A low-overhead energy-efficient routing method suitable for large-scale mobile ad hoc networks

Abstract

The application is a low-overhead energy-saving routing method suitable for large-scale mobile ad hoc networks, comprising the following steps: first step: node state information interaction and maintenance; second step: node capability index calculation; third step: node capability value evaluation based on dynamic weight; fourth step: cluster head election and distributed clustering; fifth step: cluster maintenance; sixth step: gateway node selection; seventh step: inter-cluster on-demand routing construction; the core idea of the method is: combining node residual energy, node connectivity and average link persistence and other indexes, the comprehensive capability of the node is evaluated, and the CRITIC objective weighting method is used to realize the adaptive adjustment of the index weight; on this basis, the DPC density peak clustering idea is introduced, the cluster head election and distributed clustering are completed by combining the local density and separation degree of the node; then, through the cluster member maintenance, cluster head maintenance and inter-cluster on-demand routing construction mechanism, the control message overhead and node energy consumption in the large-scale MANET are reduced.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication network technology, specifically relating to a low-overhead, energy-saving routing method suitable for large-scale mobile ad hoc networks (MANETs). Background Technology

[0002] MANET is a distributed network composed of multiple wireless nodes that operate in a self-organizing manner without the need for fixed infrastructure. Network nodes can act as source or destination nodes for data transmission and also possess data forwarding capabilities. Due to node mobility, the network topology constantly changes. Traditional planar on-demand routing methods, such as Ad hoc On-Demand Distance Vector Routing (AODV), typically require flooding of routing requests over a large area or even the entire network. This can easily lead to increased routing control overhead and energy consumption as the network grows.

[0003] To alleviate the flooding overhead and energy consumption problems of planar routing in large-scale MANETs, ​​clustering mechanisms are widely used for MANET routing optimization. By dividing the network into several local clusters, the propagation of control messages that would otherwise flood the entire network can be confined to a local scope, and hierarchical forwarding paths can be constructed using cluster heads, thereby reducing the overall overhead of route discovery and maintenance. However, existing clustering routing protocols still have shortcomings in the clustering process, cluster structure maintenance, and inter-cluster communication. First, in the clustering process, most existing methods rely on centralized global information or simple local competition rules, and the weights of the cluster head election indicators are usually fixed, unable to adaptively adjust according to changes in network status, easily leading to unbalanced cluster head distribution and unreasonable cluster partitioning. Second, in the cluster structure maintenance phase, insufficient consideration is given to local topology changes such as member switching caused by node movement and inter-cluster overlap, resulting in poor cluster structure stability. Finally, in the inter-cluster communication phase, if the on-demand routing selection between cluster heads is unreasonable, it will not only generate additional control overhead but also further exacerbate node energy consumption.

[0004] Therefore, existing technologies lack a clustering routing method that can adapt to dynamic topology changes in large-scale MANETs and take into account cluster head election, cluster structure maintenance, and low-overhead inter-cluster communication, making it difficult to simultaneously meet the requirements for low control overhead and energy-efficient communication in large-scale MANET scenarios.

[0005] To address the aforementioned problems, this invention proposes a distributed clustering low-overhead energy-saving routing method for large-scale MANETs (hereinafter referred to as the DC-LOEER method). This method first designs indicators such as node remaining energy, node connectivity, and average link persistence, and combines them with an objective weighting method (Criteria Importance Through IntercriteriaCorrelation, CRITIC) to achieve adaptive evaluation of node comprehensive capabilities. Subsequently, it introduces a density peak clustering (DPC) method, combining local density and separation to complete cluster head election and distributed clustering. Based on this, through cluster maintenance and on-demand inter-cluster routing construction mechanisms, it reduces control overhead and node energy consumption in large-scale MANETs. Summary of the Invention

[0006] Technical Problem: This invention proposes a distributed clustering low-overhead energy-saving routing method for large-scale mobile ad hoc networks (MANETs). The core idea of ​​this method is to evaluate the overall capability of nodes by combining indicators such as remaining node energy, node connectivity, and average link persistence, and to adaptively adjust the indicator weights using the CRITIC objective weighting method. Based on this, the method introduces the DPC density peak clustering concept, combining node local density and separation to complete cluster head election and distributed clustering. Subsequently, through cluster member maintenance, cluster head maintenance, and on-demand inter-cluster routing construction mechanisms, the method reduces control message overhead and node energy consumption in large-scale MANETs.

[0007] Technical solution: The present invention includes the following steps

[0008] Step 1: Node State Information Exchange and Maintenance: Each node in the network periodically broadcasts a HELLO message (a special routing reply message used for node state exchange). The message contains fields such as node identifier, current location, speed, direction of movement, and remaining energy, and is used to publish its own state information to its one-hop neighbors. After receiving a HELLO message from a neighboring node, each node updates its local neighbor table to record the state information of the corresponding neighboring node.

[0009] Step 2: Node Capability Metric Calculation: Based on the neighbor node status information recorded in the neighbor table, the current node calculates its remaining energy metric, connectivity metric, load level metric, and average link sustainability metric, and then normalizes each metric. This includes the following steps:

[0010] (1) The local state information of a node includes its remaining energy, location coordinates, movement speed and direction, and cache queue length. The neighbor table records the number of neighboring nodes and the location coordinates, movement speed and direction, and cache queue length of each neighboring node. First, the node at time... The remaining energy in the local state information is defined as the node's remaining energy. Then, the number of neighboring nodes in the node's neighbor table is defined as the node connectivity. Secondly, the length of the node's cache queue is defined as the node's load level. Finally, based on the node's own location coordinates, movement speed, and direction in its local state information, and the location coordinates, movement speed, and direction of neighboring nodes in the neighbor table, nodes whose distances to each neighbor are less than the node's communication radius are selected. The average time is defined as the average link duration. The calculation is as follows: Let the node be... and nodes At any moment The positions are respectively and The speeds are respectively and ,make:

