Unmanned aerial vehicle key distribution method and system based on post-quantum cryptography

Abstract

The application belongs to the technical field of unmanned aerial vehicle cluster communication security, and relates to an unmanned aerial vehicle key distribution method and system based on a post-quantum cipher, which comprises the following steps: obtaining a post-quantum key encapsulation ciphertext stream and analyzing a memory structure, stripping to obtain a common algebraic matrix and a node characteristic vector of a member node in a corresponding cluster; constructing a perception matrix reflecting the spatial topology and channel bandwidth capacity of the cluster; establishing an asymmetric mapping logic of the common algebraic matrix and the node characteristic vector on a physical channel, delivering the common algebraic matrix to an omnidirectional low-frequency multicast channel, and routing the node characteristic vector to a directional high-frequency channel; through cross-layer perception and decoupling of the ciphertext algebraic structure, the application eliminates the retransmission redundancy of the common load in the cluster key distribution process, solves the broadcast storm caused by large-volume ciphertext under a dynamic fading channel, and improves the convergence speed of the security protocol under a poor channel environment.

Description

Technical Field

[0001] This invention belongs to the field of UAV swarm communication security technology, and relates to a UAV key distribution method and system based on post-quantum cryptography, which is used for key security distribution of UAV swarms under dynamic fading channels, and can improve the efficiency and reliability of key distribution. Background Technology

[0002] Currently, communication security for drone swarms is crucial for ensuring collaborative operations, but existing technologies still have limitations in handling quantum attack resistance, dynamic synchronization encryption, and infrastructure decoupling. First, based on lattice cryptography and physically unclonable functions (PUCs), Chinese invention patent CN119421150B discloses a quantum-resistant UAV swarm authentication key negotiation method. This method combines PUCs with the NTRU algorithm for bidirectional authentication and uses a Latin square design to reduce authentication frequency. However, this scheme has the following drawbacks: the scalability of the node scale is limited, requiring the addition of virtual nodes to match the specific power-law requirement, and it is highly dependent on the consistency of the hardware-generated response values, which can easily lead to authentication failure under harsh operating conditions.

[0003] Second, regarding encrypted communication technology based on dynamic matrix transformation, Chinese invention patent CN117395002B discloses a UAV swarm communication method based on a dynamic encryption matrix strategy. By establishing a time node function matrix with a unified time origin, the encryption and decryption matrix can be dynamically transformed. The drawback of this scheme is that the communication process has a rigid requirement for global time synchronization accuracy, the accumulation of timing errors will lead to encryption and decryption logic mismatch, and the real-time calculation of the inverse matrix consumes a lot of processor resources, making it difficult to deploy on micro UAV nodes with limited computing power.

[0004] Third, regarding communication encryption technology based on quantum random numbers, Chinese invention patent application CN118215022A discloses a method for encrypting communication in UAV swarms based on quantum communication. This method switches communication modes according to the distance between the UAV and the base station and environmental information, and uses a quantum random number generator to generate session keys. The drawbacks of this scheme are: it is highly dependent on external infrastructure such as quantum communication base stations, and cannot be applied in remote scenarios without base station coverage. Furthermore, when determining the order of communication nodes, it is necessary to process multiple weighted evaluation coefficients, including distance, remaining capacity, and interference values. In cluster scenarios with high-frequency topology changes, there is a time delay feedback lag, making it difficult to maintain the stability and integrity of data transmission.

[0005] Therefore, how to construct a quantum-resistant communication architecture that is independent of infrastructure, adaptable to any node size, and has low computing power overhead is the technical problem to be solved by this invention. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and solve the technical problems of large post-quantum ciphertext size leading to high bandwidth pressure and severe broadcast storms in wireless ad hoc networks, and to provide a UAV key distribution method based on post-quantum cryptography.

[0007] To achieve the above-mentioned objectives, this invention provides a UAV key distribution method based on post-quantum cryptography, comprising the following steps: Step S1, stripping to obtain the first algebraic matrix and the second algebraic vector: the master node obtains the quantum key encapsulated ciphertext stream, parses the memory structure of the quantum key encapsulated ciphertext stream to strip to obtain the first algebraic matrix and the second algebraic vector corresponding to each member node in the cluster. Step S2, constructing the spatial topology state matrix and the channel state awareness matrix: The master control node collects the signal-to-noise ratio parameters fed back by each wireless link in the cluster, constructs the spatial topology state matrix reflecting the physical relative position of the cluster, and extracts the channel state awareness matrix characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands. Step S3, Configure omnidirectional low-frequency multicast and multiple high-frequency directional channel queues: Based on the spatial topology state matrix and the channel state awareness matrix, the master control node configures a transmission channel queue consisting of omnidirectional low-frequency multicast channels and multiple high-frequency directional channels. Step S4: Establish asymmetric mapping logic and map it to the transmission channel for distribution: The master control node establishes asymmetric mapping logic between the first algebraic matrix and the second algebraic vector on the physical channel, encapsulates the first algebraic matrix into a common payload message and maps it to the omnidirectional low-frequency multicast channel for wide-area delivery, and encapsulates each of the second algebraic vectors into independent payload messages and maps them to the multiple high-frequency directional channels for relay distribution according to the routing path determined by the spatial topology state matrix. Step S5, splicing and restoring the first algebraic matrix and the second algebraic vector: The member node listens to the omnidirectional low-frequency multicast channel and the multiple high-frequency directional channel at the physical layer, extracts the common payload message and the independent payload message corresponding to the member node, splices and restores the first algebraic matrix and the second algebraic vector in the memory address space, and uses the locally pre-stored private key to decapsulate and restore the global session key. Step S6, Broadcast Matrix Missing Beacon and Receive Forward to Recover Key: When the member node detects that the received common payload message is missing, it broadcasts the matrix missing beacon to neighboring nodes in the neighborhood and receives the first algebra matrix forwarded from the local cache by the neighboring node to complete the recovery of the global session key.

