Method for collaborative election of low-delay cluster head and gateway in clustering ad hoc network

Abstract

The invention relates to a method for collaborative election of a low-delay cluster head and a gateway in a clustering ad hoc network, and belongs to the technical field of wireless communication. The method comprises the following steps: firstly, constructing an initial clustering multi-level self-organizing network model which comprises a cluster structure, node distribution and a communication channel model; secondly, performing cluster head election by adopting a combined weighting method, and constructing a utility function to select an optimal cluster head by comprehensively considering a node degree, an average routing hop count and an inter-node distance; and finally, designing a gateway node election optimization problem, and ensuring that the elected gateway node has the optimal cross-cluster communication efficiency by minimizing the end-to-end time delay between adjacent clusters and utilizing the improved ant colony algorithm to solve. Compared with a traditional method, the cluster head management capability and the gateway node cross-cluster communication efficiency are effectively improved, the network time delay is remarkably reduced, and the method is suitable for military communication, environment monitoring, intelligent traffic, disaster rescue and other scenes.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and specifically to a method for low-latency cluster head and gateway collaborative election in a clustered ad hoc network. Background Technology

[0002] Ad hoc networks, as a wireless network technology that can autonomously form networks without infrastructure support, have broad application prospects in fields such as military communications, environmental monitoring, intelligent transportation, and disaster relief. However, due to their dynamic topology and resource constraints, ad hoc networks face many challenges in network performance optimization, especially in large-scale clustered networks. How to efficiently elect cluster heads (CHs) and gateways (GWs) to reduce network latency has become a key research issue.

[0003] Ad hoc networks typically employ a clustering architecture to optimize network performance. A clustering architecture divides the network into multiple clusters, each consisting of a cluster head and multiple regular nodes. The cluster head is responsible for coordinating intra-cluster communication and forwarding data to other cluster heads or gateways. Clustering effectively reduces redundant communication in the network, lowers energy consumption, and improves network scalability. However, the performance of a clustering architecture largely depends on the election mechanism for cluster heads and gateways.

[0004] Traditional cluster head election algorithms primarily rely on factors such as a node's hop count and the number of its neighboring nodes. While using a single criterion based solely on hop count or the number of neighboring nodes simplifies the election process to some extent, it also has significant limitations. For example, an election method based solely on hop count might select a node that is close to the gateway but has high latency as the cluster head, resulting in high latency for routes passing through that node and impacting packet transmission. Similarly, an election method based solely on the number of neighboring nodes might select a node with high connectivity but excessive load as the cluster head, causing that node to become a bottleneck in network performance due to handling too much data forwarding.

[0005] Gateways are critical nodes for communication between adjacent clusters in an ad hoc network, and their performance directly impacts network latency and throughput. As a bridge connecting different clusters, gateways not only need to efficiently forward inter-cluster data but also coordinate communication between cluster head nodes to ensure reliable and real-time data transmission. Traditional gateway election mechanisms are typically independent of cluster head election, focusing only on the connection quality between the gateway and adjacent clusters, such as signal strength and bandwidth availability, while neglecting coordination with cluster head nodes. This fragmented election approach can lead to excessively long data transmission paths and increased latency, especially in large-scale networks where latency issues are particularly pronounced. Due to the large network size, data transmission from the source node to the target cluster may require multiple hops, each adding latency. If the communication path between the gateway and cluster head nodes is not optimized, the total data transmission latency can significantly increase, affecting the network's real-time performance and reliability.

[0006] In recent years, with the rapid development of the Internet of Things (IoT) and the Industrial Internet, low-latency communication has become a crucial requirement for ad hoc networks. For example, in industrial automation scenarios, sensor nodes need to transmit collected data to the control center in real time; any communication delay could lead to system malfunction. Therefore, how to achieve collaborative election between cluster heads and gateways to optimize data transmission paths and reduce end-to-end latency has become a key research focus. Summary of the Invention

[0007] This invention addresses the problems of low communication efficiency and network congestion in large-scale self-organizing networks by combining a clustered multi-level network architecture with resource scheduling optimization technology. Through joint optimization of cluster head election and gateway node election, it improves network management efficiency and communication performance.

[0008] The technical solution adopted in this invention is as follows:

[0009] A method for low-latency cluster head and gateway collaborative election in a clustered ad hoc network includes the following steps:

[0010] Step 1: Construct an initial clustered multi-level self-organizing network model, including a cluster structure model, a node distribution model, and an inter-node communication channel model. The cluster structure model and node distribution model divide the nodes in the self-organizing network into multiple clusters, each cluster containing multiple nodes, with the nodes in the overlapping area of ​​two clusters serving as candidate gateway nodes. The inter-node communication channel model adopts the free space path loss model.

