Transmission method and system of large-scale unmanned cluster sensor data collection network

Abstract

The present application relates to the technical field of unmanned cluster wireless communication, and provides a transmission method and system of a large-scale unmanned cluster sensor data collection network, a minimum spanning tree structure constructed by initial clustering, first round merging and second round merging, which is accurately mapped into a specific concurrent time slot scheduling scheme by a graph coloring method, and the concurrent transmission link in each time slot fully plays the potential of the previously reserved parallel transmission path. In the method, the initial clustering realizes data aggregation in the cluster to reduce the transmission times, the first round merging eliminates unnecessary redundant paths to reduce the network complexity, and the second round merging ensures that the reserved parallel paths are superior to the merging scheme in terms of throughput. Finally, the graph coloring time slot scheduling maps these optimized paths into a specific concurrent transmission schedule, realizing cross-layer joint optimization from network layer routing design to link layer time slot scheduling.

Description

Technical Field

[0001] This invention belongs to the field of unmanned cluster wireless communication technology, and relates to a transmission method and system for a large-scale unmanned cluster sensor data acquisition network. Background Technology

[0002] When large-scale unmanned swarms (such as drone swarms, unmanned vehicle swarms, and robot formations) perform tasks such as regional environmental monitoring, target monitoring, collaborative mapping, or disaster monitoring, the numerous unmanned nodes distributed within the swarm need to transmit the sensor data collected by their respective onboard sensors (such as optical cameras, infrared detectors, multispectral sensors, meteorological sensors, and sonar) back to designated data aggregation nodes or ground control centers in real time via wireless networks for subsequent data fusion and situational analysis. In these application scenarios, due to the large number of nodes and the vast operating area, and the limited communication distance of individual nodes, large-scale unmanned swarm sensor data acquisition networks typically employ a "multi-hop relay" (i.e., multi-hop) method to transmit sensor data from source nodes to data aggregation nodes.

[0003] The characteristics of large-scale unmanned swarm sensor data acquisition networks include: limited communication distance of a single wireless node; multi-hop transmission allows data to be transmitted through intermediate nodes, significantly expanding the sensor monitoring coverage area of ​​the swarm and breaking through the limitation of single-point communication radius; nodes do not require high-power signal transmission, reducing power consumption and extending swarm endurance; and lowering the requirements for node hardware performance to save system costs. If a node fails or the link is interrupted, data can be transmitted through other paths, improving the swarm's anti-interference capability and fault tolerance in complex environments. Wireless multi-hop routing from the source node to the data aggregation node is crucial for the stable and efficient transmission of swarm sensor data; therefore, many existing technologies focus on proposing different routing protocols for different targets. From the perspective of network topology, these protocols can be divided into two categories: planar routing protocols and hierarchical routing protocols. However, existing hierarchical routing protocols cannot maximize channel utilization and achieve optimal throughput. Summary of the Invention

[0004] To address the problems existing in the aforementioned traditional technologies, this invention proposes a transmission method and system for a large-scale unmanned swarm sensor data acquisition network, which can significantly improve the end-to-end throughput of the large-scale unmanned swarm sensor data acquisition network, thereby increasing the frequency of sensor data acquisition.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions: On the one hand, a method for transmitting data in a large-scale unmanned swarm sensor data acquisition network is provided, including the following steps: With the data aggregation node as the root node and each link weight value of 1, a minimum spanning tree A is constructed for a large-scale unmanned cluster sensor data acquisition network based on the minimum spanning tree algorithm. Perform initial clustering on the minimum spanning tree A, dividing all source nodes on each branch into a separate cluster; Perform the first round of cluster merging on the clusters after initial clustering; starting from the leftmost branch, if there is no parallel transmission link between the current branch and its right branch, and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune the right branch. After traversing all branches, obtain the minimum spanning tree B that retains only the parallel transmission path. Perform a second round of cluster merging on the minimum spanning tree B. Starting from the leftmost branch, use graph coloring to calculate the throughput after merging the current branch with its right branch and the throughput of independent concurrent transmission of the two branches. If the throughput after merging is better than the throughput of independent concurrent transmission and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune the branch to obtain the minimum spanning tree C. After configuring the time slot scheduling order for data transmission by each unmanned node using the graph coloring method on the minimum spanning tree C, each unmanned node transmits sensor data in a time-division multiple access manner according to the time slot scheduling order.

[0006] In one embodiment, before constructing the minimum spanning tree A, the method further includes: Initiate a flooding protocol at the data aggregation node to query the one-hop neighbor information table of all unmanned nodes in the large-scale unmanned cluster sensor data acquisition network; the one-hop neighbor information table contains the hierarchy of each unmanned node relative to the data aggregation node and the one-hop neighbor information of the next level.

[0007] In one embodiment, the vertices in the minimum spanning tree A correspond to the unmanned nodes in the large-scale unmanned cluster sensor data acquisition network. The edges between two vertices that belong to the same branch and are directly connected are represented by solid lines, while the edges between two vertices that do not belong to the same branch but are directly connected are represented by dashed lines, indicating that there is a mutual interference relationship between the two vertices.

[0008] In one embodiment, data aggregation includes application-layer aggregation and network-layer aggregation; The application layer aggregation involves integrating and refining data collected from multiple sensors, removing redundant information, and refining it into compact data. Network layer aggregation combines small data packets collected by multiple sensors into a large data packet for transmission within the same time slot.

[0009] In one embodiment, during the first round of cluster merging, if there is at least one parallel transmission link between the current branch and the right branch, then cluster merging is not performed between the current branch and the right branch.

[0010] In one embodiment, pruning involves deleting all non-cluster head leaf nodes on the branch to be merged.

[0011] In one embodiment, the process of configuring the time slot scheduling order for data transmission from each unmanned node using a graph coloring method includes: An interference matrix is ​​generated based on the current minimum spanning tree; the element values ​​of the interference matrix indicate whether there is interference between two corresponding sending nodes. An undirected connected graph is constructed based on the interference matrix; each sending node is treated as a vertex in the undirected connected graph, and if there is interference between two sending nodes, the corresponding two vertices in the undirected connected graph are connected by an edge. Solve the graph coloring problem for an undirected connected graph. Use the minimum number of colors obtained as the minimum number of time slots required for each source node in the current minimum spanning tree to send data to the data aggregation node, and obtain the time slot scheduling order of each sending node.

[0012] In one embodiment, the maximum message length that a cluster head node can transmit in each timeslot is equal to the product of the timeslot length and the transmission rate.

[0013] On the other hand, a transmission system for a large-scale unmanned cluster sensor data acquisition network is provided, including: Multiple unmanned nodes, each equipped with sensor data acquisition devices to periodically collect sensor monitoring data; Data aggregation nodes are used to receive sensor data from unmanned nodes transmitted via a multi-hop wireless network; The data aggregation node is configured with itself as the root node and all link weights are 1. A minimum spanning tree A is constructed for the large-scale unmanned swarm sensor data acquisition network based on the minimum spanning tree algorithm. Initial clustering is performed on minimum spanning tree A, dividing all source nodes on each branch into a cluster. The first round of cluster merging is performed on the clusters after the initial clustering. Starting from the leftmost branch, if there are no parallel transmission links between the current branch and its right-hand branch, and the cluster head node of the current branch can aggregate all the data generated by the right-hand branch, then the cluster of the right-hand branch is merged into the current branch, and pruning is performed on the right-hand branch. After traversing all branches, only parallel transmission paths are retained. Find the minimum spanning tree B; perform a second round of cluster merging on the minimum spanning tree B; starting from the leftmost branch, use graph coloring to calculate the throughput of the current branch after merging with its right branch and the throughput of the two branches transmitting independently and concurrently. If the throughput after merging is better than the throughput of independent concurrent transmission and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune it to obtain the minimum spanning tree C; after configuring the time slot scheduling order of data transmission by each unmanned node on the minimum spanning tree C using graph coloring, each unmanned node transmits sensor data in time-division multiple access mode according to the time slot scheduling order.