[0008] In step S1 of this invention, the process of obtaining the first algebraic matrix and the second algebraic vector includes: step S11, identifying the lattice cryptographic mathematical model corresponding to the post-quantum key encapsulation algorithm, and locating the high-dimensional coefficient matrix data in the post-quantum key encapsulation ciphertext stream; step S12, based on linear error correction coding logic, extracting the common basis polynomial data shared by the cluster from the high-dimensional coefficient matrix data as the first algebraic matrix; and step S13, removing the deviation component from the high-dimensional coefficient matrix data relative to the common basis polynomial data through matrix elimination operation, and using the deviation component as the independent second algebraic vector for each member node.

[0009] In step S2 of the present invention, constructing the spatial topology state matrix includes: step S21, obtaining the real-time three-dimensional coordinates and relative velocity vectors of each UAV node, and calibrating the connectivity weight of the cluster link through the one-way communication delay between the member nodes; step S22, constructing an adjacency matrix reflecting the adjacency relationship between each UAV node based on the connectivity weight, and performing dimensionality reduction processing on the adjacency matrix to generate the spatial topology state matrix.

[0010] In step S4 of the present invention, relaying the independent payload message includes: step S41, determining the routing depth of each member node in the wireless ad hoc network based on the spatial topology state matrix; step S42, encapsulating the second algebra vector corresponding to each member node into an independent short payload data packet through the multiple high-frequency directional channels in order from shallow to deep routing depth, and relaying it step by step.

[0011] In step S6 of the present invention, broadcasting a matrix missing beacon to neighboring nodes includes: step S61, the member node verifies the parity bit of the received first algebraic matrix to determine whether there is any missing data bit; step S62, when it is determined that there is missing data, the member node broadcasts a repair request instruction; step S63, the neighboring node that receives the repair request instruction pushes the missing fragment of the first algebraic matrix to the member node according to its own cache state.

[0012] In step S2 of the present invention, obtaining the channel state perception matrix includes: step S23, real-time monitoring of the signal-to-noise ratio and Doppler frequency shift parameters of wireless signals in different frequency bands; step S24, calculating the path loss gradient of different frequency bands within the coverage area of ​​the UAV cluster, and constructing the channel state perception matrix that reflects the channel penetration capability.

[0013] In step S4 of the present invention, before mapping the second algebraic vector to the multiple high-frequency directional channel, the method further includes: performing Huffman compression on the second algebraic vector to remove redundant bits in the second algebraic vector, so as to reduce the resource occupancy rate of the multiple high-frequency directional channel.

[0014] In step S4 of the present invention, delivering the first algebraic matrix includes: the master control node dynamically adjusts the transmission power of the omnidirectional low-frequency multicast channel according to the channel state perception matrix, so as to maintain the signal coverage radius of the first algebraic matrix in a state that can cover the edge nodes in the spatial topology state matrix.

[0015] In step S6 of the present invention, after the matrix repair is completed, it further includes: step S64, when the number of neighborhood collaboration recovery attempts exceeds the preset upper limit and the first algebra matrix is ​​still incomplete, the member node sends a retransmission request to the master node, requesting a single full ciphertext retransmission to the member node.

[0016] This invention also provides a UAV key distribution system based on post-quantum cryptography, which is used to implement a UAV key distribution method based on post-quantum cryptography, including: The master control node module includes: a parsing and stripping unit, used to obtain the post-quantum key encapsulated ciphertext stream, parse the memory structure of the post-quantum key encapsulated ciphertext stream to strip out a first algebraic matrix and second algebraic vectors corresponding to each member node module in the cluster; a topology sensing unit, used to collect the signal-to-noise ratio parameters fed back by each wireless link in the cluster, and construct a spatial topology state matrix reflecting the physical relative position of the cluster; a channel scheduling unit, used to extract the channel state sensing matrix characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands, and configure a transmission channel queue composed of an omnidirectional low-frequency multicast channel and multiple high-frequency directional channels based on the spatial topology state matrix and the channel state sensing matrix; and a splitting and delivery unit, used to establish the asymmetric mapping logic between the first algebraic matrix and the second algebraic vector on the physical channel, encapsulate the first algebraic matrix into a common payload message and map it to the omnidirectional low-frequency multicast channel for wide-area delivery, and encapsulate each of the second algebraic vectors into independent payload messages and map them to the multiple high-frequency directional channels for relay distribution. The member node module includes: a cooperative receiving unit, used to listen to the omnidirectional low-frequency multicast channel and the multiple high-frequency directional channels at the physical layer, extract the common payload message and the independent payload message corresponding to the member node module, and reconstruct the first algebraic matrix and the second algebraic vector in the memory address space, and decapsulate using the locally pre-stored private key to recover the global session key; and a cooperative recovery unit, used to broadcast a matrix missing beacon to neighboring node modules in the neighborhood when the received common payload message is detected to be missing, and to receive the first algebraic matrix forwarded from the local cache by the neighboring node module to complete the recovery of the global session key.

[0017] Compared with the prior art, the present invention has at least the following beneficial effects: 1. In UAV key distribution, by analyzing the asymmetric construction features of the post-quantum cryptographic ciphertext at the algebraic level, the ciphertext payload is decoupled into a common algebraic matrix shared across nodes and a specific residual matrix for each member node. This breaks the encapsulation inertia of treating large-volume ciphertext as an atomized black box payload in traditional digital communication protocols. This mechanism allows the common part, which occupies the main volume of the ciphertext payload, to only need to perform a single wide-area transmission, while the specific part, which is independently carried by each member node, maintains a small data size. This eliminates the linear expansion of the total amount of cluster key negotiation data as the number of nodes increases, and effectively prevents the generation of broadcast storms in restricted wireless channels.

