UAV cluster self-organizing network communication system and method based on broadband and narrowband fusion

Abstract

This invention proposes a self-organizing network communication system and method for drone swarms based on broadband and narrowband fusion. The technical solution includes: employing a multi-channel broadband and narrowband fusion approach, utilizing individual radio channels to switch MAC protocols during different periods within the cluster; a cluster formation module utilizing the KMeans algorithm and a weighted clustering algorithm to establish a self-organizing network for drone swarms; constructing a topology map for each cluster and maintaining topology information within the cluster; a time slot allocation module utilizing the cluster head node to collect all data transmission requests for the next cycle and allocate time slots for the next cycle; and a multi-hop forwarding module utilizing load-aware and topology-aware methods to select forwarding paths. This invention enables normal communication between individual and multiple clusters within a drone network, improving the communication performance of drone swarm networks in large-scale, highly mobile cluster scenarios.

Description

Technical Field

[0001] This invention belongs to the field of wireless communications technology, and more specifically, to a system and method for defining routing set members in a drone cluster ad hoc network based on broadband and narrowband convergence. This invention can be applied to ad hoc network communications between drone clusters composed of multiple drones, meeting the requirements for drone cluster management and maintenance under dynamic topology changes, and enabling efficient dynamic allocation of time slot resources in real time based on transmission services and optimal path selection. Background Art

[0002] A Flying Ad Hoc Network (FANET) is a self-organizing network composed of unmanned aerial vehicles (UAVs). With the rapid development of civilian drone technology, FANET has attracted widespread attention, sparking a wave of related research and applications. In recent years, FANET has been widely applied and explored in fields such as agriculture, forestry, environmental monitoring, logistics distribution, and areas with limited communication coverage. Due to the high speed and frequent topology changes of UAV nodes in FANET, traditional ad hoc networking protocols are no longer suitable for the highly dynamic mobile nodes. Current challenges faced by FANET at the media access (MAC) layer include dynamic topology, multi-channel operation, and unstable links. Therefore, a protocol solution more suitable for FANET is needed.

[0003] Chongqing University disclosed a multi-hop TDMA time slot allocation method suitable for UAV self-organizing networks in its patent application document "A multi-hop TDMA time slot allocation method suitable for UAV self-organizing networks" (application number: 202310096038.2, application date: 2023.02.09, application publication number: CN 116131927 A). The implementation steps of the method disclosed in the patent are: first, each TDMA time frame is divided into reservation subframes and data subframes according to time sequence, and a control short frame, a service long frame, and a time slot occupancy table are defined; second, each node that has completed network access sends a reservation application frame and a reservation notification frame to the data time slot in its occupied reservation period according to the current service demand and time slot occupancy; third, after the reservation subframe of a TDMA time frame ends, the nodes send service long frames to communicate in the reserved data time slots. The shortcoming of this method is that in an environment where the drone topology changes rapidly, this method does not take into account the dynamic topology changes of the cluster when processing multi-hop routing. Dynamic topology changes will lead to communication network instability and network information congestion. Therefore, this method cannot be applied to the time slot allocation problem of drone clusters with highly dynamic topology changes.

[0004] The China Shipbuilding Industry Systems Engineering Research Institute, in its patent application, "Dynamic Allocation Method for Time Slot Resources in a Self-Organizing Communication Network for Unmanned Aerial Vehicle Clusters" (Application Number: 202210140692.4, Application Date: February 16, 2022, Application Publication Number: CN114650603A), discloses a method for dynamically allocating time slot resources in a self-organizing communication network for unmanned aerial vehicle clusters. The method disclosed in this patent involves the following steps: first, network member nodes calculate node degree centrality, betweenness centrality, remaining battery coefficient, and signal-to-noise ratio estimation, and count node traffic; second, the head node calculates the weighted importance coefficients of member nodes; and third, the head node calculates the weighted importance of member nodes and time slot allocation parameters, counts the time slot resources to be allocated across the entire network, and thus calculates the number of time slots to be allocated to each member node. A shortcoming of this method is that it fails to consider the scenario of large-scale clusters with highly dynamic topology changes, where a large number of allocated time slots will result in higher latency, making it unsuitable for the transmission of time-sensitive information.

[0005] Nanjing University of Aeronautics and Astronautics has disclosed a MAC protocol switching system for tactical UAV networks in its patent application, "MAC Protocol Switching Method and System for Tactical UAV Networks" (Application Number: 202410212520.2, Application Date: 2024.02.27, Application Publication Number: CN 117938979 A). The system disclosed in this patent consists of a receiving module and a decision module. The receiving module is used to drive the cluster head UAV to collect status information of all UAVs in the cluster, and the decision module is used to make protocol decisions based on the calculated number of sending UAVs and network traffic load. A shortcoming of this system is that it does not take into account that in large-scale cluster scenarios, the status information of UAVs and network traffic load will change rapidly and dynamically. Frequent protocol switching will bring additional overhead, requiring a large amount of bandwidth for frequent protocol switching and cluster maintenance, and wasting communication resources. Summary of the Invention

[0006] The purpose of the present invention is to address the shortcomings of the above-mentioned existing technologies and propose a drone cluster self-organizing network communication system and method based on wide-narrowband fusion, which is used to solve the problems of network information congestion caused by dynamic cluster topology changes, high communication delay in large-scale clusters, and waste of communication resources caused by occupying a large amount of bandwidth to maintain the cluster.