[0014] In one embodiment, before constructing the minimum spanning tree A, a flooding protocol is initiated at the data aggregation node to query the one-hop neighbor information table of all unmanned nodes in the large-scale unmanned cluster sensor data acquisition network; the one-hop neighbor information table includes the hierarchy of each unmanned node relative to the data aggregation node and the next level.

[0015] One of the above technical solutions has the following advantages and beneficial effects: The aforementioned transmission method and system for a large-scale unmanned swarm sensor data acquisition network utilizes a minimum spanning tree structure jointly constructed by initial clustering, first-round merging, and second-round merging. This structure is precisely mapped to a specific concurrent time slot scheduling scheme using a graph coloring method. The concurrent transmission links within each time slot fully leverage the potential of previously retained parallel transmission paths. Specifically, initial clustering achieves intra-cluster data aggregation, reducing the number of transmissions; the first-round merging eliminates unnecessary redundant paths, reducing network complexity; the second-round merging ensures that the retained parallel paths outperform the merging scheme in terms of throughput; and the final graph coloring time slot scheduling maps these optimized paths to specific concurrent transmission timetables, achieving cross-layer joint optimization from network layer routing design to link layer time slot scheduling. By comprehensively employing data aggregation and multi-path concurrent data transmission, the end-to-end throughput of the wireless multi-hop network can be significantly improved when large-scale unmanned swarms perform collaborative sensing and data acquisition tasks, thereby enhancing the acquisition frequency and real-time transmission performance of sensor data from each unmanned node within the swarm. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram of a large-scale unmanned swarm sensor data acquisition network; Figure 2 This is a flowchart illustrating the transmission method of a large-scale unmanned cluster sensor data acquisition network in one embodiment; Figure 3 This is a schematic diagram of the minimum spanning tree A in one embodiment; Figure 4 This is a schematic diagram of the minimum spanning tree B in one embodiment; Figure 5 This is an example of an interference matrix obtained from the minimum spanning tree B. Figure 6 This is an example of an undirected connected graph obtained from an interference matrix. Figure 7 This is a flowchart of the overall algorithm for improving the throughput of an unmanned cluster sensor network using data aggregation and concurrent transmission in one embodiment. Figure 8 This is a Gantt chart illustrating a 4-slot concurrent transmission scheduling scheme for a large-scale unmanned cluster sensor data acquisition network transmission method in one embodiment. Figure 9 This is a performance comparison chart of a transmission method for a large-scale unmanned swarm sensor data acquisition network in one embodiment and an existing solution; wherein, Figure 9 (a) Comparison of the number of time slots required to complete the aggregation of data from all 5 source nodes. Figure 9 (b) is a comparison chart of normalized throughput; Figure 10 This is a schematic diagram of the transmission system architecture of a large-scale unmanned cluster sensor data acquisition network in one embodiment. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention.

[0019] It should be noted that, in this document, the reference to "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The presentation of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will understand that the embodiments described herein can be combined with other embodiments. The term "and / or" as used herein refers to any combination of one or more of the associated listed items, and all possible combinations, including such combinations.

[0020] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0021] Figure 1 The diagram illustrates a typical large-scale unmanned swarm sensor data acquisition network. This network comprises five unmanned nodes (such as small drones) from S1 to S5, each equipped with sensor acquisition devices. Because each unmanned node is located far from the data aggregation node (N10, which can be situated at a ground station or a long-endurance airborne platform), the sensor monitoring data periodically collected by each node needs to be transmitted through multiple intermediate nodes via multi-hop forwarding to reach the data aggregation node. The data aggregation node then transmits the data back to the backend data processing platform via broadband networks such as the Internet, mobile networks, and local area networks.

[0022] In the above process, wireless multi-hop routing from the source node to the data aggregation node is crucial for the stable and efficient transmission of cluster sensor data. In planar routing protocols, all network nodes are equal, with no hierarchical differences. Each source node independently finds and establishes a route to the data aggregation node. The advantages of planar routing are its simplicity and scalability; it requires no network topology maintenance, is less prone to bottlenecks, and therefore has good robustness. The disadvantage of planar routing is the lack of a management node in the network, resulting in a lack of orderly management of communication resources and an inability to fully and optimally utilize limited communication resources.

[0023] In hierarchical routing protocols, networks are typically divided into clusters, each containing a cluster head node and several cluster member nodes. The cluster head node is determined by a specific mechanism and coordinates data transmission among its member nodes. Under the unified coordination of the cluster head node, each member node only needs to send its received sensor data to the cluster head node, which then aggregates this data and sends it directly to the data aggregation node or forwards it through multiple cluster head nodes to the data aggregation node. The earliest proposed sensor network clustering algorithm was LEACH, and its basic idea permeates many subsequent clustering routing protocols. Compared to planar routing protocols, a significant advantage of hierarchical routing protocols is that by aggregating data from multiple sensors before transmission, they can effectively reduce contention for shared wireless resources during independent transmission, thereby improving the utilization of the wireless channel and achieving efficient data collection.

[0024] like Figure 1 As shown, based on the basic idea of ​​hierarchical routing protocols, N2 is elected as the cluster head node according to the principle of maximizing the number of nodes with one-hop connections. In this case, cluster head node N2 coordinates nodes N1, N3, N4, and N6 to send their respective sensor data to cluster head node N2 in time slots 1 to 4 in TDMA (Time Division Multiple Access). Cluster head node N2 aggregates these data along with its own collected data into one data packet and sends it to node N5 in time slot 5. Node N5 forwards the packet to node N8 in time slot 6. Finally, node N8 sends the sensor data of all five nodes to data aggregation node N10 in time slot 7.

[0025] If a planar routing protocol is used instead, nodes N1, N2, N3, N4, and N6 must each forward their data to the data aggregation node N10 independently. During this process, the transmission of these five data packets will compete with each other, requiring much more than seven time slots to complete the transmission of all data. Therefore, the transmission efficiency is much lower than that of the hierarchical routing protocol.

[0026] The above example illustrates that for special multi-hop wireless networks such as unmanned swarm sensor data acquisition networks, which have numerous source nodes, a single destination node, and highly aggregated data traffic, the method of using a hierarchical routing protocol, where the cluster head node collects data from multiple source nodes and aggregates it into a single packet for transmission, can greatly reduce mutual interference and competition caused by the independent and disordered transmission of multiple data packets compared to a planar routing protocol. Therefore, it is more suitable for this type of system.

