Node task execution method, apparatus, device, storage medium, and program product

Abstract

This application provides a node task execution method, apparatus, device, storage medium, and program product. The method includes: acquiring digital identity information and physical identity information of multiple nodes; the digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity characteristics of the node in the physical perception domain; performing consistency matching on the digital identity information and physical identity information of each node to obtain a matching result; determining a set of trusted candidate neighbor nodes based on the matching result; and executing a decision task based on the set of trusted candidate neighbor nodes. The solution of this application addresses the problem in the prior art where candidate forwarding nodes in UAV ad hoc networks are derived from ordinary neighbor discovery results, lacking verification of node identity trustworthiness.

Description

Technical Field

[0001] This application relates to the field of mobile communication technology, specifically to a node task execution method, apparatus, device, storage medium, and program product. Background Technology

[0002] Routing methods for self-organizing networks based on reinforcement learning or Q-learning can dynamically optimize the route establishment and maintenance process according to link quality, node status, network load, or topology changes, offering advantages in improving path selection flexibility and network adaptability. However, these methods typically assume that the set of neighboring nodes participating in routing decisions is inherently trustworthy, and their candidate forwarding nodes generally still originate from ordinary neighbor discovery results, lacking verification of node identity trustworthiness. In other words, these methods primarily address the problem of "how to choose a route" in dynamic network environments, without further addressing the problem of "who can participate in route selection." When a drone ad hoc network is subjected to Sybil attacks or other identity spoofing attacks, spoofed nodes may still enter the candidate next-hop set, further affecting the correctness of route establishment, the reliability of path selection, and the stability of data transmission. Summary of the Invention

[0003] At least one embodiment of this application provides a node task execution method, apparatus, device, storage medium, and program product to solve the problem in the prior art that candidate forwarding nodes in UAV ad hoc networks are derived from ordinary neighbor discovery results and lack verification of node identity credibility.

[0004] To solve the above-mentioned technical problems, this application is implemented as follows:

[0005] In a first aspect, embodiments of this application provide a node task execution method, including:

[0006] Acquire digital and physical identity information of multiple nodes; the digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity features of the node in the physical perception domain.

[0007] The digital identity information and physical identity information of each node are matched for consistency to obtain a matching result;

[0008] Based on the matching results, a set of reliable candidate neighbor nodes is determined;

[0009] The decision-making task is performed based on the set of trusted candidate neighbor nodes.

[0010] Optionally, a consistency match is performed on the digital identity information and the physical identity information of each node to obtain a matching result, including:

[0011] Calculate the similarity between the node and the target data template based on the digital identity information and the physical identity information;

[0012] Based on the similarity, the matching results are obtained;

[0013] The target data template is the data template corresponding to the digital identity information.

[0014] Optionally, based on the digital identity information and the physical identity information, the similarity between the node and the target data template is calculated, including:

[0015] Based on the digital identity information, determine the IP address of the node;

[0016] If the IP address of the node is verified, obtain the target data template;

[0017] Extract the digital identity features from the digital identity information and the physical identity features from the physical identity information;

[0018] Based on the digital identity features and the physical identity features, determine the fusion features;

[0019] Calculate the similarity between the digital identity features, the physical identity features, and the fused features and the target data template.

[0020] Optionally, a consistency match is performed on the digital identity information and the physical identity information of each node to obtain a matching result, including:

[0021] Based on the digital identity information and physical identity information of the multiple nodes, physical domain features and digital domain features are determined;

[0022] Calculate the similarity between the physical domain features and the digital domain features;

[0023] Based on the similarity, a fusion association cost matrix of the physical domain features and the digital domain features is constructed;

[0024] Based on the node authenticity coefficient and the fusion association cost matrix, the digital identity information and the physical identity information are matched for consistency to obtain a matching result; the node authenticity coefficient is related to the rate of change of the node movement speed.

[0025] Optionally, based on the digital identity information and physical identity information of the plurality of nodes, physical domain features and digital domain features are determined, including:

[0026] Based on the digital identity information and physical identity information of the multiple nodes, obtain a set of physical domain nodes and a set of digital domain nodes within a preset time period; the number of nodes in the set of digital domain nodes is greater than or equal to the number of nodes in the set of physical domain nodes.

[0027] Based on the set of physical domain nodes, determine the physical domain characteristics; based on the set of digital domain nodes, determine the digital domain characteristics.

[0028] Optionally, based on the set of trusted candidate neighbor nodes, a decision-making task is performed, including:

[0029] The set of candidate neighbor nodes is used as the candidate action space for the current node or the current agent.

[0030] In the candidate action space, routing decisions, data forwarding control, or agent decision control are performed.

[0031] Optionally, the method further includes:

[0032] Obtain a set of multiple candidate neighbor nodes at consecutive time points;

[0033] Based on the multiple candidate neighbor node sets, determine the stability index of the candidate neighbor node sets;

[0034] The execution of the decision-making task is adjusted based on the stability index.

[0035] Optionally, the digital identity information includes at least one of the following:

[0036] IP address, MAC address, link layer identifier, session identifier, identity field in control messages, node number in neighbor discovery messages, and radio frequency identity characteristics;

[0037] The physical identity information includes at least one of the following:

[0038] Position, velocity, heading, trajectory continuity, relative distance, relative velocity, azimuth, pitch, altitude, nodal appearance features, structural difference features, attitude Doppler features.

[0039] Secondly, embodiments of this application provide a node task execution device, including:

[0040] The acquisition module is used to acquire digital identity information and physical identity information of multiple nodes; the digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity features of the node in the physical perception domain.

[0041] The matching module is used to perform consistency matching on the digital identity information and the physical identity information of each node to obtain a matching result;

[0042] The determination module is used to determine a set of trustworthy candidate neighbor nodes based on the matching results;

[0043] The execution module is used to perform decision-making tasks based on the set of trusted candidate neighbor nodes.

