Node-disjoint dual-path routing method based on graph theory calculation

Abstract

The invention relates to the technical field of routing, in particular to a node-disjoint dual-path routing method based on graph theory calculation. According to the technical scheme, the node-disjoint dual-path routing method based on graph theory calculation comprises the following steps that S1, a main path is calculated, an aggregation node calculates the shortest path from a source node to the aggregation node based on a network topological graph by adopting a Dijkstra algorithm to serve as the main path, and a relay node set of the main path is recorded; and S2, generating a suboptimal path, deleting a relay node and an associated edge of the main path, generating an incomplete topological graph, calculating the suboptimal path from the source node to the sink node by adopting the Dijkstra algorithm again, and ensuring that the node of the suboptimal path is not intersected with the node of the main path. By constructing a double-path transmission mechanism, the problems of node faults, signal interference and the like are effectively solved, the performance of monitoring data transmission of the power distribution network is remarkably improved, the requirement for continuous development of the intelligent power distribution network is met, and safe, stable and efficient operation of the power distribution network is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of routing technology, and in particular to a node-disjoint two-path routing method based on graph theory computation. Background Technology

[0002] In the ongoing development of smart grids, the intelligent upgrading of distribution networks is crucial. Wireless sensor networks (WSNs), with their low cost, ease of deployment, and self-organizing characteristics, are widely used in distribution network monitoring. By deploying a large number of sensor nodes at key locations in the distribution network, WSNs can collect key data such as the operating status of power equipment, power parameters, and environmental information in real time, providing important data support for the safe, stable operation and precise control of the distribution network.

[0003] However, the distribution network environment is complex, and data transmission faces numerous challenges. Existing routing protocols can meet some communication needs under normal circumstances; however, in the complex and ever-changing distribution network environment, the limitations of single-path data transmission gradually become apparent. When a node fails, data transmission is easily interrupted due to the lack of a backup path, disrupting the continuity of monitoring data and making it difficult for maintenance personnel to grasp the distribution network's operating status in a timely manner, leading to delays in fault diagnosis and repair.

[0004] In areas with strong signal interference, such as around substations or locations with numerous electromagnetic interference sources, data transmitted via a single path is susceptible to interference, resulting in packet loss or errors, making it difficult to guarantee data accuracy and integrity. This can lead to incorrect judgments in distribution networks that rely on precise data for decision-making, affecting their safe and stable operation. Furthermore, the distribution network topology dynamically changes due to equipment commissioning, maintenance, and other factors. Single-path routing protocols lack flexibility and cannot quickly adapt to topology changes, resulting in reduced data transmission timeliness and failing to meet the high real-time requirements of monitoring scenarios. This invention addresses this technological application background by designing a graph theory-based dual-path routing protocol to meet the routing protocol needs of distribution network monitoring. Summary of the Invention

[0005] This invention proposes a node-disjoint dual-path routing method based on graph theory computation, which solves the problem that the existing single-path routing protocol is difficult to meet the requirements of data transmission stability, reliability and timeliness of the distribution network when WSN is used for distribution network monitoring.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] The graph theory-based method for non-intersecting bipath routing includes the following steps:

[0008] Step S1: Calculate the main path. Based on the network topology, the sink node uses Dijkstra's algorithm to calculate the shortest path from the source node to the sink node as the main path, and records the set of relay nodes of the main path.

[0009] Step S2: Suboptimal path generation. Delete the relay nodes and associated edges of the main path to generate an incomplete topology graph. Use Dijkstra's algorithm again to calculate the suboptimal path from the source node to the sink node, ensuring that the suboptimal path does not intersect with the nodes of the main path.

[0010] Step S3: Dual-path deployment. The primary path and the secondary path are distributed to relevant nodes, so that the node routing table stores the next-hop information of both paths simultaneously, and the independence of the two paths is verified based on the node relevance factor, where the node relevance factor δ = 0.

[0011] Furthermore, the implementation of Dijkstra's algorithm in step S1 includes the following sub-steps:

[0012] Initialize the adjacency matrix W, where the edge weight is the equivalent signal distance d, calculated using the following formula:

[0013] Where A is the signal strength constant at a reference distance of 1 meter, and n is the path loss factor;

[0014] Define distance vector D and visit marker vector U. Initially, only the distance to the source node is 0, and the distances to the other nodes are infinite.

[0015] Iteratively select the vertex v with the smallest distance among the unvisited nodes. k Mark it as visited and update the distance values ​​of its neighboring nodes:

[0016] d j =min(d j d k +ω(v k ,v j Once the target node has been visited or all nodes have been traversed, output the shortest path and its set of relay nodes.

[0017] Furthermore, the calculation of the equivalent signal distance is based on the logarithmic path loss model:

[0018] RSSI=A-10nlg(d)

[0019] A is calibrated through actual measurement, and n is dynamically adjusted according to the electromagnetic interference intensity in the power distribution network environment. The higher the interference intensity, the larger the value of n.

[0020] Furthermore, the incomplete topology graph is generated as follows:

[0021] Delete all relay nodes in the main path except for the source node and the aggregation node;

[0022] Remove the edges that are directly connected to the relay node, including all edges where the relay node is an endpoint;

[0023] Preserve the original connections of the remaining nodes to form a new adjacency matrix W. f Used for suboptimal path calculation.

[0024] Furthermore, the formula for calculating the node relevance factor δ is as follows:

[0025] Where n is the number of common relay nodes in the two paths, and n0 is the total number of relay nodes in the main path; when δ = 0, the two paths are determined to be non-intersecting, and the relay nodes of the suboptimal path have no intersection with the main path.

[0026] Furthermore, the dual-path distribution method in step S3 is as follows:

[0027] The aggregation node sends the suboptimal path information to the source node through the reverse routing of the primary path;

[0028] The node routing table uses a two-column structure to store the following information for the primary path and the secondary path respectively:

[0029] hop count, next hop address, predecessor list;

[0030] Time to Live (TTL), route tag, RSSI value;

[0031] The route tag field is used to identify that the primary path is active and the secondary path is in standby status.

[0032] Furthermore, the construction of the network topology graph includes the following steps:

[0033] The sensor node periodically broadcasts HELLO messages, which are extended to carry a transmit power field;

[0034] The neighboring node reads the RSSI value by calling the physical layer interface based on the received HELLO message, and calculates the equivalent distance by combining it with the transmit power;

[0035] The equivalent distance is used as the edge weight and uploaded to the aggregation node to form a global topology matrix.

