Edge computing power network route generation method and system based on SDN (Software Defined Network) and Raft algorithm election mechanism

Abstract

The invention provides an edge computing power network route generation method and system based on an SDN (Software Defined Network) and Raft algorithm election mechanism. The edge computing power network route generation method comprises the following steps: adding a security duration range, a dynamic election area controller and a global controller in a Raft distributed consensus algorithm; when the regional controllers are valid, the regional controllers collect intra-domain network information and perform interaction so as to construct a global network topology view; setting QoS (Quality of Service) primary and secondary optimization targets for the flow based on the global network topology view and the type of the flow to perform layered optimization to obtain an optimal path and a standby path; and when the regional controller fails, the SDN nodes in the domain generate a routing path by adopting an on-demand routing mechanism. According to the invention, the dynamic adaptability of the network can be improved, the network security is enhanced, and the QoS guarantee is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of computing power networks, and particularly to a method and system for generating an edge computing power network routing based on an SDN and Raft algorithm election mechanism. Background Art

[0002] With the progress of new network technologies, as a new type of productive force, the computing power network is developing towards the integration of cloud resources and edge nodes. The computing power network interconnects distributed computing and storage resources and realizes on-demand and real-time invocation through a unified scheduling mechanism, thereby improving the utilization efficiency of overall resources. In edge operations, due to the wide distribution of combat units, challenges such as limited bandwidth, node mobility, and demand changes are often faced. By deploying computing resources to edge nodes, the computing power network can push the computing ability to the place closest to the combat nodes, reduce transmission latency, and ensure quick response. The on-demand allocated computing resources can effectively support data processing and transmission in tactical communications, improve the efficiency and real-time performance of tactical execution, and ensure the successful completion of combat missions. Currently, mobile ad hoc networks (MANETs) are mainly used for network organization in battlefield mobile communications. Due to the highly distributed nature of MANETs, each node can only achieve local path optimality based on local information and it is difficult to achieve global optimality. Even for service quality (QoS) routing schemes for different traffic, problems such as QoS requirement conflicts and difficulty in changing traffic management strategies will occur. In recent years, some research has combined software-defined network (SDN) technology with tactical MANETs. Due to the existence of a global view, SDN can consider the network traffic distribution when making routing decisions, reasonably regulate the transmission paths of different application traffic, reduce the possibility of network congestion while providing high-quality routing decisions; secondly, SDN can reduce the control overhead of each node maintaining a routing table or the long latency of source node route finding by issuing flow tables.

[0003] Prior art solution 1: For tactical ad hoc networks, the SDN architecture is introduced, aiming to design a flexible network control architecture that can operate in a distributed manner in the event of controller failure. By local agents sensing the state changes of neighbor links (such as link availability, bandwidth, latency, etc.), it dynamically selects alternative paths and completes traffic forwarding while minimizing message propagation overhead. Local agents support path selection and forwarding decisions by maintaining multiple tables (status table, local forwarding table, and OpenFlow forwarding table). When the source node needs to know the link state of the alternative path and when a certain alternative link fails, it needs to notify the source node. Instead of using the node broadcast mechanism, a subscription mechanism is proposed. The head node of each link maintains a subscription table to record which nodes have subscribed to the status information of the link. When the link fails, the head node only needs to send the status to the subscribed nodes to control the overhead. However, this technical solution only considers the case of a single SDN controller with a fixed location and does not consider the fault tolerance of the SDN controller; in addition, the routing optimization considered in this solution is relatively simple and does not consider the guarantee of QoS. Different combat missions have different requirements for QoS, and a routing selection strategy based on QoS optimization needs to be designed.

[0004] Prior art solution 2: Aiming at the unreliability of a single controller, a multi - controller collaborative routing mechanism is designed. A collaborative mechanism based on link information is proposed for the problem of difficult collaboration between multiple controllers, which is divided into two parts: network information synchronization and routing calculation. The controllers directly use link information for synchronization. The controllers are divided into a leader controller and follower controllers. The leader controller mainly plays a role of conveying update information to other controllers and receiving confirmation information. To control the overhead in the collaborative process, the compressible part of the network information is compressed using a Bloom filter. Considering that traditional routing algorithms are unavailable after compression, a breadth - first search routing algorithm is proposed for routing calculation. However, in this solution, the arrangement position of the controllers is fixed. Tactical MANET nodes have high mobility and need to frequently switch controllers to re - establish control connections and synchronize control information. At the same time, once each controller fails due to a fault or enemy attack, there is no consideration of a backup controller, lacking a flexible SDN controller fault - tolerance mechanism; and when uploading link information to the controller, the QoS - related parameters of the link are not uploaded, and QoS optimization is not considered during the routing process. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method and system for generating an edge computing power network routing based on an SDN and Raft algorithm election mechanism to eliminate or improve one or more defects existing in the prior art.

[0006] On the one hand, the present invention provides a method for generating an edge computing power network route based on an SDN and Raft algorithm election mechanism. The edge computing power network combines a mobile ad hoc network and an SDN architecture. The method includes the following steps:

[0007] Based on the Raft distributed consensus algorithm, elect a regional controller for each area of the mobile ad hoc network, and elect a global controller from the regional controllers; set a secure term range in the Raft distributed consensus algorithm to prevent malicious nodes from conducting elections;

[0008] When the regional controller is valid, within each area, the regional controller collects network information uploaded by each SDN node in the domain, constructs and maintains a local topology view; between each area, synchronize the network information between the regional controllers through link information, update their respective local topology information, and generate a global network topology view; the global controller includes all the functions of the regional controller, and in addition has the function of managing the synchronization between each controller, ensuring the consistency of the network information between each regional controller;

[0009] According to the global network topology view, adopt a hierarchical game mechanism to classify traffic, set primary and secondary optimization objectives for service quality according to the category of traffic for hierarchical optimization, and obtain the optimal path and the backup path;

[0010] Generate flow table rules according to the global network topology view, the optimal path, and the backup path, and issue them to the switches along the path.

[0011] In some embodiments of the present invention, setting a secure term range in the Raft distributed consensus algorithm includes:

[0012] Set the secure term range according to the network state, the node's own capabilities, and the historical election behavior. The calculation formula is:

[0013] T max =T base ×(1+F stability +F resources +F trust );

[0014] Among them, T max represents the upper limit of the secure term range; T base represents the basic term value; F stability represents the network stability factor; F resources represents the node resource factor; F trust represents the node trust factor.