[0011] (1),

[0012] Then the node can be obtained. and nodes Link duration between:

[0013] (2),

[0014] Set nodes The set of neighbor nodes is The number of neighboring nodes included is Then the average link duration can be obtained, that is:

[0015] (3),

[0016] (2) Calculate the node's remaining energy index. Normalize the node's remaining energy and define the node. The remaining energy index is:

[0017] (4),

[0018] in, For nodes The remaining energy. and Let be the maximum and minimum remaining energy values ​​of the current neighboring nodes, respectively, which can be expressed as:

[0019] (5),

[0020] (6),

[0021] (3) Calculate the node connectivity index. Normalize the node connectivity and define the node. The connectivity index is:

[0022] (7),

[0023] in, For nodes The connectivity. and Each is the current The maximum and minimum connectivity of neighboring nodes can be expressed as follows:

[0024] (8),

[0025] (9),

[0026] (4) Calculate the node load level index. Normalize the node load level and define the node. The load level index can be expressed as:

[0027] (10),

[0028] in, For nodes The load level, This represents the maximum length of the node's cache queue.

[0029] (5) Calculate the average link persistence metric. (This involves placing nodes...) Average link duration normalization defines the average link persistence metric. for:

[0030] (11),

[0031] in, For nodes The average link duration. and These are the maximum and minimum average link durations of the current neighboring nodes, respectively, and can be expressed as:

[0032] (12),

[0033] (13),

[0034] Step 3: Node capability evaluation based on dynamic weights. Using normalized indicators, the node capability value is calculated, and the weights of different indicators are adaptively adjusted using the CRITIC objective weighting method. The specific steps are as follows:

[0035] (1) Based on the three normalized indices defined in the second step, construct the node capability value function:

[0036] (14),

[0037] in, , , , These are the remaining energy indices of the nodes. Node connectivity index Node load level indicators and average link persistence metrics The corresponding weighting coefficients can be determined by the CRITIC objective weighting method.

[0038] (2) The CRITIC objective weighting method is used to achieve adaptive adjustment of the weights of each indicator in the node capability value function. Let the node contain nodes... Its own local node set is The set contains a total of nodes. Then, a node index matrix can be constructed, namely:

[0039] (15),

[0040] in, , , Each corresponding to any node The remaining energy index of the node, the connectivity index of the node, the load level index of the node, and the average link sustainability index.

[0041] Calculate the first The standard deviation of an indicator in a local set :

[0042] (16),

[0043] in, Identifier The average value of an indicator within a local set. Standard deviation. It reflects the degree of dispersion of the indicator within a local set. The larger the value, the stronger the indicator's ability to distinguish different nodes.

[0044] To measure the degree of information overlap between different indicators, the following calculation is further performed: The first indicator and the first Correlation coefficients among indicators :

[0045] (17),

[0046] in, If an indicator has a low correlation with other indicators, it means that it can provide more independent information, and its importance should be increased accordingly.

[0047] After obtaining the standard deviation and correlation coefficient, define the first... Information content of each indicator for:

[0048] (18),

[0049] Based on the information content of each indicator, the objective weights of the three indicators can be obtained, namely:

[0050] (19),

[0051] Step 4: Cluster Head Election and Distributed Clustering. Based on the capabilities of each node, and combining the DPC (Density Peak Clustering) approach, the local density and separation of the nodes are calculated. This is used to calculate the cluster head competition value. Nodes with higher local density and separation are preferentially selected as cluster heads. Ordinary nodes are joined to the cluster of the corresponding cluster head according to the proximity principle, completing the distributed clustering process. The specific steps are as follows:

[0052] (1) Calculate the predicted effective distance between nodes, assuming the nodes are... and nodes At any moment The coordinates are respectively and The distance between the two nodes is... It can be represented as:

[0053] (20),

[0054] Meanwhile, the relative position component and the relative velocity component are defined as follows:

[0055] (twenty one),

[0056] (twenty two),

[0057] Based on this, the radial relative velocity of the two nodes along the connecting line direction. It can be represented as:

[0058] (twenty three),

[0059] If the HELLO message sending cycle As a short-term prediction window, the node and nodes Predicted effective distance between It can be represented as:

[0060] (twenty four),

[0061] (2) A Gaussian kernel function is used to apply distance decay weighting to the influence of neighbors to construct a local density index for nodes. This considers the node capability value. The nodes have been taken into account. The node has the capacity to act as a cluster head, therefore... The local density is defined as:

[0062] (25),

[0063] in, This represents the cutoff distance, which is the node's communication radius. Half of it.

[0064] (3) Calculate the separation degree of the nodes. At the nodes... local sets Within, the local density is higher than that of the nodes. The set of nodes can be represented as:

[0065] (26),

[0066] like If not empty, then the node Separation Defined as the minimum effective prediction distance to nodes with a density higher than its own within a local set, i.e.:

[0067] (27),

[0068] like If it is empty, it means that the node It possesses the strongest cluster head competitiveness within the current local area. At this point, the node... The separation degree is defined as the maximum effective predicted distance from a node to other nodes within the local set, i.e.:

[0069] (28),

[0070] (4) Calculate the cluster head contention value of the node. Based on the local density and separation of the node, the cluster head contention value of the node is defined. For the node... Then we have:

[0071] (29),

[0072] (5) Distributed Clustering. If a node has the largest cluster head contention value within the local set, then that node becomes the cluster head within the local set and broadcasts a CH_CLA message (Cluster Head Claim) carrying its own ID and location information, declaring its cluster head identity and providing a basis for surrounding nodes to join the cluster. After receiving the CH_CLA message, other ordinary nodes select the nearest cluster head and join the corresponding cluster in a one-hop manner, and send a JOIN_REQ message (JoinRequest) to the corresponding cluster head. After confirmation, the cluster head returns a JOIN_ACK message (Join Acknowledgement), thereby gradually completing the clustering process and forming a one-hop route within the cluster.