[0011] Step 2: Taking into account the node degree, average route hops, and distance between nodes in each cluster, construct a weighted utility function and select the node with the smallest utility value as the cluster head;

[0012] Step 3: Based on the jitter, packet loss rate and end-to-end delay on a certain route during communication between adjacent clusters, design an optimization problem for gateway node election. By minimizing the end-to-end delay between adjacent clusters, an improved ant colony algorithm is used to solve the problem between adjacent clusters, and the candidate gateway node traversed by the optimal route is selected as the gateway node.

[0013] Furthermore, the cluster structure model and node distribution model in step 1 specifically involve dividing the nodes in the self-organizing network into multiple clusters, with the node set of each cluster represented as follows: in, Let i represent the i-th node in the k-th cluster, where 1 ≤ i ≤ N. k N k This represents the total number of nodes in the k-th cluster, where K is the number of clusters. Nodes existing in the overlapping area of ​​two clusters are candidate gateway nodes.

[0014] Furthermore, in step 1, the inter-node communication channel model uses a free-space path loss model to describe the inter-node communication channel, with channel gain h. i,j The calculation formula is:

[0015]

[0016] In the formula, G t G represents the transmit antenna gain of node i. r Let λ represent the receiving antenna gain at node j, λ represent the wavelength, λ = c / f, c represent the speed of light, f represent the frequency, and d represent the frequency. i,j This represents the distance between node i and node j.

[0017] Furthermore, the weight-based utility function in step 2 is:

[0018]

[0019] In the formula, a, b, and c are the weight factors corresponding to the three factors, and a+b+c=1, 0<a,b,c<1. The weight factors are set according to the specific network environment and business requirements. This represents the degree difference of node i within the k-th cluster: The expression is: in, It is a binary variable. Let d(i,j) represent the distance between node i and node j, and r represent the maximum communication range of the nodes. Let C be the distance between node i and node j. k The set of nodes within a cluster; The average number of routing hops for node i in the k-th cluster is expressed as: in, This represents the number of hops between node i and node j in the k-th cluster; The average distance between node i and node j in the k-th cluster is expressed as: in, and Let represent the number of all one-hop and two-hop neighbor nodes of node i, respectively. This represents all one-hop nodes of node i in the k-th cluster. and Let i, j, and z represent the coordinates of nodes i, j, and z in the two-dimensional plane, respectively, and let z be the two-hop neighbor of node i.

[0020] Furthermore, the optimization problem of gateway node election in step 3 is as follows:

[0021]

[0022] The constraints include the time delay constraint T. t,q (p)≤T max Jitter constraint J t,q (p)≤J max and packet loss rate constraint α t,q (p)≤α max ;T t,q (p) represents the neighboring cluster C t and C q When communicating between routes, the end-to-end delay of the p-th route is expressed as: in, P t Let δ represent the transmit power of the transmitting node, δ represent the noise power, S represent the size of the transmitted data packet, B represent the transmission bandwidth, h(e) represent the channel gain, and ε represent the transmission bandwidth. t,q (p) represents cluster C t and C q edge set, N t,q (p) represents cluster C t and C q The set of nodes, T t,q (n) represents the node delay, T max Let P be the maximum delay for route transmission, and P be the set of paths from the source node to the destination node; J t,q (p) represents the neighboring cluster C t and C q During communication between them, the jitter on the p-th route, J max α represents the maximum jitter in the routed transmission. t,q (p) represents the neighboring cluster C t and C q When communicating between routes, the packet loss rate α on the p-th route is... max This represents the maximum packet loss rate for routed transmission.

[0023] Furthermore, the specific process of using the improved ant colony algorithm to solve the problem among adjacent clusters in step 3 is as follows:

[0024] First, initialize the parameters required by the algorithm; then, construct the path selection strategy: for each ant k, starting from the source cluster head node, the set of neighbor nodes of the current node i is N. i According to the probability formula Choose the next hop node j, where τ ij For pheromone concentration, η ij Heuristic information is defined as the reciprocal of latency, meaning that the path with the smaller the latency, the greater its attraction to ants. α and β are the pheromone importance factor and the heuristic information importance factor, respectively.