[0044] Thirdly, embodiments of this application provide a node task execution device, including: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the node task execution method as described above.

[0045] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program, which, when executed by a processor, implements the steps of the node task execution method described above.

[0046] Fifthly, embodiments of this application provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the node task execution method described above.

[0047] Compared with existing technologies, the node task execution method, apparatus, device, storage medium, and program product provided in this application embodiment screens untrusted nodes among multiple nodes by matching the digital and physical identity information of the acquired nodes to determine the set of trustworthy candidate neighbor nodes. Under the constraints of the set of trustworthy candidate neighbor nodes, decision-making tasks such as routing decisions, data forwarding control, or agent decision control are then performed. The solution in this application addresses the problem in existing technologies where candidate forwarding nodes in UAV ad hoc networks are derived from ordinary neighbor discovery results, lacking verification of node identity trustworthiness. Attached Figure Description

[0048] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0049] Figure 1 This is a schematic diagram illustrating the steps of the node task execution method in an embodiment of this application;

[0050] Figure 2 This is a schematic diagram of the iterative optimization process for dynamic network environments according to an embodiment of this application;

[0051] Figure 3 This is a schematic diagram of the module of the node task execution device according to an embodiment of this application;

[0052] Figure 4 This is a schematic diagram of the structure of a network device according to an embodiment of this application. Detailed Implementation

[0053] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, without limiting the number of objects; for example, the first object can be one or more. Furthermore, "or" in this application indicates at least one of the connected objects. For example, "A or B" covers three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B. The character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0054] To enable those skilled in the art to better understand the embodiments of this application, the following description is provided first:

[0055] Flying Ad-hoc Networks (FANETs) are a type of distributed mobile communication network autonomously constructed by multiple unmanned aerial vehicle (UAV) nodes via wireless links. Compared to traditional terrestrial ad-hoc networks, FANETs are characterized by high-speed three-dimensional movement, rapid network topology changes, short link durations, and fluctuating inter-node connectivity, thus placing higher demands on neighbor discovery, link evaluation, route establishment, and maintenance mechanisms. In scenarios such as emergency communication, disaster relief, military reconnaissance, and collaborative operations, FANETs typically need to achieve rapid network deployment and stable transmission in the absence of fixed infrastructure support. Therefore, balancing routing efficiency, link stability, and node reliability in dynamic environments has become a crucial research challenge in this field.

[0056] As described in the background section, in the prior art, the candidate forwarding nodes of the UAV ad hoc network are derived from the results of ordinary neighbor discovery, which lacks verification of the node's identity credibility. To solve the above problems, this application provides a node task execution method, which can reduce or avoid the occurrence of the above situations, improve the communication reliability of the UAV ad hoc network in complex dynamic environments, and improve the user experience.

[0057] This application provides a method and apparatus for executing node tasks. The method and apparatus are based on the same concept, and since they solve problems in similar principles, their implementations can be mutually referenced; repeated details will not be repeated.

[0058] like Figure 1 As shown in the figure, an embodiment of this application provides a node task execution method, which includes the following steps:

[0059] Step 101: Obtain digital identity information and physical identity information of multiple nodes; the digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity characteristics of the node in the physical perception domain.

[0060] Step 102: Perform consistency matching on the digital identity information and physical identity information of each node to obtain the matching result;

[0061] Step 103: Determine a set of trustworthy candidate neighbor nodes based on the matching results;

[0062] Step 104: Execute a decision-making task based on the set of trusted candidate neighbor nodes.

[0063] The node task execution method of this application embodiment filters untrusted nodes from multiple nodes by matching the digital and physical identity information of the acquired nodes to determine the set of trustworthy candidate neighbor nodes. Under the constraints of this set of trustworthy candidate neighbor nodes, it performs decision-making tasks such as routing decisions, data forwarding control, or agent decision control. This application's solution solves the problem in the prior art where candidate forwarding nodes in UAV ad hoc networks are derived from ordinary neighbor discovery results, lacking verification of node identity trustworthiness.

[0064] Optionally, the digital identity information corresponds to auditory domain information in the communication sensing link; it includes at least the communication identity identifier of the node, and may further include radio frequency identity features that can reflect individual differences of the transmitting end communication equipment;

[0065] The physical identity information corresponds to the visual domain information in the physical perception link; the physical identity information includes at least the node's position, speed, trajectory status or spatial motion characteristics, and may further include appearance and structural features that can reflect the differences between UAV entities.

[0066] The acquisition of digital identity information in the auditory domain can be categorized into two types: passive listening and active interaction. The former extracts identity fields from control messages, broadcast messages, or data messages sent by neighboring nodes via wireless listening. The latter acquires the currently declared communication identity of a node through handshake, probe, query, or link maintenance interaction processes. Through these methods, an explicit identity representation of the node in the communication domain can be obtained.

[0067] Optionally, the digital identity information includes, but is not limited to: IP address, MAC address, link layer identifier, control message identity field and / or radio frequency fingerprint features, as well as node number, timestamp and other network protocol fields that can characterize the logical identity of a node in a neighbor discovery message.

[0068] In this embodiment of the application, to enhance the ability of digital identity information to distinguish between forged, copied, or replayed identities, radio frequency (RF) features can be further introduced as supplementary descriptions. Unlike identity representations that rely solely on protocol fields, address fields, or link layer declaration values, RF features are directly derived from physical layer waveforms and can reflect, to a certain extent, the minor differences introduced by transmitter hardware non-ideals.

[0069] Optionally, acquiring the radio frequency signal includes:

[0070] The original sequence is obtained by performing complex baseband sampling on the node link signal using an RF receiving front-end;

[0071] Based on the original sequence, the valid transmission segments are determined.