[0018] 2. The method establishes a mapping relationship between the volume distribution of cryptographic algebraic matrices and the penetration capability of physical channels. Utilizing the node spatial topology state matrix that reflects the physical connectivity of the cluster, the common algebraic matrix is ​​delivered to an omnidirectional low-frequency multicast channel with high penetration characteristics. Simultaneously, the specific residual matrix is ​​mapped to a high-bandwidth high-frequency directional channel for routing distribution. This cross-level asymmetric channel mapping logic leverages the differentiated advantages of physical links to carry algebraic features of different dimensions, enabling high-dimensional, large-capacity post-quantum cryptographic data to adapt to highly mobile, strongly fading wireless ad hoc network environments, thereby improving the protocol convergence speed and transmission reliability of digital security information under harsh topology conditions.

[0019] 3. The method constructs an asynchronous distributed matrix repair path and uses the shared algebraic matrix cached by neighboring nodes to provide local compensation transmission for edge nodes, changing the inherent mode of traditional retransmission protocols that absolutely rely on the source node for full retransmission. When the shared matrix is ​​lost due to Doppler shift or transient interference in a local link, member nodes obtain the missing part by triggering point-to-point collaborative repair logic within the neighborhood. This technical coupling based on the high-density topology within the cluster avoids large-volume global retransmission requests for encrypted data caused by the loss of a single data slice, effectively blocking the path of communication nodes falling into continuous retransmission deadlock and ensuring deterministic latency in the safe state synchronization process. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the post-quantum key distribution process of the ciphertext algebraic decoupling and asymmetric mapping of the present invention; Figure 2 This is a schematic diagram of the system architecture for the heterogeneous channel collaboration and distributed matrix collaborative repair function of this invention. Detailed Implementation

[0021] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings.

[0022] Example 1: This embodiment discloses a method for unmanned aerial vehicle (UAV) key distribution based on post-quantum cryptography, including the following steps: Step S1, stripping to obtain the first algebraic matrix and the second algebraic vector: the master node obtains the quantum key encapsulated ciphertext stream, parses the memory structure of the quantum key encapsulated ciphertext stream to strip to obtain the first algebraic matrix and the second algebraic vector corresponding to each member node in the cluster. Step S2, constructing the spatial topology state matrix and the channel state awareness matrix: The master control node collects the signal-to-noise ratio parameters fed back by each wireless link in the cluster, constructs the spatial topology state matrix reflecting the physical relative position of the cluster, and extracts the channel state awareness matrix characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands. Step S3, Configure omnidirectional low-frequency multicast and multiple high-frequency directional channel queues: Based on the spatial topology state matrix and the channel state awareness matrix, the master control node configures a transmission channel queue consisting of omnidirectional low-frequency multicast channels and multiple high-frequency directional channels. Step S4: Establish asymmetric mapping logic and map it to the transmission channel for distribution: The master control node establishes asymmetric mapping logic between the first algebraic matrix and the second algebraic vector on the physical channel, encapsulates the first algebraic matrix into a common payload message and maps it to the omnidirectional low-frequency multicast channel for wide-area delivery, and encapsulates each of the second algebraic vectors into independent payload messages and maps them to the multiple high-frequency directional channels for relay distribution according to the routing path determined by the spatial topology state matrix. Step S5, splicing and restoring the first algebraic matrix and the second algebraic vector: The member node listens to the omnidirectional low-frequency multicast channel and the multiple high-frequency directional channel at the physical layer, extracts the common payload message and the independent payload message corresponding to the member node, splices and restores the first algebraic matrix and the second algebraic vector in the memory address space, and uses the locally pre-stored private key to decapsulate and restore the global session key. Step S6, Broadcast Matrix Missing Beacon and Receive Forward to Recover Key: When the member node detects that the received common payload message is missing, it broadcasts the matrix missing beacon to neighboring nodes in the neighborhood and receives the first algebra matrix forwarded from the local cache by the neighboring node to complete the recovery of the global session key.

[0023] In step S1 of this embodiment, the process of extracting the first algebraic matrix and the second algebraic vector includes: step S11, identifying the lattice cryptographic mathematical model corresponding to the post-quantum key encapsulation algorithm, and locating the high-dimensional coefficient matrix data in the post-quantum key encapsulation ciphertext stream; step S12, based on linear error correction coding logic, extracting the common basis polynomial data shared by the cluster from the high-dimensional coefficient matrix data as the first algebraic matrix; and step S13, removing the deviation component from the high-dimensional coefficient matrix data relative to the common basis polynomial data through matrix elimination operation, and using the deviation component as the independent second algebraic vector for each member node.

[0024] In step S2 of this embodiment, constructing the spatial topology state matrix includes: step S21, obtaining the real-time three-dimensional coordinates and relative velocity vectors of each UAV node, and calibrating the connectivity weight of the cluster link through the one-way communication delay between the member nodes; step S22, constructing an adjacency matrix reflecting the adjacency relationship between each UAV node based on the connectivity weight, and performing dimensionality reduction processing on the adjacency matrix to generate the spatial topology state matrix.

[0025] In step S4 of this embodiment, relaying the independent payload message includes: step S41, determining the routing depth of each member node in the wireless ad hoc network based on the spatial topology state matrix; step S42, encapsulating the second algebra vector corresponding to each member node into an independent short payload data packet through the multiple high-frequency directional channels in order from shallow to deep routing depth, and relaying it step by step.

[0026] In step S6 of this embodiment, broadcasting a matrix missing beacon to neighboring nodes includes: step S61, where the member node verifies the parity bit of the received first algebraic matrix to determine whether any data bits are missing; step S62, where, if data is missing, the member node broadcasts a repair request instruction; and step S63, where, based on its own cache state, the neighboring node that receives the repair request instruction pushes the missing fragment of the first algebraic matrix to the member node.

[0027] In step S2 of this embodiment, obtaining the channel state awareness matrix includes: step S23, real-time monitoring of the signal-to-noise ratio and Doppler frequency shift parameters of wireless signals in different frequency bands; step S24, calculating the path loss gradient of different frequency bands within the coverage area of ​​the UAV cluster, and constructing the channel state awareness matrix that reflects the channel penetration capability.

[0028] In step S4 of this embodiment, before mapping the second algebraic vector to the multiple high-frequency directional channel, the method further includes: performing Huffman compression on the second algebraic vector to remove redundant bits in the second algebraic vector, so as to reduce the resource occupancy rate of the multiple high-frequency directional channel.