[0073] Step 5: Cluster Maintenance. During network operation, when a node's movement changes its connection to the original cluster head, the cluster maintenance mechanism is triggered. The specific steps are as follows:

[0074] (1) The cluster head periodically broadcasts a cluster maintenance message HELLO_CH (Cluster Head Hello), which includes the cluster head identifier, cluster identifier, and location information. After receiving the HELLO_CH message, the cluster member nodes return a response message HELLO_CM (Cluster Member Hello) to the corresponding cluster head, which includes the member node identifier, current location, and current cluster identifier.

[0075] (2) For any cluster member node Then at time With the current cluster head Distance between It can be represented as:

[0076] (30),

[0077] in, For nodes coordinates For cluster head The coordinates. If the node and the current cluster head satisfy... This indicates that the node is still within the communication coverage area of ​​the current cluster head and can maintain its original cluster affiliation; if If this is the case, it means that the node can no longer communicate directly with the current cluster head and needs to be adjusted in terms of cluster affiliation.

[0078] (3) The HELLO message sending interval of different nodes is adaptively adjusted according to their distance from the current cluster head. Let the node... At any moment The actual HELLO message sending interval is Then it can be expressed as:

[0079] (31),

[0080] in, This is the standard sending interval for HELLO messages. The closer a node is to the cluster head, the longer its HELLO message sending interval; conversely, the closer a node is to the cluster edge, the shorter its sending interval.

[0081] (4) When a cluster member node leaves the coverage area of ​​the current cluster head, first determine whether it has entered the coverage area of ​​another cluster head. Let node... The set of neighboring cluster heads that can receive HELLO_CH messages is For any candidate cluster head If the following conditions are met:

[0082] (32),

[0083] This indicates the node You can join directly If a node is within the coverage area of ​​multiple cluster heads, the cluster head to which it belongs is selected according to the principle of minimum distance.

[0084] (5) After completing the selection of the new cluster head, the node A cluster switch request (SWITCH_REQ) message is sent to the target cluster head. After confirming that the node meets the access conditions, the target cluster head returns a cluster switch acknowledgment (SWITCH_ACK) message and sets the node... Add to the new cluster membership table. For the original cluster head, if no node is received within three consecutive message cycles... If a node receives a HELLO_CM message, it is considered that the node has been removed from the current cluster's coverage area and is deleted from the cluster member table.

[0085] (6) When a cluster head detects another cluster head node in its neighbor list, the two cluster heads first exchange their respective cluster head contention values ​​and member information, and retain the cluster head with the higher contention value. If more than half of the members of the cluster head with the lower contention value are already in the one-hop neighbor set of the other cluster head, the cluster head with the lower contention value sends a cluster merge request message MERGE_REQ (Merge Request) to the other cluster head. After the other cluster head confirms the request, it returns a cluster merge confirmation message MERGE_ACK (Merge Acknowledgement). Subsequently, the cluster head with the lower contention value exits the cluster head role and becomes a cluster member. Its original member nodes are reassigned to the retained cluster head according to the one-hop access principle.

[0086] (7) If a cluster member node does not receive a HELLO_CH message from the cluster head node within three consecutive HELLO message cycles, the cluster head node is determined to have failed. In this case, a clustering process is triggered within the local set of nodes to quickly rebuild the cluster structure.

[0087] Step 6: Gateway Node Selection. This is only performed when the distance between cluster heads is greater than the communication radius. Nodes located in the inter-cluster boundary region act as gateway nodes to participate in relay forwarding, thereby reducing the propagation range of control messages. The specific steps are as follows:

[0088] (1) Only when the distance between cluster heads is greater than the communication radius Only when this condition is met are inter-cluster nodes located in the cluster boundary region allowed to participate in local relay forwarding as gateway nodes. At this time, in adjacent clusters... and cluster Only one inter-cluster node is selected as the gateway node to assist in relay forwarding. Let this inter-cluster node be... If they simultaneously satisfy:

[0089] (33),

[0090] (34),

[0091] This indicates inter-cluster nodes Capable of clustering and cluster A communication link is established between them, thereby obtaining a set of candidate gateway nodes. .

[0092] (2) Since the gateway node needs to undertake additional forwarding tasks, the candidate gateway node is required to... load The load should be no less than the average load of the adjacent cluster heads at both ends to ensure that the gateway node has continuous forwarding capability.

[0093] (3) In order to avoid routing loops and continuation during inter-cluster forwarding, and to ensure that data always converges gradually towards the destination cluster head, the source cluster head is required to... Point to the gateway node vector With the next jump cluster head vector The angle between It is an acute angle, that is:

[0094] (35),

[0095] (4) From the set of candidate gateway nodes In the middle, select the node with the lowest load. As the sole gateway node between the two clusters, the gateway node The selection of can be defined as:

[0096] (36),

[0097] Step 7: On-demand Inter-cluster Route Construction. After clustering is completed, inter-cluster routes are established on demand. Route request messages are only propagated between the cluster head and the gateway node. Based on the inter-cluster topology, the transmission path with the shorter cumulative distance is selected as the inter-cluster forwarding path. The specific steps are as follows:

[0098] (1) Inter-cluster on-demand route discovery process is triggered. When the source cluster head needs to send data to the destination cluster head and there is no available inter-cluster path, the source cluster head initiates the inter-cluster route discovery process and generates an RREQ message.

[0099] (2) Inter-cluster multi-hop route discovery. This method abstracts the current network topology into a directed weighted graph. The vertex set It consists of the current cluster head and gateway nodes of the network; edge set This represents the actual communication links between the nodes. For any two nodes in the diagram... and If and only if the distance between two nodes At that time, establish directed edges .