[0007] The idea for achieving the purpose of the present invention is: the system of the present invention adopts a channel access module, a time slot allocation module and a multi-hop forwarding module based on multi-channel wide- and narrowband fusion, and manages the transmission and reception of different types of data between drones through wide- and narrowband radio respectively. Wideband radio is used for data transmission operations, and narrowband radio is used for control, management and data confirmation operations. The cluster head node collects and maintains the topology information within the cluster and between adjacent clusters, which solves the problem that traditional single broadband communication networking systems need to occupy a large amount of bandwidth for frequent cluster maintenance and topology updates in the face of large-scale clusters and highly dynamic topology changes. It has the advantages of large coverage area of ​​narrowband and high-speed data transmission of broadband. The method of the present invention adopts load-aware and topology-aware methods. During the time slot request phase, the cluster head node receives and summarizes the transmission requests and load backlog of each node through narrowband radio. Based on the cluster topology, the node degree and node degree change rate of each drone, the multi-hop forwarding routing problem within the cluster is transformed into a multi-objective selection problem that satisfies the highest node degree, lowest node change rate, and lightly loaded path node selection. The multi-hop transmission path with the most stable link and lowest latency is selected. The narrowband is used to maintain the large-scale cluster topology and communication conflicts in the cluster overlapping area, solving the problem that traditional broadband short-distance networks can only feedback transmission status through multi-hop forwarding, resulting in network information congestion. The method of the present invention adopts a hybrid access method of CSMA / CA and MC-TDMA on the broadband link and a low-latency time slot allocation optimization model. Each radio uses a different MAC protocol at different times in the cluster. The cluster head node dynamically allocates time slots based on the transmission request tasks within each cycle, avoiding the waste of channel resources caused by long time slot reservation time and increasing the task load of a single time slot. This solves the problem that traditional large-scale networks using TDMA for access occupy a lot of time slot resources, resulting in resource waste and excessive latency.

[0008] To achieve the above objectives, the drone cluster self-organizing network communication system of the present invention includes: a channel access module, a cluster formation module, a cluster maintenance module, a time slot allocation module, and a multi-hop forwarding module, wherein:

[0009] The channel access module is used to manage the protocol switching between CSMA / CA and MC-TDMA for each radio channel at different times in the cluster;

[0010] The cluster formation module performs self-organizing networking on all free drone nodes to obtain the head node of the cluster;

[0011] The cluster maintenance module builds a topology map for each cluster, maintains the topology information within the cluster, and performs handshake interactions for nodes to join and leave the cluster. It shares collision interference information between two adjacent clusters to maintain topology information, and allocates dynamic radio frequencies in parallel to the head nodes in each cluster.

[0012] The time slot allocation module uses the cluster head node to collect all data transmission requests for the next cycle and allocate the time slots for the next cycle;

[0013] The multi-hop forwarding module selects a forwarding path using load perception and topology perception methods.

[0014] The drone cluster self-organizing network communication method of the present invention comprises the following steps:

[0015] Step 1: The channel access module manages each radio channel to switch between CSMA / CA and MC-TDMA protocols at different times in the cluster;

[0016] Step 2: The cluster formation module conducts self-organizing networking for all free drone nodes to obtain the head node of the cluster;

[0017] Step 3: Build a topology map for each cluster, maintain the topology information within the cluster, and perform handshake interactions for nodes to join and leave the cluster.

[0018] Step 4: Share the collision interference information between two adjacent clusters to maintain topology information, and allocate dynamic radio frequencies in parallel by the head node in each cluster.

[0019] Step 5: The time slot allocation module uses the cluster head node to collect all data transmission requests for the next cycle and allocate the time slots for the next cycle;

[0020] Step 6: The multi-hop forwarding module selects a forwarding path using load awareness and topology awareness methods.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] First, the system of the present invention adopts a channel access module, a time slot allocation module and a multi-hop forwarding module based on multi-channel wide- and narrowband fusion, which overcomes the defect that the traditional single broadband communication networking system needs to occupy a large amount of bandwidth for frequent cluster maintenance and topology updates in the face of large-scale clusters and highly dynamic topology changes. It utilizes the advantages of large coverage area of ​​narrowband and high-speed data transmission of broadband, so that the present invention has the advantages of reducing communication overhead and improving the overall communication performance of the network.

[0023] Second, the method of the present invention adopts load perception and topology perception methods to select a multi-hop transmission path with the most stable link and the lowest latency, and maintains large-scale cluster topology and communication conflicts in cluster overlapping areas through narrowband, overcoming the defect that traditional broadband short-distance networks can only feedback transmission status through multi-hop forwarding, resulting in network information congestion. The present invention improves network communication stability and end-to-end successful delivery rate.

[0024] Third, the method of the present invention adopts a hybrid access method of CSMA / CA and MC-TDMA and a low-latency time slot allocation optimization model on the broadband link, overcoming the defects of traditional large-scale networks using TDMA for access, which occupy a lot of time slot resources, resulting in resource waste and excessive delay. This enables the present invention to give full play to the flexibility of CSMA / CA and the reliability of TDMA in drone cluster networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a schematic diagram of the UAV self-organizing network communication process under the wide- and narrow-band fusion of the present invention;

[0026] Figure 2 This is a schematic diagram of multi-channel communication of a drone cluster according to the present invention;

[0027] Figure 3 Schematic diagram of protocol switching on each channel of the cluster at different times of the present invention;

[0028] Figure 4 Schematic diagram of the UAV motion model of the present invention;

[0029] Figure 5 This is a handshake flow chart of nodes joining and exiting the cluster in the present invention;

[0030] Figure 6 This is a flow chart of merging time slots within the cluster MC-TDMA of the present invention;

[0031] Figure 7 This is a schematic diagram of the interaction of the multi-hop forwarding protocol within a cluster of the present invention;

[0032] Figure 8 This is a schematic diagram of the interaction between the cluster multi-hop forwarding protocol of the present invention;

[0033] Figure 9 This is a schematic diagram of inter-cluster multi-hop forwarding according to the present invention;

[0034] Figure 10 This is a schematic diagram of multi-hop forwarding within a cluster of the present invention. DETAILED DESCRIPTION

[0035] The present invention is further described in detail below with reference to the accompanying drawings and examples.

