A service-aware multimode communication link adaptive selection system and method

Abstract

This application discloses a service-aware multi-mode communication link adaptive selection system and method, belonging to the field of communication technology. The method includes: acquiring command service data and performing redundant fragmentation coding; generating redundant coded fragments by dynamically adjusting the redundancy factor based on real-time link status using a fountain code algorithm; constructing a three-dimensional air-space-ground topology map based on node location and link quality evaluation matrix; predicting link status using Kalman filtering and triggering path backup in advance; responding to path backup instructions or link failures, redirecting fragments to backup links for transmission according to a load balancing strategy; and receiving fragments from multiple heterogeneous links at the receiving end, performing fountain code decoding to restore the data after reaching a decoding threshold. This application integrates heterogeneous networks through multi-source access, combined with service-aware dynamic scheduling and self-healing reconfiguration mechanisms, to achieve efficient utilization of heterogeneous link resources and seamless switching during communication interruptions, improving the continuity and reliability of command and dispatch in emergency scenarios.

Description

Technical Field

[0001] This application belongs to the field of communication technology, specifically relating to a service-aware multimode communication link adaptive selection system and method. Background Technology

[0002] With the rapid development of emergency communication technologies, building an integrated air-space-ground collaborative networking architecture has become an important means to ensure the continuity of command and dispatch in extreme environments. At the scene of natural disasters or emergencies, integrating heterogeneous resources such as satellite communication, UAV relay, and ground-based self-organizing networks can effectively compensate for communication interruptions caused by damage to traditional ground infrastructure. This multi-dimensional communication support system not only improves the coverage of on-site command but also provides a physical foundation for data transmission in complex environments.

[0003] Among these, service-aware multi-mode communication link adaptive selection is the core of achieving efficient collaboration in heterogeneous networks. Its goal is to ensure stable transmission of command service data through dynamic scheduling of link resources at each layer. In practical application scenarios, the system needs to monitor the link quality of space-based, air-based, and ground-based systems in real time and optimize resource allocation according to service priorities. This places extremely high demands on the system's topology management capabilities, data encoding efficiency, and link reconstruction speed.

[0004] However, existing multimode communication solutions often treat each link layer as an independently operating siloed system, lacking a unified resource scheduling framework and dynamic backup mechanism. This results in an inability to seamlessly switch command and control data volumes when a single link is damaged or interfered with. Furthermore, traditional scheduling methods provide insufficient redundancy for critical data, making it difficult to cope with drastic fluctuations in link quality and prone to data loss or interruptions in on-site command and control. In addition, existing systems have a single dimension of perception for heterogeneous links, failing to accurately correlate data fragmentation with path performance in complex and dynamic environments, leading to low utilization of multimode resources.

[0005] Therefore, a service-aware adaptive selection scheme for multimode communication links is desired. Summary of the Invention

[0006] The purpose of this invention is to provide a service-aware multimode communication link adaptive selection system and method, which can effectively solve the problems in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Firstly, a service-aware multi-mode communication link adaptive selection method is applied to an integrated air-space-ground collaborative network including a dynamic topology manager, comprising the following steps: The command service data to be transmitted is acquired, and redundant fragmentation coding is performed on the command service data. The redundant fragmentation coding adopts the fountain code algorithm and dynamically adjusts the redundancy factor according to the real-time link status to generate multiple associated redundant coded fragments. The real-time link status includes packet loss rate and bit error rate. A three-dimensional air-space-ground topology map is constructed. The three-dimensional air-space-ground topology map is generated based on the three-dimensional spatial location information of each link node and the link quality evaluation matrix. The three-dimensional spatial location information is obtained through differentiated positioning methods, and the link quality evaluation matrix includes channel state parameters that characterize the link status. Based on the aforementioned three-dimensional topology map, the Kalman filter algorithm is used to predict the future state of the link. When it is predicted that the link is about to be interrupted, a path backup command is triggered in advance. In response to the path backup command or a link failure is detected, a portion of the fragments are selected from the plurality of associated redundant coded fragments, and the selected fragments are redirected to the backup link for transmission according to a preset load balancing strategy. At the receiving end, any part of the multiple associated redundant coded fragments is received from multiple heterogeneous links. When the total number of received fragments reaches the decoding threshold, the fountain code decoding algorithm is executed to restore the complete command service data. The decoding threshold is equal to the number of original data blocks K.

[0008] Preferably, constructing a three-dimensional topological map of space, air, and ground includes: The system acquires the three-dimensional spatial geographic coordinates of all active nodes. For space-based low-Earth orbit satellite nodes, the instantaneous orbital position is calculated by reading satellite ephemeris data. For airborne tethered UAV nodes, altitude information is corrected using airborne barometers and inertial navigation units based on satellite positioning. For ground-based nodes, differential positioning technology is used to improve coordinate accuracy. The link status is obtained by periodically sending probe frames, and the received signal strength indication, signal-to-noise ratio, round-trip time delay, and delay jitter are measured and obtained as channel status parameters. The transmission period of the probe frame is preset to 100 milliseconds, and the probe frame format includes a transmission timestamp, sequence number, and link identifier field. A three-dimensional topology graph is constructed using a graph data structure. The node set stores the three-dimensional coordinates and node type identifier of each node, while the edge set stores the source node identifier, destination node identifier, and corresponding channel state parameters of each link.

[0009] Preferably, the Kalman filter algorithm is used to predict the future state of the link, including: The three-dimensional position coordinates of the node and the key quality indicators of the link are defined as state variables of the Kalman filter algorithm. A uniform motion model is used as the state transition equation. The current positioning result and the link quality measurement value are used as the observation vector to recursively predict the position coordinates of the node and the key quality indicators of the link within a preset time period in the future. The preset time period is dynamically set according to the node's moving speed, and the typical value is 0.5 seconds to 2 seconds. When the predicted link quality index is lower than the preset threshold or the predicted distance between nodes exceeds the upper limit of the communication distance, it is determined that the link is about to be interrupted.

[0010] Preferably, redundant fragmentation coding is performed on the command and control data, including: The command and control data is divided into K raw data blocks of equal length; The pre-stored degree distribution function library is invoked, and a degree value d is randomly generated according to the soliton distribution algorithm. d data blocks are randomly selected from the K original data blocks, and an XOR operation is performed on the selected d data blocks to generate a redundant coded fragment. The degree value d ranges from 1 to K, and a robust soliton distribution is adopted with parameters c=0.02 and δ=0.5. The XOR operation is performed byte-aligned. Repeat the above steps for generating redundant coded fragments until N redundant coded fragments are generated, where the ratio of N to K is determined by the real-time sensed redundancy factor. A header structure is constructed for each generated redundant coded fragment. The header structure includes fragment sequence number, coding degree value, original block index mask, and cyclic redundancy check code.

[0011] Preferably, it further includes: before performing redundant fragmentation coding on the command and control data, constructing a multi-source access module to integrate the access capabilities of various heterogeneous networks, including low-orbit satellites, tethered unmanned aerial vehicles, ground mesh ad hoc networks, wireless local area network authentication and security infrastructure, and mobile communication networks; The construction of the multi-source access module includes: establishing a heterogeneous protocol conversion gateway; for low-Earth orbit satellite links, estimating the Doppler frequency shift and correcting the local carrier frequency in real time through a frequency offset estimation and compensation module; for tethered UAV links, dynamically adjusting the cyclic prefix length according to the Rice factor of the air-to-ground channel; and establishing a unified virtual link layer interface to abstract the link characteristics of different physical media and present them to the upper layer as a unified link access method.

