Unmanned aerial vehicle cluster dynamic routing method based on adaptive fuzzification coefficient dynamic clustering

Abstract

The invention belongs to the technical field of unmanned aerial vehicle communication, and discloses an unmanned aerial vehicle cluster dynamic routing method based on adaptive fuzzification coefficient dynamic clustering, which is suitable for a flight ad hoc network with topology height dynamic and limited resources. According to the method, node clustering is carried out by adopting a fuzzy C-means algorithm of multiple fuzzy coefficients, and a hierarchical communication structure of cluster members, cluster heads and gateways is constructed. An improved ant colony algorithm is introduced in the routing stage, multi-dimensional pheromones such as residual energy, link stability, congestion and hop count are fused, and efficient path selection is achieved. And in combination with a self-adaptive HELLO packet mechanism and a pheromone exponential decay strategy, the energy balance and robustness of routing are effectively improved, and the control overhead is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle (UAV) communications, and in particular relates to a dynamic routing method for UAV clusters based on dynamic clustering of adaptive fuzzy coefficients. Background Art

[0002] Unmanned aerial vehicle (UAV) swarms communicate and collaborate through self-organizing networks (FANETs), showing great potential in wide-area surveillance, search and rescue, emergency communications, and other fields. However, UAV swarm networks have the following challenges:

[0003] 1) Highly dynamic topology: Drones maneuver at high speed in three-dimensional space, causing frequent and drastic changes in network topology and the easy interruption of links between nodes. Traditional routing protocols based on the assumption of stable topology (such as DSDV) are difficult to apply.

[0004] 2) High communication overhead: Some dynamic routing protocols (such as AODV) rely on network-wide broadcasts ("flooding") during route discovery, which can easily cause "broadcast storms" in densely populated drone clusters, consuming a large amount of limited bandwidth and energy resources.

[0005] 3) Resource Constraints: UAVs have limited energy resources, so routing protocols must take energy efficiency into consideration to avoid premature failure of a few key nodes due to excessive forwarding tasks, which would affect the life cycle of the entire cluster network.

[0006] 4) Difficulty in Network Organization: To cope with topology changes and reduce overhead, network clustering strategies are often employed. However, traditional "hard clustering" methods (such as K-means) require each node to clearly belong to only one cluster. This can lead to frequent cluster reconfiguration at cluster boundaries where nodes move rapidly, causing network instability. While "fuzzy clustering" (such as FCM) can improve this problem, traditional FCM algorithms use a single fuzzification coefficient, making them difficult to adapt to the complex situation of uneven node distribution in the network (dense in some areas, sparse in others).

[0007] Therefore, there is an urgent need for a dynamic routing method for drone clusters that can intelligently perceive network status, adaptively organize network structure, and make routing decisions efficiently, reliably, and with low energy consumption. Summary of the Invention

[0008] The present invention aims to overcome the shortcomings of the prior art by providing a method for dynamic routing of drone swarms based on adaptive dynamic clustering of fuzzy coefficients. This method combines a multi-factor fuzzy C-means clustering algorithm with an improved ant colony optimization routing algorithm to construct a hierarchical drone ad hoc network topology. Furthermore, it uses a multidimensional pheromone model and an adaptive maintenance mechanism to achieve reliable and efficient routing.

[0009] The application provides a dynamic routing method for a UAV cluster based on adaptive fuzzy coefficient dynamic clustering, comprising the following steps:

[0010] S1. Network initialization;

[0011] Initial parameter configuration: configure the communication radius, maximum speed v i and energy threshold E L for each UAV.

[0012] Topology discovery: obtain the state information of the UAV node and the neighbor node, wherein the state information comprises the geographic position information of the UAV node, the link congestion degree, the node residual energy, the link signal strength and the path hop count, and a unified multidimensional pheromone model is constructed based on the state information.

[0013] Time synchronization: complete the network clock alignment by using the GPS or PTP protocol.

[0014] S2. Cluster head election;

[0015] Based on the multidimensional pheromone model containing the state information in step S1, the UAV node is subjected to multiple fuzzy factor fuzzy C-means clustering, then the node is divided into a cluster member node, a cluster head node and a gateway node according to the membership degree, and a hierarchical communication structure is constructed.