[0036] Furthermore, the structure of the HELLO message includes:

[0037] Grouping type, reserved bits, skipped number field;

[0038] Routing request message identifier, destination node IP address, source node IP address;

[0039] Transmit power value, source node sequence number;

[0040] The transmit power value and RSSI value are used together for equivalent distance calculation.

[0041] Furthermore, it also includes dynamic maintenance steps:

[0042] When a node detects a link failure, it switches to the suboptimal path to transmit data and sends a RERR message to the aggregation node. The RERR message contains the IP address and sequence number of the faulty node.

[0043] The aggregation node filters the affected node set B based on the RERR message, where set B consists of all upstream nodes whose routing tables contain the faulty node.

[0044] The aggregation node sends a Route Check Message (RCHE) to the nodes in set B. The RCHE message structure includes:

[0045] Grouping type, reserved bits, number of unreachable destination nodes;

[0046] Faulty node IP address, faulty node serial number;

[0047] After receiving RCHE, a normal node replies with a CREP message, which includes the source node's IP address, the faulty node's verification field, and the link status identifier.

[0048] The aggregation node updates the topology map based on the CREP response, recalculates the dual paths, and distributes the update information via reverse routing through the suboptimal path.

[0049] Furthermore, the triggering conditions for topology graph updates in the dynamic maintenance step include:

[0050] Electromagnetic interference caused RSSI fluctuations to exceed a preset threshold;

[0051] Node failures caused a continuous packet loss rate exceeding 10%.

[0052] When the network topology changes, the relevance factor δ>0, the two paths are forcibly recalculated.

[0053] The positive effects of this invention are:

[0054] By constructing a dual-path transmission mechanism, problems such as node failures and signal interference can be effectively addressed, significantly improving the performance of power distribution network monitoring data transmission to meet the ever-evolving needs of smart power distribution networks and ensure the safe, stable, and efficient operation of the power distribution network.

[0055] This invention fundamentally solves the problem of data transmission interruption caused by node failures or signal interference in traditional single-path routing protocols by constructing a dual-path transmission mechanism with non-overlapping nodes. There are no shared relay nodes between the primary and secondary paths. When the primary path fails due to strong electromagnetic interference around the substation or equipment maintenance, it can instantly switch to the backup path, avoiding data loss and significantly improving the continuity and integrity of distribution network monitoring data. Especially in complex and ever-changing distribution network environments, the independent design of the dual paths effectively resists the impact of single-point failures on global communication, ensuring the real-time and stable transmission of operational status data of critical power equipment.

[0056] Through dynamic topology pruning and intelligent routing maintenance mechanisms, this invention can quickly respond to network topology changes. When a link interruption or node failure is detected, the aggregation node proactively triggers local network diagnosis, updating paths only for the affected nodes rather than reconstructing the entire network, significantly reducing computational and communication overhead. Simultaneously, dynamic weight adjustment based on a signal equivalent distance model enables routing decisions to adapt to fluctuations in environmental interference intensity (such as sudden electromagnetic noise), avoiding frequent manual intervention. This closed-loop maintenance mechanism of "precise diagnosis - local update - adaptive optimization" significantly shortens network fault recovery time, meeting the stringent real-time requirements of distribution network monitoring.

[0057] Traditional routing protocols often require the deployment of numerous backup nodes or redundant data transmissions when designing path redundancy, leading to wasted energy and bandwidth resources. This invention achieves highly reliable transmission using existing network resources without adding hardware nodes through global topology graph optimization and dual-path collaborative distribution. The parallel storage and rapid switching mechanism of dual-path information in the node routing table reduces reliance on redundant hardware and extends the battery life of sensor nodes. Furthermore, the lightweight topology collection and maintenance mechanism based on HELLO messages reduces the communication load during protocol operation, making it particularly suitable for large-scale, high-density distribution network monitoring scenarios, providing a low-cost communication guarantee solution for the long-term stable operation of smart grids. Attached Figure Description

[0058] Figure 1 This is a flowchart of the AODV workflow in an embodiment of the present invention;

[0059] Figure 2 This is a schematic diagram of the AODV route discovery phase in an embodiment of the present invention;

[0060] Figure 3 This is a schematic diagram of the AODV routing maintenance phase in an embodiment of the present invention;

[0061] Figure 4 This is a schematic diagram illustrating the interrelationships of multiple paths in an embodiment of the present invention;

[0062] Figure 5 This is a flowchart of the D-Dijkstra algorithm in an embodiment of the present invention;

[0063] Figure 6 This is a schematic diagram of the optimal path calculation results in an embodiment of the present invention;

[0064] Figure 7 This is an incomplete topology diagram in an embodiment of the present invention;

[0065] Figure 8 This is a schematic diagram of the dual-path calculation results in an embodiment of the present invention;

[0066] Figure 9-14 The graphs a, b, c, d, e, and f are the results of calculating the dual paths using the D-Dijkstra algorithm for some nodes in this embodiment of the invention.

[0067] Figure 15 This is a flowchart of the RPWNCN route establishment process in an embodiment of the present invention;

[0068] Figure 16 This is a flowchart of the RPWNCN routing maintenance process in an embodiment of the present invention; Detailed Implementation

[0069] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0070] Example 1

[0071] The graph theory-based method for non-intersecting bipath routing includes the following steps:

[0072] Step S1: Calculate the main path. Based on the network topology, the sink node uses Dijkstra's algorithm to calculate the shortest path from the source node to the sink node as the main path, and records the set of relay nodes of the main path.

[0073] Step S2: Suboptimal path generation. Delete relay nodes and associated edges of the main path to generate an incomplete topology graph. Use Dijkstra's algorithm again to calculate the suboptimal path from the source node to the sink node, ensuring that the suboptimal path does not intersect with the nodes of the main path.

[0074] Step S3: Dual-path deployment. The primary path and the secondary path are distributed to the relevant nodes, so that the node routing table stores the next-hop information of both paths at the same time, and the independence of the two paths is verified based on the node relevance factor, where the node relevance factor δ = 0.

[0075] Main path calculation: The aggregation node periodically collects HELLO messages from all nodes in the network to obtain the neighbor relationships and received signal strength (RSSI) of each node.

[0076] Based on the equivalent distance formula, RSSI is converted into equivalent distance, and an adjacency matrix is ​​constructed.

