This benchmark threat level Characterization The advantage of listening under baseline conditions; for example, the advantage of listening under baseline conditions. Can be defined as The ratio of the reference signal power received from Alice to the reference interference power received from Carol, or a metric related to the signal-to-interference ratio (SIR):
[0071] ;
[0072] in: and For reference transmission power; and Alice to With Carol The squared equivalent channel gain modulus, and Alice to With Carol distance, This is the path loss factor. for Noise power.
[0073] In calculation Baseline threat level of each eavesdropper Subsequently, the eavesdropping threat assessment module 205 normalizes the data to obtain the final threat weighting factor. Its expression is: ;
[0074] The eavesdropping threat assessment module 205 ensures all threat weighting factors are considered. All are non-negative real numbers and satisfy the normalization constraint: ;
[0075] If the scenario is determined to be statistically known, the area security assessment module 203 is activated. This module is based on preset non-security zones. The geometric shape and the listening method in Based on the spatial distribution assumptions within the domain, the probability density function (PDF) of the key variables is derived.
[0076] In a specific implementation, the non-safe zone It is defined as an inner radius centered on the covert sender Alice (origin (0, 0)). Outer radius is A ring-shaped area. The listening party. exist The spatial distribution within is assumed to be uniform. Based on this, the eavesdropping party... Spatial probability density function Defined as: ;
[0077] Regional security assessment module 203 then derives the critical ratio. probability density function This key ratio Defined as: ;
[0078] in For eavesdropping party random coordinates, Determine the coordinates of the friendly interference drone Carol (assuming Alice is located at the origin (0, 0)). This is the path loss factor.
[0079] Derivation Regional security assessment module 203 calculates The cumulative distribution function (CDF) , The expression is:
[0080] ;
[0081] In obtaining After obtaining the analytical or numerical solution, the regional security assessment module 203 performs the following steps: Taking the derivative, we obtain the final probability density function. : ;
[0082] S302, Single eavesdropper detection performance analysis: Step S302 of this method is executed by the single eavesdropper detection performance analysis unit in the concealment performance analysis module 206. This step is used to establish the detection model of the eavesdropper and derive the detection model of any single eavesdropper. Optimal detection performance.
[0083] From the listening side From this perspective, the detection problem is modeled as a binary hypothesis test. Attempting to receive signals through it To determine whether the covert sender Alice exists.
[0084] Assumption: Alice, the covert sender, transmits nothing. Under this assumption, In the Received signal within one symbol period Includes only jamming signals and additive noise from the friendly jamming drone Carol:
[0085] ;
[0086] Assumption: Alice, the covert sender, is transmitting data. Under this assumption, Received signal This includes covert signals from Alice, interference signals from Carol, and additive noise:
[0087] ;
[0088] in: This refers to Alice's transmission power. For Carol's interference power, this power comes from Randomly selected from a uniform distribution; and Alice to With Carol The distance; This refers to the path loss factor. and Alice to With Carol The channel matrix; and These are the pre-encoded vectors for Alice and Carol, respectively; and Alice and Carol were in the 1st and 2nd years respectively. The concealed signal and interference signal transmitted in one symbol period, of which:
[0089] ,
[0090] eavesdropping party A power decision strategy is employed to detect the received signal. This strategy operates within a range that includes... Within a detection window of one symbol period, the cumulative received power is calculated. This test statistic is used as a test statistic. The calculation formula is: ;
[0091] in To detect the length of the window, For the first Received signal within one symbol period.
[0092] This ruling rule utilizes the principle of... Assuming the received signal Include The assumption includes additional signal power components from the covert sender Alice that are not present in the data.
[0093] Assuming concealed signal and interference signals Statistically independent and with a mean of zero. For simplicity, the squared magnitude of the equivalent channel gain after beamforming is defined as... and .
[0094] Therefore, under both assumptions, the expected average power of the received signal (i.e. time They are respectively:
[0095] Expected average power under the assumption : ;
[0096] Expected average power under the assumption : ;
[0097] eavesdropping party Put it in The actual cumulative received power within one symbol period (its statistical mean is) The lower is greater than the upper (Below), with a preset decision threshold. Compare them.
[0098] like ,but judgment Established (i.e., Alice has transmitted data); if ,but judgment Established (i.e., Alice has no transmission); this decision threshold This is the threshold that distinguishes between these two power statistical characteristics.