[0015] In some embodiments of the present invention, the election method of the regional controller or the global controller includes:

[0016] The node initiates an election request. If the term value of the node is within the secure term range and the behavior score of the node is greater than a preset score threshold, it is determined that the node is a normal node. If it obtains votes from more than half of the other nodes, it becomes the regional controller or the global controller;

[0017] If the term value of the node is not within the secure term range, or the behavior score of the node is not greater than the preset score threshold, it is determined that the node is a malicious node, and the node is prohibited from participating in the election, and the term value is reset.

[0018] In some embodiments of the present invention, within each region, the regional controller collects network information uploaded by each SDN node in the domain, constructs and maintains a local topology view, including:

[0019] Each SDN node regularly monitors the one-way link status information between itself and its neighbor nodes, encapsulates the one-way link status information into a Hello message for broadcasting, and at the same time receives the Hello message sent by the neighbor node to itself to obtain two-way link status information;

[0020] Each SDN node encapsulates the two-way link status information into a TC message and unicasts it to the regional controller regularly to construct the local topology view.

[0021] In some embodiments of the present invention, the synchronization method between the regional controller and the global controller includes:

[0022] When the global controller initiates information synchronization, the global controller broadcasts the update information to each regional controller. After the regional controller confirms the synchronization, it sends a confirmation message to the global controller;

[0023] When the regional controller initiates information synchronization, the regional controller sends the update information to the global controller. The global controller updates according to the update information, and at the same time broadcasts the update information to each regional controller. After each regional controller confirms the synchronization, it sends a confirmation message to the global controller; after the global controller receives the confirmation messages from all regional controllers, it sends a confirmation message to the regional controller that initiated the synchronization request.

[0024] In some embodiments of the present invention, according to the global network topology view, a hierarchical game mechanism is adopted to classify traffic, and primary and secondary optimization objectives for service quality are set according to the category of traffic for hierarchical optimization, including:

[0025] According to the global network topology view, a routing path is selected based on link available capacity, end-to-end delay, and transmission cost;

[0026] Classify traffic into delay-sensitive traffic, high-capacity traffic, and low-cost traffic;

[0027] For the delay-sensitive traffic, the primary optimization objective is to optimize the end-to-end delay, and the secondary optimization objectives are to optimize the link available capacity and transmission cost; for the high-capacity traffic, the primary optimization objective is to optimize the link available capacity, and the secondary optimization objectives are to optimize the end-to-end delay and transmission cost; for the low-cost traffic, the primary optimization objective is to optimize the transmission cost, and the secondary optimization objectives are to optimize the link available capacity and end-to-end delay.

[0028] In some embodiments of the present invention, the primary and secondary optimization objectives are hierarchically optimized to obtain an optimal path and a backup path, including:

[0029] Determine the primary optimization objective according to the traffic class, and construct a primary utility function with the goal of maximizing the utility according to the primary optimization objective; use Yen's K-Shortest Paths algorithm to generate K optimal paths from the source to the destination node, and screen out a set of primary optimization paths from the K paths according to the preset threshold conditions of the primary optimization objective;

[0030] Construct a secondary utility function for the secondary optimization objective, sort the paths in the set of primary optimization paths according to the utility value of the secondary utility function, and use the path with the highest utility value as the optimal path and the path with the second highest utility value as the backup path.

[0031] In some embodiments of the present invention, when the area controller fails, the SDN nodes in the local area generate a routing path for traffic based on a distributed on-demand routing discovery mechanism, including:

[0032] When the source node needs to communicate with the destination node and no path is cached locally, trigger a routing request mechanism to generate a routing request message; the source node broadcasts the routing request message to its neighbor nodes;

[0033] The neighbor nodes add their own link state information to the routing request message; update the local routing table, record the path information from the source node to the current node, and broadcast the updated routing request message;

[0034] When the target node receives the updated routing request message, generate an optimal path according to the link state information in the message; the target node generates a routing reply message and returns it to the source node along the optimal path;

[0035] The source node sends service data along the optimal path.

[0036] In some embodiments of the present invention, the distributed on-demand routing discovery mechanism uses a shortest path algorithm to calculate the optimal path, including:

[0037] Based on the link available capacity, end-to-end delay, and transmission cost, with the goal of minimizing the total path weight, a target function is constructed, and the calculation formula is as follows:

[0038] minW path = ∑ (i,j∈path) α × p ij + β × TC ij + γ × L ij ;

[0039] Wherein, W path represents the total path weight; α, β, and γ all represent weight coefficients, and α + β + γ = 1; p ij represents the link available capacity between nodes i and j; TC ij represents the end-to-end delay; L ij represents the transmission cost.

[0040] On the other hand, the present invention provides an edge computing power network routing generation system based on the SDN and Raft algorithm election mechanism. When the system is executed, it realizes the steps of any one of the methods mentioned above. The system includes:

[0041] Data plane; in the data plane, each SDN node regularly monitors the one-way link status information between itself and its neighbor nodes, encapsulates the one-way link status information into a Hello message for broadcasting, and at the same time receives the Hello message sent by the neighbor nodes to itself to obtain the two-way link status information; encapsulates the two-way link status information into a TC message and unicasts it to the area controller regularly; when the area controller fails, the SDN node generates a routing path for traffic based on the distributed on-demand routing discovery mechanism;

[0042] Control plane; the control plane includes an area election module, an intra-domain topology management module, a controller coordination module, a global topology management module, a routing calculation and optimization module, and a traffic management and flow table distribution module;

[0043] Among them, the area election module is used to elect an area controller for each area of the mobile ad hoc network based on the improved Raft distributed consensus algorithm;

[0044] The intra-domain topology management module is used to collect network information uploaded by each SDN node within the domain in each area, and construct and maintain a local topology view;

[0045] The controller cooperation module is used to synchronize network information among area controllers through link information among areas, update their respective local topology views; and elect a global controller from among area controllers based on the improved Raft distributed consensus algorithm.

[0046] The global topology management module is used to interact with the intra-domain topology management module and the controller cooperation module, and construct a global network topology view based on the maintained intra-domain network topology information and the network topology information of other domains received by the global controller.

[0047] The routing calculation and optimization module is used to classify traffic according to the global network topology view by adopting a hierarchical game mechanism, set primary and secondary optimization objectives for quality of service according to the type of traffic, and perform hierarchical optimization to obtain an optimal path and a backup path.

[0048] The traffic management and flow table distribution module is used to generate flow table rules according to the global network topology view, the optimal path, and the backup path, and distribute them to switches along the path.

[0049] The present invention provides a method and system for generating an edge computing power network route based on an SDN and Raft algorithm election mechanism, which is oriented to the edge tactical scenario in the computing power network. First, a tactical edge network architecture based on SDN multi-controllers is proposed, and its control plane includes an area election module, an intra-domain topology management module, a controller cooperation module, a global topology management module, a routing calculation and optimization module, and a traffic management and flow table distribution module.