[0077] Dijkstra's algorithm is used to calculate the shortest path from the source node to the sink node. The algorithm uses a greedy strategy to select the unvisited node with the smallest current distance, iteratively updates the distance values ​​of neighboring nodes, and finally outputs the main path and its set of relay nodes.

[0078] Suboptimal path generation:

[0079] Delete relay nodes (such as V2 and V5) and their associated edges in the main path to generate an incomplete topology graph.

[0080] Rerun Dijkstra's algorithm in the incomplete topology graph to ensure that suboptimal paths do not intersect with primary path nodes.

[0081] Dual-path deployment:

[0082] The aggregation node distributes primary and secondary path information to relevant nodes through reverse routing.

[0083] The node routing table is expanded into a two-column structure, recording the hop count, next-hop address, RSSI value, and time to live for each path.

[0084] The complete independence of the two paths is verified by using a node relevance factor δ=0, ensuring that the suboptimal path has no shared nodes.

[0085] The implementation of Dijkstra's algorithm in step S1 includes the following sub-steps:

[0086] Initialize the adjacency matrix W, where the edge weight is the equivalent signal distance d, calculated using the following formula:

[0087] Where A is the signal strength constant at a reference distance of 1 meter, and n is the path loss factor;

[0088] Define distance vector D and visit marker vector U. Initially, only the distance to the source node is 0, and the distances to the other nodes are infinite.

[0089] Iteratively select the vertex v with the smallest distance among the unvisited nodes. k Mark it as visited and update the distance values ​​of its neighboring nodes:

[0090] d j =min(d j d k +ω(vk ,v j Once the target node has been visited or all nodes have been traversed, output the shortest path and its set of relay nodes.

[0091] initialization:

[0092] The weights of the adjacency matrix W are based on the equivalent signal distance, where:

[0093]

[0094] A represents the signal strength calibrated at 1 meter, and n is dynamically adjusted according to environmental interference (e.g., n = 3.5 in substation areas).

[0095] The distance vector D is initialized to 0 for the source node and infinity for the other nodes; the visit marker vector U is initialized to all zeros.

[0096] Iterative updates:

[0097] Each time, select the vertex v with the smallest distance among the unvisited nodes. k Marked as visited (U[k] = 1).

[0098] Update v k Distance values ​​for all neighbors:

[0099] d j =min(d j d k +ω(v k ,v j ))

[0100] The update process is represented in matrix form:

[0101]

[0102] Termination conditions:

[0103] Output the shortest path when the target node is visited or all nodes have been traversed.

[0104] The calculation of the equivalent signal distance is based on the logarithmic path loss model:

[0105] RSSI=A-10nlg(d)

[0106] A is calibrated through actual measurement, and n is dynamically adjusted according to the electromagnetic interference intensity in the power distribution network environment. The higher the interference intensity, the larger the value of n.

[0107] Logarithmic path loss model:

[0108] Relationship between received power and distance:

[0109]

[0110] Simplified to RSSI expression:

[0111] RSSI=A-10nlg(d)

[0112] Where A is the measured reference signal strength at 1 meter, and X is Gaussian noise.

[0113] Dynamic parameter adjustment:

[0114] In high-interference areas such as substations, the value of n increases with the intensity of electromagnetic interference (e.g., n=4 when the interference is strong, and n=2 when the interference is weak) to more accurately reflect the link quality.

[0115] The incomplete topology graph is generated as follows:

[0116] Delete all relay nodes in the main path except for the source node and the aggregation node;

[0117] Remove edges that are directly connected to relay nodes, including all edges where the relay node is an endpoint;

[0118] Preserve the original connections of the remaining nodes to form a new adjacency matrix W. f Used for suboptimal path calculation.

[0119] Delete relay node:

[0120] The relay nodes of the main path (such as V2, V5) are removed from the vertex set V (Document 3). Figure 4-7 ).

[0121] Remove edges that are directly connected to these nodes (such as V1-V2, V2-V5, V5-V7) to prevent suboptimal paths from passing through nodes in the original path.

[0122] Constructing an incomplete topology graph:

[0123] Preserve the connection relationships of the remaining nodes (e.g., V1, V3, V4, V6, V7) to form an adjacency matrix W. f .

[0124] Recalculate paths in the incomplete topology graph to ensure that suboptimal paths are completely independent.

[0125] The formula for calculating the node relevance factor δ is:

[0126] Where n is the number of common relay nodes in the two paths, and n0 is the total number of relay nodes in the main path; when δ = 0, the two paths are determined to be non-intersecting, and the relay nodes of the suboptimal path have no intersection with the main path.

[0127] Node relevance factor δ:

[0128] The formula is defined as:

[0129] Where n is the number of common relay nodes in the dual paths, and n0 is the total number of relay nodes in the main path.

[0130] When δ = 0, the suboptimal path and the primary path have no shared nodes (Document 3). Figure 4-8 ), satisfying the requirement that nodes do not intersect.

[0131] The distribution method for the dual paths in step S3 is as follows:

[0132] The aggregation node sends the suboptimal path information to the source node through the reverse routing of the primary path;

[0133] The node routing table uses a two-column structure to store the following information for the primary path and the secondary path respectively:

[0134] hop count, next hop address, predecessor list;

[0135] Time to Live (TTL), route tag, RSSI value;

[0136] The route tag field is used to identify that the primary path is active and the secondary path is in standby status.

[0137] Reverse route distribution:

[0138] The aggregation node sends the suboptimal path information to the source node through the reverse path of the primary path, for example, by transmitting path information through RREP messages.

[0139] Two-column routing table:

[0140] The routing table has been expanded to two columns, storing the following information separately:

[0141] Main path: hop count, next hop address (e.g., V2), RSSI value (-70dBm), time to survival (30s).

[0142] Suboptimal path: number of hops, next hop address (e.g., V3), RSSI value (-75dBm), time to survival (25s).

[0143] The route tag field identifies the primary path as "active" and the secondary path as "alternate".

[0144] The construction of a network topology diagram includes the following steps:

[0145] Sensor nodes periodically broadcast HELLO messages, which are extended to carry a transmit power field.

[0146] The neighboring node reads the RSSI value by calling the physical layer interface based on the received HELLO message, and calculates the equivalent distance by combining it with the transmit power;

[0147] The equivalent distance is used as the edge weight and uploaded to the aggregation node to form a global topology matrix.