[0036] Reference Figure 1 , the system of the present invention is described in further detail.

[0037] The system of the present invention includes a channel access module, a cluster formation module, a cluster maintenance module, a time slot allocation module, and a multi-hop forwarding module, wherein:

[0038] The channel access module is used to manage each radio channel to switch between CSMA / CA and MC-TDMA protocols at different times in the cluster.

[0039] The cluster forming module performs self-organizing networking on all free drone nodes to obtain the head node of the cluster.

[0040] The cluster maintenance module constructs a topology map for each cluster, maintains the topology information within the cluster, and performs handshake interactions for nodes to join and leave the cluster. Between two adjacent clusters with collision interference, the collision interference information is shared to maintain the topology information, and dynamic radio frequency points are allocated in parallel by the head node in each cluster.

[0041] The time slot allocation module uses the cluster head node to collect all data transmission requests for the next cycle and allocates time slots for the next cycle.

[0042] The multi-hop forwarding module selects a forwarding path using load perception and topology perception methods.

[0043] Reference Figure 2 , further describes the multi-channel communication process of the drone cluster of the present invention.

[0044] Figure 2 The figure contains two independent clusters, among which the red mark indicates two central drones, the colored mark indicates a relay drone, the pure blue mark indicates nine member drones and a free drone in the upper left corner; the green straight line indicates the wide band data channel (WBDC), the yellow dotted line indicates the intra-cluster narrow band management channel (IntraC-NBMC), the blue dotted line indicates the inter-cluster narrow band management channel (InterC-NBMC), and the red dotted line indicates the cluster data auxiliary channel (NBADC).

[0045] Each drone in the system is equipped with one wideband radio (WBR) and two narrowband radios (NBR), labeled NBR1 and NBR2. The wideband radio (WBR) serves as a data channel, primarily used for high-speed data frame transmission and bidirectional data transmission between cluster member nodes. The narrowband radio (NBR1) is primarily used for the transmission of control and management frames within the cluster, which require lower throughput. It serves as a narrowband management channel within the cluster, suitable for both uplink and downlink control and management transmissions between the cluster head node and cluster member nodes. For narrowband radio NBR2, it differs depending on the role of the node in the cluster. If the node is a cluster head node or a free node that does not belong to any cluster, NBR2 is used as a management channel for inter-cluster communication. It is the basis for the cluster head node to manage inter-cluster communication and broadcast the existence of the cluster, or for free nodes to discover and join the cluster. The frequency band of this channel is the same in all clusters; if the node is a cluster member node, NBR2 is used as a narrowband channel to assist cluster data communication. It can be used for operations such as data frame confirmation. In order to reduce interference with other surrounding nodes, the transmission range of NBR2 is adjusted to be consistent with the transmission range of broadband.

[0046] The communication method of the present invention comprises the following steps:

[0047] Step 1: The channel access module manages each radio channel to switch between CSMA / CA and MC-TDMA protocols at different times in the cluster.

[0048] Reference Figure 3 , the protocol switching of each channel in the cluster at different times of the present invention is further described in detail.

[0049] Figure 3 In the figure, the gray part represents CSMA / CA and the white part represents MC-TDMA, which further describes the steps of MAC protocol switching. The left side of the dotted line represents the initial stage of cluster formation or the period before joining the cluster, and the right side of the dotted line represents the period after the cluster is stable or after joining the cluster.

[0050] When a drone does not join a cluster, the free drone uses the CSMA / CA competitive channel reservation method for neighbor discovery, data aggregation, and data confirmation.

[0051] During the cluster's stable period, the cluster head node primarily uses MC-TDMA for centralized cluster scheduling. During time slot reservation and topology maintenance, each node in the cluster uses CSMA / CA free contention transmission on the WBDC data channel. For the cluster head node, the narrowband channel InterC-NBMC is primarily used for inter-cluster communication, so CSMA / CA is used to adapt to the flexible and changing environment.

[0052] Step 2: The cluster formation module conducts self-organizing networking for all free drone nodes to obtain the head node of the cluster.

[0053] A cluster temporary head node is selected from the free drones randomly distributed in the free distribution space of all drones. The MAC layer of the communication protocol stack of each free drone is set with a cluster temporary head announcement timer. The initial timing value of the announcement timer is random. The drone that expires the earliest broadcasts a temporary cluster head announcement frame to its one-hop neighborhood, thus becoming the cluster temporary head node; the cluster temporary head node uses a narrowband channel to summarize the data of all drone nodes, uses the KMeans algorithm to divide the collected data information into clusters, and uses the weighted clustering algorithm to select a fixed cluster head node.

[0054] Reference Figure 4 , the KMeans algorithm in cluster division and the weighted clustering algorithm in cluster head node selection in the present invention are further described in detail.

[0055] Figure 4 The three vertical dashed lines in the figure represent the three coordinate axes in the spatial coordinate system. The two gray nodes represent the two positions of the drone in the air and mark the spatial velocity v of the drone. j .

[0056] The steps of using the KMeans algorithm to divide the clusters are as follows:

[0057] The first step is to calculate the distance between the i-th UAV and the j-th UAV according to the following formula:

[0058]

[0059] Among them, p ix 、p iy 、p iz represents the spatial coordinate position of the i-th UAV, p jx 、p jy 、p jz Represents the spatial coordinate position of the j-th UAV.

[0060] In the second step, the average distance between the i-th UAV and other UAVs in the one-hop neighborhood is calculated according to the following formula:

[0061]

[0062] in, represents the set of all one-hop neighbors within the broadband transmission range of the i-th UAV, |·| represents the size of the value, and d ij represents the distance between the i-th UAV and the j-th UAV.