[0100] Let a candidate inter-cluster path from the source cluster head to the base station be... The total distance of the path can then be expressed as the sum of the distances of each hop link in the path, i.e.:

[0101] (37),

[0102] The path with the minimum cumulative distance is selected as the final inter-cluster multi-hop path. :

[0103] (38),

[0104] (3) After selecting an inter-cluster multi-hop path, the destination cluster head returns an RREP message in the opposite direction of the path. The RREP message carries the confirmed path information and the corresponding next-hop node information. After receiving the RREP message, the source cluster head writes the corresponding path into its local routing table, thereby completing the inter-cluster route establishment.

[0105] Beneficial effects: The DC-LOEER method proposed in this invention has the following beneficial effects:

[0106] (1) This invention combines multiple indicators such as node remaining energy, node connectivity and average link persistence, and adopts the CRITIC objective weighting method to achieve adaptive evaluation of node competitiveness, thereby improving the rationality and flexibility of cluster head election.

[0107] (2) This invention introduces the idea of ​​DPC density peak clustering, and combines local density and separation to complete cluster head election and distributed clustering. It can take into account the comprehensive performance of nodes and spatial distribution characteristics, and improve the problems of uneven distribution of cluster heads and unreasonable cluster division.

[0108] (3) The present invention uses a mechanism that combines cluster member maintenance and cluster head maintenance to make local adjustments to the local topology changes caused by node movement, thereby reducing the control overhead caused by frequent global reconstruction and helping to maintain the continuity of the cluster structure.

[0109] (4) By constructing inter-cluster on-demand routing, the present invention limits the propagation range of control messages to the cluster head and necessary boundary auxiliary nodes, which can effectively reduce the routing control overhead and node energy consumption in large-scale MANETs. Detailed Implementation

[0110] Based on the maximum cache queue length of the node Node communication radius HELLO message sending cycle The DC-LOEER method of this invention will be described using an example.

[0111] Step 1: Node State Information Exchange and Maintenance: Each node in the network periodically broadcasts a HELLO message (a special routing reply message used for node state exchange). The message contains fields such as node identifier, current location, speed, direction of movement, and remaining energy, and is used to publish its own state information to its one-hop neighbors. After receiving a HELLO message from a neighboring node, each node updates its local neighbor table to record the state information of the corresponding neighboring node.

[0112] Step 2: Node Capability Metric Calculation: Based on the neighbor node status information recorded in the neighbor table, the current node calculates its remaining energy metric, connectivity metric, load level metric, and average link sustainability metric, and then normalizes each metric. This includes the following steps:

[0113] (1) The local state information of a node includes its remaining energy, location coordinates, movement speed and direction, and cache queue length. The neighbor table records the number of neighboring nodes and the location coordinates, movement speed and direction, and cache queue length of each neighboring node. First, the node at time... The remaining energy in the local state information is defined as the node's remaining energy. Then, the number of neighboring nodes in the node's neighbor table is defined as the node connectivity. Secondly, the length of the node's cache queue is defined as the node's load level. Finally, based on the node's own location coordinates, movement speed, and direction in its local state information, and the location coordinates, movement speed, and direction of neighboring nodes in the neighbor table, nodes whose distances to each neighbor are less than the node's communication radius are selected. The average time is defined as the average link duration. The calculation is as follows: Let the node be... and nodes At any moment The positions are respectively and The speeds are respectively and ,make:

[0114] (1),

[0115] Then the node can be obtained. and nodes Link duration between:

[0116] (2),

[0117] Set nodes The set of neighbor nodes is The number of neighboring nodes included is Then the average link duration can be obtained, that is:

[0118] (3),

[0119] (2) Calculate the node's remaining energy index. Normalize the node's remaining energy and define the node. The remaining energy index is:

[0120] (4),

[0121] in, For nodes The remaining energy. and Let be the maximum and minimum remaining energy values ​​of the current neighboring nodes, respectively, which can be expressed as:

[0122] (5),

[0123] (6),

[0124] (3) Calculate the node connectivity index. Normalize the node connectivity and define the node. The connectivity index is:

[0125] (7),

[0126] in, For nodes The connectivity. and Each is the current The maximum and minimum connectivity of neighboring nodes can be expressed as follows:

[0127] (8),

[0128] (9),

[0129] (4) Calculate the node load level index. Normalize the node load level and define the node. The load level index can be expressed as:

[0130] (10),

[0131] in, For nodes The load level, This represents the maximum length of the node's cache queue.

[0132] (5) Calculate the average link persistence metric. (This involves placing nodes...) Average link duration normalization defines the average link persistence metric. for:

[0133] (11),

[0134] in, For nodes The average link duration. and These are the maximum and minimum average link durations of the current neighboring nodes, respectively, and can be expressed as:

[0135] (12),

[0136] (13),

[0137] Step 3: Node capability evaluation based on dynamic weights. Using normalized indicators, the node capability value is calculated, and the weights of different indicators are adaptively adjusted using the CRITIC objective weighting method. The specific steps are as follows:

[0138] (1) Based on the three normalized indices defined in the second step, construct the node capability value function:

[0139] (14),

[0140] in, , , , These are the remaining energy indices of the nodes. Node connectivity index Node load level indicators and average link persistence metrics The corresponding weighting coefficients can be determined by the CRITIC objective weighting method.

[0141] (2) The CRITIC objective weighting method is used to achieve adaptive adjustment of the weights of each indicator in the node capability value function. Let the node contain nodes... Its own local node set is The set contains a total of nodes. Then, a node index matrix can be constructed, namely:

[0142] (15),

[0143] in, , , Each corresponding to any node The remaining energy index of the node, the connectivity index of the node, the load level index of the node, and the average link sustainability index.