[0099] Based on the listening party The judgment rule determines the detection performance based on the false detection probability. and the probability of missed detection Jointly determined. False alarm probability. This means Alice did not send a signal. The probability of it being sent and the probability of it being missed are determined. This means Alice sent a signal. The probability of not being detected.
[0100] For simplicity, it is assumed that the received power can be approximated as the instantaneous power within one symbol period. Then, for a given decision threshold... The specific expressions for these two probabilities are:
[0101] ;
[0102] ;
[0103] in, and Alice to With Carol The square of the equivalent channel gain; For listening nodes The noise power; This refers to the path loss factor. , Alice to With Carol The distance; This is the threshold for the judgment.
[0104] eavesdropping party Total False Detection Probability (DEP) Defined as the sum of the probability of a false detection and the probability of a false negative, i.e. .
[0105] eavesdropping party An optimal decision threshold will be selected. To minimize its total false detection probability .
[0106] Through analysis Compared to The functional relationship can be used to deduce that makes Minimum optimal decision threshold The optimal decision threshold Exactly equal to Under the assumption that the received signal contains no interference power components, the expected value is given.
[0107] The optimal threshold is when At that time, that is ;
[0108] in, For listening nodes The optimal decision threshold; Noise power; This refers to Alice's transmission power. For Alice The square of the equivalent channel gain of the link; For Alice The distance between them; This is the path loss factor.
[0109] This optimal decision threshold Substitution and From the expression, we can obtain that under this optimal threshold, Minimum probability of error detection that can be achieved , The expression is:
[0110] ;in
[0111] Minimum error detection probability Characterizes the eavesdropping party The maximum eavesdropping capability under optimal decision conditions; the larger the value, the more advantageous it is for covert communication. This value represents... The greatest threat capability.
[0112] S303, Global Covertness Performance Quantification (Dual-Track System): Step S303 in this method is executed by the global detection performance analysis unit of the eavesdropper in the covertness performance analysis module 206. When the scenario is determined to be completely known, the system needs a unified metric to characterize all... The overall threat from eavesdropping parties.
[0113] In this scenario, global stealth performance is determined by the globally weighted error detection probability. This metric incorporates the threat weighting factor determined by the eavesdropping threat assessment module 205 in step S301. And the minimum error detection probability of a single listener derived in step S302 .
[0114] Global weighted error detection probability The calculation formula is defined as follows: ;
[0115] in:
[0116] The total number of eavesdroppers; For the first One listening party The threat-weighted factor, which satisfies the normalization constraint. ;
[0117] for At its optimal decision threshold The minimum error detection probability that can be achieved.
[0118] Because the system cannot know the listening party The precise instantaneous location and CSI are used to measure the global stealth performance of the system by statistical averaging.
[0119] In this scenario, the system assumes Among all eavesdropping parties, the one that poses the greatest threat to covert communication (i.e., has the lowest false detection probability) is the one that does not. The smallest (smallest) eavesdropping party is The system's global stealth performance is determined by the average minimum false detection probability of the strongest eavesdropper. This is used as a metric, representing the system's average stealth performance under worst-case conditions.
[0120] Average minimum error detection probability The calculation requires Integrate the probability density function of all possible random variables to find the expected value.
[0121] In statistically certain scenarios, the minimum probability of false detection It can be viewed as a random variable The function, where It integrates the randomness of channel gain and distance ratio. For example, It can be the channel gain ratio. With distance ratio product .
[0122] Area security assessment module 203 combined with eavesdropping parties The spatial distribution model (e.g., uniform distribution) and the channel model (e.g., Gamma distribution) of the eavesdropping channel evaluation module 204 are used to derive... and joint probability density function Or obtained through transformation probability density function .
[0123] Average minimum error detection probability The final calculation is performed on the random variable. probability density function Integrating and calculating the expected value yields: ;
[0124] in: for Regarding random variables The function form; For random variables The probability density function, this value This reflects the worst-case scenario faced by the system in a statistical sense (i.e., the strongest eavesdropper). The average concealment performance under ( ).
[0125] S304. Problem Construction: Step S304 of this method is executed by the parameter optimization module 207. The core function of this module is to construct an optimization problem, the goal of which is to maximize the communication rate from the covert sender Alice to the legitimate receiver Bob, while satisfying the subsequently defined covert performance constraints.