[0148] HELLO message broadcast:

[0149] The node periodically broadcasts a HELLO message, which carries the transmit power value.

[0150] Equivalent distance calculation:

[0151] Neighboring nodes read RSSI values ​​through the physical layer interface and calculate the equivalent distance by combining them with the transmit power.

[0152] For example, node V1 receives a HELLO message from V2, RSSI = -70dBm, transmit power = 20dBm, and calculates d = 10^((A+70) / 10n).

[0153] Construction of topological matrix:

[0154] The aggregation node integrates the neighbor relationships and equivalent distances of all nodes to construct a global adjacency matrix W.

[0155] The structure of the HELLO message includes:

[0156] Grouping type, reserved bits, skipped number field;

[0157] Routing request message identifier, destination node IP address, source node IP address;

[0158] Transmit power value, source node sequence number;

[0159] The transmit power value and RSSI value are used together for equivalent distance calculation.

[0160] HELLO message structure:

[0161] The group type field is identified as a HELLO message.

[0162] It includes the transmit power value (e.g., 20dBm), the source node IP (e.g., 192.168.1.2), and the serial number.

[0163] The receiving node calculates the equivalent distance based on the transmit power and RSSI.

[0164] It also includes dynamic maintenance steps:

[0165] When a node detects a link failure, it switches to the suboptimal path to transmit data and sends a RERR message to the aggregation node. The RERR message contains the IP address and sequence number of the faulty node.

[0166] The aggregation node filters the affected node set B based on the RERR message. Set B consists of all upstream nodes whose routing tables contain the faulty node.

[0167] The aggregation node sends a Route Check message (RCHE) to the nodes in set B. The RCHE message structure includes:

[0168] Grouping type, reserved bits, number of unreachable destination nodes;

[0169] Faulty node IP address, faulty node serial number;

[0170] After receiving RCHE, a normal node replies with a CREP message, which includes the source node's IP address, the faulty node's authentication field, and the link status identifier.

[0171] The aggregation node updates the topology map based on the CREP response, recalculates the dual paths, and distributes the update information via reverse routing through the suboptimal path.

[0172] Fault detection and switching:

[0173] When a node detects a link failure (such as three consecutive packet losses), it switches to a suboptimal path and sends a RERR message.

[0174] RCHE / CREP interaction:

[0175] The aggregation node filters out the set B of upstream nodes affected by the faulty node (Section 4.4.2 of Document 1) and sends an RCHE message.

[0176] Normal nodes reply with CREP messages (Document 2, Table 4-6), which include verification fields for faulty nodes (such as CRC checksums).

[0177] Topology update and path recalculation:

[0178] The aggregation node updates the topology map based on the CREP response, recalculates the dual paths, and distributes the update information via reverse routing through the suboptimal path.

[0179] The triggering conditions for topology graph updates in the dynamic maintenance process include:

[0180] Electromagnetic interference caused RSSI fluctuations to exceed a preset threshold;

[0181] Node failures caused a continuous packet loss rate exceeding 10%.

[0182] When the network topology changes, the relevance factor δ>0, the two paths are forcibly recalculated.

[0183] Triggering conditions:

[0184] RSSI fluctuation threshold: When the RSSI change exceeds ±5dB, path switching is triggered.

[0185] Packet loss rate threshold: A continuous packet loss rate >10% indicates a link failure.

[0186] Forced update if δ>0: If a topology change causes δ>0 (e.g., adding a node that shares a relay with the main path), recalculate the dual path.

[0187] Further details are described in Example 2.

[0188] Example 2

[0189] Based on Example 1, this example proposes a node-disjoint bipath routing method based on graph theory computation. The specific implementation method is as follows:

[0190] The AODV protocol works based on the distance vector algorithm, but unlike traditional distance vector routing protocols, AODV does not require periodically broadcasting routing update messages, thus saving network resources such as bandwidth and energy consumption.

[0191] In power distribution monitoring networks, changes in network topology can occur due to line maintenance, the addition of new monitoring points, equipment failures, or the presence of mobile intelligent terminals. If proactive routing is used for network configuration, a large number of topology messages are generated during the route discovery phase. However, with the AODV routing protocol, when a source node needs to communicate with a destination node, it only initiates a route request process if a valid route does not exist in its routing table. It broadcasts a query to inquire whether a route to the destination node exists in the network, without calculating all routing information across the entire network. This saves significant network resources. Furthermore, power distribution monitoring networks are large-scale, containing numerous nodes. Data from these dispersed nodes needs to be transmitted to the data center via multiple hops. The AODV routing protocol supports multi-hop networks and includes optimal path selection during transmission path selection. Optimizing the optimal path further improves the protocol's network performance. Each message in the AODV protocol contains a sequence number field. This mechanism allows nodes to distinguish between new and old messages, thereby updating the routing information in their routing tables. The sequence number also effectively solves the problems of routing loops and infinite counting. The specific workflow of AODV is shown in the diagram below. Figure 1 As shown.

[0192] AODV message message and routing table format

[0193] In wireless sensor networks, "control packets" specifically refer to the information used in the routing layer to maintain the operation of routing protocols. Under the AODV protocol framework, control packets specifically include four types: Route Request (RREQ), Route Reply (RREP), Hello messages, and Route Error (RERR). These control packets together support the route discovery and maintenance process of the AODV protocol.

[0194] Routing request messages and routing maintenance packets

[0195] The RREQ (Route Request Message) packet, whose structure is detailed in Table 4-1, is used in the route discovery phase to determine the location of the destination node. The destination sequence number in this message reflects the recentity of establishing a route to the destination address using this packet, while the source sequence number indicates the recentity of establishing a route to the source address using this packet. RREQ packets propagate in the same direction as the data packets are transmitted, and their sending and receiving nodes are also the same.

[0196] Table 4-1 RREQ Structure Table

[0197]

[0198] Routing reply message

[0199] RREP, or Route Reply Message, primarily functions to provide the source node initiating the request with the specific location information of the destination node. Its structural details are shown in Table 4-2. The transmission path of RREP is the reverse of that of the DATA data packet; that is, the sender of the RREP is the intended receiver of the DATA packet, while the destination of the RREP is the sender of the DATA packet. The destination node of the RREP is precisely the location of the source node of the DATA packet.