[0144] Calculate the first The standard deviation of an indicator in a local set :

[0145] (16),

[0146] in, Identifier The average value of an indicator within a local set. Standard deviation. It reflects the degree of dispersion of the indicator within a local set. The larger the value, the stronger the indicator's ability to distinguish different nodes.

[0147] To measure the degree of information overlap between different indicators, the following calculation is further performed: The first indicator and the first Correlation coefficients among indicators :

[0148] (17),

[0149] in, If an indicator has a low correlation with other indicators, it means that it can provide more independent information, and its importance should be increased accordingly.

[0150] After obtaining the standard deviation and correlation coefficient, define the first... Information content of each indicator for:

[0151] (18),

[0152] Based on the information content of each indicator, the objective weights of the three indicators can be obtained, namely:

[0153] (19),

[0154] Step 4: Cluster Head Election and Distributed Clustering. Based on the capabilities of each node, and combining the DPC (Density Peak Clustering) approach, the local density and separation of the nodes are calculated. This is used to calculate the cluster head competition value. Nodes with higher local density and separation are preferentially selected as cluster heads. Ordinary nodes are joined to the cluster of the corresponding cluster head according to the proximity principle, completing the distributed clustering process. The specific steps are as follows:

[0155] (1) Calculate the predicted effective distance between nodes, assuming the nodes are... and nodes At any moment The coordinates are respectively and The distance between the two nodes is... It can be represented as:

[0156] (20),

[0157] Meanwhile, the relative position component and the relative velocity component are defined as follows:

[0158] (twenty one),

[0159] (twenty two),

[0160] Based on this, the radial relative velocity of the two nodes along the connecting line direction. It can be represented as:

[0161] (twenty three),

[0162] If the HELLO message sending cycle As a short-term prediction window, the node and nodes Predicted effective distance between It can be represented as:

[0163] (twenty four),

[0164] (2) A Gaussian kernel function is used to apply distance decay weighting to the influence of neighbors to construct a local density index for nodes. This considers the node capability value. The nodes have been taken into account. The node has the capacity to act as a cluster head, therefore... The local density is defined as:

[0165] (25),

[0166] in, This represents the cutoff distance, which is the node's communication radius. Half of it.

[0167] (3) Calculate the separation degree of the nodes. At the nodes... local sets Within, the local density is higher than that of the nodes. The set of nodes can be represented as:

[0168] (26),

[0169] like If not empty, then the node Separation Defined as the minimum effective prediction distance to nodes with a density higher than its own within a local set, i.e.:

[0170] (27),

[0171] like If it is empty, it means that the node It possesses the strongest cluster head competitiveness within the current local area. At this point, the node... The separation degree is defined as the maximum effective predicted distance from a node to other nodes within the local set, i.e.:

[0172] (28),

[0173] (4) Calculate the cluster head contention value of the node. Based on the local density and separation of the node, the cluster head contention value of the node is defined. For the node... Then we have:

[0174] (29),

[0175] (5) Distributed Clustering. If a node has the largest cluster head contention value within the local set, then that node becomes the cluster head within the local set and broadcasts a CH_CLA message (Cluster Head Claim) carrying its own ID and location information, declaring its cluster head identity and providing a basis for surrounding nodes to join the cluster. After receiving the CH_CLA message, other ordinary nodes select the nearest cluster head and join the corresponding cluster in a one-hop manner, and send a JOIN_REQ message (JoinRequest) to the corresponding cluster head. After confirmation, the cluster head returns a JOIN_ACK message (Join Acknowledgement), thereby gradually completing the clustering process and forming a one-hop route within the cluster.

[0176] Step 5: Cluster Maintenance. During network operation, when a node's movement changes its connection to the original cluster head, the cluster maintenance mechanism is triggered. The specific steps are as follows:

[0177] (1) The cluster head periodically broadcasts a cluster maintenance message HELLO_CH (Cluster Head Hello), which includes the cluster head identifier, cluster identifier, and location information. After receiving the HELLO_CH message, the cluster member nodes return a response message HELLO_CM (Cluster Member Hello) to the corresponding cluster head, which includes the member node identifier, current location, and current cluster identifier.

[0178] (2) For any cluster member node Then at time With the current cluster head Distance between It can be represented as:

[0179] (30),

[0180] in, For nodes coordinates For cluster head The coordinates. If the node and the current cluster head satisfy... If so, it means that the node is still within the communication coverage area of ​​the current cluster head and can maintain its original cluster affiliation; if If this is the case, it means that the node can no longer communicate directly with the current cluster head and needs to be adjusted in terms of cluster affiliation.

[0181] (3) The HELLO message sending interval of different nodes is adaptively adjusted according to their distance from the current cluster head. Let the node... At any moment The actual HELLO message sending interval is Then it can be expressed as:

[0182] (31),

[0183] in, This is the standard sending interval for HELLO messages. The closer a node is to the cluster head, the longer its HELLO message sending interval; conversely, the closer a node is to the cluster edge, the shorter its sending interval.

[0184] (4) When a cluster member node leaves the coverage area of ​​the current cluster head, first determine whether it has entered the coverage area of ​​another cluster head. Let node... The set of neighboring cluster heads that can receive HELLO_CH messages is For any candidate cluster head If the following conditions are met:

[0185] (32),

[0186] This indicates the node You can join directly If a node is within the coverage area of ​​multiple cluster heads, the cluster head to which it belongs is selected according to the principle of minimum distance.

[0187] (5) After completing the selection of the new cluster head, the node A cluster switch request (SWITCH_REQ) message is sent to the target cluster head. After confirming that the node meets the access conditions, the target cluster head returns a cluster switch acknowledgment (SWITCH_ACK) message and sets the node... Add to the new cluster membership table. For the original cluster head, if no node is received within three consecutive message cycles... If a node receives a HELLO_CM message, it is considered that the node has been removed from the current cluster's coverage area and is deleted from the cluster member table.