[0126] The objective function of this optimization problem is defined as the covert transmission capacity. This capacity, based on Shannon's formula, represents the capacity at Alice's transmit power of... At that time, the theoretical maximum transmission rate achievable at Bob's location is expressed by the objective function as follows: ;
[0127] in: This refers to Alice's transmission power. This represents the distance from Alice to Bob. This refers to the path loss factor. This represents the noise power at Bob.
[0128] One premise for this formula to hold true is that the interference from the friendly jamming drone Carol to the covert receiver Bob has been effectively suppressed.
[0129] To achieve the above objectives, the parameter optimization module 207 performs joint optimization on a multi-dimensional set of variables including power and deployment location. This set of optimization variables specifically includes: ;
[0130] in: Alice's transmit power, this variable directly determines the objective function. Size; The maximum jamming power limit for the friendly jamming drone Carol; Let Carol be the deployment position of the drone in a two-dimensional polar coordinate system with Alice as the origin, where Radial distance, This is the azimuth angle. This position variable can also be equivalently represented as Cartesian coordinates. Carol's deployment location determines her distance from each listening party. This profoundly affects the contribution of its interference signal to the overall concealment performance.
[0131] The optimization problem (P) constructed in step S304 of this method consists of an objective function and a series of constraints. These constraints ensure that while pursuing the maximum covert transmission capacity, the system's preset covertness requirements and the operational limitations of physical devices are met.
[0132] Based on the scene recognition results in step S301, the optimization problem has different hidden constraints. The complete optimization problem (P) is constructed as follows: ;
[0133] This maximization problem is subject to the following constraints:
[0134] (C1) Global Hidden Constraints:
[0135] This is the core constraint ensuring the overall concealment of the system, and its specific form switches between two modes based on the scenario determination result. For a fully known scenario, this constraint requires the globally weighted error detection probability calculated in step S303. It must be higher than a preset threshold, the expression of which is as follows:
[0136] ;
[0137] in It is a very small positive number close to 0 (e.g., or The value represents the required global concealment level for the system. This constraint ensures that the overall concealment performance of the system meets the requirements after considering the weighted threat from all eavesdroppers. The closer it is to 1, the higher the requirement for concealment.
[0138] In a statistically known scenario, this constraint requires the average minimum false detection probability of the strongest eavesdropper to be calculated in step S303. Above a corresponding threshold: ;
[0139] in This is a concealment level parameter set for statistical scenarios. This constraint ensures that even in the worst-case scenario (i.e., against the strongest eavesdropper), the system's average concealment performance meets the standard.
[0140] (C2) Individual Hidden Constraints (only for fully known scenarios):
[0141] This constraint is an additional safeguard designed for fully known scenarios, intended to prevent any single eavesdropper (even with its threat-weighted factor) from... The detection error probability (which is relatively low) is too low. This requires that for each listening party... Its minimum error detection probability All must satisfy a single threshold, expressed as follows:
[0142] ;
[0143] in, It is the collection of all eavesdroppers. This is the concealment level set for a single eavesdropping party; its value can be equal to or different from... .
[0144] (C3) Power Constraint:
[0145] This constraint reflects the physical hardware limitations of the covert transmitter Alice and the friendly jamming drone Carol. Alice's transmit power The expression must be within the allowed range, as follows:
[0146] ;in This is the maximum output power limit of the Alice transmitter.
[0147] Carol's maximum interference power It also cannot exceed its physical limit, as expressed below:
[0148] ;in That is the maximum output power of the Carol jammer.
[0149] (C4) Deployment constraints:
[0150] This constraint is used to limit the legal airspace, i.e., the safe zone, for the friendly jamming drone Carol. Its expression is as follows: ;
[0151] in, It is the radial distance of Carol relative to Alice, and It is a pre-defined safe zone centered on Alice. The radius. This constraint ensures that Carol operates within her designated safe area and does not enter no-fly zones or danger zones. Azimuth angle There are usually no restrictions, and it is permissible to... Choose freely within the range.
[0152] S305, Solving the optimization problem (dual-track system): Step S305 in this method is executed by the parameter optimization module 207. In scenarios deemed completely known, the constraints (C1) and (C2) of the optimization problem (P) contain complex nonlinear functions concerning the optimization variables, resulting in a highly non-convex, multivariate coupled problem. This embodiment employs a particle swarm optimization (PSO)-based intelligent algorithm for global optimization.