[0072] Specifically, since background noise, blank periods, and invalid sampling segments may coexist in the original wireless environment, the system first extracts valid transmission segments from the continuous sampling stream using energy detection or burst detection algorithms, and then extracts the radio frequency transmission content that is truly relevant to the current communication session from the continuous sampling stream.

[0073] In this embodiment of the application, in order to enhance the stability and comparability of RF features, the following preprocessing operations are performed on the effective transmission segment:

[0074] (1) Remove DC components to eliminate DC bias introduced by the receiving link;

[0075] (2) Amplitude normalization to reduce the impact of different reception intensities on model training and feature comparison;

[0076] (3) Frequency offset correction to compensate for the carrier frequency offset between the receiver and the transmitter;

[0077] (4) Time alignment and slicing to ensure that the samples entering the subsequent processing units are consistent in time scale.

[0078] Considering that a single signal representation is insufficient to fully characterize RF identity information, this embodiment constructs three complementary representations from the same preprocessed signal, including:

[0079] I / Q sequence representation, short-time Fourier representation, and logarithmic spectrogram representation;

[0080] Through the above-mentioned multi-representation construction, the digital identity information in the auditory domain can include not only the explicit identity identifier at the protocol layer, but also the implicit RF identity description reflecting the physical emission characteristics of the device, thereby improving the reliability of subsequent dual-identity (digital identity information and physical identity information) matching.

[0081] If the RF encoding is completed, feature extraction branches can be constructed for the three RF representations mentioned above, and a unified RF identity representation can be formed in the future.

[0082] Visual domain information can be acquired based on visual devices. The system acquires image sequences corresponding to the current communication session through external camera equipment, ground monitoring equipment, or airborne visual sensors.

[0083] Optionally, the physical identity information of the visual domain can be obtained through various sensing devices. The types of information provided by different sensing sources may be different, but they can all serve as the basis for observing the existence and state evolution of physical domain entities.

[0084] Specifically, visual domain information can be acquired based on radar or integrated sensing devices. Specifically, information such as relative distance, relative velocity, azimuth, pitch angle, altitude, direction of motion, and rotor-related micro-motion characteristics of the target node can be extracted using electromagnetic wave propagation delay, Doppler frequency shift, angle estimation, or micro-Doppler analysis results. Among these, distance and velocity information can be used to characterize the instantaneous state of the target node in physical space, angle and altitude information can be used to assist in establishing the spatial positional relationships of the node, and micro-Doppler and motion continuity information can be used to enhance the ability to distinguish real flying entities. By correlating continuous observation results at adjacent moments, further descriptions of physical behaviors such as node trajectory continuity, acceleration change trends, and motion smoothness can be formed.

[0085] In one embodiment of this application, the node includes an intelligent agent, such as a drone.

[0086] Specifically, during the operation of the UAV ad hoc network, each UAV (the node) acquires the auditory domain digital identity information of its neighboring nodes through its onboard wireless communication module, and acquires the visual domain information of itself and its neighboring nodes through an onboard integrated radar and communication front-end, millimeter-wave radar, camera, lidar, GNSS, IMU, or flight control telemetry interface. The visual domain information includes relative distance, relative speed, azimuth angle, pitch angle, three-dimensional position, altitude, heading, and / or motion continuity characteristics.

[0087] Since the original images often contain complex backgrounds and non-target areas, to ensure the consistency of the subsequently extracted visual information with the current communication object, this embodiment first uses a drone detector to obtain candidate target bounding boxes, and then obtains the target trajectory consistent with the current session through multi-frame tracking. This isolates the visual observation from the complex environmental background, forming a more targeted physical identity input for the drone entity. After obtaining the cropped target image sequence, this embodiment can also perform size normalization, pixel normalization, and enhancement processing on the visual samples to form a visual input tensor, thereby extracting visual identity description information such as the aircraft outline, dimensions, color distribution, rotor layout, attitude changes, and short-term dynamic features.

[0088] In addition to radar and visual methods, in this embodiment, the physical identity information of the visual domain can also be obtained through navigation and flight control interfaces. Specifically, the three-dimensional position, heading angle, attitude angle, velocity vector, rate of climb, acceleration, and other motion state variables of the node can be obtained through GNSS, RTK, IMU, magnetometer, barometer, or flight control telemetry interfaces. This type of information can be used as direct input to the local state, or it can be used to obtain auxiliary physical identity descriptions of neighboring nodes through link broadcasting, cooperative sensing, or shared state announcements. When visual observation is affected by occlusion or radar observation is uncertain, navigation and flight control state variables can be used as supplementary information to participate in the construction of the physical domain identity representation, thereby improving the completeness and robustness of the source of the visual domain physical identity information.

[0089] Since the digital identity information in the auditory domain and the physical identity information in the visual domain originate from the communication link and the perception link, respectively, the two types of information usually differ in terms of sampling frequency, time reference, coordinate representation method, and observation noise characteristics. Therefore, before entering the subsequent dual-identifier matching module, this embodiment further performs unified processing on the acquired dual-identifier information to form standardized dual-identifier input data. The unified processing includes at least the following steps: time synchronization, spatial coordinate unification, observation window alignment, anomaly removal, feature normalization, and confidence calibration.

[0090] Time synchronization is used to map data observations from different sources to the same time slot or time window to ensure temporal comparability between the entity states in the visual domain and the identity claims in the auditory domain. For periodic broadcast messages, RF sampling segments, radar observation results, visual image sequences, and navigation state quantities, unified alignment can be performed based on timestamps, sampling times, or synchronization triggering mechanisms. Spatial coordinate unification is used to transform the azimuth, distance, altitude, geographic coordinates, or local coordinates output by different physical sensors to a unified reference coordinate system. For example, it can be converted to a local three-dimensional coordinate system with the current node as the reference origin, or to a unified geographic coordinate expression, to facilitate the subsequent construction of spatial relationships between nodes.