[0025] After each iteration, based on the path chosen by the ants, the path with the minimum latency between adjacent clusters is found, and the pheromone is updated: τ ij =(1-ρ)τ ij +ρΔτ ij ,in, ρ represents the pheromone evaporation factor, Q is the total amount of pheromone carried by each ant, and cost(p) is the cost function of the current path p.

[0026]

[0027] Where ω1, ω2 and ω3 are the weights of latency, jitter and packet loss rate, respectively, and T(p), J(p) and α(p) are the latency, jitter and packet loss rate on path p, respectively. By repeatedly going through the path selection and pheromone update process, after multiple rounds of iteration, the ant colony gradually converges to the path with the minimum latency, satisfying the requirement that there is only one gateway node between adjacent clusters.

[0028] The advantages of this invention compared to the prior art are as follows:

[0029] This invention proposes a method for low-latency cluster head and gateway collaborative election in clustered ad hoc networks. Based on an initial clustered multi-level ad hoc network, a combined weighted method is used to comprehensively consider three factors—node degree, average routing hop count, and inter-node distance—to elect the cluster head, ensuring efficient management of nodes within the cluster. An optimization problem for gateway node election is designed and solved using an improved ant colony algorithm to minimize end-to-end latency between adjacent clusters. Compared to traditional methods, this invention effectively improves cluster head management capabilities and gateway node cross-cluster communication efficiency, demonstrating significant application value. Attached Figure Description

[0030] Figure 1 This is a network diagram of the self-organizing network clustering structure of the present invention.

[0031] Figure 2 This is a schematic diagram of the cluster head election results under the self-organizing network clustering structure of the present invention.

[0032] Figure 3 This diagram illustrates the comparison between cluster head election in this invention and other schemes, comparing the total transmission delay from all nodes within the cluster to the gateway.

[0033] Figure 4 This is a schematic diagram of the gateway election results after cluster head election in the self-organizing network clustering structure of the present invention.

[0034] Figure 5 This diagram illustrates the comparison between the present invention and other solutions, showing the transmission delay under the same route. Detailed Implementation

[0035] The present invention will be further explained below with reference to the accompanying drawings.

[0036] A method for low-latency cluster head and gateway collaborative election in a clustered ad hoc network includes a pre-defined cluster structure, cluster member nodes, and candidate gateway nodes.

[0037] In the divided cluster structure, there are ordinary cluster member nodes. Cluster member nodes are a low-level network structure in self-organizing networks. Their functions are relatively simple. The cluster head needs to maintain communication within the cluster and coordinate the interaction between clusters to provide a basis for further gateway election. In addition, candidate gateway nodes are located in the overlapping area of ​​two intersecting clusters and are responsible for information transmission between different clusters.

[0038] The specific steps are as follows:

[0039] Step 1: Construct an initial clustered multi-level self-organizing network model based on the actual scenario, including a cluster structure model, a node distribution model, and an inter-node communication channel model.

[0040] The cluster structure model and node distribution model in step 1 are specifically as follows: the existing N nodes in the self-organizing network are divided into K clusters, and the node set of each cluster can be represented as follows: in, N represents the i-th node in the k-th cluster. k This represents the total number of nodes in the k-th cluster; nodes existing in the overlapping area of ​​two clusters are candidate gateway nodes. For communication between nodes within a cluster, it is assumed that communication between nodes within a cluster occurs at most twice. This assumption helps simplify the network model while ensuring communication efficiency and reliability.

[0041] The inter-node communication channel model is as follows: the channel gain between node i and node j is represented by h. i,j ,in Considering that communication between nodes in a wireless ad hoc network is point-to-point, a free-space path loss model is used to describe the communication channel between nodes:

[0042]

[0043] In the formula, G t G represents the transmit antenna gain of node i. r Let λ represent the receiving antenna gain at node j, λ be the wavelength, and λ = c / f, where c represents the speed of light, f represents the frequency, and d i,j This represents the distance between node i and node j.

[0044] Step 2: Use a combined weighted method to elect cluster heads. Consider the three factors of node degree, average routing hops and distance between nodes in each cluster, construct a weighted utility function, and select the node with the smallest utility value as the cluster head to ensure that it has the ability to efficiently manage nodes in the cluster.

[0045] The construction of the cluster head election utility function in step 2 is as follows: Nodes obtain information about their one-hop and two-hop neighbors by sending Hello messages, and then calculate the number of directly connected neighbor nodes, i.e., the node degree; the node degree of node i in the k-th cluster is:

[0046]

[0047] In the formula, It is a binary variable. This indicates that node j is a neighbor of node i, otherwise it is not; d(i,j) is the distance between node i and node j, r is the maximum communication range of the nodes, and C k The set of nodes within a cluster;

[0048] To achieve a more balanced load distribution among the elected cluster head nodes in the network, node differentials are introduced:

[0049]

[0050] In the formula, This represents the degree difference of node i within the k-th cluster.