[0012] Preferred, preset load balancing strategies include: Utilize deep message inspection technology to identify the service type of command and control data; Priority weights and service quality requirement indicators are assigned based on the identified business types; Based on the real-time link status, the analytic hierarchy process (AHP) combined with a weighted summation model is used to transform the priority weights and service quality requirement indicators into weight coefficients of the constraints. The comprehensive score of each candidate path is calculated, and the path with the highest score is selected as the optimal transmission path. The transformation is achieved by constructing a judgment matrix and calculating the weight coefficients of each indicator. The comprehensive score is calculated using a weighted summation formula. Based on the remaining available bandwidth, current latency, and reliability score of each physical link, the allocation weight of each link at the current moment is dynamically calculated, and the allocation weight determines the proportion of redundant coding fragments to be allocated by that link.

[0013] Preferably, detecting a link failure includes: A full-mesh or tree-structured heartbeat detection link is established between all active nodes. Each node periodically sends heartbeat detection packets to its neighboring nodes. The heartbeat detection packets contain the sender's timestamp, current battery status, and a local link table summary. If no response is received from a specific node within a preset number of consecutive detection cycles, or if the link signal-to-noise ratio carried in the received detection packet is lower than a preset threshold, it is determined that the link associated with that node has been interrupted or has suffered severe quality degradation. The preset number is 3 cycles, and the preset threshold is set according to the service type. The threshold for core command instructions is 15dB, and the threshold for general data is 8dB. Severe quality degradation refers to the link quality degrading to a signal-to-noise ratio lower than the preset threshold for more than 1 second.

[0014] Preferably, a subset of fragments is selected from the plurality of associated redundant coded fragments, and the selected fragments are redirected to the backup link for transmission according to a preset load balancing strategy, including: When no heartbeat response is received for three consecutive detection cycles or the received signal-to-noise ratio is lower than 8dB, an interruption alarm is triggered and an interruption alarm is sent to the dynamic topology manager. The faulty node and its associated logical edge are marked as in failure in the three-dimensional topology map of air, space and ground. The failure mark includes the failure timestamp and failure reason field. The weight of the failure edge is set to infinity in the subsequent path calculation. Based on the updated air-space-ground 3D topology map, the optimal alternative path for the affected command and control data is recalculated among the remaining available links. The new routing policy and link switching instructions are broadcast to all nodes in the network through the backup channel. The instructions include the backup link identification information corresponding to the affected command and control service data. After detecting an interruption in the primary transmission link, the receiving device automatically switches to the designated backup link according to the link switching instruction and continues to receive the remaining redundant coded fragments.

[0015] Preferably, after the fountain code decoding algorithm is executed at the receiving end to restore the complete command and control data, the method further includes: The restored original data blocks undergo dual integrity verification using cyclic redundancy check and hash signature verification. If the verification passes, the data will be submitted to the upper-layer application. If the verification finds that the data is corrupted and cannot be decoded correctly after 100 iterations of the belief propagation algorithm, it is determined that the redundancy fragmentation in the fountain code decoding process cannot be corrected. The original data block number corresponding to the corrupted data is recorded, and a retransmission request is generated. The retransmission request includes the sequence number of the specific redundant coded fragment or the original data block number that needs to be retransmitted. The retransmission request is sent through the link with the highest current overall link quality score calculated based on the link quality evaluation matrix, in order to obtain retransmission fragments until the data is complete and error-free.

[0016] Secondly, a service-aware multi-mode communication link adaptive selection system, applied to integrated air-space-ground collaborative networking, includes: Multi-source access module, used to construct physical connections and protocol conversion mechanisms for heterogeneous links; A redundant fragmentation coding unit is used to acquire command service data to be transmitted, and to perform redundant fragmentation coding on the command service data. The redundant fragmentation coding adopts the fountain code algorithm and dynamically adjusts the redundancy factor according to the real-time link status to generate multiple associated redundant coded fragments. A dynamic topology manager is used to construct a three-dimensional air-space-ground topology map. The three-dimensional air-space-ground topology map is generated based on the three-dimensional spatial location information of each link node and the link quality evaluation matrix. The three-dimensional spatial location information is obtained through differentiated positioning methods, and the link quality evaluation matrix includes channel state parameters that characterize the link status. Based on the three-dimensional air-space-ground topology map, the Kalman filter algorithm is used to predict the future state of the link. When the link is predicted to be about to be interrupted, a path backup command is triggered in advance. The resource coordination scheduler is used to identify service types and perform link selection and fragmentation allocation; The self-healing reconfiguration module is used to select a portion of the fragments from the plurality of associated redundant coded fragments in response to the path backup command or the detection of a link failure, and redirect the selected fragments to the backup link for transmission according to a preset load balancing strategy. The data fusion and restoration module is used to receive any part of the multiple associated redundant coded fragments from multiple heterogeneous links at the receiving end. When the total number of received fragments reaches the decoding threshold, the fountain code decoding algorithm is executed to restore the complete command and control business data. The data integrity verification module is used to verify data integrity and trigger retransmission.

[0017] In summary, this application includes at least one of the following beneficial technical effects: 1. This application integrates various heterogeneous networks such as low-orbit satellites, tethered UAVs, and ground mesh self-organizing networks by constructing a multi-source access module, and establishes a unified virtual link layer interface to abstract the link characteristics of different physical media and provide a unified network view to the upper layer. This solves the problems of independent operation of heterogeneous links and lack of a unified resource scheduling framework in the prior art, and realizes the collaborative scheduling and efficient utilization of multi-mode communication resources.

[0018] 2. This application uses the fountain code algorithm to perform redundant fragmentation encoding on command service data, dynamically adjusts the redundancy factor based on the real-time link status, and combines deep message detection technology to identify service types. Differentiated transmission paths are allocated to different services according to service priority and service quality requirements, which not only ensures extremely low latency and high reliability of core instructions, but also makes full use of the transmission capacity of high-bandwidth links, realizing resource optimization configuration under service awareness.

[0019] 3. This application uses a dynamic topology manager to obtain node locations and link status in real time, constructs a three-dimensional topology map of air, space, and ground, and uses the Kalman filter algorithm to predict the future state of the links, triggering path backup in advance when the link is about to be interrupted; at the same time, a heartbeat monitoring mechanism is established to quickly identify faults and automatically trigger a self-healing reconstruction process to redirect data fragments to backup links. The receiving end does not need to wait for the faulty link to recover, but can restore the data through any sufficient number of fragments, realizing seamless service switching in link failure scenarios and improving the continuity and reliability of communication in extreme environments. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical architecture of the multimode communication link adaptive selection method of this application; Figure 2 This is a logical flowchart illustrating the redundancy fragmentation coding and multi-path collaborative scheduling of command business data in this application. Figure 3 This is a schematic diagram illustrating the multi-level interaction relationship between the three-dimensional dynamic topology management and self-healing reconstruction of air, space, and ground in this application. Figure 4 This is a schematic diagram illustrating the principle of business-aware heterogeneous link load balancing and path optimization in this application. Detailed Implementation

[0021] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the following embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] This embodiment provides a service-aware multi-mode communication link adaptive selection system, including a multi-source access module, a redundant fragmentation coding unit, a dynamic topology manager, a resource collaborative scheduler, a self-healing reconfiguration module, a data fusion and restoration module, and a data integrity verification module. The multi-source access module is used to construct the physical connection and protocol conversion mechanism for heterogeneous links; the redundant fragmentation coding unit is used to perform redundant fragmentation coding on command service data; the dynamic topology manager is used to obtain node locations and link states and construct a three-dimensional topology map; the resource collaborative scheduler is used to identify service types and perform link selection and fragmentation allocation; the self-healing reconfiguration module is used to monitor link states and perform fault recovery; the data fusion and restoration module is used to integrate multi-path fragments and decode and restore them; and the data integrity verification module is used to verify data integrity and trigger retransmission. The following is in conjunction with the appendix... Figures 1 to 4 The working method of this system is explained in detail.