[0050] Based on the above modules, an edge computing power network route generation method is implemented, including: adding a secure tenure range in the Raft distributed consensus algorithm, dynamically electing area controllers in the mobile MANET area, ensuring that the area controllers can always be in the core position of the network as the combat team moves, and preventing malicious nodes from interfering to enhance network security.

[0051] When the area controllers are valid, each area controller collects intra-domain network information and conducts interactions to construct a global network topology view, and proposes an adaptive routing mechanism based on multi-objective hierarchical game, considering multiple optimization objectives such as link available capacity, end-to-end delay, and transmission cost. Adopting a hierarchical game mechanism, the QoS objectives are divided into two layers for optimization, generating routing paths, optimizing QoS guarantee, providing differentiated services for different traffic types, and improving the communication quality and resource utilization efficiency of the tactical network.

[0052] When the area controller fails, that is, during the re-election phase of the area controller, the distributed SDN nodes in the data plane temporarily take over the traffic management tasks, adopt an on-demand routing mechanism to complete the routing path optimization, enhance the continuity and stability of the network in a dynamic environment, improve the network's dynamic adaptability, and effectively solve the problem of network function interruption after the controller fails in the prior art.

[0053] Additional advantages, objects, and features of the present invention will be partially described below and will become partially apparent to those of ordinary skill in the art after studying the following. Or they can be learned through the practice of the present invention. The objects and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the specification and the drawings.

[0054] Those skilled in the art will understand that the objects and advantages that can be achieved by the present invention are not limited to the above specifically described, and the above and other objects that the present invention can achieve will be more clearly understood according to the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and do not limit the present invention. In the drawings:

[0056] Figure 1 It is a schematic diagram of the steps of a method for generating an edge computing power network route based on an SDN and Raft algorithm election mechanism in an embodiment of the present invention.

[0057] Figure 2 It is a structural diagram of an edge computing power network based on an SDN and Raft algorithm election mechanism in an embodiment of the present invention.

[0058] Figure 3 It is an overall flowchart of a method for generating an edge computing power network route based on an SDN and Raft algorithm election mechanism in an embodiment of the present invention.

[0059] Figure 4 It is a schematic diagram of the principle of node transformation for improving the Raft algorithm in an embodiment of the present invention.

[0060] Figure 5 It is a schematic diagram of the process of the area controller election mechanism for the improved Raft algorithm in an embodiment of the present invention.

[0061] Figure 6 It is a flowchart of information synchronization between the area controller and the global controller in an embodiment of the present invention.

[0062] Figure 7 It is a flowchart of a route generation method when the area controller is effective in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings. Herein, the illustrative embodiments of the present invention and the descriptions thereof are used to explain the present invention, but do not limit the present invention.

[0064] Herein, it should also be noted that in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution according to the present invention are shown in the drawings, while other details less related to the present invention are omitted.

[0065] It should be emphasized that the term "comprising / including" when used herein refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0066] Herein, it should also be noted that if not otherwise specified, the term "connection" in this document can not only refer to direct connection, but also represent indirect connection with an intermediate.

[0067] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0068] It should be emphasized here that the step numbers mentioned hereinafter do not limit the order of the steps. It should be understood that the steps can be executed in the order mentioned in the embodiments, or different from the order in the embodiments, or several steps can be executed simultaneously.

[0069] In order to solve the problems existing in the existing routing optimization technical solutions, such as the fixed position of the SDN controller, the lack of a controller fault tolerance mechanism, and the simple routing optimization method without considering QoS optimization, the present invention provides an edge computing power network routing generation method based on the SDN and Raft algorithm election mechanism, as Figure 1 shown, the method includes the following steps S101 to S104:

[0070] Step S101: Based on the Raft distributed consensus algorithm, elect a regional controller for each area of the mobile ad hoc network, and elect a global controller from the regional controllers; set a secure tenure range in the Raft distributed consensus algorithm to prevent malicious nodes from conducting elections.

[0071] Step S102: When the regional controller is valid, within each region, the regional controller collects the network information uploaded by each SDN node within the domain, constructs and maintains a local topology view; between regions, the network information between regional controllers is synchronized through link information to update their respective local topology information and generate a global network topology view; the global controller ensures the consistency of network information between local controllers.

[0072] Step S103: Adopt a hierarchical game mechanism to classify traffic, and set primary and secondary optimization objectives for service quality according to the global network topology view and the type of traffic for hierarchical optimization to obtain the optimal path and the backup path.

[0073] Step S104: Generate flow table rules according to the global network topology view, the optimal path, and the backup path, and distribute them to the switches along the path.

[0074] Corresponding to the above method, the present invention also provides an edge computing power network routing generation system based on the SDN and Raft algorithm election mechanism, as Figure 2 shown, the edge computing power network architecture diagram provided by the present invention. The system includes a data plane and a control plane. In the control plane, it further includes a regional election module, an intra-domain topology management module, a controller coordination module, a global topology management module, a routing calculation and optimization module, and a traffic management and flow table distribution module.

[0075] As Figure 3 shown, it shows the functions implemented by each module, and then the edge computing power network routing generation method based on the SDN and Raft algorithm election mechanism is realized as a whole. The following further explains the method and the system.

[0076] In step S101, in the regional election module, based on the Raft distributed consensus algorithm, a regional controller is elected for each region of the mobile ad hoc network (MANET).

[0077] Considering the potential problem of the regional controller as a single point of failure, it may fail due to faults, disconnections, or moving out of the communication range. When the regional controller in a region fails, the network needs to quickly detect this situation through a distributed mechanism and initiate the regional controller election process to quickly select a new regional controller to ensure the continuous operation of the network. The dynamic selection of the regional controller can be achieved through the Raft distributed consensus algorithm. In the Raft consensus protocol, to initiate a leader election, a candidate will increase its term value and claim to be the leader of this new term. However, during the election process, since any node has the right to initiate an election and interrupt the current term, this provides an opportunity for malicious nodes. By forging a higher term value, malicious nodes can easily interfere with the leader node election process, mislead other nodes in the Raft consensus cluster, and thus successfully usurp the leadership position.

[0078] Therefore, the present invention proposes an improved Raft algorithm election mechanism, adding a secure term range. By all nodes determining an interval for term growth, this interval is based on the expected term change range of the system under normal operating conditions. This interval is dynamically adjustable to adapt to network latency, changes in node response time, and fluctuations in other operating conditions. When a node receives an election request or heartbeat information, it first verifies whether the term value of the node is within the secure term range, thus effectively screening out illegal requests for forged term attacks.