[0027] However, research has found that existing hierarchical routing protocols cannot maximize channel utilization to achieve optimal throughput because: (1) The cluster head election rules of existing hierarchical routing protocols are determined independently by each node, and the cluster head election mechanism is not designed from a global perspective of maximizing channel utilization. For example, the LEACH protocol stipulates that each node randomly selects a value between 0 and 1. If the selected value is less than a certain threshold, then this node becomes the cluster head node. Obviously, the cluster head node elected by this rule cannot guarantee that the channel utilization will be optimal, and may even be very poor. (2) The cluster head nodes in existing hierarchical routing protocols are not fixed, but rotated in time. This means that even if a node that initially achieves optimal or good channel utilization becomes the cluster head, the new cluster head node after rotation still cannot guarantee high channel utilization. (3) Existing hierarchical routing protocols do not design multi-hop routes between cluster head nodes and data aggregation nodes from the perspective of maximizing channel utilization. For example, the LEACH protocol simply assumes that each cluster head node can send data directly to the data aggregation node in one hop without the need for other nodes to forward it. Some subsequent hierarchical routing protocols, even if they consider the case of multi-hop forwarding between cluster heads, mostly aim to reduce the transmit power of cluster head nodes. A few of them also take into account the channel utilization index, but they only consider this problem from the coarse-grained routing level, without precise quantitative design from the fine-grained time slot scheduling level. Therefore, they cannot obtain the optimal channel utilization.

[0028] In one embodiment, such as Figure 2 As shown, a method for transmitting data in a large-scale unmanned swarm sensor data acquisition network is provided, which may include the following processing steps S10 to S18: S10, with the data aggregation node as the root node and each link weight value being 1, constructs the minimum spanning tree A for the large-scale unmanned cluster sensor data acquisition network based on the minimum spanning tree algorithm.

[0029] Understandable. For ease of explanation, we can continue with... Figure 1 For example, the weight values ​​of the links between each unmanned node are all set to 1. Then, based on the minimum spanning tree algorithm, a tree-like route from the data aggregation node N10 to each source node is constructed with the data aggregation node N10 as the root node. The constructed tree-like route is denoted as the minimum spanning tree A. Setting the weight value of each link to 1 ensures that the minimum spanning tree algorithm constructs the shortest path tree from the data aggregation node to each source node with the minimum hop count as the optimization objective.

[0030] S12, perform initial clustering on the minimum spanning tree A, and divide all source nodes on each branch into a cluster; wherein, the source node with the highest level on the branch is the cluster head node, and the cluster head node aggregates the sensor data of each cluster member node in its own cluster along with the data it collects into a data packet.

[0031] This can be understood as performing initial clustering on the minimum spanning tree A. The principle of clustering is to group all source nodes on each branch into one cluster and determine the highest-level source node on that branch as the cluster head node. For example... Figure 3 As shown, the routes from data aggregation node N10 to nodes N4 and N1 are on the same branch, the route from data aggregation node N10 to node N2 is on the second branch, and the routes from data aggregation node N10 to nodes N6 and N3 are on the third branch. Therefore, nodes N1 and N4 are divided into cluster 1, with node N4 as the cluster head node; node N2 is divided into cluster 2, with node N2 as the cluster head node; and nodes N3 and N6 are divided into cluster 3, with node N6 as the cluster head node.

[0032] Each cluster member node in a cluster only needs to send its collected data to the cluster head node to upload it to the data aggregation node. The cluster head node aggregates the data from all cluster member nodes along with its own data into a single data packet, and then uploads it to the data aggregation node along a tree-structured route. Data aggregation can greatly improve data transmission efficiency. Taking the transmission between nodes N1 and N4 as an example, without data aggregation, node N1 needs 3 hops to upload its own data to the data aggregation node N10, and node N4 needs 2 hops, for a total of 5 hops. However, with aggregation, node N1 only needs to send its own data to node N4 in 1 hop, and node N4 aggregates node N1's data with its own data, then sends it to the data aggregation node N10 in 2 hops, for a total of only 3 hops, improving channel utilization by 40%.

[0033] The initial clustering steps described above use the branch structure of the minimum spanning tree as the natural grouping basis, giving nodes within each cluster a natural aggregation advantage in the routing topology. This means that the transmission path from cluster member nodes to the cluster head node is short and branch interference can be eliminated. Selecting the highest-level source node as the cluster head places the cluster head node at the downstream end of the data convergence path, maximizing aggregation efficiency and laying the foundation for subsequent inter-cluster concurrent transmission and global throughput optimization.

[0034] S14, perform the first round of cluster merging on the clusters after the initial clustering; starting from the leftmost branch, if there is no parallel transmission link between the current branch and its right branch, and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune the right branch. After traversing all branches, obtain the minimum spanning tree B that retains only the parallel transmission path.

[0035] It is understandable that the first round of cluster merging is performed on the clusters after the initial clustering. Starting from the leftmost branch i (i=1) in the minimum spanning tree A, it is checked whether the current branch i has a parallel transmission link with the branch to its right. If there is no parallel transmission link, it is further checked whether the cluster head of the current branch i can aggregate all the data generated by the branch to its right; if the cluster head of the current branch i can aggregate all the data generated by the branch to its right, the cluster of the branch to its right is merged into the current branch i. After merging, all non-cluster head leaf nodes on the branch to its right are deleted, i.e., a pruning operation is performed. If there is at least one parallel transmission link between the current branch i and the branch to its right, the cluster merging of the two branches is not performed. After the merging judgment of the current branch i is completed, the next branch to its right i+1 is further examined, and the above process is repeated until the cluster merging of all branches is completed. At this time, the original minimum spanning tree A is pruned into a minimum spanning tree B with only parallel transmission paths.

[0036] In actual operation, the data aggregation node is determined based on the edges between the vertices of the minimum spanning tree A and the mutual interference relationships between them. For example... Figure 3 As shown, there are no parallel transmission links between the leftmost branch N10-N7-N4-N1 and its rightmost branch N10-N8-N5-N2. For example, when node N1 transmits data to node N4, any transmissions from nodes N2, N5, and N8 will interfere with node N4's reception. Therefore, the clusters on branch N10-N8-N5-N2 can be merged into the cluster on branch N10-N7-N4-N1, that is, nodes N2, N1, and N4 are merged into one cluster, with node N4 as the cluster head node. After merging, the other non-cluster head leaf nodes N5 and N8 on branch N10-N8-N5-N2 are deleted, thus completing the pruning operation for this branch. Next, we examine the next branch N10-N9-N6-N3. Since the transmission from node N1 to node N4 does not interfere with the transmissions from node N3 to node N6, node N6 to node N9, or node N9 to node N10 (i.e., there are parallel transmission links), we do not merge these two branches, nor do we perform any pruning operations. The final minimum spanning tree B is as follows: Figure 4 As shown.

[0037] Thus, if no parallel transmission links exist and aggregation capability allows, the two clusters are merged to eliminate unnecessary redundant transmission nodes, reducing forwarding hops and slot overhead. If parallel transmission links exist, the independent clusters are retained to provide conditions for subsequent concurrent transmission. This mechanism based on the existence of parallel transmission links ensures that the minimum spanning tree B obtained after the first round of merging retains only the path structure with parallel transmission capability. Merging non-parallel links in the network into a single branch reduces invalid independent transmission contention, while retaining links capable of parallel transmission to create conditions for multi-path concurrency. By merging branches without parallel capabilities, the overall number of slots required is effectively reduced, which, together with the subsequent second round of merging and slot scheduling, significantly improves network throughput.

[0038] S16, perform the second round of cluster merging on the minimum spanning tree B; starting from the leftmost branch, use graph coloring to calculate the throughput of the current branch after merging with its right branch and the throughput of the two branches independently and concurrently. If the throughput after merging is better than the throughput of independent concurrent transmission and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune the branch to obtain the minimum spanning tree C.