[0023] In the service-aware multi-mode communication link adaptive selection method provided in this embodiment, step S1 constructs a multi-source access module to integrate the access capabilities of various heterogeneous networks, specifically including low-Earth orbit satellites, tethered UAVs, terrestrial mesh ad hoc networks, Wireless LAN Authentication and Security Infrastructure (WAPI), and mobile communication technology signals, and establishes a physical connection and protocol conversion mechanism for heterogeneous links based on this. This is specifically achieved through the following steps: Step S101: Establish heterogeneous protocol conversion and physical connection.

[0024] First, through the heterogeneous protocol conversion gateway configured in the multi-source access module, a physical connection and protocol conversion mechanism with various heterogeneous links are established. The heterogeneous protocol conversion gateway runs in the central processing unit, working between the data link layer and the network layer, and is responsible for performing deep decapsulation and recapsulation operations of protocol messages.

[0025] For low-Earth orbit satellite links, the heterogeneous protocol conversion gateway adapts to the dedicated satellite-to-ground link protocol and processes signals with high propagation delay and Doppler frequency shift characteristics. The heterogeneous protocol conversion gateway integrates a frequency offset estimation and compensation module. This module estimates the Doppler frequency shift based on the preamble sequence or pilot symbol of the received signal and corrects the local carrier frequency in real time through a digitally controlled oscillator to achieve pre-compensation for carrier frequency offset.

[0026] For tethered UAV links, the heterogeneous protocol conversion gateway integrates air-to-ground radio communication protocols and supports orthogonal frequency division multiplexing modulation. The heterogeneous protocol conversion gateway dynamically adjusts the cyclic prefix length according to the Ricean fading characteristics of the air-to-ground channel. Specifically, the gateway measures the ratio of the power of the direct component to the power of the multipath component in the received signal, i.e., the Ricean factor (the ratio of the power of the direct signal to the power of the multipath signal). When the Ricean factor is low, a longer cyclic prefix is ​​selected to resist multipath delay spread, and when the Ricean factor is high, a shorter cyclic prefix is ​​selected to reduce transmission overhead.

[0027] For terrestrial mesh ad hoc networks, the heterogeneous protocol conversion gateway runs a cluster-head-based self-organizing routing protocol, maintains a neighbor relationship table between nodes, and manages the dynamic joining and leaving handshake processes of nodes based on the cluster member information broadcast by the cluster head. When a new node joins the network, it sends a joining request to the cluster head. After verifying the node's identity, the cluster head assigns an intra-cluster address and updates the routing table. When a node leaves the network, the cluster head removes it from the routing table after detecting a heartbeat timeout.

[0028] Step S102: Configure the multimode RF front-end subsystem.

[0029] At the physical hardware level, the multi-source access module contains multiple independent RF front-end subsystems, each corresponding to a different communication link, ensuring physical isolation and optimal performance for signal transmission and reception.

[0030] The first radio frequency front-end subsystem corresponds to a preset high-frequency band, such as the Ku band or the Ka band, and is equipped with a high-gain parabolic antenna or a phased array antenna to maintain narrow-beam communication with low-orbit satellites. This subsystem adjusts the antenna pointing through a beam tracking algorithm to ensure beam alignment during satellite movement.

[0031] The second RF front-end subsystem corresponds to a preset operating frequency band, such as 2.4 GHz or 5.8 GHz, and is mainly used for high-bandwidth relay communication of tethered drones. This subsystem adopts a multiple-input multiple-output antenna configuration to improve air interface transmission capacity.

[0032] The third RF front-end subsystem supports multi-band adaptive frequency hopping for anti-interference transmission of terrestrial mesh self-organizing networks in complex electromagnetic environments. This subsystem dynamically switches the operating frequency based on channel quality assessment results to avoid interference-prone frequencies.

[0033] Each RF front-end subsystem includes a low-noise amplifier, a power amplifier, a bandpass filter, and a high-performance digital-to-analog converter and analog-to-digital converter.

[0034] Low-noise amplifiers are used to amplify weak received signals, power amplifiers are used to amplify signals to be transmitted to meet transmission power requirements, bandpass filters are used to suppress out-of-band interference, and digital-to-analog converters and analog-to-digital converters are used to convert between analog signals and digital baseband signals.

[0035] These RF front-end subsystems connect to the central processing unit via high-speed serial peripheral interfaces or general-purpose parallel interfaces to enable real-time interaction of baseband signals.

[0036] Step S103: Construct a unified virtual link layer interface.

[0037] To shield the differences between the underlying heterogeneous physical media, the multi-source access module establishes a unified virtual link layer interface. This interface abstracts the link characteristics of different physical media and presents them to the upper layer as a unified link access method.

[0038] All heterogeneous signals that are accessed, including low-orbit satellite signals, tethered drone signals, ground mesh self-organizing network signals, and mobile communication technology signals, are encapsulated into standard Internet Protocol (IP) data packets at the virtual link layer.

[0039] During the encapsulation process, the system adds a unified global link identifier, source node physical address, destination node physical address, and link type label to each data packet. The global link identifier is used to uniquely identify a physical link, the source node physical address and the destination node physical address are used to identify the communication endpoints, and the link type label is used to distinguish different access technology types. Through standardized processing, it provides unified data support for subsequent routing and resource scheduling.

[0040] Through the steps S1 described above, a multi-source access module integrating space-based, air-based, and ground-based communication resources is constructed. This not only achieves physical connection and protocol interoperability between heterogeneous links, but more importantly, it provides a unified network view to the upper layers through a virtual link layer interface. This allows subsequent link selection and scheduling to be performed within a global, standardized framework, without needing to concern themselves with the underlying physical details.

[0041] In the above method, step S2 performs redundant fragmentation coding, redundantly coding the command service data to be transmitted according to the link status, generating multiple associated redundant coded fragments using the fountain code algorithm, and adding check information and sequence numbers to each fragment to improve the reliability of data transmission. This is specifically achieved through the following steps: Step S201: Raw data segmentation and encoding preparation.

[0042] The redundant fragmentation coding unit receives command service data from the application layer and stores the command service data into a high-speed circular buffer.

[0043] The data in the high-speed circular buffer is divided into K original data blocks of equal length according to the preset data block size. The preset value of the data block size is set based on a combination of system processing capabilities and link transmission unit size, ranging from 512 bytes to 2048 bytes. The data block size is selected based on the minimum value of the link MTU (Maximum Transmission Unit) to avoid fragmentation of encoded fragments at the transport layer. In a typical scenario, the ground link MTU is 1500 bytes, so the default value is 1024 bytes. The value of K is the total length of the original data stream divided by the data block size and rounded up. The capacity of the circular buffer is preset to twice the total amount of original data to prevent data from being prematurely overwritten during the encoding process.