[0063] The third step is to calculate the minimum value of the intra-cluster sum of squares in the same cluster according to the following formula:

[0064]

[0065] Where X represents the set of observation values ​​synthesized by the average relative speed and the average distance, X={x1,x2,…,x N},||·|| is the usual sense of L 2 Norm, |C i |It is C i The size of C is the initial k cluster set C={C1,C2,C3,…,C k}.

[0066] According to the above algorithm, all drone nodes in the space are divided into different relatively stable clusters.

[0067] The steps of selecting a cluster head node using a weighted clustering algorithm are as follows:

[0068] The first step is to calculate the average ARV of the relative speed between each drone in the cluster and other drones in the one-hop neighborhood. i , ARV i The smaller the value, the more likely the drone is to become the head node of the cluster, ensuring that the cluster head node is updated less frequently.

[0069] The second step is to calculate the absolute value of the number of drones in the one-hop neighborhood within the broadband transmission range of each drone as the broadband node degree. Calculate the absolute value of the number of drones in the one-hop neighborhood within the narrowband transmission range of each drone as the narrowband node degree

[0070] The third step is to calculate the broadband node change rate and narrowband node change rate of the i-th UAV according to the following formula:

[0071]

[0072]

[0073] Where M represents the number of times each UAV node degree change is observed, 3≤M≤10, CRND irepresents the average value of the node degree change rate of the i-th UAV after M observations, ΔND i,k (t) represents the change in node degree within the interval Δt at the kth observation.

[0074] The fourth step is to calculate the fitness weight of each cluster according to the following formula:

[0075]

[0076] Where ω represents the fitness weight of each cluster calculated individually, m represents the cluster fitness weight number, 1≤m≤4, n represents the total number of drones in the cluster, ln(·) represents the logarithmic operation with the natural constant e as the base, and x′ ij Represents the normalized value of the observed value of the j-th cluster fitness weight in the i-th UAV.

[0077] Step 5: Calculate the comprehensive score of each drone in the cluster according to the following formula:

[0078]

[0079] Among them, f i represents the fitness value of the i-th UAV, ω is the cluster fitness weight,

[0080] Step 6: According to the above algorithm, the node with the highest comprehensive score is set as the cluster head node.

[0081] Step 3: Build a topology map for each cluster, maintain the topology information within the cluster, and perform handshake interactions for nodes to join and leave the cluster.

[0082] Each vertex in the topology graph corresponds to each drone in the cluster, and each edge in the topology graph corresponds to the connection between two drones that can communicate normally. Within the cluster, cluster member nodes send node information to the cluster head node through the narrowband link within the cluster. This information includes the current drone's location information, speed information, broadband and narrowband node degree, and broadband and narrowband node degree change rate.

[0083] Reference Figure 5 , further describes the handshake interaction of joining and leaving nodes in the cluster of the present invention.

[0084] Figure 5 In the figure, black nodes represent cluster head nodes, white nodes represent free nodes, gray nodes represent cluster member nodes, and dotted arrows represent the direction of data transmission.

[0085] In the first step, when a free node receives Beacon frame information from a nearby cluster through the InterC-NBMC narrowband common management channel, if this node wants to join this cluster, it can send a Join Cluster Request (JCREQ) to the head node of the nearby cluster through InterC-NBMC.

[0086] In the second step, after identity authentication, the cluster head node can agree to join the node as a cluster member node, and at the same time reply a Join Cluste Response (JCREP) through InterC-NBMC.

[0087] In the third step, when a member node in the cluster wants to leave the cluster, it sends an Exit Cluster Request (ECREQ) to the cluster head node through the intra-cluster narrowband management channel IntraC-NBMC.

[0088] In the fourth step, the head node replies with an Exit Cluster Response (ECREP) to confirm.

[0089] The cluster head node is also responsible for maintaining a lifetime timer (CMNLT) for each member node in the cluster. The cluster head node resets the timer every time it sends a message to a cluster member node. If a member's lifetime timer expires, the cluster head node sends a pending confirmation frame to the member node. If the member node does not send a confirmation within the specified time, the cluster head node removes the member node from the member list.

[0090] Step 4: Between two adjacent clusters with collision interference, the collision interference information is shared to maintain topology information, and the head node in each cluster allocates dynamic radio frequencies in parallel.

[0091] When two clusters with the same channel number approach each other, collision interference may occur at the cluster edge nodes. At this time, the edge node notifies this message to the cluster head node. The two cluster head nodes negotiate a new node through the inter-cluster narrowband link. After the negotiation is completed, it is broadcast within the cluster, and each node in the cluster adjusts the frequency of the wideband and narrowband channels.

[0092] The parallel allocation of dynamic radio frequencies by the head nodes in each cluster is further described.

[0093] The dynamic frequency allocation algorithm runs in parallel in each cluster head node. In each round, the cluster head nodes dynamically adjust the allocated frequencies by sending and receiving information. The algorithm ends successfully only when all channels used by all clusters are no longer conflicting. The cluster head node then broadcasts the index number of the latest frequency f throughout the cluster and obtains confirmation from all cluster members. The entire cluster then adjusts both broadband and narrowband channels (excluding the InterC-NBMC inter-cluster common management channel) to the frequency of the target index.

[0094] Step 5: The time slot allocation module uses the cluster head node to collect all data transmission requests for the next cycle and allocate time slots for the next cycle.

[0095] Reference Figure 6 , the allocation of time slots of the next cycle by the cluster head node is further described.

[0096] Figure 6 The eleven grey nodes in the figure represent the eleven cluster member nodes, and the edges with arrows represent the directions of data transmission requests.