[0016] The vector after clustering is a complex number, wherein the real part of the complex number is the geographic position information of the UAV node, and the imaginary part is the comprehensive pheromone representing the communication and routing performance of the UAV node.

[0017] The cluster head node is responsible for the cluster internal routing maintenance, the gateway node maintains the fuzzy membership degree to two or more clusters and undertakes the cross-cluster routing forwarding function.

[0018] S3. Routing discovery;

[0019] That is, when the source node needs to send data, a routing request packet carrying the link stability, the node residual energy, the link congestion degree and the path hop count is generated and broadcasted.

[0020] S4. Intermediate forwarding of the routing request packet;

[0021] That is, after the intermediate node receives the routing request packet, if the residual energy of the current intermediate node is lower than the energy threshold E U , the routing request packet is discarded; if the residual energy of the current intermediate node is higher than the energy threshold E i , the link pheromone is calculated and selectively forwarded according to the pheromone threshold condition. The pheromone threshold is the lower limit of the path quality set in the routing optimization stage, and is used for filtering low-quality links.

[0022] S5. Routing establishment;

[0023] That is, after the destination node receives the routing request packet, it selects a candidate path according to the accumulated pheromone value and returns a routing reply packet along the candidate path to establish a data transmission path;

[0024] S6. Exponential decay evaporation and positive feedback gain are performed on the total pheromone of the transmission path during data transmission to dynamically maintain the path weight;

[0025] S7. Path selection;

[0026] During path selection, the node preferentially selects the path with the highest pheromone concentration for data transmission;

[0027] At the same time, the node backs up suboptimal paths in the routing table according to the received multiple RREP information, so as to quickly switch when the main path fails;

[0028] S8. Route maintenance;

[0029] The HELLO handshake packet sending interval is adaptively adjusted according to the node moving speed and neighbor node density to realize real-time link maintenance;

[0030] In the routing table maintenance stage, the path record with a pheromone concentration lower than the maintenance threshold is deleted to reduce the size of the routing table and improve the search efficiency. The maintenance threshold is different from the pheromone threshold in step S4. The pheromone threshold is used to filter candidate paths during the route discovery process to prevent low-quality links from participating in the establishment of new routes. The maintenance threshold in this stage is mainly used in route maintenance. By periodically removing paths with excessively low pheromone concentrations, invalid paths are avoided from occupying routing table resources.

[0031] Further, in the step S2, the fuzzy coefficient of the multiple fuzzy factor fuzzy C-means clustering is dynamically adjusted according to the node neighborhood density and the node energy state, so that the membership degree of the nodes in the dense area tends to be determined and the nodes in the sparse area retain multiple cluster membership degrees;

[0032] The calculation expression of the fuzzy coefficient is:

[0033]

[0034] In the formula, m i is the fuzzy coefficient of node i; m L , m U are the minimum value and the maximum value of the fuzzy coefficient, respectively, for controlling the variation range of the coefficient; δ i represents the average distance of node i and its adjacent nodes, which is regarded as the density index of the area where node i is located; δ min is the minimum average distance of all nodes in the network; and δ maxis the maximum average distance of all nodes in the network; a is an exponential parameter that adjusts the sensitivity of the fuzzy coefficient to the change of the node neighborhood density.

[0035] Further, in the step S2, the step of constructing the hierarchical communication structure is:

[0036] When the source node and the destination node are located in the same cluster, intra-cluster routing is performed, and the routing request is performed in the cluster;

[0037] When the source node and the destination node are located in different clusters, inter-cluster routing is performed, and the routing request is completed by limited range forwarding between cluster head nodes.

[0038] Further, in the step S4, the calculation expression of the link pheromone is:

[0039]

[0040] PL ij represents the link pheromone between node i and node j; LS ij represents the link stability between node i and node j; AERP is the average node residual energy percentage on the path; is the hop penalty factor, used to balance the influence of path length; MCG represents the maximum congestion degree on the path.

[0041] Further, in the step S6, the calculation expression of the link pheromone performing exponential decay volatility is as follows:

[0042]

[0043] PL ij (t) represents the pheromone concentration of link ij at time t; PL ij (t+n) represents the pheromone concentration of link ij at time t+n; v max is the maximum moving speed of the node; m is the number of nodes or segments on the path; r is the basic volatility coefficient.