[0044] The redundant segmented coding unit calls the pre-stored degree distribution function library, which uses the soliton distribution algorithm to guide the subsequent coding process.

[0045] Step S202: Generate redundant coded fragments.

[0046] During the encoding process, a degree value d is first randomly generated according to the soliton distribution algorithm. This degree value represents the number of original data blocks participating in this encoding operation. The value of the degree value d ranges from 1 to K (the total number of original data blocks). The soliton distribution adopts the robust soliton distribution, and its parameters c and δ are set to c=0.02 and δ=0.5 respectively to balance the encoding overhead and decoding complexity.

[0047] Then, d data blocks are randomly selected from the K original data blocks, and these data blocks are XORed to generate a redundant coded fragment. The XOR operation is performed byte-aligned, that is, the bytes at the same offset position in the selected d original data blocks are XORed byte by byte to generate the coded bytes at the corresponding positions.

[0048] The above process is repeated until N redundant coded fragments are generated. The value of N is not fixed, but is determined by the redundancy factor sensed by the system in real time. The redundancy factor reflects the redundancy strength required by the current link environment.

[0049] The encoding logic executed by the redundant fragmented encoding unit follows the following formula: in, This represents the j-th redundant coded fragment generated. This represents the i-th original data block. This represents the XOR operator. This represents the set of original data block indices selected for the j-th redundant coding fragment according to the degree distribution function. The set of original data block indices contains d elements, each element corresponding to the index value of a selected original data block.

[0050] In summary, this embodiment completes the conversion from raw data to redundant coded fragments, preparing redundant data units for subsequent link distribution.

[0051] Step S203: Construct the fragment header structure.

[0052] To enhance the self-descriptiveness and traceability of redundant coded fragments during transmission, the redundant fragmentation coding unit constructs a dedicated header structure for each generated redundant coded fragment. The header structure contains the following fields: Fragment sequence numbers are used to sort and reassemble redundant coded fragments at the receiving end.

[0053] The coding degree value records how many original data blocks were generated from the redundant coding fragment through an XOR operation, i.e., the degree value d mentioned above.

[0054] The original block index mask, in the form of a bitmap, indicates the location of the original data block that participates in the XOR operation, making it easier for the receiving end to locate the required original block.

[0055] Cyclic Redundancy Check (CRC) codes are used at the receiving end to detect whether bit errors have occurred during the transmission of fragments.

[0056] In addition, this embodiment will also embed a priority flag in the header according to the importance level of the service, so that the subsequent scheduling module can perform differentiated processing according to the priority.

[0057] Step S204: Dynamically adjust redundancy.

[0058] The redundant fragment coding unit has dynamic redundancy adjustment capability. The system monitors the average packet loss rate and bit error rate of each heterogeneous link in real time. These parameters are provided by the dynamic topology manager through the link quality evaluation matrix constructed in subsequent steps.

[0059] When a deterioration in the link environment is detected, such as when a ground mesh ad hoc network is subjected to strong multipath interference leading to an increased packet loss rate, the redundant fragment coding unit automatically increases the redundancy factor, that is, increases the ratio of N to K. This indicates that the system will generate more redundant coding fragments N when the number of original data blocks K remains unchanged.

[0060] By increasing the number of redundant fragments, it is ensured that even if some fragments are lost during transmission, the complete original data block sequence can be reconstructed as long as the total number of any received fragments reaches K. In other words, the decoding threshold equals the number of original data blocks, K. Specifically, the receiver uses Gaussian elimination or belief propagation algorithms to perform decoding operations on any K received fragments to reconstruct the complete original data block sequence.

[0061] When the number of received fragments is small (e.g., less than K+10), Gaussian elimination is used to ensure decoding success rate; when the number of fragments is sufficient (e.g., greater than or equal to K+50), the belief propagation algorithm is used to reduce computational complexity. The maximum number of iterations for the belief propagation algorithm is set to 100, and the early stopping condition is that there are no variable node updates for 5 consecutive iterations.

[0062] The above step S2 realizes a redundant fragmentation coding unit with dynamic adaptive capability. The original data is transformed into redundant fragments that can be flexibly distributed through the fountain code algorithm. At the same time, the manageability of redundant coding fragments is enhanced through header design, and the redundancy level is dynamically adjusted according to the real-time link quality, achieving a balance between transmission efficiency and reliability, and laying the data foundation for subsequent multi-link collaborative scheduling and self-healing reconstruction.

[0063] In the above method, step S3 constructs a three-dimensional air-space-ground topology map. Based on the physical location and channel quality of each layer of links, node coordinates and channel state parameters are obtained in real time to construct a logically unified three-dimensional air-space-ground topology map, providing a global view for subsequent resource scheduling and path selection. This is specifically achieved through the following steps: Step S301: Obtain the three-dimensional spatial location information of the node.

[0064] The dynamic topology manager acquires the three-dimensional spatial geographic coordinates of all active nodes in the system through the positioning interface integrated in the positioning system receiver at a preset sampling frequency. These coordinates include longitude, latitude, and altitude.

[0065] Different positioning methods are used for different types of nodes to ensure accuracy. Specifically: For space-based low-orbit satellite nodes, the instantaneous orbital position of the satellite in space can be calculated by reading satellite ephemeris data and combining it with the current time.

[0066] For airborne tethered UAV nodes, in addition to satellite positioning, airborne barometers are used to measure atmospheric pressure to estimate altitude, and attitude and acceleration data from the inertial navigation unit are used for auxiliary correction, thereby improving the accuracy of altitude information.

[0067] For ground-based mobile terminals and mesh self-organizing network nodes, differential positioning technology is used to eliminate common errors through correction data provided by ground reference stations, thereby improving coordinate accuracy to within the preset error range.

[0068] Step S302: Construct the link quality evaluation matrix.

[0069] The dynamic topology manager internally constructs a link quality evaluation matrix, which is a dynamically updated two-dimensional data structure. The rows and columns of the link quality evaluation matrix represent the source node and the destination node in the system, respectively. Each element in the link quality evaluation matrix is ​​used to store the comprehensive quality score of the corresponding link.

[0070] Link status is acquired by periodically sending probe frames. Specifically, the source node sends timestamped probe messages to the target node, and the target node responds with relevant measurement information. Based on this, the system measures and acquires the following channel state parameters in real time: Received signal strength indicator, this parameter directly reflects the loss and path attenuation of the signal during its propagation in space.

[0071] Signal-to-noise ratio (SNR) is used to quantify the power ratio of the useful signal to the noise in a link, reflecting the link's noise immunity.

[0072] Round-trip delay is calculated by comparing the sending and receiving timestamps of the probe frames to determine the total time it takes for the signal to travel from the source node to the target node.

[0073] Latency jitter reflects the stability of a link by statistically analyzing the variation in round-trip latency.

[0074] The original measurement parameters are normalized to map each parameter to the same numerical range, and then weighted and summed according to preset weight coefficients to form a comprehensive quality score for each link. The preset weight coefficients can be set according to the sensitivity of different services to various indicators in the application scenario and are dynamically updated in the evaluation matrix. The initial values ​​of the preset weight coefficients are set as follows: Received signal strength indication weight 0.3, signal-to-noise ratio weight 0.4, round-trip time delay weight 0.2, and delay jitter weight 0.1, which can be dynamically adjusted according to the service type.

[0075] Step S303: Construct a three-dimensional topology map of space, air, and ground.