[0097] Cluster member nodes 2, 3, 6, 7, and 9 submit time slot reservation requests to the cluster head node. Assume that TR(A,B) represents a transmission request from node A to node B. In this section, nodes A and B are reachable within one hop, and only one time slot is requested. According to the figure above, the requested transmission tasks are: TR(2,1), TR(3,4), TR(6,3), TR(7,4), and TR(9,10). It's easy to analyze that member nodes 2 and 3, as well as member nodes 2 and 9, are mutually exposed terminals, while member nodes 3 and 7 are mutually hidden terminals. In a traditional TDMA allocation scheme, the head node allocates a time slot to each of these five transmission requests, requiring five time slots to complete the transmission. However, using a topology-aware time slot allocation scheme, nodes with non-interfering transmission requests are assigned to the same time slot for transmission. In the figure, TR(2,1) and TR(3,4) can be assigned to the same time slot, TR(6,3) and TR(7,4) can be assigned to the same time slot, and TR(9,10) can be assigned to any time slot without conflict. This approach maintains transmission quality while reducing transmission latency by 50%, significantly improving channel resource utilization. Furthermore, when requesting a time slot from the head node, the member node also specifies the transmission deadline for the task. The head node strives to complete the transmission task before the deadline.

[0098] The low-latency time slot allocation optimization model of the present invention is described in detail below:

[0099] The link relationship between each node in the cluster is abstracted into a graph network (undirected graph) G = (V, E), where V is the set of nodes in the cluster and E is the set of link relationships between nodes. When node u sends a transmission request to the head node to node v, the edge e = (u, v) is changed to<u,v> , that is, due to the time slot reservation request, the original undirected graph is transformed into a hybrid graph. At the same time, each time slot reservation request is an event and is also a directed edge. Combining the above content, the problem is to include a set of directed edges in the hybrid graph G Indicates that multiple nodes have issued N event A transmission request; also contains a set of time slots in represents the number of time slots actually required; thus, the problem is transformed into how to arrange as many transmission requests in e as possible to In each time slot, the time delay should be minimized and the time slot utilization should be maximized as much as possible, which is a multi-objective optimization problem.

[0100] The first step is to use the link matrix Represents the one-hop link relationship between n nodes in the cluster, using the transmission request matrix TR n×n To indicate which nodes have sent time slot requests to the head node.

[0101] Step 2: Generate a conflict matrix To specify the transmission requests that cannot be assigned to the same time slot. The graph generated by the conflict matrix is ​​G C =(V C ,E C ), where the transmission request e i Abstract graph G C Vertex v in i , v i ∈V C , conflicting transmission requests e i and e j Equivalent to graph G C An edge in (v i ,v j )∈E C .

[0102] The third step is to transmit the request e i The problem of allocating time slots can be partially equivalent to a graph coloring problem. Introducing a 0-1 decision variable μ k and λ ik , k <N slot， in

[0103]

[0104]

[0105] Assume that the start time of time slot 1 is a1 and the time slot interval is t, then the start time of time slot k is a k =a1+(k-1)·t. And assuming β i (β i ≤t) is the transmission request e i The expected transmission time, γ i For transmission request e i The absolute deadline, let ξ ik =α k +β i -γ i For transmission request e i The time relaxation index is defined as follows: ik :

[0106]

[0107] From the above formula, we can see that δ ik Smaller values ​​indicate lower latency for transmission requests. When the value is greater than 0, more penalty terms are added using quadratic terms to ensure that transmission tasks with high latency requirements are scheduled in the time slot before the test as much as possible, which also makes the algorithm converge faster. Based on the constraints and target requirements mentioned above, the following mathematical model is obtained:

[0108]

[0109] Where 1≤i≠j≤N event ,1≤k≤N slot .

[0110] In the above optimization model, the constraint relationship means that the purpose of the first constraint is to use the earliest time slot possible. If the previous time slot is not used, that is, μ k = 0, the next time slot, μ, cannot be used k+1 = 1. The second constraint indicates that the transmission requests in each time slot cannot conflict with each other, that is, the two adjacent vertices v in the conflict matrix i ,v j , meaning that a pair of conflicting transmission requests cannot be simultaneously allocated to k time slots. The third constraint is an additional constraint on time slot allocation, and the fourth constraint states that each transmission request can only be allocated to one time slot. The first objective function is to minimize the total number of time slots allocated to transmission requests, and the second objective function is to prioritize the allocation of time slots to transmission requests with higher latency requirements, minimizing overall transmission latency. Ultimately, the optimal time slot allocation solution is obtained from the solution set of this optimization problem. The cluster head node notifies the member nodes via narrowband broadcast, and each member node completes data transmission in the next TDMA period.

[0111] Step 6: The multi-hop forwarding module selects a forwarding path using load awareness and topology awareness methods.

[0112] When a node's transmission request targets a node within the current cluster but outside its own one-hop range, the cluster member node sends a transmission request reservation to the cluster head node. The cluster head node searches the cluster topology map for the next-hop address of the service and allocates a time slot. When the next-hop node receives the data frame and confirms that the frame needs to be forwarded, it requests a time slot from the cluster head node again and repeats the above steps until the frame is transmitted to the target node. After each transmission, an ACK is sent on the narrowband link to confirm the completion of the transmission, ensuring the reliability of data transmission.

[0113] When the target node of a node transmission request is outside the current cluster, the cluster head node broadcasts an inter-cluster multi-hop transmission request to surrounding clusters via a narrowband channel. After receiving the request, the target cluster head node replies to the source cluster head node with a cluster edge topology table. After receiving this information, the source cluster head node selects the lowest-latency path and gateway node and allocates time slots, decomposing the inter-cluster transmission task into an intra-cluster multi-hop forwarding request and an inter-cluster single-hop transmission request. The gateway node adjusts the broadband data link and the inter-cluster narrowband link to the target cluster channel for inter-cluster single-hop transmission and completes the inter-cluster data transmission via the CSMA / CA competitive access channel. After receiving the ACK confirmation from the target node, the broadband data link and narrowband control link are adjusted back to the original cluster state and a phased ACK confirmation is sent to the cluster head node. The transmission process of this frame within the target cluster is the same as the above process. After the last hop intra-cluster transmission is completed, the destination cluster head node replies with a full ACK confirmation to the source cluster head node via the inter-cluster narrowband link, thus completing the inter-cluster multi-hop task transmission.