[0039] Understandably, the minimum spanning tree B is checked to see if further cluster merging can be performed. Starting from the leftmost branch j (j=1) in the minimum spanning tree B, the graph coloring method is used to check whether the throughput of the current branch j merged with its right-hand branch is better than the throughput of the two branches transmitting independently and concurrently. If the throughput after merging is better than the throughput of independent concurrent transmission, then it is further checked whether the cluster head of the current branch j can aggregate all the data generated by the right-hand branch; if the cluster head of the current branch j can aggregate all the data generated by the right-hand branch, then the clusters of the right-hand branch are merged into the current branch j, and pruning is performed after merging, that is, all non-cluster head leaf nodes on the right-hand branch are deleted. If the cluster head of the current branch j cannot aggregate all the data generated by the right-hand branch, then the cluster merging of the two branches is not performed. After completing the merging judgment of the current branch j, the next right-hand branch j+1 is checked, and the above process is repeated until the cluster merging of all branches is completed. At this time, the original minimum spanning tree B is pruned into minimum spanning tree C.

[0040] by Figure 4 Taking the minimum spanning tree B as an example, the throughput of independent concurrent transmission between two branches is first calculated using graph coloring. Specifically, the operation involves generating a graph from the minimum spanning tree B as shown below. Figure 5 The interference matrix shown indicates whether interference exists between two corresponding transmitting nodes; a value of 1 indicates the presence of interference. Secondly, based on the interference matrix, further results are obtained as follows: Figure 6The undirected connected graph shown treats each transmitting node in the interference matrix as a vertex. If the interference value between any two different transmitting nodes in the interference matrix is ​​1, then the corresponding two vertices in the undirected connected graph can be connected by an edge, indicating that these two vertices are adjacent. Third, regarding... Figure 6 The undirected connected graph shown can be solved using existing methods such as backtracking or heuristics to address the graph coloring problem. The minimum number of colors required is found to be 4, meaning the minimum number of time slots required for the five source nodes in the minimum spanning tree B to send their collected data to the data aggregation node is 4, resulting in a normalized throughput of 1.25. This normalized throughput is calculated by dividing the number of source nodes by the minimum number of time slots (i.e., 5 / 4 = 1.25). One possible time slot allocation scheme with a minimum of 4 time slots is as follows: Time slot 1, nodes N1 and N6 send data simultaneously; Time slot 2, nodes N2 and N6 send data simultaneously; Time slot 3, nodes N4 and N9 send data simultaneously; Time slot 4, node N7 sends data.

[0041] Next, the same graph coloring method can be used to calculate the normalized throughput after merging the two branches, yielding a minimum of 5 time slots and a normalized throughput of 1.0. Based on the above analysis, for... Figure 4 The minimum spanning tree B shown has a throughput of 1.0 after merging two branches, which is lower than the throughput of 1.25 when the two branches are transmitted independently and concurrently. Therefore, in this embodiment, the two branches do not need to be merged, and the final minimum spanning tree C is actually the same as the minimum spanning tree B.

[0042] S18. After configuring the time slot scheduling order for data transmission by each unmanned node using the graph coloring method on the minimum spanning tree C, each unmanned node transmits sensor data in a time-division multiple access manner according to the time slot scheduling order.

[0043] It is understandable that, for the minimum spanning tree C, the graph coloring method is used to configure the time slot scheduling order for data transmission from each unmanned node in the minimum spanning tree C. Specifically, in the case where the minimum spanning tree C is the same as the minimum spanning tree B, the time slot allocation scheme for the minimum spanning tree C is also the same as that for the minimum spanning tree B.

[0044] For example, the specific time slot scheduling order can be as follows: In time slot 1, node N1 sends its sensor data to node N4 (N1→N4), while node N3 sends its sensor data to node N6 (N3→N6); In time slot 2, node N2 sends its sensor data to node N4 (N2→N4), while node N6 aggregates its sensor data and the sensor data received from node N3 in time slot 1 into a single data packet and sends it to node N9 (N6→N9); In time slot 3, node N4 aggregates its sensor data, the sensor data received from node N1 in time slot 1, and the sensor data received from node N2 in time slot 2 into a single data packet and sends it to node N7 (N4→N7), while node N9 sends the sensor data received from node N6 in time slot 2 to node N10 (N9→N10); In time slot 4, node N7 sends the sensor data received from node N4 in time slot 3 to node N10 (N7→N10).

[0045] It should be noted that in the above time slot scheduling process, node N6 needs to complete two transmissions in two time slots: one transmission is the raw data collected by its own sensors, and the other transmission is the cluster data received from node N3 and aggregated. (These are node N6's own data and the aggregated cluster data, respectively). Therefore, they do not belong to the repeated transmission of the same data stream, nor are they subject to the restriction of "the same node cannot transmit simultaneously" in the interference matrix. That is, the two transmissions belong to different time slots, but they just happen to be transmitted by node N6. In the interference matrix modeling and graph coloring solution, they are treated as two independent transmission events.

[0046] In summary, by employing concurrent transmission across multiple links on top of data aggregation, all five sensor data points can be sent to the data aggregation node N10 in just four time slots, resulting in a normalized network throughput of 5 / 4 = 1.25 packets / time slot. In contrast, the existing clustering routing protocol, which only considers data aggregation, has a normalized network throughput of 5 / 7 ≈ 0.714 packets / time slot. The former represents a 75% improvement over the latter, demonstrating that simultaneously employing data aggregation and concurrent multi-route transmission can significantly enhance channel utilization.

[0047] The transmission method of the aforementioned large-scale unmanned swarm sensor data acquisition network, constructed through initial clustering, first-round merging, and second-round merging, forms a minimum spanning tree structure. This structure is then precisely mapped to a specific concurrent time slot scheduling scheme using a graph coloring method. The concurrent transmission links within each time slot fully utilize the potential of previously retained parallel transmission paths. Specifically, initial clustering achieves intra-cluster data aggregation to reduce the number of transmissions; the first-round merging eliminates unnecessary redundant paths, reducing network complexity; the second-round merging ensures that the retained parallel paths outperform the merging scheme in terms of throughput; and the final graph coloring time slot scheduling maps these optimized paths to specific concurrent transmission timetables, achieving cross-layer joint optimization from network layer routing design to link layer time slot scheduling.

[0048] In one embodiment, such as Figure 7 As shown, before constructing the minimum spanning tree A, the following processing can also be included: Initiate a flooding protocol at the data aggregation node to query the one-hop neighbor information table of all unmanned nodes in the large-scale unmanned cluster sensor data acquisition network; the one-hop neighbor information table contains the hierarchy of each unmanned node relative to the data aggregation node and the one-hop neighbor information of the next level.

[0049] It is understandable that the purpose of initiating the flooding protocol at the data aggregation node is to obtain complete topology information of the entire large-scale unmanned swarm sensor data acquisition network. The basic working process of the flooding protocol is as follows: the data aggregation node broadcasts a query message to all neighboring nodes within its one-hop range; the node receiving the query message records the source node information of the message and appends its own identifier and known neighboring node information to the message and continues to broadcast it to its neighboring nodes; through hop-by-hop diffusion, the data aggregation node is eventually able to collect the hierarchical information and adjacency relationships of all nodes in the network.