[0029] In step S4 of this embodiment, delivering the first algebraic matrix includes: the master control node dynamically adjusts the transmission power of the omnidirectional low-frequency multicast channel according to the channel state perception matrix, so as to maintain the signal coverage radius of the first algebraic matrix in a state that can cover the edge nodes in the spatial topology state matrix.

[0030] In step S6 of this embodiment, after the matrix repair is completed, it also includes: step S64, when the number of neighborhood collaboration recovery attempts exceeds the preset limit and the first algebra matrix is ​​still incomplete, the member node sends a retransmission request to the master node, requesting a single full ciphertext retransmission for the member node.

[0031] This embodiment also provides a drone key distribution system based on post-quantum cryptography, including: The master control node module includes: a parsing and stripping unit, used to obtain the post-quantum key encapsulated ciphertext stream, parse the memory structure of the post-quantum key encapsulated ciphertext stream to strip out a first algebraic matrix and second algebraic vectors corresponding to each member node module in the cluster; a topology sensing unit, used to collect the signal-to-noise ratio parameters fed back by each wireless link in the cluster, and construct a spatial topology state matrix reflecting the physical relative position of the cluster; a channel scheduling unit, used to extract the channel state sensing matrix characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands, and configure a transmission channel queue composed of an omnidirectional low-frequency multicast channel and multiple high-frequency directional channels based on the spatial topology state matrix and the channel state sensing matrix; and a splitting and delivery unit, used to establish the asymmetric mapping logic between the first algebraic matrix and the second algebraic vector on the physical channel, encapsulate the first algebraic matrix into a common payload message and map it to the omnidirectional low-frequency multicast channel for wide-area delivery, and encapsulate each of the second algebraic vectors into independent payload messages and map them to the multiple high-frequency directional channels for relay distribution. The member node module includes: a cooperative receiving unit, used to listen to the omnidirectional low-frequency multicast channel and the multiple high-frequency directional channels at the physical layer, extract the common payload message and the independent payload message corresponding to the member node module, and reconstruct the first algebraic matrix and the second algebraic vector in the memory address space, and decapsulate using the locally pre-stored private key to recover the global session key; and a cooperative recovery unit, used to broadcast a matrix missing beacon to neighboring node modules in the neighborhood when the received common payload message is detected to be missing, and to receive the first algebraic matrix forwarded from the local cache by the neighboring node module to complete the recovery of the global session key.

[0032] Example 2: In this embodiment, the application scenario involves a UAV swarm formation operating in an environment with strong electromagnetic interference and high-frequency topological movement. The public key and ciphertext generated by the post-quantum lattice cryptography algorithm are typically presented as high-dimensional polynomial matrices on the order of kilobytes. When the swarm communication nodes face load power consumption limitations and the physical constraints of narrowband fading in the wireless channel, if the ciphertext is considered as an atomized, encapsulated data packet, the loss of any data fragment within the channel will trigger a global retransmission request, thereby causing a broadcast storm and link deadlock during multi-node concurrent negotiation. The master node constructs a spatial topology state matrix reflecting the physical relative position of the swarm by collecting the signal-to-noise ratio parameters fed back from each wireless link within the swarm, and extracts a channel state awareness matrix characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands. Based on graph theory, the master node calculates the eigenvalues ​​and eigenvectors of the adjacency matrix to extract the eigenvalue diagonal matrix. The eigenvalue cumulative variance contribution rate truncation threshold is set to 0.85, and the master node filters... The M largest eigenvalues ​​and their corresponding eigenvectors that satisfy the truncation threshold are combined to form a projection matrix, where M is a positive integer. The master node performs matrix multiplication between the adjacency matrix and the projection matrix, outputting the dimensionality-reduced low-dimensional feature state as the spatial topology state matrix. A deterministic mapping relationship is established between physical nodes and the low-dimensional state space. After generating the spatial topology state matrix, the master node recovers the topological association chain reflecting the relative connectivity weights between nodes by calculating the projection intensity of the low-dimensional feature vectors in the original physical node coordinate system. The routing depth is determined based on the hierarchical adjacency logic established by this projection intensity. That is, the master node maps the dimensionality-reduced topology state back to the physical link layer of the wireless ad hoc network. By finding the node cluster with the closest Euclidean distance in the low-dimensional feature space, a spanning tree rooted at the master node is constructed. The level of each member node in the spanning tree is defined as the routing depth, thereby guiding the high-frequency directional channel in the relay distribution process of the second-generation vector. The process involves step-by-step delivery of the quantum key, followed by parsing by the master node, which encapsulates the ciphertext stream in memory and separates it to obtain the first algebraic matrix. and the second algebra vectors corresponding to each member node in the cluster. ;in This represents a polynomial matrix. Indicates the first The specific residual components of each member node, with subscripts The index parameter is an integer greater than 0; specifically, the first algebra matrix. The extraction process is based on a linear mapping performed on a pre-shared set of generator polynomials in a cluster, while the second algebra vector... The stripping process, while maintaining the homomorphic properties of lattice cryptography, utilizes error correction coding logic to align the ciphertext offsets. In matrix elimination, the master node cancels out the contribution terms of the common basis polynomial by performing a subtraction based on modulo-Q operations on the received ciphertext matrix and the local copy. The remaining residual components are identified as the unique deviation features of each member node. This process transforms the large-volume common algebraic structure into a single-broadcast common payload by performing arithmetic reconstruction on the ciphertext space. Each member node only needs to extract the tiny residual components related to its private key to complete the decapsulation. This method, through algebraic structure decoupling, splits the originally uniformly encapsulated post-quantum ciphertext payload into a common part occupying more than 70% of the total data volume and multiple extremely small specific payloads, eliminating the redundant costs caused by the isolation between the security layer data and the transport layer state.