[0200] HELLO packets are special RREP packets with a fixed hop count of 1. The core function of HELLO packets is to monitor the connectivity and activity status of single-hop links between adjacent nodes. This packet mechanism ensures that nodes in the network can promptly understand the reachability of their direct neighbors, thereby maintaining and updating routing information.

[0201] Table 4-2 RREP Structure Table

[0202]

[0203] Routing error message

[0204] RERR, or Routing Error Packet, is responsible for notifying relevant nodes of route breaks. Its structure is shown in Table 4-3. The sequence number of the unreachable destination within the RERR packet represents the latest information about the broken path. RERR indicates that there are no longer valid paths between certain nodes in the network, thus triggering a route rediscovery or maintenance process.

[0205] Table 4-3 RERR Structure Table

[0206]

[0207] Routing table format

[0208] The primary function of a routing table is to indicate the next-hop forwarding path for data packets; its structure details can be found in Table 4-4. The destination sequence number in the table reflects the up-to-dateness of the routing information from the current node to the destination address, while the predecessor node list in the routing table compiles all nodes originating from that route. The routing table not only records the best paths to each destination but also records all source nodes that depend on these paths.

[0209] Table 4-4 Routing Structure Table

[0210]

[0211] Route discovery phase

[0212] like Figure 2 As shown, this is the AODV route discovery phase.

[0213] When a data transmission request is initiated from source node S to destination node D, S first checks its routing table, searching for a path to D. If a route to the destination node exists in the routing table, the data is sent directly along the next hop of that path. If no path to the destination node is found, S initiates a route discovery mechanism. During route discovery, S broadcasts a Route Request (RREQ) to neighboring nodes. Nodes receiving the RREQ check the destination address within the packet. If their own address matches that of the destination node D, they send a Route Response (RREP) to source node S. If their address differs from that of the destination node D, they continue forwarding the RREQ. This process continues until the destination node D receives the RREQ and responds with an RREP. Once source node S receives the RREP from destination node D, it establishes a route to D, thus completing the route discovery process.

[0214] Routing maintenance phase

[0215] In the AODV protocol's operation, nodes maintain neighboring node routes by periodically sending HELLO messages. Once these messages are received by neighboring nodes, they trigger them to update the route lifetimes in their local routing tables, ensuring the timeliness and accuracy of routing information. When a route's lifetime reaches zero, it indicates that the route is invalid, and nodes periodically clear invalid routes to achieve network link maintenance.

[0216] The routing maintenance process is as follows Figure 3As shown, when a link failure occurs between nodes c and D in the network, the path to node D in node c's routing table expires due to the node failure. When node c detects the routing failure, it sends a Routing Error (RERR) message. The predecessor node of the failed node that receives this message will relay the RERR until the message reaches the source node. Once the source node receives the RERR, it will identify and clear all related failed routes to ensure that the network's routing information remains up-to-date.

[0217] WSN Graph Theory Algorithm

[0218] Graph theory, a branch of mathematics dedicated to exploring the structure and properties of graphs, can be applied to WSNs to simplify some key network models. The WSN network topology can be likened to a graph in graph theory, with sensor nodes representing vertices and connections between nodes representing edges. By utilizing this correspondence, the routing decision problem in WSNs can be optimized into an optimization problem in graph theory, thus leveraging graph theory and algorithms to optimize WSNs.

[0219] Dijkstra's algorithm

[0220] Dijkstra's algorithm calculates the shortest paths from a starting vertex to all other vertices in a graph. Because of its simplicity and efficiency, Dijkstra's algorithm is frequently used in complex network architectures such as WSNs to improve routing decisions and enhance data transmission efficiency and stability.

[0221] Dijkstra's algorithm flow

[0222] At the beginning, Dijkstra's algorithm sets the source node in the network as the starting point in the graph. Nodes that are neighbors in the network are connected by edges in the graph. At this time, the metric from the source node to other nodes is represented by the weight of the edge in the graph. Obviously, in the initial stage, the distance from the source node to itself is zero. Since path calculation has not yet started, the metric from the source node to other unreachable nodes is set to infinity. These metrics are stored in the path information, and each node is marked with a state. This state can indicate whether the current node has been visited. The purpose of this is to avoid repeatedly visiting nodes in the network and forming a loop.

[0223] During the pathfinding phase, the source node selects the node with the smallest metric value from the set of unvisited nodes. This node represents the closest node to the starting point in the graph. After the first node selection, the node is saved in the path information, and the selected node's identifier is changed to "visited." At this point, the shortest path from the source node to this node is confirmed. Then, the distances of the source node's neighboring nodes to the source node are known. These distances are compared with the existing distance values ​​in the path information; if the distance is smaller, the path and distance values ​​are updated. This completes the first round of pathfinding. After the first round of pathfinding, the algorithm returns to the source node selection state, selecting the node with the smallest metric value from the unvisited nodes and repeating the first round until all nodes have been visited. The algorithm terminates when all nodes have been visited or the shortest path from the source node to the target node has been determined.

[0224] The specific process of Dijkstra's algorithm is as follows:

[0225] (1) Definition of a graph

[0226] Vertex set: V = {v1, v2, ..., v3} n There are n vertices in total.

[0227] Edge set: The number of edges is denoted as |E|.

[0228] Weighting function: w: E→R + , representing the non-negative weight of each edge, w(v i ,v j ) is an edge (v i ,v j The weight of ). If Define w(v) i ,v j ) = +∞.

[0229] Define the adjacency matrix W to represent the weight function w:

[0230]

[0231] (2) Initialization

[0232] Distance vector: D = [d1, d2, ..., d n ] Τ , where d i This represents the distance from the source node s to the vertex v. i The current known shortest distance. Initial value:

[0233]

[0234] Access marker vector: U = [u1, u2, ..., u n ] Τ , where u i =0 means v i Not visited, u i =1 means v i This page has been visited. The initial value is: U = [0,0,...,0] Τ

[0235] (3) Algorithm Core

[0236] First, a greedy selection algorithm is used, choosing the vertex with the minimum distance to the source node. That is, each time the set of unvisited vertices is traversed, a vertex v that satisfies the following condition is selected. k :

[0237]

[0238] Then v k Mark as visited and set u k =1.

[0239] Update the distance vector D for each unvisited v k The neighboring vertex update formula is:

[0240] d j =min(d j ,d k +w(v i ,v j )) (0-4)

[0241] For each j, written in matrix form:

[0242] D (t+1) =min(D (t) D k (t) +W[k,:]) (0-5)

[0243] In the formula, D k (t) It is a scalar d k W[k,:] is the k-th row of matrix W.