[0153] The implementation details of the PSO algorithm are as follows:
[0154] First, define the structure of the particle swarm optimization problem. Each particle represents a potential solution to the optimization problem. Specifically, define the... The particle in the first The four-dimensional position vector at the next iteration for:
[0155] ;
[0156] in, The Cartesian coordinates represent the position of the drone Carol. Simultaneously, a corresponding four-dimensional velocity vector is defined for each particle. During the initialization phase ( Randomly generated within the search space defined by constraints (C3) and (C4). The initial position and velocity of each particle.
[0157] Next, a fitness function is constructed to evaluate the quality of each particle. This embodiment employs a fitness function based on a Brick-wall type penalty factor. This function uses a constraint condition as a hard threshold: a particle's position vector... When all constraints (C1), (C2), (C3), and (C4) defined in step S304 are satisfied simultaneously, the fitness value is equal to the objective function value at this position vector. Otherwise, if any constraint is not satisfied, its fitness value is set to 0. Its mathematical expression is:
[0158] ;
[0159] Subsequently, the algorithm enters the iterative solution phase. In each iteration... In, for each particle The algorithm will perform the following operations:
[0160] Based on the particle's current position Calculate its fitness value The fitness value is compared with its own historical best fitness value. If the current fitness value is better, the individual's best position is updated. The fitness value is compared with the global best fitness value of the entire particle swarm. If the current fitness value is better, the global best position is updated. Update the velocity and position of each particle according to the following rules to prepare for the next iteration:
[0161] Speed update rules:
[0162] ;
[0163] in, and They are respectively and The speed of the particles, It is the inertia weight, used to balance global and local search capabilities; and These are acceleration constants, representing the weights by which particles learn towards their individual and global optima, respectively. and These are two random numbers that are uniformly distributed within the interval.
[0164] Location update rules: ;
[0165] This iterative process continues until the preset maximum number of iterations is reached, or the globally optimal fitness value is achieved. The algorithm converges when there is no further significant improvement after multiple iterations. After the iterations are complete, the parameter optimization module 207 outputs the final global optimal solution. The solution is the optimal combination of parameters that maximizes the covert transmission capacity while satisfying all constraints in a fully known scenario.
[0166] Step S305 of this method is executed by the parameter optimization module 207. When the scenario is determined to be statistically known, the objective function can be discovered by analyzing the structural characteristics of the optimization problem (P). Only with Alice's transmission power Related to, and related to Carol's deployment location Irrelevant. However, the average minimum false detection probability in the hidden constraint (C1) Simultaneously with power variables and deployment location Related.
[0167] Subproblem P2.1 (Position Optimization):
[0168] The goal of this subproblem is for a friendly force to interfere with the deployment location of the drone Carol. This aims to maximize the system's stealth performance. Specifically, given power parameters, it aims to maximize the average minimum false detection probability. By maximizing this concealment metric, the most relaxed constraints can be created for the subsequent power optimization phase, thereby maximizing Alice's transmit power. It is possible to reach even higher values, and its expression is: ;
[0169] right Analyzing the expression, we can find that its value varies with the distance ratio. It increases as it increases.
[0170] To maximize its statistical expectation, we need to choose a value that allows... Throughout the unsafe zone The statistically largest deployment location. It can be proven through derivation that this is true for any location within the non-safe zone if and only if Carol is deployed directly above Alice (i.e., at the origin). Internal eavesdropping The distance from it to Carol All are equal to their distance from Alice This makes the distance ratio... The value is always equal to 1, thus statistically maximizing the system's concealment performance and avoiding the interference weakness caused by Carol's deviation from the center. Therefore, the optimal solution to this subproblem is: ;
[0171] Subproblem P2.2 (Power Optimization):
[0172] After determining the optimal deployment location Then, the original optimization problem (P) simplifies to a problem involving only the transmit power. and The problem is expressed as: ;
[0173] The constraints are: ;
[0174] ;
[0175] By analyzing the monotonicity of this problem, we can see that the objective function... It is about It is a monotonically increasing function.
[0176] Hidden constraints It is about It is a monotonically decreasing function, but rather about It is a monotonically increasing function.