[0076] Based on the physical location information obtained in step S301 and the channel state parameters obtained in step S302, the dynamic topology manager constructs a three-dimensional air-space-ground topology map using a graph data structure. In the graph data structure, the node set stores the three-dimensional coordinates and node type identifier of each node, and the edge set stores the source node identifier, destination node identifier, corresponding comprehensive quality score, and channel state parameters of each link.

[0077] In the three-dimensional topology diagram of air, space, and ground, each node in the network is abstracted as a vertex in three-dimensional space, and its spatial coordinates are determined by the result of step S301. Each communication link between nodes is abstracted as an edge connecting two vertices. The topology manager adopts an asymmetric link modeling method to record the quality difference between the uplink and downlink for each edge, that is, to store the link quality scores from node A to node B and from node B to node A respectively.

[0078] Each edge is assigned multiple weight attributes, which correspond to various indicators in the link quality evaluation matrix, including overall quality score, received signal strength indication, signal-to-noise ratio, round-trip time, and time jitter. The update frequency of the air-space-ground three-dimensional topology map is set to a preset time interval to ensure that the topology status can reflect the changes in the physical environment of the emergency site in real time.

[0079] Step S304: Perform link prediction and proactive early warning.

[0080] The dynamic topology manager also executes a link prediction algorithm. Based on the historical trajectory of node movement and the changing trend of channel quality, the system uses the Kalman filter algorithm to predict the future state of the link.

[0081] The state variables of the Kalman filter algorithm are defined as the three-dimensional position coordinates of the nodes and the key quality indicators of the link. The state transition equation adopts a uniform motion model, assuming that the nodes move in uniform linear motion between adjacent time steps. The state transition matrix F is set as... ,in For the prediction period (e.g., 0.1 seconds), the observation matrix H is set to... The x-axis and y-axis positions are observed respectively. The observation equation uses the current positioning result and link quality measurement value as the observation vector.

[0082] The system maintains a Kalman filter for each node and each link. Using the node's position sequence and link quality parameter sequence over a past period as input, it recursively predicts the node's position coordinates and key link quality indicators within a preset timeframe through two steps: state prediction and observation update. The preset timeframe is dynamically set based on the node's movement speed, typically ranging from 0.5 to 2 seconds, with a smaller value used for high-speed moving nodes (such as drones).

[0083] The Kalman filter recursive process includes: state prediction equation Covariance prediction equation Kalman gain State update equation Covariance update equation .

[0084] Where Q is the process noise covariance matrix (set as a diagonal matrix with diagonal elements of 0.01), and R is the observation noise covariance matrix (with diagonal elements of 0.1).

[0085] The path backup command uses JSON format and includes the following fields: {"cmd_type":"path_backup","link_id":"link identifier","predict_failure_time":timestamp","affected_flows":[list of business flow identifiers]}.

[0086] Based on the prediction results, it is determined whether a link is about to be interrupted due to physical obstruction, signal attenuation, or exceeding the communication distance. The judgment is based on whether the predicted link quality index is lower than the preset threshold or the predicted distance between nodes exceeds the upper limit of the communication distance.

[0087] If a link is predicted to be about to fail, the dynamic topology manager marks the link as being in a warning state in advance on the three-dimensional topology map and immediately triggers the resource coordination scheduler to perform pre-path backup. The resource coordination scheduler reserves resources or establishes backup paths on the backup link in advance, so that it can quickly switch over when the link is actually interrupted, thereby reducing the instantaneous jitter and data packet loss caused by link switching.

[0088] Through the above step S3, a dynamic, accurate and predictive three-dimensional topology map of air, space, and ground is constructed and maintained. This map not only reflects the current spatial structure and link quality of the network, but also proactively predicts link risks. It provides a reliable data foundation for the path selection and load balancing of the resource coordination scheduler in the subsequent step S4, so that the path selection under business awareness can be based on a comprehensive understanding of the network status.

[0089] In the above method, step S4 performs resource collaborative scheduling, obtains the priority and bandwidth requirements of command and control service data, and allocates redundant coding fragments to multiple paths such as satellites, UAVs, and ground mesh ad hoc networks according to a predetermined ratio, thereby achieving path optimization and efficient resource utilization under service awareness. This is specifically achieved through the following steps: Step S401: Business feature identification and classification.

[0090] The resource coordination scheduler includes a service feature recognition submodule, which performs the service feature recognition function. When command service data enters the scheduling queue, the service feature recognition submodule uses Deep Packet Inspection (DPI) technology to extract features from the packet payload.

[0091] This system has a pre-established business type feature library, which stores matching rules for various businesses. The business feature identification submodule accurately identifies the business type of the current data stream by matching the header fields of the data packets, such as protocol type, source port number and destination port number, as well as traffic statistics features, such as the arrival interval of data packets and the distribution pattern of packet size.

[0092] The types of services that this system can recognize include real-time voice calls, high-definition video dispatch streams, core command instructions, environmental monitoring sensor data, and general file transfers.

[0093] Step S402: Mapping business priorities to service quality.

[0094] Based on the service type identified in step S401, the resource coordination scheduler assigns different priority weights and service quality requirement indicators to each service.

[0095] For example, core command instructions are given the highest priority, requiring extremely low transmission latency and extremely high reliability. Typically, latency must not exceed milliseconds and packet loss is not allowed. High-definition video scheduling streams require high bandwidth and low jitter to ensure smooth video playback. Bandwidth requirements are typically above several megabits per second, and jitter must be controlled within tens of milliseconds. Environmental monitoring sensor data, on the other hand, are relatively insensitive to latency and can be transmitted using a best-effort transmission method.

[0096] The resource coordination scheduler uses these priority weights and service quality indicators as constraints for subsequent path selection. Simultaneously, based on the weight of each quality requirement in the service quality indicators, it presets corresponding service adaptation weight coefficients for different service types. These coefficients are used for the weighted calculation of each link parameter during subsequent load balancing. Latency-sensitive services receive increased weights for latency and jitter, while bandwidth-sensitive services receive increased weights for bandwidth and reliability scores. Example of service adaptation weight coefficients: For latency-sensitive services (such as voice calls), the latency weight is set to 0.5, bandwidth weight to 0.2, and reliability weight to 0.3; for bandwidth-sensitive services (such as video streaming), the bandwidth weight is set to 0.5, latency weight to 0.2, and reliability weight to 0.3.

[0097] Step S403: Path optimization based on multiple constraint objectives.

[0098] The resource coordination scheduler executes a path optimization algorithm based on multiple constraint objectives, according to the real-time link status provided by the dynamic topology manager.

[0099] The path optimization algorithm is implemented by combining the analytic hierarchy process (AHP) with a weighted summation model. The judgment matrix of the AHP is constructed through expert experience or historical data statistics. Taking four indicators, latency, bandwidth, jitter, and packet loss rate, as examples, a 4×4 judgment matrix A is constructed, where A(i,j) represents the importance ratio of indicator i to indicator j (using the 1-9 scale). The weight vector w is obtained by solving the eigenvector corresponding to the largest eigenvalue of A and normalizing it. When the consistency ratio CR is less than 0.1, the weight allocation is considered reasonable.