[0079] In the improved Raft algorithm, the steps for implementing term management include verification of the term value, detection and response to abnormal term growth, and synchronization and management of historical terms. When the term value received by a node exceeds the secure term range, the request will be directly rejected to prevent malicious nodes from illegally usurping power by forging term values. The detection of abnormal term growth depends on the combination of the historical term table and the global view. Each node needs to locally maintain a historical term table and store the following information: the term value of each election, the timestamp of the election, the status of whether the election was successful, etc. Nodes will regularly synchronize the historical term table to the current leader (regional controller). The leader maintains a global historical view by summarizing the historical records of all nodes. This global view is not only used to dynamically calculate the secure threshold for term growth but also can quickly synchronize the global term growth rules to newly joined nodes to ensure that all nodes in the cluster are consistent with the secure term growth range, thus maintaining the consistency and stability of the cluster.

[0080] As Figure 4 shown, it is the node transformation process of the improved Raft algorithm.

[0081] In the improved Raft algorithm election mechanism, according to the behavior and status of nodes in the network, they are divided into initial nodes, normal nodes, and malicious nodes. The definitions and processing rules for each node type are as follows:

[0082] An initial node refers to a node newly added to the cluster. When it joins, the system initializes it, such as the term value and index value of the node. The initial identity state of all initial nodes is Follower, and a neutral behavior score is assigned by default. The initial node synchronizes the global historical view, term range rules, and the current cluster state from the leader, including the log and the term value of the current leader. During the initialization phase, the permissions of the initial node are restricted, and it cannot become a candidate or a leader, and is only allowed to participate in voting. After the observation period, if the initial node has no abnormal behavior during this period and its behavior score reaches the safety threshold, it is upgraded to a normal node.

[0083] A normal node refers to a node that can join the cluster normally and participate in processes such as election, voting, and log replication according to the improved Raft algorithm rules. Normal nodes are eligible to become candidates or leaders.

[0084] A malicious node refers to a node that tries to illegally obtain the leader identity by forging the term value or other improper behaviors after joining the cluster. The identification of malicious nodes depends on the term value verification mechanism and the behavior scoring system. When the behavior score of a node is lower than the preset score threshold, or its term proposal exceeds the safe term range, the leader marks it as a malicious node and broadcasts the list of malicious nodes across the network. The node marked as malicious will be isolated and cannot participate in elections and voting, and is only allowed to act as an ordinary forwarding node. After isolation, the malicious node enters the observation period (penalty period). If there is no record of abnormal behavior during the observation period, its behavior score will gradually recover and may eventually regain the eligibility to participate in elections.

[0085] After a node joins the cluster, a random election timeout is set to trigger the election process. During this process, the identities of the nodes in the cluster may change. For example, some nodes may change from followers to candidates.

[0086] When a node enters the election phase, it first calculates the safe term range to ensure that the proposed term value satisfies the constraint conditions such as formula (1):

[0087] 1 ≤ T request ≤ T max ; (1)

[0088] Where, T request represents the proposed term value; [1, T max represents the safe term range.

[0089] In some embodiments, the safe term range is calculated comprehensively according to the network state, the node's own capabilities, and historical election behaviors, as shown in formula (2):

[0090] T max = T base×(1+F stability +F resources +F trust ); (2)

[0091] Among them, T max Indicates the upper limit of the safe term range; T base Indicates the basic term value; F stability Represents the network stability factor, which is used to dynamically evaluate the overall stability of the current network. The higher the value, the more unstable the network is. The election requires a larger term range to tolerate dynamic changes in topology. resources F represents the node resource factor, which reflects the resource sufficiency of the current node (such as battery, power, computing power). The more sufficient the resources are, the more capable the node is of assuming the role of leader for a long time, and its maximum term range should also increase accordingly. trust Represents the node trust factor, which is used to dynamically evaluate the credibility of the node, and is set based on the node's behavior score and historical election records. The higher the credibility, the more reliable the term value proposed by the node, allowing a larger term range.

[0092] Specifically, the network stability factor is defined as formula (3):

[0093]

[0094] Among them, N neighbors is the number of neighbors of the current node, indicating the connection status of the node. When the number of neighbors is small, the network connectivity is poor and the stability decreases. Its reciprocal is used to dynamically reflect connectivity; VAR (D net ) represents the variance of network delay, reflecting the delay fluctuation. The more drastic the fluctuation, the larger the value. k1 and k2 are weight parameters.

[0095] In MANET, the number of neighbors a node is connected to directly determines the connectivity of the network. When the number of neighbors is small, the network may be in a partitioned state, and a larger term value is required to tolerate the impact of this partition. Networks with large delay fluctuations will interfere with the consistency election. Increasing the weight of the fluctuation impact can improve adaptability to dynamic environments.

[0096] The node resource factor is defined as formula (4):

[0097]

[0098] Among them, R uesd Indicates the resources used by the node, such as remaining power, CPU load, etc.; R total Indicates the total resource capacity of the node, such as total power, computing power, etc.; Indicates resource utilization; k3 is the weight parameter.

[0099] When node resources are sufficient, a longer tenure range can be allocated to improve the stability of elections and reduce the election frequency. When resources are exhausted (such as when the battery level is close to 0), the node should try to avoid long tenures and avoid becoming the leader frequently to relieve resource pressure.

[0100] The node trust factor is defined as in formula (5):

[0101]

[0102] Among them, B node represents the node behavior score, reflecting its past reliability, as defined in formula (6):

[0103]

[0104] Among them, N valid represents the total number of all compliant behaviors of the node, such as log synchronization, data forwarding, heartbeat response, etc.; N total represents the total number of all operations of the node, including valid and invalid behaviors; N malicious represents the number of times the node is judged to have malicious behavior; k4 and k5 are weight parameters.

[0105] If the behavior score of the node is low or the historical compliant behaviors are few, the trust factor value is low, and the proposals of the node will be strictly restricted within a smaller tenure range. By increasing the tenure range of nodes with high credibility, the overall efficiency and security of the election process can be improved.