[0244] (4) Termination conditions

[0245] When the target vertex is visited or all vertices are visited (i.e., U = [1, 1, ..., 1)). Τ When ), the algorithm ends, and vector D... k That is, the distance from the source node s to the vertex v k The shortest path length.

[0246] (5) Output results

[0247] The resulting vector D represents the shortest path length from the source vertex s to all vertices in the graph. If the paths themselves are needed, the path matrix P can be stored.

[0248] P[v i ] = argminpathvia{v k} (0-6)

[0249] In the formula, P[v i ] is from source node s to vertex v i The set of vertices with the shortest path.

[0250] D-Dijkstra algorithm based on node independence

[0251] Dijkstra's algorithm can find the shortest path. Considering the comprehensive weights, it can obtain the optimal path required in a WSN, achieving the goal of minimizing network resource waste. However, the algorithm itself can only traverse a single-source shortest path and cannot meet the redundancy requirements of transmission paths in WSNs. For the same topology graph, it cannot find a suboptimal path as a backup path, and it cannot solve the redundancy problem in the network. Therefore, an iterative improvement is made to the algorithm. The original network topology is defined as a complete topology, and the topology graph with some nodes and edges removed from the complete topology is called an incomplete topology.

[0252] Interrelationships of multiple paths

[0253] Multi-path interrelationships such as Figure 4 As shown, multipath routing is divided into three categories based on the degree of sharing between paths. For example... Figure 4 (a) Paths without common nodes or links are called non-intersecting paths. These paths will not affect each other when they encounter failures and can withstand failures of up to n-1 paths, thus having the best fault tolerance. Figure 4 (b) Link-disjoint paths allow for the existence of common nodes between paths, but no shared links; therefore, the failure of one node can affect multiple paths. In contrast, Figure 4 (c) Related multipath routes have low fault tolerance because these paths may share nodes or links, reducing their independence. Therefore, the failure of any node or link may disrupt multiple paths.

[0254] Relevance factor

[0255] In routing protocols with multiple transmission paths, the interrelationships between these paths affect the reliability of the protocol. Data should ideally be transmitted via the optimal path. However, when the optimal path fails, a backup path takes over. The correlation between the backup and optimal paths represents the likelihood that the backup path will remain unaffected when the optimal path fails. This means that if multiple nodes are shared between the optimal and backup paths, a failure in the optimal path will highly likely affect the backup path as well, preventing data transmission. Only when the paths are mutually exclusive can the impact of node failures on network stability be minimized. To quantify this link-to-link difference, the concept of a node "relevance factor" is introduced. This factor measures the similarity between the backup and optimal paths using the number of shared nodes. The effectiveness of backup paths can be evaluated by calculating the node relevance factor, thus ensuring the reliability of backup paths during path selection. The node relevance factor δ is defined as:

[0256]

[0257] In the formula, n is the number of common relay nodes for the backup path and the optimal path, and n0 is the total number of relay nodes for the main route.

[0258] From equation (4-7), we can see that when δ = 0, the relay nodes of the optimal path and the backup path are completely unrelated; when δ = 1, the relay nodes of the optimal path and the backup path are completely related, that is, they are the same path. The lower the correlation factor δ, the higher the redundancy of the backup path.

[0259] Improved algorithm flow

[0260] To improve network redundancy by finding backup paths in the topology, the original Dijkstra algorithm is improved to obtain the D-Dijkstra algorithm. Figure 5 This is a flowchart of the D-Dijkstra algorithm. After obtaining the optimal path using the original Dijkstra algorithm, this algorithm requires a second solution to the topology. However, solving the same topology will only yield the same optimal solution. Therefore, it is necessary to perform iterations of pruning on the complete topology and a second solution on the incomplete topology to obtain the suboptimal path. It is known that when the node relevance factor δ = 0, the relay nodes of the main path and the backup path are completely unrelated, and the redundancy of the backup path reaches its maximum. Therefore, based on the complete topology, removing the relay nodes and associated edges of the main path in the initial calculation yields an incomplete topology. Using this incomplete topology as input, the obtained suboptimal path guarantees that the relay nodes of the main path and the backup path are completely unrelated.

[0261] The optimal path calculation results are as follows Figure 6As shown, in a complete topological graph, Dijkstra's algorithm finds the shortest path from V1 to V7 as {V1, V2, V5, V7}. After removing elements from this complete topology, we obtain the following: Figure 7 The incomplete topological graph is used as the input for the secondary computation.

[0262] The results of the dual-path calculation are as follows Figure 8 As shown, after four traversals, the suboptimal path from V1 to V7 is {V1, V3, V6, V7}, and the node relevance factor δ = 0 between it and the optimal path from V1 to V7 {V1, V2, V5, V7}.

[0263] Simulation verification was performed using MATLAB 2022b, and the D-Dijkstra algorithm was used to calculate the bipath results for some nodes, as shown below. Figure 9-14 As shown, an optimal transmission path and a suboptimal transmission path were found for each node. The red path is the optimal transmission path, serving as the best route for the network node, while the green path is the suboptimal transmission path, serving as the backup route for the network node. Furthermore, the relationship between the two paths is based on node-independent paths, so when an intermediate node of the optimal route fails, it does not affect the connectivity of the backup route at all.

[0264] RPWNCN Protocol Design

[0265] In the ecosystem of wireless sensor networks, two important structural components exist: sensor nodes and aggregation nodes. Sensor nodes act as data collectors, active at the forefront of environmental monitoring, responsible for capturing subtle changes in the surrounding environment. Meanwhile, aggregation nodes act as information hubs. Their core responsibility is to aggregate and process the data collected by the sensor nodes and efficiently transmit this refined information, ensuring unimpeded information flow throughout the monitoring network. Aggregator nodes possess certain computing and communication capabilities and have sufficient power, allowing them to understand the network topology and thus enabling the use of graph theory algorithms to calculate globally optimal data transmission paths.