[0044] Further, in the step S6, the calculation expression of the link pheromone performing positive feedback gain is as follows:

[0045]

[0046] PL count is the number of successfully confirmed messages on the path.

[0047] Further, in the step S7, the calculation expression of the HELLO handshake message sending interval is:

[0048]

[0049] In the formula, T Hello HELLO handshake message sending interval; T0 is a basic time interval, i.e., a static default interval reference value; η represents a network density coefficient, used to measure the number of node neighbors or the degree of local congestion; representing the average speed of the node.

[0050] Further, in the step S8, the route maintenance further includes a fault recovery mechanism, when the link is interrupted, the gateway node with multiple backup cluster attributions can quickly switch to the backup cluster for routing.

[0051] Beneficial effects:

[0052] The intelligent network layering and optimized path selection are combined in the application, the network overhead is significantly reduced while the communication quality is ensured. In the network organization aspect, the application proposes a self-adaptive clustering algorithm based on multiple fuzzy coefficients, a comprehensive pheromone model is constructed by comprehensively considering node position, residual energy and link stability, the fuzzy degree is dynamically adjusted to realize stable attribution of nodes in dense areas and flexible membership of boundary nodes, thereby a robust clustering topology structure is constructed. In the route decision aspect, a multi-dimensional pheromone model is designed by fusing link stability, energy utilization rate, buffer congestion degree and hop cost, an improved ant colony algorithm is guided to perform path search and update, and global optimal path selection is realized.

[0053] Other advantages, objects, and features of the application will be set forth in part by the description that follows, and in part will become apparent to those skilled in the art upon examination of the following specification or can be learned by the practice of the application. The objects and other advantages of the application can be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 A flow chart of a kind of unmanned aerial vehicle cluster dynamic routing method based on adaptive fuzzy coefficient dynamic clustering of the application;

[0055] Figure 2 A structure schematic view of a routing request (RREQ) control message extended to carry multi-dimensional state information;

[0056] Figure 3 A schematic view of the core mechanism of pheromone fusion, hierarchical network construction and ant colony optimization routing process;

[0057] Figure 4 A specific example schematic view of an inter-cluster route discovery process. DETAILED DESCRIPTION

[0058] In order to make the technical solutions, advantages and objectives of the present application clearer, the technical solutions of the embodiments of the present application will be described clearly and completely below in conjunction with the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the described embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without any creative effort belong to the protection scope of the present application.

[0059] As shown in Figure 1 , the present application provides a dynamic routing method for a UAV cluster based on adaptive fuzzy coefficient dynamic clustering, and the overall process mainly includes seven core stages of network initialization, cluster head election, route discovery, route establishment, route selection, route maintenance and performance evaluation. The key technical links will be described in detail below:

[0060] S1. Network initialization;

[0061] Initial parameter configuration: configure each UAV with a communication radius, a maximum speed v max , and an energy threshold E th , etc.

[0062] Topology discovery: the node periodically sends a HELLO packet to collect the state information of itself and neighboring UAV nodes, and the state information includes the geographic location information of the UAV node, the link congestion, the node remaining energy, the link signal strength and the path hop count, and a unified multidimensional pheromone model is constructed based on the state information. Referring to Figure 2 , the structure of the route request (RREQ) packet is extended, and the average residual energy, the network density coefficient, the link stability and other extension fields are added. These fields enable the RREQ packet to collect comprehensive state information of the path when it propagates in the network.