[0091] Observation window alignment ensures that selected auditory and visual observations within the same time slot have consistent temporal coverage, avoiding erroneous associations due to outdated or overly recent data. Anomaly removal filters out anomalous observations that significantly deviate from normal physical motion patterns or are inconsistent with the communication session context, such as sudden, large positional changes, abrupt velocity changes, continuously missing visual targets, and significantly distorted RF sampling segments. Feature normalization reduces numerical differences between observations of different dimensions and magnitudes, facilitating subsequent unified similarity measurement and fusion processing. Confidence calibration adds confidence references to information from different sources based on observation source quality, sensor status, environmental conditions, or historical stability, enabling differentiated weighting of different features in the subsequent matching stage.

[0092] Optionally, a consistency match is performed on the digital identity information and the physical identity information of each node to obtain a matching result, including:

[0093] Calculate the similarity between the node and the target data template based on the digital identity information and the physical identity information;

[0094] Based on the similarity, the matching results are obtained;

[0095] The target data template is the data template corresponding to the digital identity information.

[0096] Optionally, calculating the similarity between the node and the target data template based on the digital identity information and the physical identity information includes:

[0097] Based on the digital identity information, determine the IP address of the node;

[0098] If the IP address of the node is verified, obtain the target data template;

[0099] Extract the digital identity features from the digital identity information and the physical identity features from the physical identity information;

[0100] Based on the digital identity features and the physical identity features, determine the fusion features;

[0101] Calculate the similarity between the digital identity features, the physical identity features, and the fused features and the target data template.

[0102] The node task execution method in this embodiment uses a similarity comparison between a pre-registered template library and the currently observed samples to achieve consistent authentication between auditory domain digital identities and visual domain physical entities. This embodiment is applicable to application scenarios where node identities can be pre-registered in the network and the system can establish binding templates for legitimate nodes, such as formation flying, task group communication, and controlled access unmanned aerial vehicle communication networks.

[0103] In this embodiment of the application, the current access request is denoted as:

[0104]

[0105] in, This is the communication source address (e.g., IP address). This indicates the received RF signal. This represents visual observations collected within the same time window;

[0106] For each registered and valid address, the system pre-creates the aforementioned data template:

[0107]

[0108] in, This is the number of the communication source address. for The corresponding radio frequency signal (digital identity information); for Corresponding visual information (physical identity information); for The information is a fusion of the corresponding digital identity information and physical identity information;

[0109] The system first performs IP pre-authentication upon startup;

[0110] If the current address fails the validity check, the subsequent process will terminate immediately; if it passes the check, the process will proceed according to the address. Read the target template from the registry. ;

[0111] Subsequently, the system preprocesses and extracts features from the RF data and visual data in the current session to obtain the features of the digit to be tested. With the physical characteristics to be measured And further construct fusion features Therefore, assuming IP address pre-authentication has been successful, the following conclusions are drawn: With template Whether the results remain consistent is defined as the binary authentication result.

[0112] ;in, Indicates matching, This indicates a mismatch.

[0113] Specifically, the current sample is compared with the template data corresponding to the current IP, without performing a global search across all templates. First, three types of similarity scores are defined: RF matching score: Image allocation score: ; and the fusion matching score: ;

[0114] in, It is a function of cosine similarity, Euclidean distance transformation, or other measurable feature similarity.

[0115] This application uses a two-level decision mechanism: the first level is IP pre-authentication, and the second level is binding consistency authentication; since IP verification is only used as an admission condition, the final output is only a binary result of "match" or "not match".

[0116] This embodiment verifies the binding relationship between "legitimate digital identity and real drone entity". It is applicable to controlled communication scenarios with pre-registration conditions. It can quickly determine whether the current request is initiated by the corresponding entity during the node access or session establishment stage, thereby filtering out abnormal nodes that only forge addresses but fail to forge both digital and physical identity information before the network is running.

[0117] Optionally, a consistency match is performed on the digital identity information and the physical identity information of each node to obtain a matching result, including:

[0118] Based on the digital identity information and physical identity information of the multiple nodes, physical domain features and digital domain features are determined;

[0119] Calculate the similarity between the physical domain features and the digital domain features;

[0120] Based on the similarity, a fusion association cost matrix of the physical domain features and the digital domain features is constructed;

[0121] Based on the node authenticity coefficient and the fusion association cost matrix, the digital identity information and the physical identity information are matched for consistency to obtain a matching result; the node authenticity coefficient is related to the rate of change of the node movement speed.

[0122] Optionally, based on the digital identity information and physical identity information of the plurality of nodes, physical domain features and digital domain features are determined, including:

[0123] Based on the digital identity information and physical identity information of the multiple nodes, obtain a set of physical domain nodes and a set of digital domain nodes within a preset time period; the number of nodes in the set of digital domain nodes is greater than or equal to the number of nodes in the set of physical domain nodes.

[0124] Based on the set of physical domain nodes, determine the physical domain characteristics; based on the set of digital domain nodes, determine the digital domain characteristics.

[0125] Optionally, the preset time period includes at least one time slot.

[0126] For example, in the nth time slot, the set of visual domain nodes (the set of physical domain nodes) is obtained: And the set of nodes in the auditory domain (the set of nodes in the digital domain): ;

[0127] in, Indicates the first The first time slot observed Each visual domain node This indicates the number of nodes in the visual domain node set;

[0128] This represents the nth auditory domain node detected in the nth time slot. This represents the number of nodes in the auditory domain node set.