[0106] Based on formulas (3), (4), and (5), the upper limit of the safe tenure range of formula (2) is adjusted to formula (7):

[0107]

[0108] Among them, the value ranges of k1, k2, k3, k4, and k5 are [0, 1], and k1 + k2 + k3 + k4 + k5 = 1. Specifically, k1 represents the weight of the neighbor quantity factor. In a stable network (such as the intra-cluster environment of MANET), the connectivity is usually good, so a lower value (such as 0.1) can be taken. In a dynamic network (such as a scenario where nodes move frequently or network partitioning occurs frequently), the connectivity is poor, and a higher value (such as 0.4 or 0.5) should be taken. k2 represents the weight of the network delay fluctuation factor. In a network with low delay fluctuation (such as a static MANET environment), the influence of the delay fluctuation factor is small, and a lower value (such as 0.2 or 0.3) should be taken. In a scenario with high delay fluctuation (such as a congested network or a highly dynamic environment), a higher value (such as 0.5 or 0.6) should be taken to significantly improve the system's tolerance to network jitter. k3 represents the weight of the node resource factor. In a scenario with limited resources (such as a network with insufficient battery power), the weight of the resource factor needs to be increased to avoid low-resource nodes frequently participating in leader elections, and a higher value (such as 0.6 or 0.7) is recommended. In an environment with sufficient resources, the weight of the resource factor can be appropriately reduced (such as 0.3 or 0.4). k4 represents the weight of the behavior scoring factor. In a scenario that requires strict defense against malicious nodes, the weight of the behavior scoring factor can be taken as a higher value (such as 0.4 or 0.5). In an environment where node behaviors are generally reliable, the weight of the behavior scoring factor can be appropriately reduced (such as 0.2 or 0.3). k5 represents the weight of the voting record factor. In a scenario with a high proportion of malicious nodes, the weight of the voting record factor should be appropriately increased (such as 0.3 or 0.4). In an environment where the election process is stable and node performances are balanced, the weight of the voting record factor can be taken as a lower value (such as 0.1 or 0.2).

[0109] In the edge tactical MANET network, the area controller regularly calculates the upper limit of the security task scope based on the current real-time network conditions and the self-historical tenure tables uploaded by each follower node, and broadcasts and synchronizes the results to all follower nodes through heartbeat signals. After the area controller fails, the nodes in the area network do not need to recalculate and directly use the pre-synchronized security task scope, enabling the nodes in the network to quickly enter the election stage without waiting for recalculation and ensuring the consistency of the security task scopes maintained by each node, that is, all nodes use the same security task scope for election, avoiding disagreements caused by different network states.

[0110] To sum up, as Figure 5As shown, when a node initiates an election request or a heartbeat message, it first checks the legality of the term value. If the term value of the node is within the safe term range and the behavior score of the node is greater than the preset score threshold, the node is determined to be a normal node. If it can obtain votes from more than half of the other nodes, it becomes the regional controller. If the term value of the node is not within the safe term range, or the behavior score of the node is not greater than the preset score threshold, the node is determined to be a malicious node, and the node is prohibited from participating in the election, and the term value is reset. The election method of the global controller is the same.

[0111] In step S102, in the intra-domain topology management module, in the data plane, when the regional controller is valid, within each region, the regional controller collects the network information uploaded by each SDN node in the domain, constructs and maintains the local topology view. When the regional controller fails, the SDN node generates a routing path for the traffic based on the distributed on-demand routing discovery mechanism.

[0112] The topology discovery of the present invention combines distributed and centralized methods. When the regional controller is valid, centralized topology management is adopted. When it fails, the distributed mechanism can ensure that the nodes in the network can still maintain the local topology information. After the regional controller recovers, it quickly reconstructs the global topology by integrating the local information of each node.

[0113] In some embodiments, the communication between the regional controller and the SDN node adopts in-band control, which does not require allocating additional spectrum or independent wireless channels for control messages. In-band control can reuse the existing data network and wireless links, does not require additional infrastructure support, and the control messages and data messages share the same communication path, with lower cost and stronger adaptability.

[0114] In some embodiments, when the regional controller is valid, each SDN node regularly monitors the unidirectional link status information between itself and its neighbor nodes, such as link available capacity, end-to-end delay, and transmission cost, etc., and encapsulates the unidirectional link status information into a Hello message for broadcasting; at the same time, it receives the Hello message sent by the neighbor node to itself to obtain the bidirectional link status information. Each SDN node encapsulates the bidirectional link status information into a TC message and unicasts it to the regional controller regularly to construct the local topology view. Among them, the frame formats of the Hello message and the TC message are shown in Table 1 and Table 2 respectively.

[0115] Table 1 Hello message frame format

[0116]

[0117] Table 2 TC message frame format

[0118]

[0119] In some embodiments, the TC message is encapsulated into an OpenFlow Experimenter message.

[0120] In the controller cooperation module, between regions, the network information between regional controllers is synchronized through link information, and their local topology views are updated, and then a global topology is constructed. Ensure the connectivity and dynamic adaptability of the entire tactical MANET network. Among them, the format of the exchanged link information packet is the same as the previously defined TC message.

[0121] In the controller cooperation module, based on the above improved Raft algorithm, a global controller is elected from the regional controllers, and the information interaction between the global controller and the regional controllers is managed to ensure the consistency of the global topology information and the effectiveness of global control.

[0122] As Figure 6 shown, it is the information synchronization flowchart of the regional controller and the global controller.

[0123] (a) The figure shows that the global controller initiates information synchronization. The global controller broadcasts the update information, that is, the new link information, to each regional controller. After the regional controller confirms the synchronization, it sends a confirmation message to the global controller, and the synchronization is completed. Thus, it ensures the state consistency of the entire network, which is particularly important in the case of regional controller failure or network partition. At the same time, during the synchronization process, each regional controller only needs to exchange incremental link information, thereby reducing communication overhead and adapting to the limited bandwidth and high latency characteristics in tactical MANET.

[0124] (b) The figure shows that the regional controller initiates information synchronization. When a regional controller senses a change in the link state (such as an increase in link delay, node disconnection, or a decrease in available capacity, etc.), it triggers the information synchronization process. The regional controller sends the update information to the global controller. The global controller updates according to the update information and at the same time broadcasts the update information to each regional controller. After each regional controller confirms the synchronization, it sends a confirmation message to the global controller; after the global controller receives the confirmation messages from all regional controllers, it sends a confirmation message to the regional controller that initiated the synchronization request, and the synchronization is completed.

[0125] In summary, the role of the controller cooperation module is not only to ensure the accurate synchronization of link information, but also to improve the distributed consistency and cooperation efficiency through the improved Raft algorithm. It can enhance the robustness and disaster tolerance of the tactical MANET network and ensure stable operation in scenarios with dynamic changes and high reliability requirements. At the same time, the cooperation module provides basic support for global QoS optimization and resource utilization through efficient information synchronization, and is a key component for realizing multi-controller cooperation.