[0100] Specifically, firstly, the service priority and service quality requirements determined in step S402 are converted into weight coefficients for each constraint. For latency-sensitive services, the weights of latency and jitter are increased; for bandwidth-sensitive services, the weights of bandwidth and reliability scores are increased. Then, for each candidate path, the remaining available bandwidth, current latency, latency jitter, and reliability scores of each link segment are accumulated or their maximum values ​​are taken to obtain a comprehensive evaluation value at the path level. Finally, the comprehensive score for each path is calculated using a weighted summation formula, which is: ,in Let be the weight coefficient of the i-th constraint. Given the normalized score of the path under this constraint, the path with the highest score is selected as the optimal transmission path.

[0101] For example, for core command instructions, the path optimization algorithm will prioritize the link with the lowest latency and highest reliability, even if the link has narrow bandwidth; for high-definition video streams, the path optimization algorithm will prioritize the link with sufficient bandwidth and low jitter.

[0102] Step S404: Weight-based load balancing and sharding allocation.

[0103] When performing specific resource allocation, the resource coordination scheduler adopts a weight-based load balancing mechanism. The system dynamically calculates the allocation weight of each link at the current moment based on the remaining available bandwidth, current latency, and reliability score of each physical link.

[0104] The weighting calculation method is as follows: First, the three parameters of remaining available bandwidth, current latency, and reliability score are normalized and mapped to the range of 0 to 1. The latency parameter is converted into a positive indicator by subtracting the normalized value from 1. Then, the weighted sum is calculated according to the preset service adaptation weight coefficient to obtain the comprehensive score of each link. Finally, the proportion of the comprehensive score in the total comprehensive score of all available links is used as the weighting of that link.

[0105] The allocation weight directly determines the proportion of redundant coding fragments that the link undertakes.

[0106] In a typical scenario, assuming the ground-based mesh ad hoc network links have sufficient bandwidth but fluctuate significantly in latency, the tethered UAV links have extremely low and stable latency but limited coverage, and the low-Earth orbit satellite links have wide coverage but limited bandwidth, the resource coordination scheduler can allocate the first portion of the real-time video stream, such as 60%, to the ground-based mesh ad hoc network path, utilizing its bandwidth advantage to carry most of the video data; allocate the second portion of the video stream, such as 30%, to the tethered UAV link, utilizing its low latency characteristics to ensure real-time synchronization of the footage; simultaneously, a complete copy or critical redundant portion of the core command signaling, such as the highest priority portion of the redundant coded portion generated in step S2, is synchronously backed up and transmitted via the satellite link to ensure the absolute reliability of the core commands.

[0107] Step S405: Flow shaping and congestion control.

[0108] The resource coordination scheduler also executes traffic shaping and congestion control strategies. The system monitors the sending queue depth of each physical link in real time, which is the number of redundant coded fragments waiting to be sent. When the queue length of a link exceeds a preset threshold, it indicates that the link is about to or has already become congested.

[0109] At this point, the resource coordination scheduler automatically triggers the flow control mechanism, dynamically reducing the allocation weight of the link, decreasing the number of redundant coded fragments allocated to the link, and redirecting fragments exceeding its carrying capacity to other relatively idle links. This process continues, ensuring that the load of each link is always maintained at a reasonable level, avoiding fragment loss and a surge in latency caused by congestion on a single link.

[0110] Through the above step S4, a service-aware dynamic resource collaborative scheduling mechanism is realized. This mechanism first identifies the service type and clarifies its service quality requirements, then performs multi-constraint path optimization based on real-time topology and link status, and finally rationally allocates redundant coding fragments to multiple heterogeneous links through dynamic load balancing and congestion control, thereby jointly realizing the optimal combination and utilization of heterogeneous communication resources, so that different types of data streams can obtain transmission quality commensurate with their importance.

[0111] In the above method, step S5 performs self-healing reconfiguration. By monitoring the node's liveness status and heartbeat signals, when a link disconnection or quality deterioration is detected, a self-healing reconfiguration command is automatically triggered to redirect the data stream, ensuring uninterrupted communication and achieving seamless service switching. This is specifically achieved through the following steps: Step S501: Establish a heartbeat monitoring mechanism.

[0112] The self-healing refactoring module is equipped with a preset level of heartbeat monitoring mechanism, which establishes a full-mesh or tree-like heartbeat detection link between all active nodes to ensure that the status of each node can be perceived by surrounding nodes in real time.

[0113] Each node periodically sends heartbeat probe packets to its neighboring nodes. The heartbeat probe packet contains the sender's timestamp, used to calculate round-trip time and determine whether the node is alive; the current power status, used to assess the node's operational continuity; and a local link table summary, used to synchronize link status information.

[0114] The self-healing reconfiguration module performs real-time statistics on the reception of heartbeat detection packets at each node, either centrally or in a distributed manner.

[0115] Step S502: Fault determination.

[0116] The self-healing reconfiguration module determines the link status based on the received heartbeat probe packets. If no response is received from a specific node within a preset number of probe cycles, the node is determined to be offline or its associated link is interrupted. The preset number is set to 3 cycles to balance fault detection speed and false positive rate. Alternatively, if the link signal-to-noise ratio carried in the received heartbeat probe packet is lower than a preset threshold, the link associated with that node is determined to have suffered severe quality degradation. The preset threshold is set according to the service type: for core command instructions, the signal-to-noise ratio threshold is set to 15dB; for general data, the threshold is set to 8dB.

[0117] Once any of the above conditions are met, fault determination is triggered, and the self-healing process begins. This allows for the rapid and accurate identification of node failures or severe link deterioration in the network without introducing excessive overhead.

[0118] Step S503: Topology update and failure marking.

[0119] Upon triggering a fault determination, the self-healing reconfiguration module immediately initiates the self-healing process. First, the self-healing reconfiguration module sends an interruption alarm to the dynamic topology manager, notifying it of the currently detected faulty node or link information. The interruption alarm data structure includes: {"alert_type":"link_failure","node_id": faulty node identifier,"link_id": faulty link identifier,"timestamp": detection time,"failure_reason":"heartbeat timeout / SNR below threshold"}.

[0120] Upon receiving an alert, the dynamic topology manager quickly updates the 3D topology graph it maintains. Specifically, the dynamic topology manager marks the faulty node itself and all its associated logical edges as in a failed state in the topology graph and returns an acknowledgment message {"status":"updated","version":topology version number}.

[0121] These logical edges include all uplinks originating from the faulty node and all downlinks pointing to the faulty node. The update of the topology graph ensures that any subsequent path calculations based on the graph will no longer use these failed resources.

[0122] Step S504: Recalculate the alternative path and issue the instruction.

[0123] Subsequently, the self-healing reconfiguration module issues a reconfiguration command to the resource coordination scheduler. Based on the updated 3D topology map by the dynamic topology manager, the resource coordination scheduler uses the shortest path algorithm or multi-path optimization algorithm to recalculate one or more optimal alternative paths for the affected command and control data among the remaining available links.

[0124] The cost metric used for path calculation is the comprehensive link quality score calculated based on the link quality evaluation matrix defined in step S303. The higher the comprehensive quality score, the lower the path cost, and the path with the highest comprehensive quality score is selected first.

[0125] During the reconstruction process, the self-healing reconstruction module broadcasts new routing policies and link switching instructions to all network nodes through a highly reliable backup channel, such as a low-Earth orbit satellite link. The instructions include backup link identification information corresponding to the affected command and control business data, as well as data reception rules that the data receiving end should follow after the switch.

[0126] Step S505: Receiver link switching and data reception.