[0129] Since visual domain nodes and auditory domain nodes originate from the physical domain and communication domain respectively, the number of nodes in the two sets may differ, and there is no natural one-to-one correspondence between them; furthermore, the digital identities of nodes are relatively easy to forge. Therefore, the number of nodes in the digital domain node set is greater than or equal to the number of nodes in the physical domain node set.

[0130] The node task execution method in this application combines physical feature differences based on relative entropy and dynamic feature weights based on popularity to calculate the similarity between dual-domain physical identifiers. The visual domain neighbor nodes and the first The bi-domain similarity between neighboring nodes in the auditory domain is:

[0131]

[0132] in, ; ; ;

[0133] Indicates the dimension of the selected physical feature; Indicates the first The observation of the nth feature of the nth neighboring node; Indicates the neighbor node number in the visual domain; It is the first The similarity measure is obtained by normalizing the JSD differences of each feature. Set its normalized weights.

[0134] Furthermore, based on the aforementioned dual-domain similarity, the fusion association cost matrix is ​​constructed as follows: ;

[0135] in, Indicates the first The visual domain node and the first The matching cost between auditory domain nodes and These represent the number of nodes in the visual domain node set and the auditory domain node set, respectively.

[0136] Generally, it can be set as follows: ;

[0137] Alternatively, other monotonically equivalent cost mapping methods can be used; the smaller the cost value, the stronger the consistency between the visual entity (physical identity information) and the communication identity (communication identity information).

[0138] In this embodiment, the node authenticity coefficient is determined by the rate of change of speed; the smoother the node speed change, the larger its authenticity coefficient; when the speed change is abnormal, the authenticity coefficient will decrease accordingly.

[0139] Since the numerical range of the fusion cost matrix may vary significantly across different network scenarios, an adaptive convergence parameter is introduced based on the cost matrix scale. First, the average scale of the fusion cost is calculated. And define the basic convergence parameters. Construct dynamic adjustment factors ;

[0140] in, ; ; ; ;

[0141] Therefore, the dynamic convergence parameter is: ;

[0142] in, This is an adjustment coefficient. To control the computational complexity of the algorithm and meet the real-time requirements of UAV ad hoc networks, the maximum number of iterations is set to [value missing]. .

[0143] During the matching process, let Indicates the first The auditory domain node at the _ The matching adjustment parameters in the next iteration, initially:

[0144]

[0145] For visual domain nodes that have not yet established associations, first calculate their comprehensive matching evaluation value with each auditory domain node:

[0146]

[0147] And select the node with the highest matching evaluation value as the candidate matching object:

[0148]

[0149] Simultaneously record the evaluation value of the suboptimal match:

[0150]

[0151] When calculating the node matching adjustment, the difference between the maximum matching evaluation value and the second-best matching evaluation value is used as the basic update value, and a small positive term is introduced to ensure stable convergence of the algorithm during the iteration process:

[0152]

[0153] in, This is the step size adjustment coefficient. Combined with the node authenticity coefficient, the final matching adjustment amount is:

[0154]

[0155] The matching relationship is then updated based on the adjustment amount of each node, and the matching adjustment parameters of the corresponding auditory domain nodes are updated accordingly.

[0156]

[0157] To prevent some nodes from being in a disadvantageous state for a long time in the early iterations due to excessively large matching adjustment parameters, a weight decay mechanism is further introduced. Let the basic decay coefficient be... The dynamic decay rate is then defined as:

[0158]

[0159] And update the node matching adjustment parameters:

[0160]

[0161] After the above iterative process, the algorithm outputs the final two-domain node matching result:

[0162]

[0163] in,

[0164] ;

[0165] when When this occurs, it indicates that the node has not successfully established a dual-domain matching relationship, and therefore the node can be identified as a potential witch node or an abnormal identity node.

[0166] Optionally, of the two methods for matching the digital identity information and the physical identity information, the first method focuses on quickly authenticating the current request by binding a template under the condition of a known set of legitimate identities, and is more suitable as a pre-step for access control, session authentication or verification of the identity of registered nodes;

[0167] The second method focuses on continuously establishing an online correspondence between physical entities and communication identities based on dual-domain observation results in the same or adjacent time slots during network operation. It is more suitable as a basic support mechanism for candidate neighbor set reconstruction, abnormal node identification, and trusted routing decision-making in UAV ad hoc networks.

[0168] The node task execution method of this application further includes: reconstructing the candidate neighbor node set;

[0169] Specifically, set nodes At any moment The set of physically connected neighbors is:

[0170]

[0171] in, Represents a node With nodes At any moment Euclidean distance, This represents the effective communication radius of a node. The set of physically connected neighbors only reflects the reachability between nodes in terms of spatial location and communication range, and does not consider the consistency and authenticity of node identities. Therefore, it may include nodes with abnormal identities, witch-like identities, or nodes that are physically reachable but not suitable for participating in trusted routing.

[0172] Based on the aforementioned matching results, consistency filtering and dynamic reconstruction of the physically connected neighbor set can yield the nodes. At any moment The set of candidate neighbor nodes: ;

[0173] in, Represents a node The dual-domain identity consistency determination result, when the match is successful. ,otherwise ; Represents a node The authenticity coefficient; This represents the threshold for authenticity.

[0174] If only binary matching results are output in some embodiments, the candidate neighbor node set can also degenerate into:

[0175]

[0176] Based on the above definition, nodes In subsequent routing or agent decision-making processes, only from the set Select a trusted next-hop node or a trusted collaborating object.

[0177] Optionally, the method further includes:

[0178] Obtain a set of multiple candidate neighbor nodes at consecutive time points;

[0179] Based on the multiple candidate neighbor node sets, determine the stability index of the candidate neighbor node sets;

[0180] The execution of the decision-making task is adjusted based on the stability index.