[0266] Signal equivalent distance model

[0267] In real-world physical environments, the Euclidean distance between WSN nodes cannot fully represent the communication quality between them, especially in situations like power distribution networks where electromagnetic interference exists. Nodes with closer Euclidean distances are not necessarily those with the highest communication quality. Therefore, using Euclidean distance to build network topology models can impact the actual optimal routing decisions. In such cases, a special variable is needed to represent the topological distance between nodes. Since the communication quality between nodes is directly affected by signal strength, and the Receiving Signal Strength Index (RSSI) is collected after nodes have been deployed and experienced a series of environmental influences, the received signal strength value more accurately reflects the communication quality between nodes. Using the RSSI value as a benchmark to calculate the equivalent signal distance as the weights in the network topology model can help routing protocols make more accurate routing decisions.

[0268] In an ideal environment, electromagnetic signals propagate directly in free space, and the expression for the received power at the receiver is:

[0269]

[0270] In the formula, P t P represents the transmit power when the distance between the transmitter and receiver is d; r (d) represents the received power; G t With G r These represent the power gain at the transmitting and receiving ends, respectively; λ is the electromagnetic signal wavelength; and L is the path loss parameter.

[0271] However, in the real world, electromagnetic signal propagation is affected by natural and man-made factors such as diffraction, multipath effects, and physical obstacles, all of which can interfere with the signal propagation path. To quantify the impact of various interferences, a logarithmic path loss model is typically used to characterize the relationship between RSSI values ​​and distance. The specific expression of this model is:

[0272]

[0273] In the formula, P r (d) and P r (d0) represents the received power at distances d and d0 from the transmitter, respectively; n is the path loss factor; d0 is usually taken as 1m, and X is a Gaussian random variable with a mean of 0, which can be ignored. P r (d0) is a constant representing the received signal strength at a distance of 1 meter, denoted as A. The propagation model is then as follows:

[0274] RSSI=A-10nlg(d) (0-10)

[0275] From equation (4-10), the equivalent signal distance can be derived as follows:

[0276]

[0277] Using the equivalent signal distance d derived from formula (4-11) to replace the Euclidean distance between nodes as the weight of the network topology can more accurately reflect the link quality of wireless communication.

[0278] RPWNCN routing protocol design

[0279] The RPWNCN routing protocol adds two new message types to the traditional AODV routing protocol for network-wide route checking and response, and expands the original routing table to accommodate the characteristics of dual transmission paths.

[0280] Routing inspection message

[0281] The Route Check (RCHE) message is a message sent by the aggregation node to the sensor node as a route check command. This message carries the address of a faulty node. The RCHE structure table is shown in Table 4-5.

[0282] Table 4-5 RCHE Structure Table

[0283]

[0284] This structure table contains the IP address and sequence number of the source node, the IP address and sequence number of the destination node, and the IP address and sequence number of the faulty node. The source node is the aggregation node. It is activated when the aggregation node receives the RERR message sent by the sensor node in the network. It serves as a network diagnostic message for the local network topology. Its purpose is to diagnose whether the link between the faulty node and other nodes is broken, thereby updating the network topology in real time and recalculating routes for the nodes affected by the fault.

[0285] Check reply message

[0286] The Check Reply (CREP) message is a message sent by a sensor node to the aggregation node after receiving a route check instruction. The structure of the Check Reply message is shown in Table 4-6.

[0287] Table 4-6 CREP Structure Table

[0288]

[0289] This structure table contains the IP address and sequence number of the source node, the IP address and sequence number of the destination node, and the IP address and sequence number of the faulty node. The IP address and sequence number of the source node are used to tell the aggregation node the origin of the packet, and the destination node is the aggregation node. The IP address and sequence number of the faulty node are used to tell the aggregation node the reason for the check and reply, so as to avoid information confusion caused by multiple node failures. The timing of its use is when the sensor node receives the routing check packet from the aggregation node, proving that the link is not damaged, and sends the link information to the aggregation node so as to determine the nodes affected by the faulty node.

[0290] Routing table extension

[0291] The RPWNCN routing table is shown in Table 4-7. The modified routing table records not only routing information to the destination node but also routing information for neighboring nodes. To store and manage the RSSI values ​​of neighboring nodes, the routing table is expanded by adding a new column specifically for recording this information. This expansion allows the routing table to not only contain traditional routing information but also reflect the signal strength of the links. Because the RPWNCN routing protocol is a dual-routing protocol, two path information entries exist in the routing table of each destination node.

[0292] Table 4-7 RPUWNCN Routing Table

[0293]

[0294] Routing protocol workflow

[0295] RPWNCN route establishment phase

[0296] (1) Network topology collection phase

[0297] After network initialization is complete, all nodes broadcast HELLO message packets. Since the forwarding lifespan of a HELLO message packet is only one hop, the other nodes that can receive a HELLO message packet broadcast by a certain node are all neighbors of that node in the topology graph. When these neighbor nodes receive the HELLO message packet, they call the physical layer interface to read the RSSI value of this node. After this process continues, all nodes will record their own neighbor nodes and their RSSI values ​​in the routing table.

[0298] (2) Flooding Constraint Routing Request Phase

[0299] After the HELLO packet sending phase ends, all nodes start at random times within a certain time period, targeting the sink node, and package their neighbor information to send to the sink node. Before sending data, each node checks its routing table. If no available route to the sink node is found, the node sends a route request packet to flood route constraint requests. Once the route is initially established, the node sends the neighbor information to the sink node.

[0300] (3) Dual route calculation and allocation stage

[0301] The RPWNCN route establishment flowchart is as follows: Figure 15 As shown. The sink node receives neighbor information from each node, stores this information in the neighbor information table, calculates the equivalent signal distance d between nodes based on the RSSI, and establishes a topology matrix using d as the metric. This matrix is ​​then used to calculate the topology map showing the distribution of node locations throughout the network. Using this topology map as input, and employing the node-independent D-Dijkstra algorithm, the sink node can calculate the optimal route and backup route for each node's data transmission, store this information in the neighbor information table, and assign the node's dual routes to each relevant node by sending the reverse routes established during the initial route to the sink node. After receiving this routing information, each node establishes a dual route with the destination address of the sink node and its reverse route in its routing table. The route establishment phase ends after the relevant nodes have established two node-independent paths.

[0302] RPWNCN routing maintenance phase

[0303] RPWNCN routing maintenance flowchart as follows Figure 16 As shown.