[0051] To quantify the congestion level of communication between different nodes acting as cluster heads, the average number of routing hops for node i in the k-th cluster is defined as:

[0052]

[0053] In the formula, This represents the number of hops between node i and node j in the k-th cluster.

[0054] Considering that the degree of signal attenuation is closely related to the transmission distance, the distance between nodes is used to quantify signal attenuation. In the k-th cluster, the average distance between node i and node j is expressed as:

[0055]

[0056] in, Let represent the number of all one-hop and two-hop neighbor nodes of node i, respectively. This represents all one-hop nodes of node i in the k-th cluster. and Let i, j, and z represent the coordinates of nodes i, j, and z in the two-dimensional plane, respectively, and let z be the two-hop neighbor of node i.

[0057] Taking into account node degree, average route hop count, and distance between nodes, the weight-based utility function is as follows:

[0058]

[0059] Where a, b, and c are the weighting factors corresponding to the three factors, and a + b + c = 1, 0 < a, b, c < 1. The weighting factors are set according to the specific network environment and business requirements to improve the rationality of cluster head election.

[0060] Step 3: Building upon Step 2, design an optimization problem for gateway node election. By minimizing the end-to-end latency between adjacent clusters, ensure that the elected gateway node has the best cross-cluster communication efficiency. An improved ant colony algorithm is used to efficiently solve the problem between adjacent clusters, selecting the candidate gateway node traversed by the optimal route as the gateway node.

[0061] The optimization problem in step 3, the election of the gateway node, is as follows:

[0062] When transmitting information between clusters, let the set of paths from the source node to the destination node be P. For a path p∈P, its edge set is denoted as ε(p), and its node set is denoted as N(p). Considering that the arrival time of the previous data packet and the subsequent data packets fluctuate when the data packets are transmitted between clusters, J is defined as... t,q (p) represents the neighboring cluster C t and C q During communication between them, the jitter on the p-th route is:

[0063]

[0064] Where, ε t,q (p) represents cluster C t and C q edge set, J t,q (e) represents the edge jitter, N t,q (p) represents cluster C t and C q The set of nodes, J t,q (n) represents the jitter on the node;

[0065] Data packets have a certain probability of being lost during transmission; neighboring cluster C t and C q When communicating between them, use α t,q (e) represents the single-hop packet loss rate, then the packet loss rate on the p-th route is:

[0066]

[0067] To measure the performance of gateway election, neighboring cluster C t and C q When communicating between them, the end-to-end delay of the p-th route is:

[0068]

[0069] Among them, T t,q (n) represents the node delay. P t δ represents the transmit power of the transmitting node, S represents the size of the transmitted data packet, B represents the transmission bandwidth, and h(e) is the channel gain.

[0070] This invention focuses on a virtual backbone network composed of candidate gateway nodes and cluster heads. By minimizing the end-to-end latency between adjacent clusters, it selects the candidate gateways traversed by the optimal route as gateway nodes. The optimization problem is:

[0071]

[0072] Where C1 represents the delay constraint in routing transmission, T max C1 represents the maximum delay of the route transmission, C2 represents the jitter constraint in the route transmission, and J represents the maximum delay of the route transmission. max C3 represents the maximum jitter in routing transmission, C3 represents the packet loss rate constraint in routing transmission, and α represents the maximum jitter in routing transmission. max This indicates the maximum packet loss rate during route transmission.

[0073] Gateway election utilizes the ant colony optimization algorithm to solve the aforementioned optimization problem, finding the path with the minimum latency between adjacent clusters and ensuring that only one gateway node is elected between two adjacent clusters. Specifically, first, in adjacent cluster C... t and C q The algorithm prioritizes selecting a node with only one candidate gateway as the gateway node. Secondly, if there are multiple common points between adjacent clusters, the improved ant colony algorithm is used to optimize cluster C. t From cluster head to cluster C q The transmission latency of the cluster head is considered, and the common node on the path with the minimum transmission latency is used as the gateway.