[0188] (6) When a cluster head detects another cluster head node in its neighbor list, the two cluster heads first exchange their respective cluster head contention values ​​and member information, and retain the cluster head with the higher contention value. If more than half of the members of the cluster head with the lower contention value are already in the one-hop neighbor set of the other cluster head, the cluster head with the lower contention value sends a cluster merge request message MERGE_REQ (Merge Request) to the other cluster head. After the other cluster head confirms the request, it returns a cluster merge confirmation message MERGE_ACK (Merge Acknowledgement). Subsequently, the cluster head with the lower contention value exits the cluster head role and becomes a cluster member. Its original member nodes are reassigned to the retained cluster head according to the one-hop access principle.

[0189] (7) If a cluster member node does not receive a HELLO_CH message from the cluster head node within three consecutive HELLO message cycles, the cluster head node is determined to have failed. In this case, a clustering process is triggered within the local set of nodes to quickly rebuild the cluster structure.

[0190] Step 6: Gateway Node Selection. This is only performed when the distance between cluster heads is greater than the communication radius. Nodes located in the inter-cluster boundary region act as gateway nodes to participate in relay forwarding, thereby reducing the propagation range of control messages. The specific steps are as follows:

[0191] (1) Only when the distance between cluster heads is greater than the communication radius Only when this condition is met are inter-cluster nodes located in the cluster boundary region allowed to participate in local relay forwarding as gateway nodes. At this time, in adjacent clusters... and cluster Only one inter-cluster node is selected as the gateway node to assist in relay forwarding. Let this inter-cluster node be... If they simultaneously satisfy:

[0192] (33),

[0193] (34),

[0194] This indicates inter-cluster nodes Capable of clustering and cluster A communication link is established between them, thereby obtaining a set of candidate gateway nodes. .

[0195] (2) Since the gateway node needs to undertake additional forwarding tasks, the candidate gateway node is required to... load The load should be no less than the average load of the adjacent cluster heads at both ends to ensure that the gateway node has continuous forwarding capability.

[0196] (3) In order to avoid routing loops and continuation during inter-cluster forwarding, and to ensure that data always converges gradually towards the destination cluster head, the source cluster head is required to... Point to the gateway node vector With the next jump cluster head vector The angle between It is an acute angle, that is:

[0197] (35),

[0198] (4) From the set of candidate gateway nodes In the middle, select the node with the lowest load. As the sole gateway node between the two clusters, the gateway node The selection of can be defined as:

[0199] (36),

[0200] Step 7: On-demand Inter-cluster Route Construction. After clustering is completed, inter-cluster routes are established on demand. Route request messages are only propagated between the cluster head and the gateway node. Based on the inter-cluster topology, the transmission path with the shorter cumulative distance is selected as the inter-cluster forwarding path. The specific steps are as follows:

[0201] (1) Inter-cluster on-demand route discovery process is triggered. When the source cluster head needs to send data to the destination cluster head and there is no available inter-cluster path, the source cluster head initiates the inter-cluster route discovery process and generates an RREQ message.

[0202] (2) Inter-cluster multi-hop route discovery. This method abstracts the current network topology into a directed weighted graph. The vertex set It consists of the current cluster head and gateway nodes of the network; edge set This represents the actual communication links between the nodes. For any two nodes in the diagram... and If and only if the distance between two nodes At that time, establish directed edges .

[0203] Let a candidate inter-cluster path from the source cluster head to the base station be... The total distance of the path can then be expressed as the sum of the distances of each hop link in the path, i.e.:

[0204] (37),

[0205] The path with the minimum cumulative distance is selected as the final inter-cluster multi-hop path. :

[0206] (38),

[0207] (3) After selecting an inter-cluster multi-hop path, the destination cluster head returns an RREP message in the opposite direction of the path. The RREP message carries the confirmed path information and the corresponding next-hop node information. After receiving the RREP message, the source cluster head writes the corresponding path into its local routing table, thereby completing the inter-cluster route establishment.

[0208] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A low-overhead, energy-efficient routing method suitable for large-scale mobile ad hoc networks, characterized in that... The method includes the following steps: Step 1: Node State Information Interaction and Maintenance: Each node in the network periodically broadcasts a HELLO message, which is a special routing reply message used for node state interaction. The message contains fields such as node identifier, current location, speed, direction of movement, and remaining energy, and is used to publish its own state information to one-hop neighbors. After receiving the HELLO message sent by the neighbor node, each node updates its local neighbor table and records the state information of the corresponding neighbor node. Step 2: Node Capability Indicator Calculation: Based on the neighbor node status information recorded in the neighbor table, the current node calculates the node remaining energy index, node connectivity index, node load level index, and average link sustainability index, and normalizes each index. Step 3: Node capability value evaluation based on dynamic weights. Calculate the node capability value using normalized indicators and adaptively adjust the weights of different indicators using the CRITIC objective weighting method. Step 4: Cluster head election and distributed clustering. Based on the capability values ​​of each node, and combined with the idea of ​​DPC density peak clustering, the local density and separation of the nodes are calculated, and the cluster head competition value is calculated accordingly. Nodes with higher local density and higher separation are preferentially selected as cluster heads. Ordinary nodes are added to the clusters of the corresponding cluster heads according to the proximity principle, thus completing distributed clustering. Step 5: Cluster maintenance. During network operation, when a node moves, causing a change in its connection with the original cluster head, the cluster maintenance mechanism is triggered. Step 6: Gateway node selection. Only when the distance between cluster heads is greater than the communication radius R, the node located in the inter-cluster boundary area will act as the gateway node to participate in relay forwarding, so as to reduce the propagation range of control messages. Step 7: On-demand inter-cluster routing construction. After clustering is completed, inter-cluster routing is established on demand. Routing request messages are only propagated between the cluster head and the gateway node. Based on the inter-cluster topology, the transmission path with the smaller cumulative distance is selected as the inter-cluster forwarding path.