[0033] The master node establishes a first-generation algebraic matrix based on the spatial topology state matrix and the channel state awareness matrix. With second-generation vectors Asymmetric mapping logic on the physical channel; the master node will use the first algebraic matrix The signal is delivered to the omnidirectional low-frequency multicast channel, utilizing the signal penetration capability of this frequency band to achieve wide-area coverage; simultaneously, the master control node distributes each second-generation vector... Each payload is mapped to multiple high-frequency directional channels and distributed point-to-point according to the connectivity path indicated by the spatial topology state matrix. This mapping logic ensures that the large shared matrix is ​​transmitted only once, while the directionally distributed payload retains only a very small number of specific components. This directly resolves the physical contradiction between high-dimensional ciphertext payloads and the bandwidth bottleneck of dynamic fading channels. Member nodes synchronously monitor the omnidirectional low-frequency multicast channel and multiple high-frequency directional channels at the physical layer to extract the first algebraic matrix. With dedicated second-generation vectors The ciphertext structure is then reconstructed within the storage address space to recover the global session key; when an edge node fails to obtain the first algebraic matrix due to sudden interference... When this happens, the node broadcasts a matrix missing beacon to its neighboring nodes; neighboring nodes that receive the repair request push the first algebra matrix to the member node based on their local cache state. The missing segments are addressed by: To prevent channel contention and backoff attenuation caused by concurrent responses from multiple nodes during distributed collaboration, member nodes activate built-in hardware timers, setting a single retransmission response time window tolerance boundary of 15ms; neighboring nodes parse the received signal strength indication data carried in the missing beacon and apply a free-space path loss model to invert the spatial distance; when the calculated spatial distance is lower than the system's set single-hop nearest neighbor, and a physical threshold is determined, neighboring nodes occupy directional channels to push missing segments, blocking blind responses from distant nodes through spatial distance filtering. This mechanism utilizes the high-density topology characteristics within the cluster to eliminate the system's dependence on the master node's retransmission link, ensuring the stability of key distribution under extreme topology fragmentation conditions.

[0034] Example 3: This embodiment verifies the effectiveness of the post-quantum ciphertext algebraic structure decoupling mechanism in suppressing link resource occupancy in a simulated UAV swarm cooperative operation under strong multipath fading and narrowband physical constraints through comparative analysis. The experimental platform adopts a distributed wireless ad hoc network simulation environment based on a discrete event simulation architecture. This environment integrates a physical layer channel model conforming to Rayleigh distribution characteristics to simulate the propagation law of signals under large-scale path loss and small-scale Doppler frequency shift interference. The experimental data source is obtained by acquiring the raw bitstream signal of the physical layer. The sampling frequency of the data acquisition unit is set to 100MHz, and its sampling accuracy is sufficient to support the capture of subtle jitter characteristics of the signal envelope. The key distribution sampling period is... The configuration needs to strike a balance between the real-time nature of cluster topology updates and the signaling overhead of the wireless channel; when the node movement speed indicated by the spatial topology state matrix... When the speed is above 15 m / s, in order to reduce the probability of routing failure caused by channel feedback lag, the sampling period is... The value tends towards the lower limit of its range, i.e., 10ms; the control group uses the traditional transmission method of encapsulating the complete lattice ciphertext into a single atomic message, while the experimental group uses a first-generation algebra matrix. With second-generation vectors The decoupling distribution method; where the first algebraic matrix For a shared polynomial matrix occupying more than 70% of the total ciphertext volume, a second-generation vector... For the corresponding number The specific residual components of each member node, with subscripts The index parameter is an integer greater than 0.

[0035] With 10 nodes and a signal-to-noise ratio Under the standard operating condition of 10dB, due to a broadcast storm in the channel caused by the control group performing full ciphertext retransmission, the consensus convergence time for the global session key was 156.4ms, at which point the link resource utilization rate was 85.2%. The experimental group achieved consensus by using the first algebraic matrix. Mapping to an omnidirectional low-frequency multicast channel ensures that the shared payload incurs only a single transmission overhead within the cluster, corresponding to the second algebra vector of each member node. The data was delivered synchronously in multiple high-frequency directional channels. The experimental data showed that the convergence time of the experimental group was reduced to 42.6ms, and the link resource utilization rate was maintained at 22.3%, which proved that the decoupling mapping mechanism reduced the signaling interaction cost under heterogeneous channels.

[0036] To verify the stability gradient of the scheme, the experiment gradually reduced the signal power sensed by the channel state sensing matrix. When the signal-to-noise ratio... When the bandwidth decreased from 20dB to 5dB, the retransmission requests in the control group increased exponentially and eventually led to a communication deadlock; the experimental group used neighboring nodes to access the first algebraic matrix. The local cache relay repair mechanism for missing segments, in terms of signal-to-noise ratio Even under extreme interference of 5dB, the key negotiation success rate is maintained at 98.4%. When the cluster size exceeds 50 nodes, due to the intensified competition and conflict in multiple high-frequency directional channels, the growth rate of convergence time changes from linear to nonlinear. This indicates that by dynamically adjusting the transmit power and time slot allocation through the channel state awareness matrix, the security protocol under the H04L architecture can maintain stable convergence performance in high-density topology environments.

[0037] Example 4: This embodiment combines Figures 1 to 2 This section describes a method and system for unmanned aerial vehicle (UAV) key distribution based on post-quantum cryptography, such as... Figure 1 As shown, the key distribution process includes sequential logical actions: Step S1, stripping to obtain the first algebraic matrix and the second algebraic vector; Step S2, constructing the spatial topology state matrix and the channel state awareness matrix; Step S3, configuring omnidirectional low-frequency multicast and multiple high-frequency directional channel queues; Step S4, establishing asymmetric mapping logic and mapping it to the transmission channel distribution; Step S5, splicing and restoring the first algebraic matrix and the second algebraic vector; Step S6, broadcasting the missing beacon in the matrix and receiving and forwarding to recover the key.