[0063] Time synchronization: complete the network clock alignment by using GPS or PTP protocol;

[0064] S2. Cluster head election;

[0065] Based on the multidimensional pheromone model containing state information of step S1, multiple fuzzy factor fuzzy C-means clustering is performed on the UAV nodes, then the nodes are divided into cluster member nodes, cluster head nodes and gateway nodes according to the membership degree, and a hierarchical communication structure is constructed according to the node pheromone concentration;

[0066] Among them, the clustered vector is a complex number, the real part of the complex number is the geographic location information of the UAV node, and the imaginary part is the comprehensive pheromone representing the communication and routing performance of the UAV node;

[0067] x i = p i + jPN i

[0068] wherein PN i is the node-level pheromone concentration, established by the following formula:

[0069]

[0070] wherein AERP i is the node residual energy; BUR i is the buffer utilization rate of node i; N(i) is the neighbor node set of node i; LS ij is the link stability between node i and neighbor node j, established by the inter-node signal strength P ij ;

[0071] The cluster head node is responsible for cluster internal routing maintenance, and the gateway node maintains the fuzzy membership to two or more clusters and undertakes the cross-cluster routing forwarding function;

[0072] In step S2, the fuzzy coefficient of the multi-fuzzy factor fuzzy C-means clustering is dynamically adjusted according to the node neighborhood density and the node energy state, so that the membership of the nodes in the dense area tends to be determined and the nodes in the sparse area retain the multi-cluster membership;

[0073] The calculation expression of the fuzzy coefficient is:

[0074]

[0075] In the formula, m i is the fuzzy coefficient of node i; m L , m U are the minimum value and the maximum value of the fuzzy coefficient respectively, used for controlling the coefficient variation range; δ i represents the average distance of node i and its adjacent nodes, and is regarded as the density index of the area where node i is located; δ min is the minimum average distance of all nodes in the network; δ max is the maximum average distance of all nodes in the network; and a is an index parameter for adjusting the sensitivity of the fuzzy coefficient to the change of the node neighborhood density.

[0076] In step S2, the step of constructing the hierarchical communication structure is:

[0077] When the source node and the destination node are located in the same cluster, the intra-cluster routing is performed, and the routing request is performed in the cluster;

[0078] When the source node and the destination node are located in different clusters, the inter-cluster routing is performed, and the routing request is completed by performing limited range forwarding between the cluster head nodes.

[0079] The structure includes backbone clusters, standard clusters, and cluster heads (CHs), member drones (CMs) and gateway drones (GWs) in each cluster as communication hubs across clusters. After clustering is completed, in each cluster, the node with the strongest comprehensive pheromone is elected as the cluster head of the cluster, and the election is completed through "cluster head announcement".

[0080] S3. Hierarchical routing discovery and establishment based on ant colony optimization;

[0081] When the source node needs to send data, a routing request packet carrying link stability, node residual energy, link congestion and path hop count is generated and broadcasted;

[0082] After the intermediate node receives the routing request packet, if the residual energy of the current intermediate node is lower than the energy threshold E th , the routing request packet is discarded; if the residual energy of the current intermediate node is higher than the energy threshold E th , the link pheromone is calculated and selectively forwarded according to the pheromone threshold condition. The pheromone threshold is the lower limit of path quality set in the routing optimization stage, which is used to filter low-quality links.

[0083] After the destination node receives the routing request packet, a candidate path is selected according to the cumulative pheromone value and a routing reply packet is returned along the candidate path to establish the data transmission path.

[0084] Referring to the "routing discovery" and "routing establishment" stages in Figure 1 , this step uses the ant colony optimization algorithm to efficiently search and establish paths on the hierarchical network structure established in S2. When a path is needed, the source node sends an RREQ message (forwarding ant) carrying multi-dimensional information. The link pheromone:

[0085]

[0086] In the formula, LS ij represents the link stability between node i and node j; AERP is the average node residual energy percentage on the path; is a hop penalty factor for balancing the influence of path length; MCG represents the maximum congestion on the path; PL ij (t) represents the pheromone concentration of link ij at time t; PL ij (t+n) represents the pheromone concentration of link ij at time t+n; v max is the maximum moving speed of the node; m is the number of nodes or segments on the path; r is the basic evaporation coefficient. Among them, the link pheromone concentration decays with time, and the decay rate is positively related to the maximum moving speed of the node; p and Q are adjustment coefficients for controlling the overall amplitude of pheromone increase.

[0087] The calculation expression of the residual energy is as follows:

[0088]

[0089] In the formula, AERP i represents the updated path average residual energy percentage of the data packet received by the ith hop node; hopcount represents the number of nodes that have been hopped; and NERP represents the residual energy percentage of the current ith node.