2. The low-overhead, energy-saving routing method for large-scale mobile ad hoc networks according to claim 1, characterized in that... The second step, the calculation of node capability indicators, specifically includes: (1) The local state information of a node includes its remaining energy, location coordinates, movement speed and direction, and cache queue length. The neighbor table records the number of neighboring nodes and the location coordinates, movement speed and direction, and cache queue length of each neighboring node. First, the node at time t is... The remaining energy in the local state information is defined as the node's remaining energy. Then, the number of neighboring nodes in the node's neighbor table is defined as the node connectivity. Secondly, the length of the node's cache queue is defined as the node's load level. Finally, based on the node's own location coordinates, movement speed, and direction in its local state information, as well as the location coordinates, movement speed, and direction of neighboring nodes in the neighbor table, the nodes with distances less than their communication radius are configured to be connected. The average time is defined as the average link duration. The calculation is as follows: Let the node be... and nodes At any moment The positions are respectively and The speeds are respectively and ,make: (1)， Then the node can be obtained. and nodes Link duration between: (2)， Set nodes The set of neighbor nodes is The number of neighboring nodes included is Then the average link duration can be obtained, that is: (3)， (2) Calculate the node remaining energy index, normalize the node remaining energy, and define the node. The remaining energy index is: (4)， in, For nodes The remaining energy, and Let be the maximum and minimum remaining energy values ​​of the current neighboring nodes, respectively, and represent them as follows: (5)， (6)， (3) Calculate the node connectivity index, normalize the node connectivity, and define the node. The connectivity index is: (7)， in, For nodes connectivity, and Each is the current The maximum and minimum connectivity of neighboring nodes are expressed as follows: (8)， (9)， (4) Calculate the node load level index, normalize the node load level, and define the node. The load level index is expressed as: (10)， in, For nodes The load level, This represents the maximum cache queue length of the node. (5) Calculate the average link persistence index, and assign nodes... Average link duration normalization defines the average link persistence metric. for: (11)， in, For nodes Average link duration, and These are the maximum and minimum average link durations of the current neighboring nodes, respectively, and are expressed as: (12)， (13)。 3. The low-overhead, energy-saving routing method for large-scale mobile ad hoc networks according to claim 1, characterized in that... The third step, the node capability value evaluation based on dynamic weights, involves the following specific steps: (1) Based on the three normalized indices defined in the second step, construct the node capability value function: (14)， in, , , , These are the remaining energy indices of the nodes. Node connectivity index Node load level indicators and average link persistence metrics The corresponding weighting coefficients are determined by the CRITIC objective weighting method. (2) Using the CRITIC objective weighting method, adaptive adjustment of the weights of each index of the node capability value function is achieved. Let the node be included. Its own local node set is The set contains a total of nodes. Then construct the node index matrix, that is: (15)， in, , , Each corresponding to any node The remaining energy index of the node, the connectivity index of the node, the load level index of the node, and the average link sustainability index. Calculate the first The standard deviation of an indicator in a local set : (16)， in, Identifier The average and standard deviation of each indicator within a local set. It reflects the degree of dispersion of the indicator within a local set. The larger the value, the stronger the indicator's ability to distinguish between different nodes. To measure the degree of information overlap between different indicators, the following calculation is further performed: The first indicator and the first Correlation coefficients among indicators : (17)， in, If an indicator has a low correlation with other indicators, it means that it can provide more independent information, and its importance should be increased accordingly. After obtaining the standard deviation and correlation coefficient, define the first... Information content of each indicator for: (18)， Based on the information content of each indicator, the objective weights of the three indicators are obtained, namely: (19)。 4. The low-overhead, energy-saving routing method for large-scale mobile ad hoc networks according to claim 1, characterized in that... The fourth step, cluster head election and distributed clustering, involves the following specific steps: (1) Calculate the predicted effective distance between nodes, assuming the nodes are... and nodes At any moment The coordinates are respectively and The distance between the two nodes is... Represented as: (20)， Meanwhile, the relative position component and the relative velocity component are defined as follows: (21)， (22)， Based on this, the radial relative velocity of the two nodes along the connecting line direction. Represented as: (23)， If based on the HELLO message sending cycle As a short-term prediction window, the node and nodes Predicted effective distance between Represented as: (24)， (2) A Gaussian kernel function is used to apply distance decay weighting to the influence of neighbors in order to construct a local density index for nodes, taking into account the node capability value. The nodes have been taken into account. The node has the capacity to act as a cluster head, therefore... The local density is defined as: (25)， in, This represents the cutoff distance, which is the node's communication radius. Half of; (3) Calculate the separation degree of the nodes. local sets Within, the local density is higher than that of the nodes. The set of nodes is represented as: (26)， like If not empty, then the node resolution Defined as the minimum effective prediction distance to nodes with a density higher than its own within a local set, i.e.: (27)， like If it is empty, it means that the node If a node possesses the strongest cluster head competitiveness within its current local area, then it will be... The separation degree is defined as the maximum effective predicted distance from a node to other nodes within the local set, i.e.: (28)， (4) Calculate the cluster head competition value of the node. Based on the local density and separation of the node, define the cluster head competition value of the node. For the node... Then we have: (29)， (5) Distributed clustering: If a node has the largest cluster head contention value in the local set, then the node becomes the cluster head in the local set and broadcasts a CH_CLA message carrying its own ID and location information to announce its cluster head identity and provide a basis for surrounding nodes to join the cluster. After receiving the CH_CLA message, other ordinary nodes select the nearest cluster head to join the corresponding cluster in a one-hop manner and send a JOIN_REQ request message to the corresponding cluster head. After the cluster head confirms, it returns a JOIN_ACK message to join the cluster, thereby gradually completing the clustering process and forming a one-hop route within the cluster.