[0050] The one-hop neighbor information table can contain two key pieces of information: first, the hierarchy of each unmanned node relative to the data aggregation node. This hierarchy represents the minimum number of hops required to reach the node from the data aggregation node, and provides a distance metric for the subsequent execution of the minimum spanning tree algorithm; second, the one-hop neighbor information of the next level down for each unmanned node, i.e., a list of neighboring nodes that are directly adjacent to the node and whose hierarchy is one level higher than the node. This information provides connectivity data for constructing the complete topology. The hierarchy of each unmanned node and its next-level one-hop neighbor information obtained by the data aggregation node using the flooding protocol are shown in Table 1. Table 1. One-hop neighbor information table maintained by the data aggregation node.

[0051] The global topology information obtained by the aforementioned flooding protocol enables data aggregation nodes to have a global network perspective. This is a prerequisite for optimizing cluster head election and routing design from a global perspective, and it solves the technical problem in existing technologies that lack global topology awareness, which leads to the inability to globally optimize throughput. At the same time, the hierarchical information in the one-hop neighbor information table directly provides input for setting the link weight to 1 and constructing the minimum spanning tree. The neighbor relationships in the one-hop neighbor information table will also play a role in the subsequent generation of the interference matrix.

[0052] In one embodiment, the vertices in the minimum spanning tree A correspond to unmanned nodes in a large-scale unmanned cluster sensor data acquisition network. A solid line connects two vertices that belong to the same branch and are directly connected, representing the edge between the two vertices. A dashed line connects two vertices that do not belong to the same branch but are directly connected, representing the mutual interference relationship between the two vertices.

[0053] It can be understood that in the minimum spanning tree A, each vertex corresponds to a wireless node in a large-scale unmanned swarm sensor data acquisition network. The construction result of the minimum spanning tree A includes two representations of connection relationships: solid line connections and dashed line connections. Two vertices belonging to the same branch and directly connected are connected by a solid line, indicating that the line between these two vertices constitutes an edge of the minimum spanning tree A, that is, the actual transmission path of data in the tree-shaped routing. Solid line connections determine the routing direction from each source node to the data aggregation node and the hop-by-hop relationship of data transmission.

[0054] Nodes that are not on the same branch but have a direct wireless connection are represented by dashed lines. This graphical representation shows the wireless interference between nodes on two different branches. Due to the broadcast nature of wireless communication, when a node is transmitting data, its wireless signal may be received by other nodes not on the same routing path. If this reception occurs simultaneously with the reception of the node itself, it will cause reception interference.

[0055] In existing technologies, routing protocols typically only concern themselves with the routing path itself, leaving interference management to the link layer independently, resulting in a lack of information sharing between the two. This embodiment achieves structured fusion storage of routing and interference information by using a dual-line representation—solid lines to record routing topology and dashed lines to record interference relationships—allowing subsequent cluster merging and time slot scheduling steps to utilize both routing and interference information simultaneously. For example, in the first round of cluster merging, the presence or absence of dashed lines can be used to determine whether parallel transmission links exist between different branches: the presence of dashed lines indicates interference, preventing parallel transmission; the absence of dashed lines indicates no interference, allowing parallel transmission.

[0056] In one embodiment, data aggregation includes application-layer aggregation and network-layer aggregation. Application-layer aggregation integrates and refines data collected from multiple sensors, removing redundant information and refining it into compact data. Network-layer aggregation merges small packets of data collected from multiple sensors into a large packet for transmission within the same time slot.

[0057] Data aggregation can be understood to be implemented in two ways: application-layer aggregation and network-layer aggregation. The basic concept of application-layer aggregation is to integrate, process, and refine data collected from multiple sensors, removing redundant information and refining it into more compact data. For example, in a monitoring cluster consisting of multiple drones equipped with infrared detectors, the average of the infrared detection data from multiple drones can be fused before transmission, thus compressing multiple packets of similar temperature and humidity data into a single packet of statistical results, significantly reducing the total amount of data that needs to be transmitted. The basic concept of network-layer aggregation is to merge small packets of data collected from multiple sensors into a large packet and transmit it within the same time slot. Both application-layer and network-layer aggregation have upper limits on their aggregation capabilities: application-layer aggregation generally cannot aggregate data from different types of sensors, while network-layer aggregation is limited by the time slot length and transmission rate. For example, if the time slot length is 1ms and the transmission rate is 100kbps, then the maximum message length that can be transmitted in each time slot is 100 bits, which is equal to the product of the time slot length and the transmission rate. If each sensor collects data at a frequency of 20 bits / ms, the maximum aggregation capability of the cluster head node is 5 sensors, meaning it can aggregate data from a maximum of 5 sensors into one packet. In this embodiment, it is assumed that all source nodes on each branch of the minimum spanning tree A can be divided into a cluster, meaning the cluster head node of that cluster can combine application-layer aggregation and network-layer aggregation to aggregate all sensor data from its branch and send it together.

[0058] It should be noted that in multi-sensor collaborative data acquisition scenarios, data collected by different sensors may be highly redundant, especially monitoring data of the same type of sensors for the same target area. Therefore, application-layer aggregation targets the semantic level of sensor-collected data. In multi-hop wireless networks, each node's data packets have a fixed header overhead. If each sensor's small data packets are sent independently, each data packet will generate its own header overhead, resulting in a waste of channel resources. Therefore, network-layer aggregation targets the packet structure level of data transmission.

[0059] In some practical applications, sensor data can first undergo semantic compression through application-layer aggregation to remove inherent redundancy; then, network-layer aggregation merges the compressed multi-packet data into a single large packet for transmission over the channel. Application-layer aggregation improves efficiency from the data semantic dimension, while network-layer aggregation improves efficiency from the channel transmission dimension; the combined effect of both is far greater than either single aggregation method. This embodiment addresses the problem in existing technologies where data aggregation is only mentioned in general terms without distinguishing specific implementation methods, leading to insufficient guidance for actual system design. The clear division and coordinated use of the aforementioned two aggregation methods provide system designers with a clear implementation path, enabling data aggregation efficiency to reach its theoretical optimum.

[0060] In one embodiment, during the first round of cluster merging, if there is at least one parallel transmission link between the current branch and the right branch, then cluster merging is not performed between the current branch and the right branch.

[0061] It is understandable that the processing in this embodiment is because parallel transmission links are the fundamental resource for subsequent implementation of multi-path concurrent transmission and should not be merged or eliminated without quantitative analysis. The significance of parallel transmission links is that they can carry two or more non-interfering data transmissions within the same time slot. If two branches with parallel transmission links are merged into a cluster, although the number of clusters can be reduced, the concurrent transmission opportunities that could have been utilized will also be eliminated, which may lead to a decrease in overall network throughput instead of an increase.

[0062] Therefore, the design principle of the first round of cluster merging is to only merge branches that are determined to have no parallel transmission links, while retaining the independent structure of branches with parallel transmission links and leaving them unprocessed. This conservative strategy means that the first round of merging is responsible for qualitatively eliminating redundant structures without parallel transmission links, while the second round of merging is responsible for quantitatively comparing and deciding on the remaining parallel transmission link structures. This division of labor between the two rounds of merging allows branches that cannot have parallel transmission to be quickly merged and simplified, while branches that may have parallel transmission are reserved for more precise graph coloring judgment, thus achieving an optimal balance between algorithm efficiency and optimization accuracy. By delegating the decision-making power to subsequent more precise quantitative analysis steps, the reliability of the optimization results is guaranteed.