[0304] After the route establishment phase is complete, each node's routing table contains two route entries to the aggregation node. When a link is interrupted, the upstream node of the failed node will activate its backup route to continue transmitting data and generate a RERR message with the aggregation node as the destination address. Upon receiving the RERR from a normal node, the aggregation node checks the routing tables of all nodes, filters out all potential upstream nodes affected by the failed node, adds them to set B, and simultaneously sends RCHE messages to all nodes in set B via the reverse path of the optimal path, waiting for CREP messages. If the node's link is normal, it will receive the RCHE message and send a CREP message. Conversely, if a node is also affected by the failure, it will not receive the RCHE message, and the aggregation node will not receive the CREP message. The node that received the CREP message is removed from set B. At this point, set B contains only upstream nodes affected by the failed node. The network topology is updated, and the optimal path without passing through the failed node and the backup path unrelated to the optimal path are calculated. The aggregation node transmits the reverse route of the suboptimal path to these relevant nodes that need route updates. Affected nodes delete their original routes and update their primary and backup routes.

[0305] The above-described embodiments are detailed and specific, illustrating preferred embodiments of the present invention. They are only used to illustrate the technical ideas and features of the present invention, with the aim of enabling those skilled in the art to understand the content of the present invention and implement it accordingly. However, they are not limited to the present invention, and the patent scope of the present invention cannot be limited by this embodiment alone. That is, any equivalent changes or modifications made to the spirit disclosed in the present invention, without departing from the structure of the present invention, such as local improvements within the system and modifications or transformations between subsystems, are still within the patent scope of the present invention.

Claims

1. A node-disjoint bipath routing method based on graph theory computation, characterized in that, Includes the following steps: Step S1: Calculate the main path. Based on the network topology, the sink node uses Dijkstra's algorithm to calculate the shortest path from the source node to the sink node as the main path, and records the set of relay nodes of the main path. Step S2: Suboptimal path generation. Delete the relay nodes and associated edges of the main path to generate an incomplete topology graph. Use Dijkstra's algorithm again to calculate the suboptimal path from the source node to the sink node, ensuring that the suboptimal path does not intersect with the nodes of the main path. Step S3: Dual-path deployment. The primary path and the secondary path are distributed to relevant nodes, so that the node routing table stores the next-hop information of both paths simultaneously, and the independence of the two paths is verified based on the node relevance factor, where the node relevance factor δ = 0.

2. The node-disjoint bipath routing method based on graph theory computation according to claim 1, characterized in that, The implementation of Dijkstra's algorithm in step S1 includes the following sub-steps: Initialize the adjacency matrix W, where the edge weight is the equivalent signal distance d, calculated using the following formula: Where A is the signal strength constant at a reference distance of 1 meter, and n is the path loss factor; Define distance vector D and visit marker vector U. Initially, only the distance to the source node is 0, and the distances to the other nodes are infinite. Iteratively select the vertex v with the smallest distance among the unvisited nodes. k Mark it as visited and update the distance values ​​of its neighboring nodes: d j =min(d j d k +ω(v k ,v j Once the target node has been visited or all nodes have been traversed, output the shortest path and its set of relay nodes.

3. The node-disjoint bipath routing method based on graph theory computation according to claim 2, characterized in that, The calculation of the equivalent signal distance is based on the logarithmic path loss model: RSSI=A-10nlg(d) A is calibrated through actual measurement, and n is dynamically adjusted according to the electromagnetic interference intensity in the power distribution network environment. The higher the interference intensity, the larger the value of n.

4. The node-disjoint bipath routing method based on graph theory computation according to claim 1, characterized in that, The incomplete topology graph is generated as follows: Delete all relay nodes in the main path except for the source node and the aggregation node; Remove the edges that are directly connected to the relay node, including all edges where the relay node is an endpoint; Preserve the original connections of the remaining nodes to form a new adjacency matrix W. f Used for suboptimal path calculation.

5. The node-disjoint bipath routing method based on graph theory computation according to claim 1, characterized in that, The formula for calculating the node relevance factor δ is as follows: Where n is the number of common relay nodes in the two paths, and n0 is the total number of relay nodes in the main path; when δ = 0, the two paths are determined to be non-intersecting, and the relay nodes of the suboptimal path have no intersection with the main path.

6. The node-disjoint bipath routing method based on graph theory computation according to claim 1, characterized in that, The dual-path distribution method in step S3 is as follows: The aggregation node sends the suboptimal path information to the source node through the reverse routing of the primary path; The node routing table uses a two-column structure to store the following information for the primary path and the secondary path respectively: hop count, next hop address, predecessor list; Time to Live (TTL), route tag, RSSI value; The route tag field is used to identify that the primary path is active and the secondary path is in standby status.

7. The node-disjoint bipath routing method based on graph theory computation according to claim 1, characterized in that, The construction of the network topology map includes the following steps: The sensor node periodically broadcasts HELLO messages, which are extended to carry a transmit power field; The neighboring node reads the RSSI value by calling the physical layer interface based on the received HELLO message, and calculates the equivalent distance by combining it with the transmit power; The equivalent distance is used as the edge weight and uploaded to the aggregation node to form a global topology matrix.

8. The node-disjoint bipath routing method based on graph theory computation according to claim 7, characterized in that, The structure of the HELLO message includes: Grouping type, reserved bits, skipped number field; Routing request message identifier, destination node IP address, source node IP address; Transmit power value, source node sequence number; The transmit power value and RSSI value are used together for equivalent distance calculation.

9. The node-disjoint bipath routing method based on graph theory computation according to claim 1, characterized in that, It also includes dynamic maintenance steps: When a node detects a link failure, it switches to the suboptimal path to transmit data and sends a RERR message to the aggregation node. The RERR message contains the IP address and sequence number of the faulty node. The aggregation node filters the affected node set B based on the RERR message, where set B consists of all upstream nodes whose routing tables contain the faulty node. The aggregation node sends a Route Check Message (RCHE) to the nodes in set B. The RCHE message structure includes: Grouping type, reserved bits, number of unreachable destination nodes; Faulty node IP address, faulty node serial number; After receiving RCHE, a normal node replies with a CREP message, which includes the source node's IP address, the faulty node's verification field, and the link status identifier. The aggregation node updates the topology map based on the CREP response, recalculates the dual paths, and distributes the update information via reverse routing through the suboptimal path.

10. The node-disjoint bipath routing method based on graph theory computation according to claim 9, characterized in that, The triggering conditions for topology graph updates in the dynamic maintenance step include: Electromagnetic interference caused RSSI fluctuations to exceed a preset threshold; Node failures caused a continuous packet loss rate exceeding 10%. When the network topology changes, the relevance factor δ>0, the two paths are forcibly recalculated.

Images