[0181] like Figure 2 As shown, the node task execution method further includes:

[0182] Constrained routing or agent decisions are made based on the set of candidate neighbor nodes, and relevant parameters are updated by combining state / reward feedback, thus forming an iterative optimization process for dynamic network environments.

[0183] To further characterize the dynamic changes in the local network environment, this embodiment introduces a trusted neighbor stability parameter based on the degree of change in the trusted neighbor set at adjacent times. Let node... At any moment and The trusted neighbor sets are respectively and Then the rate of change of reliable neighbors can be characterized by the ratio of their symmetric difference and union:

[0184]

[0185] in, This represents the symmetric difference operation between sets. Represents the cardinality of a set. To prevent small positive numbers with a denominator of zero, the rate of change... The smaller the value, the smaller the change in the set of trusted neighbors between adjacent times, and the more stable the local network topology; conversely, the larger the value, the more drastic the change in the set of trusted neighbors, and the greater the fluctuation in the local topology.

[0186] Based on this, a stability coefficient (the stability index) is further constructed:

[0187]

[0188] Obviously, When the difference between the sets of trusted neighbors at two different times is small, A relatively large value indicates that the current local network is relatively stable; when the set of trusted neighbors changes drastically, The stability coefficient is relatively small. It can be further applied to aspects of reinforcement learning such as discount terms, adaptive learning rate updates, exploration intensity control, candidate action credibility adjustment, and agent policy updates, thereby making subsequent decision-making processes more adaptable to the time-varying topology characteristics of UAV ad hoc networks.

[0189] The stability coefficient can be further applied to various stages in reinforcement learning, such as discount terms, adaptive learning rate updates, exploration intensity control, candidate action credibility adjustment, and agent policy updates, thereby making the subsequent decision-making process more adaptable to the time-varying topology characteristics of UAV ad hoc networks.

[0190]

[0191] in, This represents the sliding window length. This method can reduce the impact of sudden fluctuations at a single moment on the stability assessment results, making subsequent routing parameter updates and agent policy updates smoother.

[0192] Based on the set of trusted candidate neighbor nodes, a decision-making task is performed, including:

[0193] The set of candidate neighbor nodes is used as the candidate action space for the current node or the current agent.

[0194] In the candidate action space, routing decisions, data forwarding control, or agent decision control are performed.

[0195] In this embodiment, reinforcement learning is used to optimize hop-by-hop routing behavior. The current forwarding node can be regarded as the state carrier of reinforcement learning, and the set of candidate neighbor nodes at the current moment serves as the action space. For nodes... Its action set is defined as:

[0196]

[0197] That is, nodes The next hop can only be chosen from the current set of candidate neighbor nodes. In this way, the identity consistency result output by the dual-identity matching module is directly transformed into action space constraints for reinforcement learning, thereby restricting nodes with abnormal identities from participating in path construction.

[0198] In terms of state modeling, the current local network state of a node can be represented as:

[0199]

[0200] The state includes not only the node's own position and velocity information, but also the current set of trusted neighbors, neighbor stability coefficients, and stability, cost, and progress indicators of each candidate link. Through this design, the reinforcement learning decision-making process can directly perceive the constraints imposed on the local network structure by the dual-identifier matching results.

[0201] Taking into account the above factors, the instant reward function for hop-by-hop forwarding is defined as follows:

[0202]

[0203] in, , and Let represent the weights of the link benefit term, forwarding cost term, and forwarding advancement term, respectively, and satisfy the following:

[0204]

[0205] In an optional implementation, if the risk of packet loss or node credibility is further introduced, the reward function can be extended to:

[0206]

[0207] in, , These represent the weights for reliability and authenticity, respectively. This reward design allows nodes to simultaneously consider link stability, hop-by-hop transmission cost, destination node progress, link reliability, and identity trustworthiness when selecting the next hop.

[0208] After performing a forwarding action, the Q-value is updated based on the current reward, the optimal value of the next state, and the dynamic learning rate. The update formula is as follows:

[0209] ;

[0210] in, The current learning rate, This is a stability coefficient derived from changes in trusted neighbors. Unlike the traditional Q-Learning approach that uses a fixed discount factor, this embodiment uses... As a discount factor, the value update process can better reflect the dynamic topological changes of the UAV ad hoc network: when the local network is stable, Larger networks have greater potential for future value inheritance; when local networks experience significant fluctuations, The smaller the value, the weaker the impact of unstable future states on the current update.

[0211] In one optional implementation, UAV nodes, cluster head nodes, edge control units, or ground collaborative control terminals in the network can be abstracted as agents. The agent receives the identity consistency result, node authenticity coefficient, candidate neighbor node set, neighbor stability parameters, and link status information output by the dual-identity matching module, and selects the optimal action in the candidate action space based on a preset objective function to achieve trusted decision control for UAV ad hoc networks.

[0212] Among them, the intelligent agent at any time The observed state can be represented as:

[0213]

[0214] in, Represents the set of candidate neighbor nodes. Represents the neighbor stability parameter. Represents the set of node authenticity coefficients. Indicates link status information, This indicates the current task objective or network control objective; the link status information may include one or more of the following: link duration, hop-by-hop delay, progress, predicted packet loss rate, and reachability information.

[0215] Based on this, the set of actions of an intelligent agent can be defined as:

[0216] ;

[0217] Each action can correspond to one or more of the following: next-hop node selection, relay node switching, path reconstruction, resource allocation, transmit power adjustment, task offloading, spectrum access, or cooperative control commands. For scenarios where hop-by-hop forwarding is the primary task, the action set can directly degenerate into a candidate neighbor node set; for cooperative control scenarios, the action set can further include actions such as formation reconstruction, task handover, and link recovery.