[0127] The device at the data receiving end determines that the main transmission link has been interrupted by detecting that the heartbeat probe packet from the current main transmission link has timed out or that the link quality parameter is lower than a preset threshold. After confirming the main link interruption, the receiving end device automatically switches to the backup link specified in the instruction according to the new instruction issued by the self-healing reconstruction module, and continues to receive the remaining redundant coded fragments.

[0128] Since the system has already adopted redundant fragmentation coding in step S2, the original data is converted into multiple independent redundant coded fragments. The receiving end does not need to wait for the failed link to recover; it only needs to collect a sufficient number of fragments from other heterogeneous links that are still operating normally. For example, fragments that were originally transmitted through the UAV link can be switched to receiving the remaining fragments from both the satellite link and the ground ad hoc network link simultaneously after the link is interrupted.

[0129] Step S506: Data fusion and restoration.

[0130] The data fusion and restoration module integrates fragments from different physical paths at the receiving end. It uses the sequence number in the header of each fragment to deduplicate and sort the received fragments, avoiding duplicate reception problems that may be caused by multi-path transmission. At the same time, based on the original block index mask information in the header, the data fusion and restoration module associates each fragment with the corresponding original data block and reassembles them to generate a decoding matrix.

[0131] Subsequently, the system executes a fountain code decoding algorithm, such as the belief propagation algorithm or Gaussian elimination, to reconstruct the complete original data block sequence using any sufficient number of received fragments. The entire decoding process is completed within a very short preset time, usually controlled at the millisecond level.

[0132] Through the above step S5, a complete self-healing reconfiguration mechanism is constructed. This mechanism achieves rapid fault detection based on heartbeat monitoring, adaptive route adjustment through dynamic topology updates and path recalculation, and stateless data restoration at the receiving end by relying on redundant fragmentation coding. Thus, it jointly ensures that business data can be transmitted continuously and stably when heterogeneous links are interrupted or degraded in quality, providing reliable protection for communication continuity in extreme environments.

[0133] This embodiment is further configured with a data integrity verification module. After the data reconstruction is completed by the data fusion and restoration module at the receiving end, the data integrity verification module immediately starts to verify the integrity of the restored original data block.

[0134] The integrity verification employs a dual verification mechanism. First, the data integrity verification module performs cyclic redundancy check (CRUD) on each restored original data block, comparing the calculated CRUD code with the CRUD code appended to the fragment header in step S203 to detect whether bit errors occurred during data block transmission or decoding. Second, the data integrity verification module performs hash signature verification on the restored original data block, using a preset hash algorithm, such as a secure hash algorithm, to calculate the hash value of the data block and compare it with the hash signature pre-attached by the sender to further confirm the integrity of the data and the authenticity of its source.

[0135] If the verification passes, it indicates that the data is complete and error-free, and the system will then deliver the data to the upper-layer application.

[0136] If the verification detects errors in the data, and these errors cannot be corrected by the redundant fragments in the fountain code decoding process of step S2, the system determines that the data is corrupted. At this time, the data integrity verification module records the original data block number corresponding to the corrupted data, as well as the identifier of the original data stream to which it belongs, and generates a retransmission request. The retransmission request explicitly includes the sequence number of the specific redundant coded fragment or the original data block number that needs to be retransmitted.

[0137] The retransmission request is sent to the resource coordination scheduler. After receiving the request, the resource coordination scheduler selects the link with the highest overall link quality score from the currently available heterogeneous links as the retransmission channel, based on the real-time link quality score provided by the dynamic topology manager.

[0138] The resource coordination scheduler requests the sender to retransmit the specified redundant coded fragment through the optimal link. After receiving the retransmission request, the sender searches for the corresponding fragment in the local redundant coded fragment cache based on the redundant coded fragment sequence number or the original data block number carried in the request. If the fragment exists in the cache, it is retrieved directly. If the fragment has been overwritten in the cache, the sender re-executes the XOR operation using the corresponding original data block to generate the fragment based on the encoding parameters recorded in step S202.

[0139] After generation or retrieval is completed, the sending end retransmits the data through the optimal link specified by the resource coordination scheduler. After receiving the retransmitted fragment, the receiving end performs integrity verification again until the data is completely correct, ensuring the absolute accuracy of command data under extreme interference environments.

[0140] The method described in this embodiment is based on a software-defined network architecture that separates the system's control plane from its data plane.

[0141] Specifically, the dynamic topology manager and resource coordinator run in the control plane carried by the central processing unit, undertaking the functions of global network status perception, topology map construction and maintenance, command of service data routing calculation, and resource coordinator scheduling, as well as policy generation and distribution. The multi-source access module and redundant fragmentation coding unit run in the data plane, responsible for executing specific data forwarding, protocol conversion, redundant coding, and fragmentation processing based on the policies issued by the control plane. The control plane and the data plane communicate through a standardized southbound interface, such as using the OpenFlow protocol or its extended variants, to ensure the efficiency and reliability of policy distribution.

[0142] This control and forwarding separation architecture enables the system to flexibly upgrade and adjust the control plane's scheduling algorithm, coding strategy, and topology management logic online according to the actual needs of emergency tasks, without interrupting data plane service forwarding. This improves the system's adaptability and evolvability in complex and ever-changing environments.

[0143] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0144] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment includes only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A service-aware multimode communication link adaptive selection method, characterized in that, For applications in integrated air-space-ground collaborative networking that includes a dynamic topology manager, the following steps are included: The command service data to be transmitted is acquired, and redundant fragmentation coding is performed on the command service data. The redundant fragmentation coding adopts the fountain code algorithm and dynamically adjusts the redundancy factor according to the real-time link status to generate multiple associated redundant coded fragments. The real-time link status includes packet loss rate and bit error rate. A three-dimensional air-space-ground topology map is constructed. The three-dimensional air-space-ground topology map is generated based on the three-dimensional spatial location information of each link node and the link quality evaluation matrix. The three-dimensional spatial location information is obtained through differentiated positioning methods, and the link quality evaluation matrix includes channel state parameters that characterize the link status. Based on the aforementioned three-dimensional topology map, the Kalman filter algorithm is used to predict the future state of the link. When it is predicted that the link is about to be interrupted, a path backup command is triggered in advance. In response to the path backup command or a link failure is detected, a portion of the fragments are selected from multiple associated redundant coded fragments, and the selected fragments are redirected to the backup link for transmission according to a preset load balancing strategy. At the receiving end, any part of the fragments from multiple associated redundant coded fragments are received from multiple heterogeneous links. When the total number of received fragments reaches the decoding threshold, the fountain code decoding algorithm is executed to restore the complete command service data; the decoding threshold is equal to the number of original data blocks K.

2. The method according to claim 1, characterized in that, Constructing a three-dimensional topological map of space, air, and ground, including: The system acquires the three-dimensional spatial geographic coordinates of all active nodes. For space-based low-Earth orbit satellite nodes, the instantaneous orbital position is calculated by reading satellite ephemeris data. For airborne tethered UAV nodes, altitude information is corrected using airborne barometers and inertial navigation units based on satellite positioning. For ground-based nodes, differential positioning technology is used to improve coordinate accuracy. The link status is obtained by periodically sending probe frames, and the received signal strength indication, signal-to-noise ratio, round-trip time delay, and delay jitter are measured and obtained as channel status parameters. The transmission period of the probe frame is preset to 100 milliseconds, and the probe frame format includes a transmission timestamp, sequence number, and link identifier field. A three-dimensional topology graph is constructed using a graph data structure. The node set stores the three-dimensional coordinates and node type identifier of each node, while the edge set stores the source node identifier, destination node identifier, and corresponding channel state parameters of each link.