[0063] In one embodiment, pruning involves deleting all non-cluster head leaf nodes on the branch to be merged.

[0064] As can be understood, pruning refers to the operation of deleting all non-cluster head leaf nodes on the branch to be merged. During the cluster merging process, when the clusters on the right branch are merged into the left branch, the other leaf nodes on the right branch, except for the cluster head node, no longer have data forwarding functions. The routing tasks they originally carried have been taken over by the merged cluster head node. Therefore, these leaf nodes become redundant nodes and need to be removed from the tree structure through pruning operations.

[0065] The purpose of pruning is to simplify the network structure and reduce the computational complexity of subsequent steps. For example, in the first round of cluster merging, after the branch N10-N8-N5-N2 is merged into the branch N10-N7-N4-N1, nodes N5 and N8, as non-cluster head leaf nodes, are pruned and deleted. The merged tree structure only retains the aggregated links formed by cluster head nodes N2, N1, and N4. This simplification allows the subsequent second round of merging and graph coloring to face a simpler topology, reducing the dimension of the interference matrix and lowering computational overhead.

[0066] The cluster merging process determines which node functions are integrated, while pruning cleans up the structure of nodes that become redundant after integration. The combination of merging and pruning ensures that the minimum spanning tree not only becomes functionally reasonable during the optimization process but also remains structurally concise, providing a simple input structure for the final efficient time slot scheduling.

[0067] In one embodiment, the process of configuring the time slot scheduling order for data transmission from each unmanned node using a graph coloring method may include the following steps: An interference matrix is ​​generated based on the current minimum spanning tree; the element values ​​of the interference matrix indicate whether there is interference between two corresponding sending nodes. An undirected connected graph is constructed based on the interference matrix; each sending node is treated as a vertex in the undirected connected graph, and if there is interference between two sending nodes, the corresponding two vertices in the undirected connected graph are connected by an edge. Solve the graph coloring problem for an undirected connected graph. Use the minimum number of colors obtained as the minimum number of time slots required for each source node in the current minimum spanning tree to send data to the data aggregation node, and obtain the time slot scheduling order of each sending node.

[0068] It is understandable that the process of using graph coloring for time slot scheduling configuration can include three closely linked sub-operations: generating an interference matrix, constructing an undirected connected graph, and solving the graph coloring problem.

[0069] The first sub-operation is to generate an interference matrix based on the current minimum spanning tree. The interference matrix is ​​a square matrix with the number of rows and columns equal to the number of all sending nodes in the network. Each element in the interference matrix indicates whether there is interference between two corresponding sending nodes. Specifically, if the element in the i-th row and j-th column is 1, it indicates that there is mutual interference between the i-th and j-th sending nodes, meaning these two nodes cannot send data in the same time slot; if the element is 0, it indicates that there is no interference between the two sending nodes, and they can transmit concurrently in the same time slot.

[0070] The interference matrix is ​​generated based on the wireless interference relationships between nodes in the network: when two transmitting nodes transmit data simultaneously, if the signal transmitted by one transmitting node interferes with the target receiving node of the other transmitting node, preventing the receiving node from correctly decoding the data, then there is interference between the two transmitting nodes. This judgment process utilizes the node hierarchy and adjacency relationships recorded in the minimum spanning tree, as well as the potential interference relationships between nodes on different branches marked by dashed lines.

[0071] The second sub-operation constructs an undirected connected graph based on the interference matrix. Each sending node in the interference matrix is ​​treated as a vertex in the undirected connected graph. If the interference value between any two different sending nodes in the interference matrix is ​​1, then the corresponding two vertices in the undirected connected graph are connected by an edge, indicating that these two vertices are adjacent. This results in an undirected connected graph that fully reflects the interference relationships between all sending nodes.

[0072] The third sub-operation is solving the graph coloring problem for an undirected connected graph. The goal of graph coloring is to color all vertices in an undirected connected graph using the minimum number of colors, and no adjacent vertices can be assigned the same color. In the context of time slot scheduling, each color corresponds to a transmission time slot, and adjacent vertices cannot have the same color, which is equivalent to two transmitting nodes that are interfering with each other not being able to send data in the same time slot. The number of colors obtained is the minimum number of time slots required to complete the data transmission of all source nodes, and the set of vertices corresponding to each color is the set of nodes that can transmit concurrently within that time slot.

[0073] Solving the graph coloring problem can be achieved using existing mature methods such as backtracking algorithms or heuristic algorithms. For example, in this embodiment, the minimum number of colors obtained can be 4, that is, the minimum number of time slots is 4. A feasible time slot allocation scheme could be as follows: Time slot 1 allocates color 1 to nodes N1 and N6, time slot 2 allocates color 2 to nodes N2 and N6, time slot 3 allocates color 3 to nodes N4 and N9, and time slot 4 allocates color 4 to node N7. The specific time slot scheduling order is as follows: Time slot 1: Node N1 and Node N6 transmit simultaneously; Time slot 2: Node N2 and Node N6 transmit simultaneously; Time slot 3: Nodes N4 and N9 transmit simultaneously; Time slot 4: Node N7 sends.

[0074] Thus, by formalizing the interference constraints of the wireless network through the interference matrix, transforming time slot scheduling into a standard graph coloring problem through an undirected connected graph, and accurately obtaining the minimum number of time slots and the optimal time slot allocation scheme by solving the graph coloring problem, the three sub-operations are interconnected and progressively advance, realizing a complete, accurate, and quantifiable technical closed loop from the network physical layer interference relationship to the generation of scheduling schemes.

[0075] In one embodiment, the maximum message length that a cluster head node can transmit in each timeslot is equal to the product of the timeslot length and the transmission rate.

[0076] It is understandable that the data aggregation capability of a cluster head node is limited by two physical parameters: time slot length and transmission rate. In a time-division multiple access (TDMA) system, each node has a fixed duration within a time slot for transmitting data; this duration is called the time slot length. Within the same time slot, the node's wireless transmission module operates at a certain transmission rate. The maximum message length that a cluster head node can transmit within each time slot is equal to the product of the time slot length and the transmission rate. For example, if the time slot length is 1ms and the transmission rate is 100kbps, then the maximum message length that can be transmitted in each time slot is 100 bits, which is obtained by multiplying 1ms by 100kbps.

[0077] Furthermore, the maximum number of sensors that a cluster head node can aggregate is determined by both the maximum message length and the data acquisition rate of a single sensor. For example, if each sensor acquires data at a rate of 20 bits / ms, and the maximum message transmission capacity per time slot is 100 bits, then the maximum aggregation capacity of a cluster head node is 5 sensors. That is, the maximum amount of aggregated data that a cluster head node can transmit in one time slot is 100 bits, equivalent to 5 sensors each acquiring a total of 20 bits of data within 1 ms. Therefore, it can aggregate the data from a maximum of 5 sensors into one packet for transmission.

[0078] Therefore, before performing cluster merging, it is necessary to check whether the cluster head node of the current branch can aggregate all sensor data generated by the branches to be merged; that is, whether the total number of sensors to be merged is within the aggregation capacity limit of the cluster head node. If it exceeds the limit, merging cannot be performed. This constraint ensures the practical feasibility of the method, so that throughput optimization does not come at the expense of data integrity. This technique works closely with the cluster merging step to ensure the practicality and correctness of the optimization results.