5. The low-overhead, energy-saving routing method for large-scale mobile ad hoc networks according to claim 1, characterized in that... The fifth step, cluster maintenance, involves the following steps: (1) The cluster head periodically broadcasts the cluster maintenance message HELLO_CH, which includes the cluster head identifier, cluster identifier and location information. After receiving the HELLO_CH message, the cluster member node returns a response message HELLO_CM to the corresponding cluster head, which includes the member node identifier, current location and current belonging cluster identifier. (2) For any cluster member node Then at time With the current cluster head Distance between Represented as: (30)， in, For nodes coordinates For cluster head The coordinates of the node, if the node and the current cluster head satisfy the following conditions: If so, it means that the node is still within the communication coverage area R of the current cluster head and can maintain its original cluster affiliation; if If this is the case, it means that the node can no longer communicate directly with the current cluster head and needs to adjust its cluster affiliation. (3) The HELLO message sending interval of different nodes is adaptively adjusted according to the distance between them and the current cluster head. Let the node... At any moment The actual HELLO message sending interval is Then it is represented as: (31)， in, The standard sending interval for HELLO messages is as follows: the closer a node is to the cluster head, the longer its HELLO message sending interval is; and the closer a node is to the cluster edge, the shorter its sending interval is. (4) When a cluster member node leaves the coverage area of ​​the current cluster head, first determine whether it has entered the coverage area of ​​another cluster head. Let the node... The set of neighboring cluster heads that can receive HELLO_CH messages is For any candidate cluster head If the following conditions are met: (32)， This indicates the node You can join directly If a node is within the coverage area of ​​multiple cluster heads, the cluster head to which it belongs is selected according to the principle of minimum distance. (5) After completing the selection of the new cluster head, the node A cluster switch request message SWITCH_REQ is sent to the target cluster head. After confirming that the node meets the access conditions, the target cluster head returns a cluster switch confirmation message SWITCH_ACK and sets the node... Add to the new cluster membership table; for the original cluster head, if no node is received within three consecutive message cycles... If a node receives a HELLO_CM message, it is considered that the node has been removed from the current cluster's coverage area and is deleted from the cluster member table. (6) When a cluster head detects other cluster head nodes in the neighbor list, the two cluster heads first exchange their respective cluster head contention values ​​and member information, and retain the cluster head with the larger contention value. If more than half of the members of the cluster head with the lower contention value are already in the one-hop neighbor node set of the other cluster head, the cluster head with the lower contention value sends a cluster merge request message MERGE_REQ to the other. After the other confirms, it returns a cluster merge confirmation message MERGE_ACK. Subsequently, the cluster head with the lower contention value exits the cluster head role and becomes a cluster member. Its original member nodes are reassigned to the retained cluster head according to the one-hop access principle. (7) If a cluster member node does not receive a HELLO_CH message from the cluster head node within three consecutive HELLO message cycles, it is determined that the cluster head node has failed. At this time, a clustering process is triggered in the local set of nodes to quickly rebuild the cluster structure.

6. The low-overhead, energy-saving routing method for large-scale mobile ad hoc networks according to claim 1, characterized in that... The specific steps of the sixth step are as follows: (1) Only when the distance between cluster heads is greater than the communication radius Only when certain conditions are met is it permitted for inter-cluster nodes located in the cluster boundary region to participate in local relay forwarding as gateway nodes. In this case, in adjacent clusters... and cluster Only one inter-cluster node is selected as the gateway node to assist in relay forwarding. Let this inter-cluster node be... If they simultaneously satisfy: (33)， (34)， This indicates inter-cluster nodes Capable of clustering and cluster A communication link is established between them, thereby obtaining a set of candidate gateway nodes. ; (2) Since the gateway node needs to undertake additional forwarding tasks, the candidate gateway node is required to... load The load should be no less than the average load of the adjacent cluster heads at both ends to ensure that the gateway node has continuous forwarding capability; (3) In order to avoid routing loops and continuation during inter-cluster forwarding, and to ensure that data always converges gradually towards the destination cluster head, the source cluster head is required to... Point to the gateway node vector With the next jump cluster head vector The angle between It is an acute angle, that is: (35)， (4) From the set of candidate gateway nodes In the middle, select the node with the lowest load. As the sole gateway node between the two clusters, the gateway node The selection of can be defined as: (36)。 7. The low-overhead, energy-saving routing method for large-scale mobile ad hoc networks according to claim 1, characterized in that... The seventh step, the construction of on-demand routes between clusters, involves the following specific steps: (1) Inter-cluster on-demand route discovery process is triggered when the source cluster head needs to send data to the destination cluster head and there is no available inter-cluster path. The source cluster head initiates the inter-cluster route discovery process and generates an RREQ message. (2) Inter-cluster multi-hop route discovery: This method abstracts the current network topology into a directed weighted graph. , where the vertex set It consists of the current cluster head and gateway nodes of the network; edge set This represents the actual communication links between the nodes; for any two nodes in the diagram... and If and only if the distance between two nodes At that time, establish directed edges ; Let a candidate inter-cluster path from the source cluster head to the base station be... The total distance of the path can then be expressed as the sum of the distances of each hop link in the path, i.e.: (37)， The path with the minimum cumulative distance is selected as the final inter-cluster multi-hop path. : (38)， (3) After selecting the inter-cluster multi-hop path, the destination cluster head returns an RREP message in the opposite direction of the path. The RREP message carries the confirmed path information and the corresponding next-hop node information. After receiving the RREP message, the source cluster head writes the corresponding path into the local routing table, thereby completing the inter-cluster route establishment.

Images