[0090] Referring to the inter-cluster routing discovery example shown in Figure 3 When the source unmanned aerial vehicle S needs to communicate with the destination unmanned aerial vehicle D, the RREQ packet (purple arrow) is efficiently forwarded between the cluster heads H1 and H2, rather than being flooded throughout the network, thereby greatly reducing the routing discovery overhead. After the destination node receives the RREQ, the optimal path is selected, and the RREP packet (red arrow) is returned along the path. During the return process of the RREP packet, the pheromone of each link on the path is strengthened, so that the high-quality path has more advantages in subsequent route selection.

[0091] S4. Dynamic route selection and maintenance;

[0092] During route selection, the node preferentially selects the path with the highest pheromone concentration for data transmission. At the same time, the node can back up the suboptimal path in the routing table according to the received multiple RREP information, so as to quickly switch when the main path fails. The node continuously monitors the link state through the HELLO packet with an adaptive interval. When a link interruption is detected, the path repair mechanism is triggered immediately. Referring to Figure 4 When a path with a high pheromone concentration (for example, through H1-H2-H3) forms a failure path due to a link interruption, the mechanism of the present application can be triggered. Since the gateway node (such as H2 or H3) can have a backup cluster home, the system can quickly switch to a new backup path through another cluster head (such as H4), thereby realizing rapid failure recovery and ensuring the continuity of communication. This process is handled through the RERR (route error) packet to notify the relevant nodes to update their routing tables.

[0093] The calculation expression of the HELLO handshake packet sending interval is as follows:

[0094]

[0095] In the formula, T Hello represents the HELLO handshake packet sending interval; T0 represents a basic time interval, that is, a static default interval reference value; η represents a network density coefficient, which is used to measure the number of node neighbors or the local congestion degree; represents the average speed of the node (including the average value of the current node and the neighbor nodes).

[0096] The specific embodiment of the present application realizes a dynamic routing method capable of adaptive clustering, intelligent path searching, and having a fast fault recovery capability through the ordered steps of S1 to S4, and effectively solves the communication problem of the UAV cluster in a high dynamic environment.

[0097] It is declared that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the purpose and scope of the technical solutions, which should be covered in the scope of the claims of the present application.