[0114] Reference Figure 7 , further describes the multi-hop transmission protocol interaction process within the cluster.

[0115] Figure 7 The black node in the figure represents the cluster head node, the three gray nodes represent the three cluster member nodes, the dotted and straight arrows represent the data transmission direction, the black straight line represents the broadband link WBDC, the red straight line represents the narrowband link IntraC-NBMC, and the green dotted line represents the narrowband link NBADC.

[0116] Below is Figure 6 Taking the multi-hop request TR(1,9) as an example, the process of multi-hop forwarding within the cluster is explained in detail.

[0117] In the first step, node 1 sends a timeslot reservation request (TR-REQ) to the cluster head node through the narrowband link IntraC-NBMC, requesting the generation of a multi-hop forwarding path to node 9.

[0118] In the second step, the cluster head node broadcasts a Timeslot Reservation Response (TR-REP) over the narrowband link IntraC-NBMC, responding to the results generated by Node 1. If a path exists, the forwarding response (FREP) includes a unique forwarding identifier (FID) and the address of the next-hop node (in this figure, the next hop is Node 2), and schedules a transmission time slot for Node 1. The TR-REQ also contains a flag indicating whether the transmission request is a multi-hop transmission. If so, the head node generates a multi-hop forwarding path based on the source and destination nodes of the multi-hop forwarding request using a multi-hop forwarding routing algorithm. It then generates a unique FID based on the (source node address, destination node address) to distinguish the paths and writes the FID into the TR-REP. At this point, the multi-hop transmission service is established.

[0119] In the third step, node 1 forwards the data packet to node 2 according to the time slot scheduled in TR-REP. After node 2 receives the data from node 1 and receives it correctly, it will reply ACK to node 1 via the narrowband link NBADC to confirm the correct reception. Otherwise, node 1 will continue to reserve the time slot for this data frame and retransmit it.

[0120] In the fourth step, node 2 determines that this frame needs to be forwarded based on the destination address identifier in the data frame header. Then node 2 writes the FID in the frame header and its own address into TR-REQ and sends it to the cluster head node. Based on the FID, it finds that the next hop of this multi-hop forwarding is node 9, so it forms a transmission request TR(2,9) to participate in time slot allocation, and replies with the address of the next hop (the address of node 9) and the time slot allocation result in the form of TR-REP.

[0121] In step 5, after receiving the data frame from node 2, node 9 checks the destination address and finds that it is sent to itself. After receiving the data frame correctly, it will reply an ACK to node 2 and reply an FACK to node 1 through the cluster head node. At this point, a multi-hop transmission within the cluster is completed.

[0122] Reference Figure 8 as well as Figure 9 , further describes the interaction process of the inter-cluster multi-hop forwarding protocol.

[0123] Figure 8 The black nodes in the figure are the two cluster head nodes, and the rest are cluster member nodes. The dotted and straight arrows indicate the data transmission direction. The black straight line is the broadband link WBDC, the red straight line is the narrowband link NBADC, the green dotted line is the narrowband link IntraC-NBMC, and the blue dotted line is the narrowband link IntraC-NBMC. Figure 9 In the figure, CH1 and CH2 represent the cluster head nodes of two independent clusters, and the remaining nodes represent the cluster member nodes of the two clusters. The dotted and straight arrows indicate the data transmission direction. The green straight line represents the broadband link WBDC, the green dotted line represents the possible broadband link, the blue dotted line represents the narrowband link InterC-NBMC, and the yellow dotted line represents the narrowband link IntraC-NBMC.

[0124] Below is Figure 9 Taking the multi-hop request TR(5,20) as an example, the process of multi-hop forwarding between clusters is explained in detail.

[0125] In the first step, node 5 sends a time slot reservation request (TR-REQ) to cluster head node 1 over a narrowband link, requesting a multi-hop forwarding path to node 20. Cluster head node 1 first checks the node topology table of its cluster and finds that node 20 does not exist. It then broadcasts a multi-hop forwarding request (FREQ) to node 20 over the inter-cluster common narrowband channel InterC-NBMC.

[0126] In the second step, assuming node 20 exists within cluster 2, head node 2, after receiving the FREQ from head node 1, checks the cluster's topology table and, through InterC-NBMC, responds to head node 1 with a FREP indicating the successful target node location. This FREP also includes its own edge topology table (including location information) and channel number. Cluster head node 1 analyzes the location information in the edge topology table from head node 2 and finds that nodes 7 and 8 are candidates for the cluster gateway node (CGN). Cluster head node 1 then selects the path with the lowest latency based on its multi-hop routing algorithm. Assuming this path ends at node 7 within the cluster, node 7 is selected as the cluster gateway node.

[0127] In the third step, cluster head node 1 decomposes the inter-cluster multi-hop forwarding request of TR(5,20) into an intra-cluster multi-hop forwarding request of TR(5,7) and an inter-cluster single-hop transmission request of TR(7,12). The inter-cluster multi-hop transmission process is the same as described above. The inter-cluster single-hop transmission process is as follows: Upon receiving the reservation response FREP from the head node, node 7 adjusts the channel numbers of its own data channel (WBDC) and narrowband link 2 (NBADC) to the corresponding channels of cluster 2. It then accesses the data channel through random contention and transmits data to node 12. After receiving the ACK from node 12, the node switches its own data channel and narrowband link back to their original states and responds with a phased FACK to head node 1, marking the completion of multi-hop transmission within the cluster.

[0128] Reference Figure 10 , further describes the multi-hop forwarding situation within the cluster.