[0126] In the global topology management module, based on the network topology information maintained by the intra-domain topology management module, such as node connection status, link delay, available capacity, transmission cost, etc., and the network topology information of other domains received through the controller cooperation module, a global network topology view is constructed. By integrating the intra-domain topology information and cross-domain topology data, the global topology management module can accurately reflect the network state of the entire tactical MANET, providing an efficient data basis for the routing calculation of centralized control.

[0127] In step S103, according to the global network topology view, a hierarchical game mechanism is adopted to classify the traffic; according to the category of the traffic, the primary and secondary optimization objectives of quality of service are set for hierarchical optimization to obtain the optimal path and the backup path.

[0128] In some embodiments, based on the global network topology view, a routing path is selected based on the link available capacity, end-to-end delay, and transmission cost. The traffic is classified into delay-sensitive traffic, high-capacity traffic, and low-cost traffic. The primary and secondary optimization objectives are determined according to the classification of the traffic. Exemplarily, for real-time traffic (delay-sensitive traffic), its primary optimization objective is delay, and the secondary optimization objectives are available capacity and transmission cost. Thus, the QoS objectives are optimized in two layers. The primary objective is optimized in the first layer, and the secondary objective is optimized in the second layer, and finally the optimal path and the backup path are obtained.

[0129] The following further elaborates on the three optimization objectives of link available capacity, end-to-end delay, and transmission cost.

[0130] Link available capacity AC ij It is defined as the product of the packet delivery ratio (PDR) within a period of time, the proportion of channel idle time 1 - u(t), and the maximum capacity C(r) of the link, as shown in formula (8):

[0131]

[0132] Among them, the maximum link capacity C(r) is determined by the physical layer rate r and the link transmission delay TxDelay(r), as shown in formula (9):

[0133]

[0134] Based on the available capacities AC ij and AC ji , the available capacity p ij of a link is defined as the reciprocal of the minimum value of the two, as shown in formula (10):

[0135]

[0136] Furthermore, the available capacity P of the entire path path As shown in formula (11):

[0137] P path = ∑ (i,j∈path) p ij ; (11)

[0138] The link delay L ij Is defined as formula (12):

[0139] L ij = T pro + T que + T trans ; (12)

[0140] Wherein, T pro Represents the propagation delay, which depends on the link length and the signal propagation speed; T que Represents the queuing delay, which depends on the current load of the link; T trans Represents the transmission delay, which is related to the link rate and the packet size

[0141] Furthermore, the end-to-end delay L of the entire path end-to-end Is defined as formula (13):

[0142] L end-to-end = ∑ (i,j∈path) L ij ; (13)

[0143] The link transmission cost TC ij Is the sum of the path transmission energy consumption TE ij And the link loss LL ij As shown in formula (14):

[0144] TC ij = TE ij + LL ij ; (14)

[0145] Furthermore, the transmission cost TC of the entire path path As shown in formula (15):

[0146] TC path = ∑ (i,j∈path) TC ij ; (15)

[0147] Taking the delay-sensitive traffic mentioned above as an example, a hierarchical design is carried out for the objective function and the utility function. The objective function of the first layer is the main optimization objective, and the main optimization objective of the delay-sensitive traffic is to optimize the delay L end-to-end , Therefore, its main optimization objective function is as shown in formula (16):

[0148] minL end-to-end = ∑ (i,j∈path) L ij ; (16)

[0149] Define the utility function U1 of the first layer. The goal is to maximize the utility, that is, to minimize the delay for delay-sensitive traffic, as shown in formula (17):

[0150]

[0151] Among them, the smaller U1 is, the smaller the delay is, and the larger the utility value is. Use normalized max(L end-to-end ) to ensure the comparability of delays of different paths.

[0152] Set threshold conditions for each optimization goal to screen the paths obtained by the first-layer optimization.

[0153] For the end-to-end delay main optimization goal, the delay threshold condition is as shown in formula (18):

[0154] L end-to-end ≤ θ L × L best ; (18)

[0155] Among them, L best represents the delay size corresponding to the path with the minimum delay; θ L > 1 is an adjustment coefficient used to control the looseness of the set and can take a value of 1.5.

[0156] For the available capacity main optimization goal, the available capacity threshold condition is as shown in formula (19):

[0157] P path ≥ θ P × max(P path ); (19)

[0158] Among them, max(P path ) represents the maximum available capacity of all paths in the network; θ P ∈(0, 1] is an adjustment coefficient representing the screening ratio of available capacity and can take a value of 0.7.

[0159] For the transmission cost main optimization goal, the transmission cost threshold condition is as shown in formula (20):

[0160] TC path ≤ θ TL × min(TC path ); (20)

[0161] Among them, min(TC path) represents the minimum transmission cost of all paths in the network; θ TL > 1 is an adjustment coefficient, representing the screening range of the transmission cost. When the value is close to 1, only the paths close to the minimum transmission cost are selected, which is suitable for scenarios with strict requirements for low transmission cost; when the value is relatively large (such as 1.5), more paths are allowed to enter the set, which is suitable for scenarios that require a large optimization space.

[0162] In some embodiments, in the first-layer primary optimization objective optimization, the Yen's K-Shortest Paths algorithm is used to generate K optimal paths from the source to the destination node, and according to the preset threshold conditions of the primary optimization objective, the primary optimization path set is screened from the K paths.

[0163] Still taking the delay-sensitive traffic as an example, after the first-layer (end-to-end delay) optimization, the second-layer optimization (i.e., secondary optimization) is performed based on the obtained primary optimization path set to further evaluate the available capacity and transmission cost. The secondary optimization objective needs to find a balance between the available capacity and the transmission cost.

[0164] The available capacity optimization objective function is shown in formula (21):

[0165] maxP path =∑ (i,j∈path) p ij ; (21)

[0166] That is, to maximize the path available capacity, define its corresponding utility function U2, and the goal is to maximize the utility, as shown in formula (22):

[0167]

[0168] The transmission cost optimization objective function is shown in formula (23):

[0169] minTC path =∑ (i,j∈path) TC ij ; (23)

[0170] That is, to minimize the total transmission cost of the path, define its corresponding utility function U3, and the goal is to maximize the utility (i.e., minimize the cost), as shown in formula (24):

[0171]

[0172] In the second-layer optimization, by introducing a weight coefficient, the optimization of the available capacity and the transmission cost is comprehensively considered, as shown in formula (25):

[0173] U sond =α×U2 + β×U3; (25)

[0174] Among them, α and β are weight coefficients, and the priorities are adjusted according to different scenarios. For example: in a high-load network, the weight of α can be increased to give priority to available capacity; in an energy consumption-sensitive scenario, the value of β can be increased to give priority to reducing transmission costs. Sort the path set according to the utility value, and select the path with the highest utility value as the optimal path, and the two paths with the second-highest utility value as the backup paths.