[0177] To maximize while satisfying the hidden constraints The optimal solution must appear on the boundary of the hidden constraint, that is... At the same time, in order to give To create the greatest possible feasible space, It is set to the maximum value it can reach. Therefore, the optimal upper limit of interference power is:
[0178] ;
[0179] At this point, the hidden constraint equation becomes one that only relates to the power ratio. The relevant equations. Because about It is monotonic, and the solution that satisfies the condition can be quickly found using efficient algorithms such as binary search. The unique solution .Should This represents the maximum power ratio allowed to meet concealment requirements.
[0180] Finally, by combining the power upper limit constraint (C3), the optimal covert transmission power is obtained. for:
[0181] ;
[0182] This solution indicates that the optimal transmit power should be taken at the upper limit of the system's physical maximum transmit power. and the power limit determined by concealment constraints The smaller of the two.
[0183] S306, Parameter Execution and Closed-Loop Control: Step S306 of this method is implemented by the execution module 208. This module acts as a bridge connecting optimization decision-making and physical execution, receiving the optimal solution output by the parameter optimization module 207 in step S305, and accurately converting it into physical device control commands for the covert transmitter Alice and the friendly jamming drone Carol.
[0184] Execution module 208 first parses the received optimal parameter set. If the system is running in a fully known scenario, the received solution is the globally optimal particle position:
[0185] .
[0186] If the system operates in a statistically known scenario, the received solution is the optimal parameter set obtained through phased optimization. .
[0187] Subsequently, the execution module 208 transforms these abstract numerical values into concrete, executable instructions:
[0188] For the covert transmitter Alice, the module extracts the optimal transmit power value from the solution. (Right now or This module generates a power setting command and sends it to Alice's communication subsystem via a secure control link.
[0189] Upon receiving the instruction, Alice's communication subsystem precisely adjusted the gain of its transmit power amplifier, stabilizing the final RF output power at [value missing]. .
[0190] For the friendly jamming drone Carol, the module simultaneously configures its power and position:
[0191] Interference power configuration: Execution module 208 extracts the optimal upper limit of interference power. (Right now or A corresponding instruction is sent to Carol's interference load controller. Based on this instruction, the controller sets the operating parameters of its internal random interference generator to ensure the interference power it generates in subsequent operations. It is in the interval The samples were randomly and uniformly drawn from within.
[0192] Location deployment control: Execution module 208 extracts the two-dimensional coordinates of the optimal deployment. In a fully known scenario, these coordinates are... In a statistically known scenario, this coordinate is (0, 0). Execution module 208 compares this two-dimensional coordinate with a preset flight altitude. (For example, the height set to avoid ground obstacles or obtain a better line-of-sight path) combined to form a complete three-dimensional waypoint. The waypoint instruction is sent to Carol's flight control system. Carol's flight control system then plans and executes the flight path, autonomously flying to the target waypoint using its own navigation system (such as GPS / INS), and hovering there to complete the optimal deployment.
[0193] This method, in step S306, not only includes the one-time application of optimal parameters, but more importantly, it forms a continuously operating adaptive closed-loop control mechanism to ensure that the covert communication system maintains optimal performance in dynamically changing environments. This closed-loop control is completed collaboratively by the scene recognition module 202 and the system's central control logic.
[0194] After the initial parameters are executed, the system does not terminate but enters a continuous monitoring state. The scene recognition module 202, together with other related modules, periodically or event-triggered, reassess key parameters of the external environment. Specific monitoring content includes, but is not limited to:
[0195] Changes in communication channel conditions: The system tracks changes in the statistical properties of channel state information (CSI) by exchanging pilot signals between Alice, Bob, or Carol. For example, it monitors changes in the channel's Doppler spread, coherence time, or Rice K-factor, all of which indicate changes in the communication environment (such as scatterer movement).
[0196] Changes in the threat posture of the eavesdropping party:
[0197] In a fully known scenario, the system monitors whether the location of a known eavesdropper has moved, or whether a new eavesdropper has entered the monitored area or an old eavesdropper has left. In a statistically known scenario, the system monitors unsafe areas. Does the geometry or spatial distribution model of the eavesdropper (e.g., parameters of a uniform distribution) need adjustment? Threat weighting factor of the eavesdropper. It will also make dynamic adjustments based on external intelligence input or analysis of the eavesdropping party's behavior (such as changes in the intensity of its signal activity).