[0129] Figure 10 Each node in the figure represents a cluster member node, and the blue arrow indicates the direction of data transmission request.

[0130] by Figure 10 Taking the transmission request TR(A, G) shown in the figure as an example, when the cluster head node selects the multi-hop path between the two nodes, it needs to comprehensively consider the load squeezing degree of the nodes on each path, as well as the link stability between the nodes. Because for a dynamic topology, even if the current path has the minimum forwarding delay, if it is unstable and causes link interruption, resulting in retransmission, it will introduce more delays. Therefore, the multi-hop forwarding routing algorithm in the present invention comprehensively considers load perception and topology changes. When the cluster head node performs multi-hop path selection, in order to make the communication link sufficiently stable during the forwarding process, it needs to consider the node with the smallest broadband node degree change rate; in order to better recover and re-route in the event of link interruption during the forwarding process, it needs to select the node with the largest broadband node degree; in order to prevent congestion in the forwarding process, paths with light overall load should be considered.

[0131] To sum up, the following multi-objective optimization model is formed:

[0132]

[0133] Where, path={path1,path2,…,path θ} is a set of paths for transmission requests, path i ={v1,v2,…v ω} is one of the paths, v iis a node on the path. In the above optimization model, the constraint relationship mainly represents the value range of the independent variable. The meaning of the objective function is that the first objective function f1 represents maximizing the sum of the degrees of each broadband node on the selected path, so that it can be restored in the event of path failure. The second objective function f2 represents selecting intermediate nodes with a smaller node degree change rate as much as possible. This is to a certain extent to alleviate the impact of high dynamic topology and low expected link life caused by the high-speed movement of drones. The third objective function f3 represents the selection of lightly loaded nodes under the premise of load perception, so as to ensure that data packets can be forwarded as quickly as possible. To solve this problem, a set of reachable paths from the source node to the destination node is generated, and then the most suitable path is selected according to the target optimization algorithm.

Claims

1. A UAV cluster self-organizing network communication system based on broadband and narrowband fusion, characterized by: Each drone is equipped with a wideband radio (WBR) and two narrowband radios (NBRs), labeled NBR1 and NBR2. The WBR serves as the data channel (WBDC) for bidirectional data transmission between cluster member nodes. NBR1 serves as the intra-cluster narrowband management channel (IntraC-NBMC), suitable for uplink and downlink control and management transmission between the cluster head node and cluster member nodes with low throughput requirements over a large range within the cluster. NBR2 serves as the narrowband channel (NBADC) for auxiliary cluster data communication, used for data frame confirmation operations. To reduce interference with other surrounding nodes, the transmission range of NBR2 is adjusted to be consistent with the transmission range of the WBR. Both the wideband radio and the narrowband radio are frequency-adjustable to adjust their own frequency bands. The system includes a channel access module, a cluster formation module, a cluster maintenance module, a time slot allocation module, and a multi-hop forwarding module, among which: The channel access module is used to manage the protocol switching between CSMA / CA and MC-TDMA for each radio channel at different times in the cluster; The cluster formation module performs self-organizing networking on all free drone nodes to obtain the head node of the cluster; The cluster maintenance module builds a topology map for each cluster, maintains the topology information within the cluster, and performs handshake interactions for nodes to join and leave the cluster. It shares collision interference information between two adjacent clusters to maintain topology information, and allocates dynamic radio frequencies in parallel to the head nodes in each cluster. The time slot allocation module uses the cluster head node to collect all data transmission requests for the next cycle and allocate the time slots for the next cycle; The multi-hop forwarding module selects a forwarding path using load perception and topology perception methods.

2. The UAV cluster self-organizing network communication method based on broadband and narrowband fusion according to the self-organizing network communication system of claim 1 is characterized in that: A hybrid access method of CSMA / CA and TDMA is used for channel access, and load sensing and topology sensing methods are used for multi-hop forwarding. The communication method includes the following steps: Step 1: The channel access module manages each radio channel to switch between CSMA / CA and MC-TDMA protocols at different times in the cluster; Step 2: The cluster formation module conducts self-organizing networking for all free drone nodes to obtain the head node of the cluster; Step 3: Build a topology map for each cluster, maintain the topology information within the cluster, and perform handshake interactions for nodes to join and leave the cluster. Step 4: Share the collision interference information between two adjacent clusters to maintain topology information, and allocate dynamic radio frequencies in parallel by the head node in each cluster. Step 5: The time slot allocation module uses the cluster head node to collect all data transmission requests for the next cycle and allocate the time slots for the next cycle; Step 6: The multi-hop forwarding module selects a forwarding path using load awareness and topology awareness methods.

3. The UAV cluster self-organizing network communication method based on broadband and narrowband fusion according to claim 2 is characterized in that: Managing the protocol switching between CSMA / CA and MC-TDMA for each radio channel at different times in the cluster as described in step 1 means: When a drone does not join a cluster, it uses CSMA / CA contention channel reservation for neighbor discovery, data aggregation, and data confirmation. During the cluster's stable period, the cluster head node primarily uses MC-TDMA for centralized cluster scheduling. During time slot reservation and topology maintenance, each node in the cluster uses CSMA / CA free contention transmission on the WBDC data channel. For the cluster head node, the narrowband channel InterC-NBMC is primarily used for inter-cluster communication, so CSMA / CA is used to adapt to the flexible and changing environment.

4. The UAV cluster self-organizing network communication method based on broadband and narrowband fusion according to claim 2 is characterized in that: The self-organizing network of all free drone nodes in step 2 means: selecting a cluster temporary head node from the free drones randomly distributed in the free distribution space of all drones, and setting a cluster temporary head announcement timer in the MAC layer of the communication protocol stack of each free drone. The initial timing value of the announcement timer is random. The drone that expires the earliest broadcasts a temporary cluster head announcement frame to its one-hop neighbor, thereby becoming the cluster temporary head node; The temporary head node of the cluster uses a narrowband channel to summarize the data of all drone nodes, divides the collected data information into clusters using the KMeans algorithm, and uses the weighted clustering algorithm to select a fixed cluster head node.