[0218] After the action is performed, the environment provides a reward value:

[0219]

[0220] in, Indicates latency or cost information. This represents an indicator of task completion, network service quality, or overall control effectiveness. The reward function encourages agents to prioritize actions that are trustworthy in identity, have stable links, low cost, and effectively advance the task objective. One possible linear expression for this function is:

[0221]

[0222] in, This represents the average confidence level corresponding to the candidate action. Indicates average link stability revenue. This represents the average cost. Indicates the benefits of task progress. These are the weighting coefficients.

[0223] Furthermore, in the single-agent implementation, the current forwarding node can be considered a single agent, the set of candidate neighbor nodes can be used as its candidate action space, and reinforcement learning, deep reinforcement learning, or other policy optimization methods can be used to select the next-hop node and update the policy parameters. In this case, the aforementioned Q-learning routing instance can be regarded as a special case of the agent application instance. That is, the node acts as an agent, the environmental state is constrained by the dual-identifier matching result, the action is limited by the set of candidate neighbor nodes, and the reward is jointly determined by link stability, forwarding cost, progress, reliability, and authenticity.

[0224] Let the first An intelligent agent at time Local observations are Its local strategy is denoted as The multi-agent joint strategy can then be expressed as:

[0225]

[0226] in, This indicates the number of agents. This method not only enables hop-by-hop routing optimization but also extends the application of dual-identity consistency results to various intelligent scenarios such as cooperative relay, task allocation, spectrum access, power control, and formation communication.

[0227] In an optional embodiment, to reflect the direct constraint effect of the dual-identifier matching result on the agent's decision space, a trust-based action mask can be defined:

[0228]

[0229] in,

[0230] ;

[0231] The final set of actions that the intelligent agent can execute is:

[0232] .

[0233] Therefore, the dual-identifier consistency evaluation results are further transformed into environmental constraints that can be perceived and utilized by the agent, enabling the agent to actively avoid abnormal identity nodes, witch identity nodes, or communication identities lacking physical entity support when making decisions, thereby improving the autonomous decision-making ability, security, and robustness of UAV ad hoc networks in dynamic environments.

[0234] To enhance the algorithm's adaptability to network fluctuations, this embodiment further introduces an adaptive learning rate mechanism based on stability parameters. Let the upper and lower bounds of the learning rate be... and The learning rate will then be dynamically adjusted as follows:

[0235] ;

[0236] When the set of trusted neighbors changes significantly A smaller learning rate allows the algorithm to adapt to local topological changes more quickly; when the local neighbor structure is relatively stable, A larger value results in a smaller learning rate, thus improving the policy's convergence stability. If a sliding window stability parameter is used... Then it can also be replaced. Involve learning rate adjustment to achieve smoother parameter update results.

[0237] During the action selection phase, nodes choose actions from either a set of trusted neighbors or a set of valid actions, employing a value-driven strategy with exploratory capabilities. For example, this could be achieved using... Way:

[0238]

[0239] For agent decision-making scenarios, it can also be written as:

[0240]

[0241] in, To explore probabilities, this design can retain a certain degree of exploration capability while ensuring the utilization of the current optimal value, thereby reducing the risk of getting trapped in local suboptimal actions.

[0242] When the Q-values ​​of multiple candidate nodes or candidate actions are close, forwarding progress or task progress benefit can be introduced as auxiliary criteria, that is, actions that meet the following conditions should be given priority:

[0243]

[0244] in, To assist in the judgment criteria weighting, Indicates action At any moment Benefits from advancing the target node, completing the task, or repairing the link. By introducing auxiliary advancement items, the decision-making problem of stalemate when action values ​​are similar can be reduced.

[0245] In an optional embodiment, to avoid path oscillations caused by local loops, a set of recently visited nodes can be maintained. And when constructing the candidate set, the most recently visited node is excluded first, that is, the modified candidate action set is defined:

[0246]

[0247] After strict exclusion When this happens, you can fall back to the previous state by reducing the taboo window length or relaxing some of the recently accessed node constraints. This is to balance the need for loop suppression and accessibility.

[0248] In this embodiment, the current routing process or the current agent control process is terminated when any of the following conditions are met:

[0249] (1) The current node reaches the destination node, that is:

[0250]

[0251] (2) The current node does not have a usable next-hop node or a valid action, that is:

[0252]

[0253] (3) The current task completion rate has reached the preset target threshold, that is:

[0254]

[0255] in, This is the threshold for task completion.

[0256] Upon completion of the process, the application results are output. These outputs may include, but are not limited to: whether the destination node was successfully reached, the actual forwarding path, the number of hops, the cumulative reward, the cumulative latency, trusted neighbor coverage, task completion rate, action sequence, and whether low-trust nodes were encountered along the path. Based on these outputs, the system can further evaluate the routing performance or agent decision-making performance under dual-identity constraints, and provide a basis for subsequent parameter adjustments, offline analysis, or experimental statistics.

[0257] Using the above method, this embodiment directly applies the dual-domain identity consistency evaluation results to the dynamic reconstruction of the candidate neighbor set, and uses the reconstructed candidate neighbor node set as the action space constraint for reinforcement learning routing or agent decision-making. At the same time, the learning parameters are adaptively adjusted according to the changes in the candidate neighbor set, so that the decision-making process can be carried out under the identity consistency constraint, reducing the probability of abnormal nodes participating in path construction and collaborative control, and improving the communication stability, data transmission reliability and autonomous decision-making capability of UAV ad hoc networks in dynamic environments.

[0258] The various methods of the embodiments of this application have been described above. Apparatus for implementing the above methods will now be provided.

[0259] like Figure 3 As shown in the figure, this application embodiment also provides a node task execution device 300, including:

[0260] The acquisition module 301 is used to acquire digital identity information and physical identity information of multiple nodes; the digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity features of the node in the physical perception domain.