[0038] like Figure 2 As shown, the system architecture includes a master control node module, multiple high-frequency directional channels, an omnidirectional low-frequency multicast channel, and multiple member nodes. The master control node module is internally divided into a parsing stripping unit for generating the first algebraic matrix. The module includes a splitting and delivery unit for asymmetric mapping logic control, a topology sensing unit, and a channel scheduling unit. The topology sensing unit collects signal-to-noise ratio parameters from each wireless link within the cluster to construct a spatial topology state matrix reflecting the physical relative position of the cluster. The channel scheduling unit extracts channel state sensing matrices characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands and configures a transmission channel queue consisting of omnidirectional low-frequency multicast channels and multiple high-frequency directional channels based on the spatial topology state matrix and the channel state sensing matrix. The master control node module injects signals into the multiple high-frequency directional channels for relay distribution of second-generation vectors. and omnidirectional low-frequency multicast channels for wide-area delivery of first-generation algebraic matrices. The common payload message output via the omnidirectional low-frequency multicast channel is delivered to member node C, which is in the process of cooperative recovery. The signal output via the multiple high-frequency directional channels is directly delivered to member node B, which is the relay distribution target. At the same time, member node A, which is in the process of cooperative reception restoration, also relays the signal to member node B along the routing path. In this interaction, member node C, which is in the process of cooperative recovery, sends a matrix missing beacon broadcast to member node B, which is the relay distribution target. Member node B, which is the relay distribution target, forwards the broadcast locally to member node C, which is in the process of cooperative recovery. Finally, the data flows of member node A, which is in the process of cooperative reception restoration, member node B, which is the relay distribution target, and member node C, which is in the process of cooperative recovery, are all unified and converged to restore the global session key.

[0039] Example 5: In this embodiment, under the conditions of high-maneuverability topological changes and onboard processor memory addressing conflicts in the UAV swarm system, a bit-level extraction procedure for the post-quantum ciphertext stream is established for the master control node. The master control node locates the starting address pointer of the post-quantum key-encapsulated ciphertext stream in the static random access memory through the memory control unit. Before bit extraction, the master node reads the memory allocation table of the memory control unit and compares it with the starting address pointer. The system identifies the continuity status of blocks within a preset maximum ciphertext length range. If a non-continuous fragmentation is detected, the master node calls the direct memory access controller to rearrange memory blocks and reconstruct a continuous physical address space. The master node retrieves a preset set of post-qubit lattice cryptographic parameters to identify the quantization bit depth and array dimension of the polynomial coefficients in the ciphertext stream. The first continuous bit stream of length LA corresponds to a public key generation matrix constructed based on a fault-tolerant learning problem. The offset points to the correlation deviation vector independently generated by each member node. In the actual stripping operation, the master node follows the modulo arithmetic rules defined by the lattice cryptography algorithm to... The contiguous memory space is mapped to a coefficient array of a high-dimensional algebraic object. Weighted addressing with a logical offset step is performed on the starting memory address to achieve one-to-one attribute extraction from the physical layer bit sequence to the algebraic layer matrix features. Based on the modular fault-tolerant learning algorithm's data structure distribution principle, the ciphertext stream is constructed by concatenating a cluster-wide shared public key-generated matrix data block with each node's specific polynomial vector data block during the generation phase. The shared polynomial matrix corresponds to the globally shared public key-generated matrix data block in physical storage. The master node calculates the bit length of the shared polynomial matrix in the ciphertext data structure based on a preset lattice cryptographic security parameter set. and the offset step size of the residual components of each member node. ;in The value is determined by the product of the polynomial dimension and the quantization bits of the coefficients. Indicates the first The data start bit of each member node is relative to Logical spacing, subscript The index parameter is an integer greater than 0; the master node uses the addressing formula. Read length is A continuous bit stream and convert it into a first algebraic matrix. Based on this, the master node accumulates the offset. Extract the specific payload of each member node one by one and convert it into a second algebra vector. This precise addressing method based on bit offsets transforms the decoupling action of algebraic structures into deterministic operations on the storage address space, enabling the underlying pipeline to complete data splitting within nanosecond latency.

[0040] Regarding the threshold of node motion velocity in the spatial topological state matrix The physical calibration logic is set at 15 m / s, based on the constraint relationship between the wireless channel coherence time and the key negotiation period. Since the center frequency used in UAV swarm communication is 5.8 GHz, the channel coherence time caused by the Doppler effect shortens as the relative speed of the nodes increases. To ensure that the channel state maintains quasi-static characteristics during the key distribution of a single session, the system pre-monitors the signal gain fluctuation variance under different speed gradients. Experimental data shows that when the movement speed... When the speed reaches 15 m / s, the channel correlation coefficient drops to the critical point of 0.5, causing a nonlinear increase in the link convergence time of the traditional full retransmission scheme. Based on this calibration, the master control node obtains the instantaneous frequency shift parameters fed back from the physical layer in real time and updates the spatial topology state matrix. If the current speed is detected... If the speed exceeds the calibrated threshold of 15 m / s, the master control node will forcibly issue a control command to drive the key distribution sampling period. The process converges to a physical lower limit of 10ms. This process ensures that the transmission protocol under the H04L architecture has deterministic response accuracy under extreme dynamic conditions by quantifying and compensating for the execution deviation caused by physical movement.

[0041] In the first algebraic matrix With second-generation vectors When mapping to the transmission channel queue, the master node uses the multiple access controller to establish a priority evaluation model based on the ratio of message length to channel capacity; the master node calculates the estimated duration of the first data message in the omnidirectional low-frequency multicast channel. And the estimated duration of each second data packet in multiple high-frequency directional channels. Because the first data message carries a shared algebraic matrix that occupies more than 70% of the total ciphertext volume; The master control node assigns the highest task weight to it and places it at the top of the transmit buffer; the master control node configures the power amplifier status of the physical layer transmitter and completes the process on the omnidirectional low-frequency multicast channel. After a single blind transmission, multiple concurrent pulses from high-frequency directional channels are triggered based on the shortest connected path to deliver the corresponding second-generation vector. This explicit scheduling method for transmission timing can increase the utilization rate of the cluster system's link resources to over 75% without increasing hardware overhead, ultimately achieving a highly efficient closed loop for security state synchronization at the physical layer.