5. The UAV cluster self-organizing network communication method based on wide-narrowband fusion according to claim 4 is characterized in that: The steps of using the KMeans algorithm to divide the clusters are as follows: The first step is to calculate the distance between the i-th UAV and the j-th UAV according to the following formula: Among them, p ix 、p iy 、p iz represents the spatial coordinate position of the i-th UAV, p jx 、p jy 、p jz represents the spatial coordinate position of the j-th UAV; The second step is to calculate the average distance AD ​​between the i-th UAV and other UAVs in the one-hop neighborhood in the cluster. i : in, represents the set of all one-hop neighbors within the broadband transmission range of the i-th UAV, |·| represents the size of the value, and d ij represents the distance between the i-th UAV and the j-th UAV; The third step is to calculate the minimum value of the intra-cluster sum of squares in the same cluster according to the following formula: Where X represents the set of observation values ​​synthesized by the average relative speed and the average distance, X={x1,x2,…,x N },‖·‖ is the usual meaning of L 2 Norm, |C i |It is C i The size of C is the initial k cluster set C={C1,C2,C3,…,C k }.

6. The UAV cluster self-organizing network communication method based on wide-narrowband fusion according to claim 5 is characterized in that: The steps of selecting a cluster head node using a weighted clustering algorithm are as follows: The first step is to calculate the average ARV of the relative speed between each drone in the cluster and other drones in the one-hop neighborhood. i , ARV i The smaller the value, the more likely the drone is to become the head node of the cluster, ensuring that the cluster head node is updated less frequently; The second step is to calculate the absolute value of the number of drones in the one-hop neighborhood within the broadband transmission range of each drone as the broadband node degree. Calculate the absolute value of the number of drones in the one-hop neighborhood within the narrowband transmission range of each drone as the narrowband node degree The third step is to calculate the broadband node change rate and narrowband node change rate of each drone according to the following formula: Where M represents the number of times each UAV node degree change is observed, 3≤M≤10, CRND i represents the average value of the node degree change rate of the i-th UAV after M observations, ΔND i,k (t) represents the change in node degree within the interval Δt at the kth observation; The fourth step is to calculate the fitness weight of each cluster according to the following formula: Where ω represents the fitness weight of each cluster calculated individually, m represents the cluster fitness weight number, 1≤m≤4, n represents the total number of drones in the cluster, ln(·) represents the logarithmic operation with the natural constant e as the base, and x′ ij represents the normalized value of the observed value of the j-th cluster fitness weight in the i-th UAV; Step 5: Calculate the comprehensive score of each drone in the cluster according to the following formula: Among them, f represents the fitness value of the i-th drone, Step 6: Set the node with the highest comprehensive score as the cluster head node.

7. The UAV cluster self-organizing network communication method based on broadband and narrowband fusion according to claim 2 is characterized in that: The topology map of each cluster described in step 3 means that each vertex in the topology map corresponds to each drone in the cluster, and each edge in the topology map corresponds to the connection between two drones that can communicate normally. Within the cluster, cluster member nodes send node information to the cluster head node through the narrowband link within the cluster. This information includes the current drone's location information, speed information, broadband and narrowband node degree, and broadband and narrowband node degree change rate. The cluster head node aggregates this information to construct the topology map of the entire cluster.

8. The UAV cluster self-organizing network communication method based on broadband and narrowband fusion according to claim 2 is characterized in that: The maintenance of topology information between two adjacent clusters described in step 4 means that when two clusters with the same channel number approach each other, the cluster edge node that collides and interferes will notify the cluster head node of the collision message. The two cluster head nodes negotiate a new node through the inter-cluster narrowband link. After the negotiation is completed, it is broadcast within the cluster, and each node in the cluster adjusts the frequency of the wideband and narrowband channels.

9. The UAV cluster self-organizing network communication method based on broadband and narrowband fusion according to claim 2 is characterized in that: The forwarding path selection using the load-aware and topology-aware methods in step 6 refers to: When a node's transmission target is not within its own one-hop range, the cluster member node sends a transmission request reservation to the cluster head node. The cluster head node searches the cluster topology map for the next hop address and allocates a time slot. When the next-hop node receives the data frame and confirms that the frame needs to be forwarded, it requests a time slot from the cluster head node again and repeats the above steps until the frame is transmitted to the target node. After each transmission, an ACK is sent on the narrowband link to confirm the completion of the transmission, ensuring the reliability of data transmission. When the target node of a node transmission request is outside the current cluster, the cluster head node broadcasts an inter-cluster multi-hop transmission request to surrounding clusters via a narrowband channel. After receiving the request, the target cluster head node replies to the source cluster head node with a cluster edge topology table. After receiving this information, the source cluster head node selects the lowest-latency path and gateway node and allocates time slots, decomposing the inter-cluster transmission task into an intra-cluster multi-hop forwarding request and an inter-cluster single-hop transmission request. The gateway node adjusts the broadband data link and the inter-cluster narrowband link to the target cluster channel for inter-cluster single-hop transmission and completes the inter-cluster data transmission via the CSMA / CA competitive access channel. After receiving the ACK confirmation from the target node, the broadband data link and narrowband control link are adjusted back to the original cluster state and a phased ACK confirmation is sent to the cluster head node. The transmission process of this frame within the target cluster is the same as the above process. After the last hop intra-cluster transmission is completed, the destination cluster head node replies with a full ACK confirmation to the source cluster head node via the inter-cluster narrowband link, thus completing the inter-cluster multi-hop task transmission.

Images