[0079] In some implementations, experimental examples are also provided: The 4-time-slot concurrent transmission scheduling scheme (Gantt chart) of the above-mentioned large-scale unmanned swarm sensor data acquisition network transmission method can be described as follows: Figure 8 As shown, Figure 8 Each horizontal bar represents the transmission activity of one link. Multiple links transmit simultaneously within the same time slot. In time slot 1, N1→N4 and N3→N6 transmit concurrently. In time slot 2, N2→N4 and N6→N9 transmit concurrently. In time slot 3, N4→N7 and N9→N10 transmit concurrently. In time slot 4, N7→N10 transmits separately.

[0080] Data from the five source nodes (N1, N2, N3, N4, and N6) all arrived at the data aggregation node N10 within four time slots, verifying the effectiveness of the scheduling scheme of the present invention. The Gantt chart intuitively reflects the core design concept of data aggregation and concurrent transmission of the present invention. By using the graph coloring method, mutually non-interfering links are allocated to the same time slot, realizing the concurrent transmission of multiple links in the same time slot, thereby maximizing channel utilization.

[0081] The performance comparison chart between the above-mentioned transmission method of large-scale unmanned swarm sensor data acquisition network (hereinafter referred to as the present invention solution) and existing solutions can be seen as follows: Figure 9 As shown, Figure 9 (a) Comparison of the number of time slots required to complete the aggregation of data from all 5 source nodes: the existing scheme requires 7 time slots, while the above-mentioned scheme of the present invention only requires 4 time slots, reducing the number of time slots by 42.9%. Figure 9 (b) Comparison of normalized throughput (5 source nodes): The normalized throughput of the existing scheme is 5 / 7 ≈ 0.714 packets / time slot, while the normalized throughput of the present invention is 5 / 4 = 1.25 packets / time slot, representing a throughput improvement of 75%. This comparison verifies the significant advantages of the present invention in both time resource utilization efficiency and throughput: the reduction in the number of time slots directly means a shorter data acquisition cycle, and the increase in throughput means that more sensor data packets can be transmitted per unit time. The combined effect of these two factors significantly improves the overall data acquisition efficiency of the sensor network.

[0082] In the transmission method of the aforementioned large-scale unmanned swarm sensor data acquisition network, the minimum spanning tree pruning process can be specifically described as follows: The initial minimum spanning tree A contains 3 independent branches. Branch 1 is N10—N7—N4—N1, containing source nodes N1 and N4, with cluster head N4; Branch 2 is N10—N8—N5—N2, containing source node N2, with cluster head N2; Branch 3 is N10—N9—N6—N3, containing source nodes N3 and N6, with cluster head N6. The interference relationship between different branches can be represented by dashed lines. The first round of merging judgment process: Since there is no parallel transmission link between branch 1 and branch 2, and cluster head N4 has the ability to aggregate N2 data (the aggregation capacity is limited to 5 sensor nodes), a merging operation is performed, merging branch 2 into branch 1 and deleting redundant relay nodes N5 and N8; at the same time, there is a parallel transmission link between branch 1 and branch 3 (such as N1→N4 and N3→N6 do not interfere with each other), so two independent branches are retained. The pruned minimum spanning tree B retains only two branches: Branch 1 (after merging) is N10—N7—N4, with N4 directly connected to N1 and N2, containing source nodes N1, N2, and N4, and cluster head N4; Branch 2 (retained) is N10—N9—N6—N3, containing source nodes N3 and N6, and cluster head N6. Nodes N5 and N8 are marked as "deleted". The pruned minimum spanning tree B has a more streamlined structure, and there are parallel transmission links between the two branches, providing an optimal topology for subsequent graph coloring time slot allocation.

[0083] In summary, this invention addresses the shortcomings of existing hierarchical routing protocols for large-scale unmanned swarm ad hoc networks, which cannot adequately optimize network throughput. It proposes a novel approach that simultaneously employs data aggregation and concurrent data transmission to effectively improve network throughput. This approach considers both cluster head election and data routing from a system-wide perspective. By jointly designing end-to-end routing at the network layer and time slot scheduling at the link layer, it achieves accurate and quantifiable network throughput. This allows for quantitative guidance to increase the frequency of sensor data collection for each unmanned node, significantly improving the timeliness of environmental perception of the monitored area by the unmanned swarm.

[0084] The designed clustering mechanism and cluster head election method are designed with optimal throughput as the goal. It comprehensively employs an initial clustering mechanism based on minimum spanning trees, a first-round cluster merging mechanism based on parallel transmission links, and a second-round cluster merging mechanism based on throughput comparison, which can significantly improve network throughput. Compared with the existing technology where each unmanned node independently decides its cluster head election, this invention performs cluster head election and routing design from a global perspective of the entire network, ensuring optimal throughput.

[0085] Furthermore, it fully leverages the inherent multipath routing characteristics of large-scale unmanned swarm sensor data acquisition networks. Based on this, it proactively constructs a minimum spanning tree with parallel transmission links through pruning methods. This enables data transmission between the cluster head node and the data aggregation node, as well as between cluster member nodes and the cluster head node, to be completed concurrently as much as possible, thereby significantly improving channel utilization and network throughput. Compared to existing technologies that only roughly propose methods to improve network throughput at the routing level, this invention performs precise quantitative design at a more granular time slot scheduling level.

[0086] Given the capability for data aggregation, it's not always better to build as many clusters as possible capable of parallel transmission. In some cases, merging multiple clusters can result in higher throughput. Therefore, this invention designs a mechanism using graph coloring to determine whether merging multiple clusters is necessary. This mechanism provides a quantitative judgment on whether to merge and simultaneously offers an optimal time slot allocation scheme. By employing time-division multiple access (TDMA) to schedule the transmission order of each unmanned node, the optimal network throughput can be obtained. This mechanism has good practical operability and can guide the actual operation of large-scale unmanned swarm sensor data acquisition networks.

[0087] In one embodiment, such as Figure 10 As shown, a transmission system 100 for a large-scale unmanned cluster sensor data acquisition network is also provided, which may include: multiple unmanned nodes (such as unmanned node 1 to unmanned node M), each unmanned node is equipped with sensor acquisition equipment for periodically acquiring sensor monitoring data.

[0088] The data aggregation node receives sensor data from unmanned nodes transmitted via a multi-hop wireless network. The data aggregation node is configured with itself as the root node and all link weights set to 1. A minimum spanning tree A is constructed for the large-scale unmanned swarm sensor data acquisition network based on the minimum spanning tree algorithm. Initial clustering is performed on minimum spanning tree A, dividing all source nodes on each branch into a cluster. The first round of cluster merging is then performed on the clusters after the initial clustering. Starting from the leftmost branch, if there are no parallel transmission links between the current branch and its right-hand branch, and the cluster head node of the current branch can aggregate all the data generated by the right-hand branch, then the cluster of the right-hand branch is merged into the current branch, and pruning is performed on the right-hand branch. After traversing all branches, only parallel transmission paths are retained. Find the minimum spanning tree B; perform a second round of cluster merging on the minimum spanning tree B; starting from the leftmost branch, use graph coloring to calculate the throughput of the current branch after merging with its right branch and the throughput of the two branches transmitting independently and concurrently. If the throughput after merging is better than the throughput of independent concurrent transmission and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune it to obtain the minimum spanning tree C; after configuring the time slot scheduling order of data transmission by each unmanned node on the minimum spanning tree C using graph coloring, each unmanned node transmits sensor data in time-division multiple access mode according to the time slot scheduling order.