[0042] Example 6: In this embodiment, under the condition of heterogeneous UAV hardware collaborative deployment and differences in onboard processor storage architecture, the master control node obtains the page size parameter of the current storage unit. and byte alignment factor To determine the physical baseline of the key distribution process under different instruction set architectures, and to combine the dimension of lattice cryptography multinomials. With the number of quantization bits of the coefficients To calculate the bit length of the common polynomial matrix in the address space ,in The corresponding value is and The product is automatically rounded up by the system. The master node sends a pointer to each member node of the cluster containing the starting address, which is an integer multiple of the specified address. AND bit length The synchronous instruction frame drives each member node to align the data extraction pointers according to the local hardware addressing logic, so that the decoupling action of the algebraic structure maintains logical consistency in the heterogeneous hardware collaborative environment.

[0043] When the system faces an initial deployment environment with superimposed electromagnetic background noise and multipath interference, the master control node performs physical layer benchmark calibration before key distribution to determine the null reference of the channel state awareness matrix. It then controls the radio frequency transmitting unit to deliver a detection sequence with a preset power in the omnidirectional low-frequency multicast channel. The receiving end's detection unit calculates the fluctuation variance of the sequence through a sliding window sampling process of 1024 consecutive cycles within the 5.8 GHz band, thereby obtaining the background interference power of the current environment. The master control node, based on the received signal strength indication, and The ratio determines the initial signal-to-noise ratio of the link. This is used as the input operator for path loss compensation in the spatial topology state matrix, thus improving the key distribution sampling period during subsequent distribution processes. The convergence determination logic is based on a physical baseline.

[0044] Example 7: In this embodiment, during the pre-system calibration before the initial key distribution in the UAV swarm system, the system initiates a spatial mapping calibration process. Each member node sequentially sends an initial synchronization frame containing GPS coordinates to the master node. The master node then calculates the first synchronization frame based on the received coordinate data. The member node and the first Spatial distance between member nodes And store it in the memory buffer; the master node retrieves the normalization operator from the memory. And calculate the topology strength coefficient. , The calculation formula is: ,in, For topological strength coefficient, It is a proportionality constant. The distance between nodes. and For integer index parameters greater than 0, the master node will calculate the... The corresponding elements of the spatial topology state matrix are filled in, which provides a physical basis for the subsequent key distribution routing logic to converge when faced with coordinate sampling deviations.

[0045] After the system enters the electromagnetic environment adaptation phase, the master control node determines the proportional relationship of different parameters within the channel state sensing matrix through a sensing weight calibration process. The system monitors the signal-to-noise ratio gradient from 5dB to 30dB and the Doppler frequency shift gradient from 0Hz to 500Hz under a controlled simulation environment, collects the bit error rate distribution characteristics under different perturbation combinations, and fits the error response curve. Based on this, the master control node determines the first proportional constant characterizing the signal-to-noise ratio weight. and the second proportional constant characterizing the frequency shift weight ,in and The algebraic sum of these parameters is always equal to 1. The master node obtains real-time signal-to-noise ratio (SNR) and frequency shift parameters through the physical layer sensing unit. Based on the large-scale physical fading and dynamic frequency offset normalization mapping procedure, it calculates the difference between the real-time SNR parameter and the system's preset baseline SNR value, divides it by the upper limit of the equipment's factory-calibrated SNR range, and outputs a dimensionless SNR characteristic scalar. Simultaneously, it calculates the difference between the real-time frequency shift parameter and the static zero frequency shift value, divides it by the maximum tolerance parameter of the Doppler frequency shift, and outputs a dimensionless frequency offset characteristic scalar. The master node updates the parameter entity with the dimensionless scalar generated above and sets the real-time SNR parameter and the first proportional constant. Multiply by the real-time frequency shift parameter with the second proportional constant. Multiply the results, and finally sum the products to generate the evaluation value in the channel state awareness matrix, thus making the key distribution sampling period... The adjustment logic maintains steady state under signal fading conditions. To eliminate the fundamental difference in dimensions between the signal-to-noise ratio (SNR) and frequency shift parameters, the master control node introduces a bridging transformation process based on the coherence time decay rate through the physical layer sensing unit before summation. The system performs feature fusion between the signal amplitude probability distribution reflected by the real-time SNR and the phase jitter frequency reflected by the real-time frequency shift, utilizing the first proportional constant. Second proportional constant The contribution weights of amplitude attenuation and phase deviation to the physical layer bit error rate are respectively characterized. In this way, the originally independent physical quantities are transformed into a single dimensionless comprehensive index characterizing the channel penetration capability. This index directly reflects the effective bandwidth utilization of the channel under the current multipath interference, thus providing a unified physical criterion for the dynamic adjustment of the key distribution sampling period.