[0198] The system incorporates a change significance determination mechanism. This mechanism works by comparing the real-time monitored parameter values with the baseline values used in the previous optimization. When the change in any key parameter exceeds a pre-set threshold (e.g., the statistical mean of channel gain changes by more than 10%, or a listener's location moves beyond a specific percentage of its communication range), or when a specific key event occurs (such as the detection of a new, unknown signal source), the mechanism generates a "re-optimization" trigger signal.
[0199] Once the trigger signal is activated, the system's control flow will automatically return to step S301 and restart the entire "scenario identification - threat modeling - performance quantification - optimization solution" process using the latest acquired environmental parameters. This process will generate a new set of optimal parameter solutions that are more adapted to the current environment. Execution module 208 will then apply these new parameters to adjust Alice's transmit power and Carol's jamming power and deployment location.
Claims
1. A method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance, characterized in that, Includes the following steps: Scene identification and threat modeling: Determine the level of certainty regarding the channel state and location information of multiple eavesdroppers, and based on this, identify the current communication scene as a fully known scene or a statistically known scene, and model the threat of the eavesdroppers. Single eavesdropper detection performance analysis: Establish a detection model for a single eavesdropper and derive the minimum false detection probability for the single eavesdropper; Global stealth performance quantification: Based on the scene recognition results, the minimum error detection probabilities of multiple eavesdroppers are combined into a global stealth performance metric. When the scene is identified as a fully known scene: The threat modeling specifically involves assigning a threat weighting factor to each eavesdropper. Specifically, the global stealth performance quantification involves multiplying the threat weighting factor of each eavesdropper by the minimum false detection probability corresponding to each eavesdropper and summing the results to obtain the global weighted false detection probability, which is then used as the global stealth performance metric. When the scene is identified as a statistically known scene: The threat modeling specifically involves obtaining the spatial probability distribution of the eavesdropping party within a preset area; Specifically, the global covert performance quantification involves calculating the statistical expectation of the minimum false detection probability of a single eavesdropper based on the spatial probability distribution, and obtaining the average minimum false detection probability as the global covert performance metric. Optimization problem construction: Construct an optimization problem with the goal of maximizing covert transmission capacity, and constrained by the global covert performance metric and physical operational limitations; Optimization problem solving: Based on the scene recognition results, the optimization problem of the physical operation constraints is solved using the scene-specific optimization algorithm to obtain the optimal transmission power, interference power upper limit, and deployment position of friendly interference drones; Parameter execution and closed-loop control: Based on the solved transmit power, jamming power upper limit and the deployment position of friendly jamming UAVs, control is performed on the concealed transmitter and friendly jamming UAVs, and a closed-loop control mechanism for adaptive adjustment is established.
2. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 1, characterized in that, The threat weighting factor is determined in the following way: A baseline threat level representing the eavesdropping advantage is calculated for each eavesdropper, and the baseline threat level for all eavesdroppers is normalized.
3. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 1, characterized in that, The average minimum error detection probability is solved based on the pre-derived probability density function that characterizes the distance ratio between the eavesdropper and the covert transmitter and the friendly jamming drone.
4. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 1, characterized in that, The optimization problem is solved using a particle swarm optimization algorithm to jointly optimize the transmission power, the upper limit of the interference power, and the deployment location.
5. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 1, characterized in that, When the scene is identified as a statistically known scene, the specific steps for solving the optimization problem are as follows: decompose the optimization problem into a location optimization sub-problem and a power optimization sub-problem, and solve them in stages.
6. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 5, characterized in that, The phased solution first executes the position optimization sub-problem; the position optimization sub-problem determines the optimal deployment position of the friendly jamming drone that coincides with the position of the covert transmitter by maximizing the average minimum error detection probability.
7. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 6, characterized in that, After determining the optimal deployment location, the power optimization sub-problem is executed; the power optimization sub-problem is solved by setting the upper limit of the interference power to the maximum allowable value and under the boundary condition that the average minimum error detection probability is equal to the preset concealment threshold, to obtain the optimal transmission power.
8. The method for optimizing the communication power and deployment of multiple eavesdroppers based on UAV jamming assistance according to claim 6, characterized in that, The closed-loop control mechanism specifically includes: The system continuously monitors the statistical characteristics of communication channel conditions or changes in the threat posture of the eavesdropper. When any change exceeds a preset significance threshold, it returns to the scenario identification and threat modeling process for re-optimization.