[0261] The matching module 302 is used to perform consistency matching on the digital identity information and the physical identity information of each node to obtain a matching result;

[0262] The determining module 303 is used to determine a set of reliable candidate neighbor nodes based on the matching results;

[0263] The execution module 304 is used to perform a decision-making task based on the set of trusted candidate neighbor nodes.

[0264] The node task execution device in this application embodiment filters out untrusted nodes from multiple nodes by matching the digital and physical identity information of the acquired nodes to determine the set of trustworthy candidate neighbor nodes. Under the constraints of this set of trustworthy candidate neighbor nodes, it performs decision-making tasks such as routing decisions, data forwarding control, or agent decision control. This application's solution addresses the problem in existing technologies where candidate forwarding nodes in UAV ad hoc networks are derived from ordinary neighbor discovery results, lacking verification of node identity trustworthiness.

[0265] Another embodiment of the node task execution device of this application, such as Figure 4As shown, it includes a transceiver 410, a processor 400, a memory 420, and a program or instructions stored in the memory 420 and executable on the processor 400; when the processor 400 executes the program or instructions, it implements the various processes of the above-described node task execution method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0266] The transceiver 410 is used to receive and send data under the control of the processor 400.

[0267] Among them, Figure 4 In this context, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 400) and memory (memory 420). The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 410 may be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. The processor 400 is responsible for managing the bus architecture and general processing, and the memory 420 may store data used by the processor 400 during operation.

[0268] This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described node task execution method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0269] This application also provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, they implement the various processes of the above-described node task execution method embodiments and achieve the same technical effects. To avoid repetition, they will not be described again here.

[0270] It should be noted that the collection, gathering, updating, analysis, processing, use, transmission, and storage of user personal information involved in this disclosed technical solution all comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken to prevent unauthorized access to user personal information data and to safeguard user personal information security and network security.

[0271] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0272] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0273] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for executing node tasks, characterized in that, include: Obtain digital and physical identity information from multiple nodes; The digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity features of the node in the physical perception domain. The digital identity information and physical identity information of each node are matched for consistency to obtain a matching result; Based on the matching results, a set of reliable candidate neighbor nodes is determined; The decision-making task is performed based on the set of trusted candidate neighbor nodes.

2. The method according to claim 1, characterized in that, For each node, the digital identity information and the physical identity information are matched for consistency to obtain a matching result, including: Calculate the similarity between the node and the target data template based on the digital identity information and the physical identity information; Based on the similarity, the matching results are obtained; The target data template is the data template corresponding to the digital identity information.

3. The method according to claim 2, characterized in that, Based on the digital identity information and the physical identity information, the similarity between the node and the target data template is calculated, including: Based on the digital identity information, determine the IP address of the node; If the IP address of the node is verified, obtain the target data template; Extract the digital identity features from the digital identity information and the physical identity features from the physical identity information; Based on the digital identity features and the physical identity features, determine the fusion features; Calculate the similarity between the digital identity features, the physical identity features, and the fused features and the target data template.

4. The method according to claim 1, characterized in that, For each node, the digital identity information and the physical identity information are matched for consistency to obtain a matching result, including: Based on the digital identity information and physical identity information of the multiple nodes, physical domain features and digital domain features are determined; Calculate the similarity between the physical domain features and the digital domain features; Based on the similarity, a fusion association cost matrix of the physical domain features and the digital domain features is constructed; Based on the node authenticity coefficient and the fusion association cost matrix, the digital identity information and the physical identity information are matched for consistency to obtain a matching result; the node authenticity coefficient is related to the rate of change of the node movement speed.

5. The method according to claim 4, characterized in that, Based on the digital identity information and physical identity information of the multiple nodes, physical domain features and digital domain features are determined, including: Based on the digital identity information and physical identity information of the multiple nodes, obtain a set of physical domain nodes and a set of digital domain nodes within a preset time period; the number of nodes in the set of digital domain nodes is greater than or equal to the number of nodes in the set of physical domain nodes. Based on the set of physical domain nodes, determine the physical domain characteristics; based on the set of digital domain nodes, determine the digital domain characteristics.

6. The method according to claim 1, characterized in that, Based on the set of trusted candidate neighbor nodes, a decision-making task is performed, including: The set of candidate neighbor nodes is used as the candidate action space for the current node or the current agent. In the candidate action space, routing decisions, data forwarding control, or agent decision control are performed.

7. The method according to claim 6, characterized in that, The method further includes: Obtain a set of multiple candidate neighbor nodes at consecutive time points; Based on the multiple candidate neighbor node sets, determine the stability index of the candidate neighbor node sets; The execution of the decision-making task is adjusted based on the stability index.

8. The method according to claim 1, characterized in that, The digital identity information includes at least one of the following: IP address, MAC address, link layer identifier, session identifier, identity field in control messages, node number in neighbor discovery messages, and radio frequency identity characteristics; The physical identity information includes at least one of the following: Position, velocity, heading, trajectory continuity, relative distance, relative velocity, azimuth, pitch, altitude, nodal appearance features, structural difference features, attitude Doppler features.

9. A node task execution device, characterized in that, include: The acquisition module is used to acquire digital and physical identity information of multiple nodes; The digital identity information is used to characterize the logical identity of the node in the communication domain, and the physical identity is used to characterize the entity features of the node in the physical perception domain. The matching module is used to perform consistency matching on the digital identity information and the physical identity information of each node to obtain a matching result; The determination module is used to determine a set of trustworthy candidate neighbor nodes based on the matching results; The execution module is used to perform decision-making tasks based on the set of trusted candidate neighbor nodes.

10. A node task execution device, characterized in that, include: Transceiver, processor, memory, and programs or instructions stored in the memory and executable on the processor; When the processor executes the program or instructions, it implements the steps of the method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 8.

12. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 8.

Images