Claims

1. A method for unmanned aerial vehicle (UAV) key distribution based on post-quantum cryptography, characterized in that, Includes the following steps: Step S1: The master node acquires the post-quantum key encapsulated ciphertext stream and parses its memory structure to extract the first algebraic matrix and the second algebraic vectors corresponding to each member node in the cluster. This includes: Step S11: Identifying the lattice cryptographic mathematical model corresponding to the post-quantum key encapsulation algorithm and locating the high-dimensional coefficient matrix data in the post-quantum key encapsulated ciphertext stream; Step S12: Based on linear error correction coding logic, extracting the common basis polynomial data shared by the cluster from the high-dimensional coefficient matrix data as the first algebraic matrix; Step S13: Extracting the deviation components relative to the common basis polynomial data from the high-dimensional coefficient matrix data through matrix elimination operations, and using the deviation components as independent second algebraic vectors for each member node. Step S2: The master control node constructs a spatial topology state matrix reflecting the physical relative positions of the cluster and extracts a channel state awareness matrix characterizing the loss characteristics and bandwidth capacity of wireless channels in different frequency bands. This includes: Step S21: Obtaining the real-time three-dimensional coordinates and relative velocity vectors of each UAV node, and calibrating the connectivity weights of the cluster links through the one-way communication delay between member nodes; Step S22: Constructing an adjacency matrix reflecting the adjacency relationships between each UAV node based on the connectivity weights, and performing dimensionality reduction processing on the adjacency matrix to generate a spatial topology state matrix; Step S23: Real-time monitoring of the signal-to-noise ratio and Doppler frequency shift parameters of wireless signals in different frequency bands; Step S24: Calculating the path loss gradient of different frequency bands within the coverage area of ​​the UAV cluster, and constructing a channel state awareness matrix reflecting the channel penetration capability. Step S3: Based on the spatial topology state matrix and the channel state awareness matrix, the master control node configures a transmission channel queue consisting of an omnidirectional low-frequency multicast channel and multiple high-frequency directional channels. Step S4: The master control node establishes an asymmetric mapping logic between the first algebraic matrix and the second algebraic vector on the physical channel. The first algebraic matrix is ​​encapsulated into a common payload message and mapped to an omnidirectional low-frequency multicast channel for wide-area delivery. At the same time, each second algebraic vector is encapsulated into an independent payload message and mapped to multiple high-frequency directional channels for relay distribution according to the routing path determined by the spatial topology state matrix. Step S5: Member nodes listen to the omnidirectional low-frequency multicast channel and multiple high-frequency directional channels at the physical layer, extract the common payload message and the independent payload message corresponding to the member node, and reconstruct the first algebraic matrix and the second algebraic vector in the memory address space. They then use the locally pre-stored private key to decapsulate and restore the global session key. In step S6, when a member node detects that a received common payload message is missing, it broadcasts a matrix missing beacon to neighboring nodes in the neighborhood and receives the first algebra matrix forwarded from the local cache by the neighboring nodes to complete the recovery of the global session key.

2. The method for unmanned aerial vehicle key distribution based on post-quantum cryptography according to claim 1, characterized in that, In step S4, the relay distribution of independent payload messages includes: step S41, determining the routing depth of each member node in the wireless ad hoc network based on the spatial topology state matrix; step S42, encapsulating the second-generation vectors corresponding to each member node into independent short payload data packets through multiple high-frequency directional channels in order from shallow to deep routing depth, and relaying them level by level.

3. The method for unmanned aerial vehicle key distribution based on post-quantum cryptography according to claim 1, characterized in that, In step S6, broadcasting the matrix missing beacon to neighboring nodes includes: step S61, member nodes verify the parity bit of the received first algebra matrix to determine whether there is any missing data bit; step S62, when it is determined that there is missing data, member nodes broadcast a repair request instruction; step S63, neighboring nodes that receive the repair request instruction push the missing fragment of the first algebra matrix to member nodes according to their own cache status.

4. The UAV key distribution method based on post-quantum cryptography according to claim 1, characterized in that, In step S4, delivering the first algebraic matrix includes: the master control node dynamically adjusts the transmit power of the omnidirectional low-frequency multicast channel according to the channel state perception matrix, so as to maintain the signal coverage radius of the first algebraic matrix in a state that can cover the edge nodes in the spatial topology state matrix.

5. The method for unmanned aerial vehicle key distribution based on post-quantum cryptography according to claim 1, characterized in that, In step S6, after the matrix repair is completed, it also includes: step S64, when the number of neighborhood collaboration recovery attempts exceeds the preset limit and the first algebra matrix is ​​still incomplete, the member node sends a retransmission request to the master node, requesting a single full ciphertext retransmission to the member node.

6. A UAV key distribution system based on post-quantum cryptography, used to implement the UAV key distribution method based on post-quantum cryptography as described in claim 1, characterized in that, include: The master control node module includes: a parsing and stripping unit, used to acquire the post-quantum key encapsulated ciphertext stream, identify the lattice cryptographic mathematical model corresponding to the post-quantum key encapsulation algorithm, and locate the high-dimensional coefficient matrix data in the post-quantum key encapsulated ciphertext stream; based on linear error correction coding logic, extracting the common basis polynomial data shared by the cluster from the high-dimensional coefficient matrix data as the first algebra matrix; stripping the deviation components relative to the common basis polynomial data in the high-dimensional coefficient matrix data through matrix elimination operations, and using the deviation components as independent second algebra vectors for each member node; and a topology sensing unit, used to acquire the real-time three-dimensional coordinates and relative velocity vectors of each UAV node, calibrating the connectivity weight of the cluster link through the one-way communication delay between member nodes; and constructing an adjacency matrix reflecting the adjacency relationship between each UAV node based on the connectivity weight. The system performs dimensionality reduction on the adjacency matrix to generate a spatial topology state matrix; a channel scheduling unit is used to monitor the signal-to-noise ratio and Doppler frequency shift parameters of wireless signals in different frequency bands in real time; calculates the path loss gradient of different frequency bands within the coverage area of ​​the UAV swarm, and constructs a channel state awareness matrix that reflects the channel penetration capability; based on the spatial topology state matrix and the channel state awareness matrix, it configures a transmission channel queue consisting of an omnidirectional low-frequency multicast channel and multiple high-frequency directional channels; a split delivery unit is used to establish an asymmetric mapping logic between the first algebraic matrix and the second algebraic vector on the physical channel, encapsulates the first algebraic matrix into a common payload message and maps it to the omnidirectional low-frequency multicast channel for wide-area delivery, and encapsulates each second algebraic vector into an independent payload message and maps it to the multiple high-frequency directional channels for relay distribution; The member node module includes: a cooperative receiving unit, used to listen to the omnidirectional low-frequency multicast channel and multiple high-frequency directional channels at the physical layer, extract the common payload message and the independent payload message corresponding to the member node module, and reconstruct the first algebra matrix and the second algebra vector in the memory address space, and decapsulate using the locally pre-stored private key to recover the global session key; and a cooperative recovery unit, used to broadcast a matrix missing beacon to the neighboring node modules in the neighborhood when the received common payload message is detected to be missing, and receive the first algebra matrix forwarded from the local cache by the neighboring node modules to complete the recovery of the global session key.

Images