[0074] The process of solving the above optimization problem using the ant colony optimization algorithm is as follows: First, initialize the parameters required by the algorithm. Second, construct a path selection strategy: for each ant k, starting from the head node of the source cluster, the set of neighbor nodes of the current node i is N. i According to the probability formula Choose the next hop node j, where τ ij For pheromone concentration, η ij The heuristic information is defined as the reciprocal of the latency, meaning that paths with shorter latency are more attractive to ants. α and β are the pheromone importance factor and the heuristic information importance factor, respectively. After each iteration, the program finds the path with the shortest latency between adjacent clusters based on the path chosen by the ants and updates the pheromone level: τ ij =(1-ρ)τ ij +ρΔτ ij ,in, ρ represents the pheromone evaporation factor, Q is the total amount of pheromone carried by each ant, and cost(p) is the cost function of the current path p.

[0075]

[0076] Where ω1, ω2 and ω3 are the weights of latency, jitter and packet loss rate, respectively, and T(p), J(p) and α(p) are the latency, jitter and packet loss rate on path p, respectively. By repeatedly going through the path selection and pheromone update process, after multiple rounds of iteration, the ant colony gradually converges to the path with the minimum latency, satisfying the requirement that there is only one gateway node between adjacent clusters.

[0077] Example:

[0078] Figure 1 In the network model shown, the simulation area is set to 250m × 250m, with 64 nodes and a maximum communication range of 50m for each node. Each node has a unique ID and operates in half-duplex mode to ensure efficient communication. A stable network environment is assumed, and the topology remains unchanged during the simulation. Weights are assigned as a = 0.2, b = 0.2, and c = 0.6 to achieve initial network layering.

[0079] Figure 2 The results of the cluster head election are displayed. The elected cluster head nodes are identified by triangles. They will serve as the management center of their respective clusters, responsible for coordinating intra-cluster communication and inter-cluster interactions. Ordinary nodes are identified by circles. They will follow the instructions of the cluster head and participate in intra-cluster data transmission and communication.

[0080] Figure 3The total transmission latency of two cluster head election algorithms in randomly selected clusters was compared. The horizontal axis represents the cluster ID, and the vertical axis represents the total transmission latency from all nodes within the cluster to the gateway. In the figure, the on-demand weighted cluster head election algorithm proposed in this invention is represented in white, and the cluster head election algorithm that minimizes the number of routing hops is represented in black. By observing the graph, it is clear that the algorithm proposed in this invention can achieve lower transmission latency when electing cluster heads, thereby transmitting data more efficiently. This result highlights the importance of considering the distance between nodes in the cluster head election process and demonstrates the superiority of the on-demand weighted algorithm proposed in this invention in reducing transmission latency and improving network efficiency.

[0081] Figure 4 The results of gateway election based on end-to-end latency are presented. It is clearly observed that on the determined optimal path, specific candidate gateway nodes are selected as the final gateway nodes (represented by pentagrams). These elected gateway nodes will play a crucial role in inter-cluster communication, responsible for maintaining and optimizing cross-cluster data transmission, thus providing a solid foundation for the stable operation and efficient communication of the wireless ad hoc network.

[0082] Figure 5 This section compares the gateway election algorithm based on end-to-end latency with the election method that randomly selects gateway nodes. Transmission latency on the routes is used as a key performance indicator to compare the gateway performance under different election methods. Four routes are shown below:

[0083] Route 1: Cluster Head 2 — Cluster Head 1 — Cluster Head 3

[0084] Route 2: Cluster head 2 — Cluster head 4 — Cluster head 6

[0085] Route 3: Cluster head 4 — Cluster head 6 — Cluster head 5

[0086] Route 4: Cluster head 3 — Cluster head 5 — Cluster head 6

[0087] The horizontal axis in the graph represents four routes, and the vertical axis represents the total inter-cluster transmission delay along those routes. White represents the route delay corresponding to the end-to-end delay-based gateway election algorithm, while black represents the route delay corresponding to random gateway election. A close examination of the graph clearly shows that the proposed end-to-end delay-based gateway node election algorithm performs significantly better in reducing transmission latency, indicating that this algorithm can transmit data more efficiently. This difference highlights the potential of the algorithm in this invention to optimize cross-cluster communication efficiency in practical network applications.