[0089] The transmission system 100 of the aforementioned large-scale unmanned swarm sensor data acquisition network comprises a minimum spanning tree structure jointly constructed by initial clustering, first-round merging, and second-round merging. This structure is precisely mapped to a specific concurrent time slot scheduling scheme using a graph coloring method. The concurrent transmission links within each time slot fully utilize the potential of the previously retained parallel transmission paths. Specifically, initial clustering achieves intra-cluster data aggregation to reduce the number of transmissions; the first-round merging eliminates unnecessary redundant paths, reducing network complexity; the second-round merging ensures that the retained parallel paths outperform the merging scheme in terms of throughput; and the final graph-colored time slot scheduling maps these optimized paths to specific concurrent transmission timetables, achieving cross-layer joint optimization from network layer routing design to link layer time slot scheduling.

[0090] In one embodiment, before constructing the minimum spanning tree A, a flooding protocol is initiated at the data aggregation node to query the one-hop neighbor information table of all unmanned nodes in the large-scale unmanned cluster sensor data acquisition network; the one-hop neighbor information table includes the hierarchy of each unmanned node relative to the data aggregation node and the next level.

[0091] It is understood that the explanations of the features in the transmission system 100 of the large-scale unmanned swarm sensor data acquisition network can be understood by referring to the explanations in the embodiments of the transmission method of the large-scale unmanned swarm sensor data acquisition network, and will not be repeated here.

[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of protection of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and all such modifications and improvements fall within the scope of protection of the present invention.

Claims

1. A method for transmitting data in a large-scale unmanned cluster sensor data acquisition network, characterized in that, Including the following steps: With the data aggregation node as the root node and each link weight value of 1, a minimum spanning tree A is constructed for a large-scale unmanned cluster sensor data acquisition network based on the minimum spanning tree algorithm. Perform initial clustering on the minimum spanning tree A, dividing all source nodes on each branch into a separate cluster; Perform the first round of cluster merging on the clusters after initial clustering; starting from the leftmost branch, if there is no parallel transmission link between the current branch and its right branch, and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune the right branch. After traversing all branches, obtain the minimum spanning tree B that retains only the parallel transmission path. Perform a second round of cluster merging on the minimum spanning tree B. Starting from the leftmost branch, use graph coloring to calculate the throughput after merging the current branch with its right branch and the throughput of independent concurrent transmission of the two branches. If the throughput after merging is better than the throughput of independent concurrent transmission and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune the branch to obtain the minimum spanning tree C. After configuring the time slot scheduling order for data transmission by each unmanned node using the graph coloring method on the minimum spanning tree C, each unmanned node transmits sensor data in a time-division multiple access manner according to the time slot scheduling order.

2. The transmission method of the large-scale unmanned cluster sensor data acquisition network according to claim 1, characterized in that, Before constructing the minimum spanning tree A, the following steps are also included: Initiate a flooding protocol at the data aggregation node to query the one-hop neighbor information table of all unmanned nodes in the large-scale unmanned cluster sensor data acquisition network; the one-hop neighbor information table contains the hierarchy of each unmanned node relative to the data aggregation node and the one-hop neighbor information of the next level.

3. The transmission method of the large-scale unmanned cluster sensor data acquisition network according to claim 1 or 2, characterized in that, In the minimum spanning tree A, the vertices correspond to the unmanned nodes in the large-scale unmanned cluster sensor data acquisition network. The edges between two vertices that belong to the same branch and are directly connected are represented by solid lines, while the edges between two vertices that do not belong to the same branch but are directly connected are represented by dashed lines, indicating that there is a mutual interference relationship between the two vertices.

4. The transmission method of the large-scale unmanned cluster sensor data acquisition network according to claim 1, characterized in that, Data aggregation includes application-layer aggregation and network-layer aggregation; The application layer aggregation involves integrating and refining data collected from multiple sensors, removing redundant information, and refining it into compact data. Network layer aggregation combines small data packets collected by multiple sensors into a large data packet for transmission within the same time slot.

5. The transmission method for a large-scale unmanned cluster sensor data acquisition network according to claim 3, characterized in that, In the first round of cluster merging, if there is at least one parallel transmission link between the current branch and the branch to its right, then the current branch and the branch to its right will not be merged into a cluster.

6. The transmission method of the large-scale unmanned cluster sensor data acquisition network according to claim 3, characterized in that, Pruning involves deleting all non-cluster head leaf nodes on the branches to be merged.

7. The transmission method of the large-scale unmanned cluster sensor data acquisition network according to claim 1, characterized in that, The process of configuring the time slot scheduling order for data transmission from each unmanned node using the graph coloring method includes: An interference matrix is ​​generated based on the current minimum spanning tree; the element values ​​of the interference matrix indicate whether there is interference between two corresponding sending nodes. An undirected connected graph is constructed based on the interference matrix; each sending node is treated as a vertex in the undirected connected graph, and if there is interference between two sending nodes, the corresponding two vertices in the undirected connected graph are connected by an edge. Solve the graph coloring problem for an undirected connected graph. Use the minimum number of colors obtained as the minimum number of time slots required for each source node in the current minimum spanning tree to send data to the data aggregation node, and obtain the time slot scheduling order of each sending node.

8. The transmission method of the large-scale unmanned cluster sensor data acquisition network according to claim 3, characterized in that, The maximum message length that a cluster head node can transmit in each time slot is equal to the product of the time slot length and the transmission rate.

9. A transmission system for a large-scale unmanned cluster sensor data acquisition network, characterized in that, include: Multiple unmanned nodes, each equipped with sensor data acquisition devices to periodically collect sensor monitoring data; Data aggregation nodes are used to receive sensor data from unmanned nodes transmitted via a multi-hop wireless network; The data aggregation node is configured with itself as the root node and all link weights are 1. A minimum spanning tree A is constructed for the large-scale unmanned swarm sensor data acquisition network based on the minimum spanning tree algorithm. Initial clustering is performed on minimum spanning tree A, dividing all source nodes on each branch into a cluster. The first round of cluster merging is performed on the clusters after the initial clustering. Starting from the leftmost branch, if there are no parallel transmission links between the current branch and its right-hand branch, and the cluster head node of the current branch can aggregate all the data generated by the right-hand branch, then the cluster of the right-hand branch is merged into the current branch, and pruning is performed on the right-hand branch. After traversing all branches, only parallel transmission paths are retained. Find the minimum spanning tree B; perform a second round of cluster merging on the minimum spanning tree B; starting from the leftmost branch, use graph coloring to calculate the throughput of the current branch after merging with its right branch and the throughput of the two branches transmitting independently and concurrently. If the throughput after merging is better than the throughput of independent concurrent transmission and the cluster head node of the current branch can aggregate all the data generated by the right branch, then merge the clusters of the right branch into the current branch and prune it to obtain the minimum spanning tree C; after configuring the time slot scheduling order of data transmission by each unmanned node on the minimum spanning tree C using graph coloring, each unmanned node transmits sensor data in time-division multiple access mode according to the time slot scheduling order.

10. The transmission system of the large-scale unmanned cluster sensor data acquisition network according to claim 9, characterized in that, Before constructing the minimum spanning tree A, a flooding protocol is initiated at the data aggregation node to query the one-hop neighbor information table of all unmanned nodes in the large-scale unmanned cluster sensor data acquisition network; the one-hop neighbor information table includes the hierarchy of each unmanned node relative to the data aggregation node and the next level.

Images