Claims

1. A dynamic routing method for a UAV swarm based on dynamic clustering with adaptive blurring coefficients, characterized in that, Comprising the following steps: S1. Network initialization; Initial parameters configuration: configuration of the communication radius, the maximum speed v max and the energy threshold E th for each drone Topology discovery: Obtain the state information of the UAV node itself and neighbor nodes, including the geographic location information of the UAV node, link congestion, node residual energy, link signal strength and path hop count, and construct a unified multi-dimensional pheromone model based on the state information; Time synchronization: Complete the network clock alignment by using GPS or PTP protocol; S2. Cluster head election; Based on the multi-dimensional pheromone model containing state information in step S1, multiple fuzzy factor fuzzy C-means clustering is performed on the UAV nodes, then the nodes are divided into cluster member nodes, cluster head nodes and gateway nodes according to the membership, and a hierarchical communication structure is constructed; Wherein, the vector after clustering is a complex number, the real part of the complex number is the geographic location information of the UAV node, and the imaginary part is the comprehensive information of the UAV node communication and routing performance; The cluster head node is responsible for cluster routing maintenance, and the gateway node maintains the fuzzy membership of two or more clusters and undertakes the cross-cluster routing forwarding function; S3. Route discovery; That is, when the source node needs to send data, a routing request packet carrying link stability, node residual energy, link congestion and path hop count is generated and broadcasted; S4. Intermediate forwarding of routing request packet; That is, after receiving the routing request packet, if the residual energy of the current intermediate node is lower than the energy threshold E th , the routing request packet is discarded; if the residual energy of the current intermediate node is higher than the energy threshold E th , the link pheromone is calculated and selectively forwarded according to the pheromone threshold condition. Wherein, the pheromone threshold is the lower limit of path quality set in the routing optimization stage, which is used to filter low-quality links; S5. Route establishment; That is, after the destination node receives the routing request packet, a candidate path is selected according to the cumulative pheromone value, and a routing reply packet is returned along the candidate path to establish a data transmission path; S6. Exponential decay and positive feedback gain are performed on the total pheromone of the transmission path during data transmission to dynamically maintain the path weight; S7. Path selection; In the route selection, the node preferentially selects the path with the highest pheromone concentration for data transmission; At the same time, the node backs up the suboptimal path in the routing table according to the received multiple RREP information, so as to quickly switch when the main path fails; S8. Route maintenance; According to the node moving speed and neighbor node density, the HELLO handshake packet sending interval is adjusted adaptively to realize real-time link maintenance; In the routing table maintenance stage, the path record with pheromone concentration lower than the maintenance threshold is deleted to reduce the size of the routing table and improve the search efficiency; The maintenance threshold is mainly used in the routing maintenance stage, and by periodically removing the paths with too low pheromone concentration, the invalid paths are avoided to occupy the routing table resources. 2.The dynamic routing method for UAV cluster based on dynamic clustering with adaptive blurring coefficients according to claim 1, characterized in that: In the step S2, the fuzzy coefficient of the multiple fuzzy factor fuzzy C-means clustering is dynamically adjusted according to the node neighborhood density and node energy state, so that the membership of the nodes in the dense area tends to be determined and the nodes in the sparse area retain multiple cluster memberships; The calculation expression of the fuzzy coefficient is: where m i is the fuzzification coefficient of node i; m L , m U are the minimum and maximum values of the fuzzification coefficient, respectively, used to control the coefficient variation range; δ i represents the average distance of node i from its neighboring nodes, which is considered as the density index of the area where node i is located; δ min is the minimum average distance of all nodes in the network; δ max is the maximum average distance of all nodes in the network; and α is an exponential parameter that adjusts the sensitivity of the fuzzification coefficient to the change in the density of the node neighborhood. 3.The dynamic routing method for UAV swarm based on dynamic clustering with adaptive blurring coefficients according to claim 2, characterized in that, In the step S2, the step of constructing the hierarchical communication structure is: When the source node and the destination node are located in the same cluster, intra-cluster routing is performed, and the routing request is performed in the cluster; When the source node and the destination node are located in different clusters, inter-cluster routing is performed, and the routing request is completed by limited range forwarding between cluster head nodes.

4. The dynamic routing method of the UAV cluster based on dynamic clustering with adaptive blurring coefficients according to claim 2, characterized in that, In the step S4, the calculation expression of the link pheromone is: wherein PL ij represents the link pheromone between node i and node j; LS ij represents the link stability between node i and node j; AERP is the average node residual energy percentage on the path; is the hop penalty factor, used to balance the influence of path length; MCG represents the maximum congestion degree on the path.

5. The dynamic routing method of the UAV cluster based on dynamic clustering with adaptive blurring coefficients according to claim 3, characterized in that, In the step S6, the calculation expression of the exponential decay and volatilization of the link pheromone is as follows: wherein PL ij (t) represents the pheromone concentration of link ij at time t; PL ij (t + n) represents the pheromone concentration of link ij at time t + n; v max is the maximum moving speed of the node; m is the number of nodes or segments on the path; and r is the basic evaporation coefficient.

6. The dynamic routing method of UAV cluster based on dynamic clustering with adaptive blurring coefficients according to claim 3, characterized in that, In the step S6, the calculation expression of the positive feedback gain of the link pheromone is as follows: where p and Q are tuning factors to control the overall magnitude of pheromone increase; ACK count is the number of successful acknowledgements of the packet by the nodes on the path.

7. The dynamic routing method of UAV cluster based on dynamic clustering with adaptive blurring coefficients according to claim 3, characterized in that, In the step S7, the calculation expression of the HELLO handshake packet sending interval is as follows: In the formula, T Hello HELLO handshake message sending interval; T0 is the base time interval, i.e. the static default interval reference value; η represents the network density coefficient, which is used to measure the number of node neighbors or the degree of local congestion; representing the average speed of the node. 8.The dynamic routing method for UAV swarm based on dynamic clustering with adaptive blurring coefficients according to claim 3, characterized in that: In the step S8, the route maintenance further includes a failure recovery mechanism. When a link is interrupted, the gateway node with multiple backup cluster homes can quickly switch to the backup cluster for routing.

Images