Claims

1. A method for low-latency cluster head and gateway co-election in a clustered ad hoc network, characterized in that, The method comprises the following steps: Step 1: constructing an initial clustering multi-level self-organizing network model, including a cluster structure model, a node distribution model and a node intercommunication channel model; wherein the cluster structure model and the node distribution model divide nodes in the self-organizing network into multiple clusters, each cluster comprising multiple nodes, and nodes in an overlapping part of two clusters being candidate gateway nodes; and the node intercommunication channel model adopts a free space path loss model; Step 2: comprehensively considering node degree, average routing hop number and node distance of nodes in each cluster, constructing a weight-based utility function, and selecting a node with a minimum utility value as a cluster head; Step 3: designing an optimization problem of gateway node election according to jitter, packet loss rate and end-to-end delay on a certain route during intercommunication between adjacent clusters, solving the optimization problem between adjacent clusters by using an improved ant colony algorithm by minimizing end-to-end delay between adjacent clusters, and selecting a candidate gateway node passed through by an optimal route as a gateway node.

2. The method of claim 1, wherein, The cluster structure model and the node distribution model in step 1 are specifically: dividing nodes in the self-organizing network into multiple clusters, and a node set of each cluster is represented as wherein, represents the i-th node in the k-th cluster, 1≤i≤N k , N k represents the total number of nodes in the k-th cluster, K is the number of divided clusters; and the node existing in the overlapping part of two clusters is a candidate gateway node.

3. The method of claim 1, wherein, The inter-node communication channel model in step 1 adopts a free space path loss model to describe the inter-node communication channel, and the channel gain h i,j The calculation formula is: where G t represents the transmit antenna gain of node i, G r represents the receive antenna gain of node j, λ represents the wavelength, λ = c / f, c represents the speed of light, and f represents the frequency, d i,j represents the distance between node i and node j.

4. The method of claim 2, wherein, The weight-based utility function in step 2 is: In the formula, a, b and c are weight factors of the three factors respectively, and a+b+c=1, 0 The node degree difference of node i in the kth cluster is represented as: The expression is: Wherein, is a binary variable, Node j is a neighbor of node i, otherwise not, d(i,j) is the distance between node i and node j, r is the maximum communication range of the node, C k is the node set in the cluster; The average routing hop number of node i in the kth cluster is represented as: Wherein, The hop number between node i and node j in the kth cluster is represented as: The average distance between node i and node j in the kth cluster is represented as: where, and denote the number of one-hop and two-hop neighbor nodes of node i, respectively, denote all one-hop nodes of node i in the kth cluster, and denote the coordinates of nodes i, j and z in a two-dimensional plane, where z is a two-hop neighbor of node i.

5. The method of claim 1, wherein, The optimization problem of gateway node election in step 3 is: The constraints include a delay constraint T t,q (p)≤T max , a jitter constraint J t,q (p)≤J max , and a packet loss rate constraint α t,q (p)≤α max ; T t,q (p) is the end-to-end delay of the pth route when communicating between adjacent clusters C t and C q , expressed as: wherein, P t denotes the transmission power of the transmitting node, δ denotes the noise power, S is the size of the transmitted data packet, B denotes the transmission bandwidth, h(e) is the channel gain, and ε t,q (p) is the edge set of clusters C t and C q , N t,q (p) is the node set of clusters C t and C q , T t,q (n) is the node delay, T max is the maximum delay of the route transmission, and P is the path set from the source node to the destination node; J t,q (p) is the jitter on the pth route when communicating between adjacent clusters C t and C q , J max is the maximum jitter of the route transmission; α t,q (p) is the packet loss rate on the pth route when communicating between adjacent clusters C t and C q , α max is the maximum packet loss rate of the route transmission.

6. The method of claim 5, wherein, The specific process of solving between adjacent clusters by using the improved ant colony algorithm in step 3 is: First, the parameters required by the algorithm are initialized; then, the path selection strategy is constructed, and for each ant k, the neighbor node set of the current node i is N i , the next hop node j is selected according to the probability formula , where τ ij is the pheromone concentration, η ij is the heuristic information defined as the reciprocal of the delay, that is, the smaller the path delay, the greater the attraction of the path to the ant, and α and β are the pheromone importance factor and the heuristic information importance factor, respectively; After each round of iteration, according to the path selected by the ants, find the path with the minimum delay between adjacent clusters, and update the pheromone: τ ij = (1 - p) τ ij + p Δ τ ij , where, p represents the pheromone evaporation factor, Q is the total amount of pheromone carried by each ant, and cost(p) is the cost function of the current path p: wherein ω1, ω2 and ω3 are weights of delay, jitter and packet loss rate respectively, T(p), J(p) and α(p) are delay, jitter and packet loss rate on a path p respectively; the process of path selection and pheromone update is repeatedly performed, and after multiple rounds of iteration, the ant colony gradually converges to a path with minimum delay, thereby meeting the requirement that there is only one gateway node between adjacent clusters.

Images