3. The method according to claim 2, characterized in that, The Kalman filter algorithm is used to predict the future state of the link, including: The three-dimensional position coordinates of the node and the key quality indicators of the link are defined as state variables of the Kalman filter algorithm. A uniform motion model is used as the state transition equation. The current positioning result and the link quality measurement value are used as the observation vector to recursively predict the position coordinates of the node and the key quality indicators of the link within a preset time period in the future. The preset time period is dynamically set according to the node's moving speed, and the typical value is 0.5 seconds to 2 seconds. When the predicted link quality index is lower than the preset threshold or the predicted distance between nodes exceeds the upper limit of the communication distance, it is determined that the link is about to be interrupted.

4. The method according to claim 1, characterized in that, Performing redundant fragmentation coding on the command and control data includes: The command and control data is divided into K raw data blocks of equal length; The pre-stored degree distribution function library is called, and a degree value d is randomly generated according to the soliton distribution algorithm. d data blocks are randomly selected from K original data blocks, and an XOR operation is performed on the selected d data blocks to generate a redundant coded fragment. The degree value d ranges from 1 to K, and a robust soliton distribution is adopted with parameters c=0.02 and δ=0.

5. The XOR operation is performed byte-aligned. Repeat the above steps for generating redundant coded fragments until N redundant coded fragments are generated, where the ratio of N to K is determined by the real-time sensed redundancy factor. A header structure is constructed for each generated redundant coded fragment. The header structure includes fragment sequence number, coding degree value, original block index mask, and cyclic redundancy check code.

5. The method according to claim 1, characterized in that, Also includes: Before performing redundant fragmentation coding on the command and control data, a multi-source access module is constructed to integrate the access capabilities of various heterogeneous networks, including low-orbit satellites, tethered UAVs, ground mesh ad hoc networks, wireless LAN authentication and security infrastructure, and mobile communication networks. The construction of the multi-source access module includes: establishing a heterogeneous protocol conversion gateway; for low-Earth orbit satellite links, estimating the Doppler frequency shift and correcting the local carrier frequency in real time through a frequency offset estimation and compensation module. For tethered UAV links, the cyclic prefix length is dynamically adjusted based on the Rice factor of the air-to-ground channel; and a unified virtual link layer interface is established to abstract the link characteristics of different physical media and present them to the upper layer as a unified link access method.

6. The method according to claim 1, characterized in that, The preset load balancing strategies include: Utilize deep message inspection technology to identify the service type of command and control data; Priority weights and service quality requirement indicators are assigned based on the identified business types; Based on the real-time link status, the analytic hierarchy process (AHP) combined with a weighted summation model is used to transform the priority weights and service quality requirement indicators into weight coefficients of the constraints. The comprehensive score of each candidate path is calculated, and the path with the highest score is selected as the optimal transmission path. The transformation is achieved by constructing a judgment matrix and calculating the weight coefficients of each indicator. The comprehensive score is calculated using a weighted summation formula. Based on the remaining available bandwidth, current latency, and reliability score of each physical link, the allocation weight of each link at the current moment is dynamically calculated, and the allocation weight determines the proportion of redundant coding fragments to be allocated by that link.

7. The method according to claim 1, characterized in that, A link failure was detected, including: A full-mesh or tree-structured heartbeat detection link is established between all active nodes. Each node periodically sends heartbeat detection packets to its neighboring nodes. The heartbeat detection packets contain the sender's timestamp, current battery status, and a local link table summary. If no response is received from a specific node within a preset number of consecutive detection cycles, or if the link signal-to-noise ratio carried in the received detection packet is lower than a preset threshold, it is determined that the link associated with that node has been interrupted or has suffered severe quality degradation. The preset number is 3 cycles, and the preset threshold is set according to the service type. The threshold for core command instructions is 15dB, and the threshold for general data is 8dB. Severe quality degradation refers to the link quality degrading to a signal-to-noise ratio lower than the preset threshold for more than 1 second.

8. The method according to claim 7, characterized in that, Selecting a subset of fragments from multiple associated redundant coded fragments, and redirecting these fragments to backup links for transmission according to a preset load balancing strategy, includes: When no heartbeat response is received for three consecutive detection cycles or the received signal-to-noise ratio is lower than 8dB, an interruption alarm is triggered and an interruption alarm is sent to the dynamic topology manager. The faulty node and its associated logical edge are marked as in failure in the three-dimensional topology map of air, space and ground. The failure mark includes the failure timestamp and failure reason field. The weight of the failure edge is set to infinity in the subsequent path calculation. Based on the updated air-space-ground 3D topology map, the optimal alternative path for the affected command and control data is recalculated among the remaining available links. The new routing policy and link switching instructions are broadcast to all nodes in the network through the backup channel. The instructions include the backup link identification information corresponding to the affected command and control service data. After detecting an interruption in the primary transmission link, the receiving device automatically switches to the designated backup link according to the link switching instruction and continues to receive the remaining redundant coded fragments.

9. The method according to claim 1, characterized in that, After the receiving end executes the fountain code decoding algorithm to reconstruct the complete command and control data, it also includes: The restored original data blocks undergo dual integrity verification using cyclic redundancy check and hash signature verification. If the verification passes, the data will be submitted to the upper-layer application. If the verification finds that the data is corrupted and cannot be decoded correctly after 100 iterations of the belief propagation algorithm, it is determined that the redundancy fragmentation in the fountain code decoding process cannot be corrected. The original data block number corresponding to the corrupted data is recorded, and a retransmission request is generated. The retransmission request includes the sequence number of the specific redundant coded fragment or the original data block number that needs to be retransmitted. The retransmission request is sent through the link with the highest current overall link quality score calculated based on the link quality evaluation matrix, in order to obtain retransmission fragments until the data is complete and error-free.

10. A service-aware multimode communication link adaptive selection system, characterized in that, Applications include integrated air-space-ground collaborative networking: Multi-source access module, used to construct physical connections and protocol conversion mechanisms for heterogeneous links; A redundant fragmentation coding unit is used to acquire command service data to be transmitted, and to perform redundant fragmentation coding on the command service data. The redundant fragmentation coding adopts the fountain code algorithm and dynamically adjusts the redundancy factor according to the real-time link status to generate multiple associated redundant coded fragments. A dynamic topology manager is used to construct a three-dimensional air-space-ground topology map. The three-dimensional air-space-ground topology map is generated based on the three-dimensional spatial location information of each link node and the link quality evaluation matrix. The three-dimensional spatial location information is obtained through differentiated positioning methods, and the link quality evaluation matrix includes channel state parameters that characterize the link status. Based on the three-dimensional air-space-ground topology map, the Kalman filter algorithm is used to predict the future status of the link. When the link is predicted to be about to be interrupted, a path backup command is triggered in advance. The resource coordination scheduler is used to identify service types and perform link selection and fragmentation allocation; The self-healing reconfiguration module is used to select a portion of the fragments from the plurality of associated redundant coded fragments in response to the path backup command or the detection of a link failure, and redirect the selected fragments to the backup link for transmission according to a preset load balancing strategy. The data fusion and restoration module is used to receive any part of the multiple associated redundant coded fragments from multiple heterogeneous links at the receiving end. When the total number of received fragments reaches the decoding threshold, the fountain code decoding algorithm is executed to restore the complete command and control business data. The data integrity verification module is used to verify data integrity and trigger retransmission.

Images