[0175] Based on the above description, as Figure 7 shown, when the area controller is effective, the route generation method includes the following steps:

[0176] Based on step S102, the area controller collects network information uploaded by each SDN node in the domain, such as link status information, etc., constructs and maintains local topology information, and exchanges topology information with other area controllers, and finally constructs a global network topology view to provide a basis for path optimization.

[0177] Classify the traffic according to application requirements or traffic characteristics, such as: delay-sensitive traffic, high-capacity (bandwidth) traffic, low-cost traffic.

[0178] Select the main optimization objective according to the traffic category, use Yen's K-Shortest Paths algorithm and the threshold condition of the main objective to obtain the main optimization path set. Based on the obtained main optimization path set, perform secondary objective optimization. For each path in the path set, calculate the comprehensive utility function of the secondary objective, sort the utility values, and select the path with the maximum comprehensive utility as the final path, and the sub-optimal one as the backup path. Exemplarily, two backup paths can be selected.

[0179] The area controller issues the calculated optimal path and backup paths to the nodes along the path through OpenFlow Flow-Mod messages to update the flow table rules of the virtual switch (OVS).

[0180] In some embodiments, when the area controller fails, that is, during the area controller election period, the SDN nodes in the local area will temporarily take over the traffic management task. The local SDN nodes use the improved distributed on-demand routing discovery mechanism AODV to establish a routing path for the traffic.

[0181] The distributed on-demand routing discovery mechanism uses the shortest path algorithm (Dijkstra algorithm) to calculate the optimal path for the traffic. Based on the link available capacity, end-to-end delay and transmission cost defined above, the goal of path selection is to minimize the total weight of the path, and the objective function is shown in formula (26):

[0182] minW path =∑ (i,j∈path) α×p ij +β×TC ij+γ×L ij ; (26)

[0183] Wherein, W path represents the total path weight; α, β, and γ all represent weight coefficients, and the value ranges of α, β, and γ are [0, 1], and α + β + γ = 1; p ij represents the available capacity of the link between nodes i and j; TC ij represents the end-to-end delay; L ij represents the transmission cost. By adjusting the weight coefficients, specific types of QoS can be optimized, such as real-time traffic, traffic with high reliability requirements, etc., to support multiple application scenarios.

[0184] The routing generation method based on the distributed on-demand routing discovery mechanism includes the following steps:

[0185] When the source node needs to communicate with the destination node and the local controller does not cache the path, trigger the routing request mechanism to generate a routing request (RREQ_A) message. The source node broadcasts the routing request message to its direct neighbor nodes.

[0186] When a neighbor node (intermediate node) receives the routing request message, first judge whether it has processed this packet according to the source node address and sequence number in the routing request message. If so, directly discard it to avoid loops; otherwise, proceed to the next step.

[0187] Each intermediate node adds its own information (link QoS parameters) to the routing request message for path performance calculation. Add the sequence number and source node address carried in the routing request message to the local identification list, and perform addition or update operations on the local routing table for the path from the source node to the current node, and at the same time broadcast the routing request message.

[0188] When the destination node receives the routing request message, select a path with the smallest objective function value, generate a routing reply (RREP) message and return it to the source node.

[0189] After the source node receives the routing reply message, confirm that the path is successfully established and cache the path in the routing table of the local controller. The source node sends service data.

[0190] As shown in Table 3, it is the frame format of the routing request (RREQ_A) message.

[0191] Table 3 Routing Request Message Frame Format

[0192]

[0193] In step S104, the traffic management and flow table distribution module realizes the refined management and optimization of the data flow in the network through the efficient distribution of traffic classification, scheduling, and control rules.

[0194] First, based on the preset policies and real-time network status, this module classifies the traffic entering the network. For example, the traffic is classified into real-time traffic, reliable traffic, low-cost traffic, etc., and corresponding priorities are assigned to different types of traffic. Subsequently, based on the network status information provided by the global topology management module in step S103, the optimal path and the alternate path selected by the routing calculation and optimization module, specific flow table rules for each traffic are generated. The flow table rules include packet matching conditions (such as IP address, port number) and forwarding actions (such as the next-hop node), and are sent to the switches (OVS) along the path through OpenFlow Flow-Mod messages to ensure that the data flow is transmitted along the optimal path.

[0195] In some embodiments, the traffic management and flow table distribution module also undertakes the tasks of dynamic adjustment and real-time update. When the network topology or link status changes (such as link interruption or excessive load), this module triggers a recalculation of the path and redistributes the updated flow table rules to ensure the reliability of the data flow and the QoS requirements. In addition, for traffic that needs to be transmitted across domains, this module exchanges flow table information with other area controllers through the controller cooperation module to ensure seamless scheduling of cross-domain traffic.

[0196] Corresponding to the above method, the present invention also provides an electronic device, which includes a computer device. The computer device includes a processor and a memory. Computer instructions are stored in the memory, and the processor is configured to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the electronic device implements the steps of the method described above.

[0197] The embodiments of the present invention also provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the foregoing edge computing server deployment method are implemented. The computer-readable storage medium may be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium known in the art.

[0198] Those of ordinary skill in the art should understand that the various exemplary components, systems, and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Specifically, whether to implement in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted via a data signal carried in a carrier wave over a transmission medium or a communication link.

[0199] It should be clear that the present invention is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present invention is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present invention.

[0200] In the present invention, the features described and / or illustrated for one embodiment can be used in the same or a similar manner in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0201] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and variations can be made to the embodiments of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for generating edge computing network routing based on SDN and Raft algorithm election mechanism, characterized in that: The edge computing network includes a mobile self-organizing network and an SDN architecture, and the method includes the following steps: Based on the Raft distributed consensus algorithm, a regional controller is elected for each area of ​​the mobile ad hoc network, and a global controller is elected from each regional controller; a safe term range is set in the Raft distributed consensus algorithm to prevent malicious nodes from conducting elections; When the regional controller is valid, in each region, the regional controller collects network information uploaded by each SDN node in the region, builds and maintains a local topology view; between regions, the network information between the regional controllers is synchronized through link information, the respective local topology information is updated, and a global network topology view is generated; the global controller ensures the consistency of network information between the regional controllers; According to the global network topology view, a hierarchical game mechanism is adopted to classify traffic, and primary and secondary optimization targets of service quality are set according to the types of traffic to perform hierarchical optimization, so as to obtain the optimal path and backup path; A flow table rule is generated according to the global network topology view, the optimal path and the backup path, and is sent to switches along the path.

2. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 1 is characterized in that: The Raft distributed consensus algorithm is provided with a safety term range, including: The safety term range is set based on the network status, node capabilities and historical election records, and satisfies the following formula: T max =T base ×(1+F stability +F resources +F trust ); Among them, T max Indicates the upper limit of the safety term range; T base Indicates the basic term value; F stability Represents the network stability factor; F resources Indicates the node resource factor; F trust Represents the node trust factor.

3. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 1 is characterized in that: The method for electing the regional controller or the global controller includes: A node initiates an election request. If the term value of the node is within the safe term range and the behavior score of the node is greater than a preset score threshold, the node is judged to be a normal node. If it obtains votes from more than half of the other nodes, it becomes the regional controller or the global controller. If the term value of the node is not within the safe term range, or the behavior score of the node is not greater than a preset score threshold, the node is judged to be a malicious node, the node parameter election is prohibited, and the term value is reset.

4. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 1 is characterized in that: In each area, the regional controller collects network information uploaded by each SDN node in the domain, builds and maintains a local topology view, including: Each SDN node regularly monitors the unidirectional link status information between itself and its neighboring nodes, and encapsulates the unidirectional link status information into a Hello message for broadcasting, and at the same time receives the Hello message sent by the neighboring node to itself to obtain the bidirectional link status information; Each SDN node encapsulates the bidirectional link status information into a TC message, and periodically unicasts it to the regional controller to construct the local topology view.

5. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 1 is characterized in that: The synchronization method between the regional controller and the global controller comprises: When the global controller initiates information synchronization, the global controller broadcasts the update information to each regional controller, and after the regional controller confirms the synchronization, it sends confirmation information to the global controller; When the regional controller initiates information synchronization, the regional controller sends the update information to the global controller. The global controller is updated according to the update information and broadcasts the update information to each regional controller. After each regional controller confirms the synchronization, it sends confirmation information to the global controller. After the global controller receives the confirmation information from all regional controllers, it sends the confirmation information to the regional controller that initiated the synchronization request.

6. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 1 is characterized in that: According to the global network topology view, a hierarchical game mechanism is adopted to classify traffic, and primary and secondary optimization targets of service quality are set according to the types of traffic to perform hierarchical optimization, including: According to the global network topology view, a routing path is selected based on link available capacity, end-to-end delay and transmission cost; Classify traffic into delay-sensitive traffic, high-capacity traffic, and low-cost traffic; For the delay-sensitive traffic, the main optimization goal is to optimize the end-to-end delay, and the secondary optimization goal is to optimize the link available capacity and transmission cost; for the high-capacity traffic, the main optimization goal is to optimize the link available capacity, and the secondary optimization goal is to optimize the end-to-end delay and transmission cost; for the low-cost traffic, the main optimization goal is to optimize the transmission cost, and the secondary optimization goal is to optimize the link available capacity and end-to-end delay.

7. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 6 is characterized in that: The primary and secondary optimization objectives are optimized hierarchically to obtain the optimal path and backup path, including: Determine the main optimization target according to the traffic category, and build a main utility function based on the main optimization target with the goal of maximizing utility; use Yen's K-Shortest Paths algorithm to generate K optimal paths from the source to the destination node, and select the main optimization path set from the K paths according to the threshold condition of the preset main optimization target; A sub-utility function of the sub-optimization target is constructed, and the paths of the main optimization path set are sorted according to the utility value of the sub-utility function, and the path with the highest utility value is used as the optimal path, and the path with the second highest utility value is used as the backup path.

8. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 1 is characterized in that: When the regional controller fails, the SDN node in the local area generates a routing path for traffic based on a distributed on-demand routing discovery mechanism, including: When the source node needs to communicate with the destination node and the path is not cached locally, the route request mechanism is triggered to generate a route request message; the source node broadcasts the route request message to its neighbor nodes; The neighbor node adds its own link state information to the route request message; updates the local routing table, records the path information from the source node to the current node, and broadcasts the updated route request message; When the target node receives the updated route request message, the target node generates an optimal path according to the link state information in the message; the target node generates a route reply message and returns to the source node along the optimal path; The source node sends service data along the optimal path.

9. The edge computing network routing generation method based on SDN and Raft algorithm election mechanism according to claim 8 is characterized in that: The distributed on-demand routing discovery mechanism uses the shortest path algorithm to calculate the optimal path, including: Based on the link available capacity, end-to-end delay and transmission cost, the objective function is constructed with the goal of minimizing the total path weight. The calculation formula is: minW path N∑ (i,j∈path) α×p ij +β×TC ij +γ×L ij 100. Among them, W path represents the total path weight; α, β, γ all represent weight coefficients, and α+β+γ=1; p ij represents the available capacity of the link between nodes i and j; TC ij represents the end-to-end delay; L ij Represents the transmission cost.

10. An edge computing network routing generation system based on SDN and Raft algorithm election mechanism, characterized in that: When the system is executed, the steps of the method according to any one of claims 1 to 9 are implemented, and the system comprises: Data plane; in the data plane, each SDN node regularly monitors the unidirectional link status information between itself and its neighboring nodes, and encapsulates the unidirectional link status information into a Hello message for broadcasting, and simultaneously receives the Hello message sent by the neighboring node to itself to obtain the bidirectional link status information; encapsulates the bidirectional link status information into a TC message, and regularly unicasts it to the regional controller; when the regional controller fails, the SDN node generates a routing path for the traffic based on a distributed on-demand routing discovery mechanism; Control plane; the control plane includes a regional election module, an intra-domain topology management module, a controller collaboration module, a global topology management module, a routing calculation and optimization module, and a traffic management and flow table delivery module; The regional election module is used to elect a regional controller for each region of the mobile ad hoc network based on an improved Raft distributed consensus algorithm; The intra-domain topology management module is used for collecting network information uploaded by each SDN node in the domain in each area, and building and maintaining a local topology view; The controller collaboration module is used to synchronize network information between regional controllers through link information between regions and update their respective local topology views; based on the improved Raft distributed consensus algorithm, a global controller is elected from the regional controllers; The global topology management module is used to interact with the intra-domain topology management module and the controller collaboration module, and construct a global network topology view based on the maintained intra-domain network topology information and the network topology information of other domains received by the global controller; The routing calculation and optimization module is used to classify traffic according to the global network topology view and adopt a hierarchical game mechanism, and to set the primary and secondary optimization targets of the quality of service according to the types of traffic for hierarchical optimization to obtain the optimal path and the backup path; The traffic management and flow table sending module is used to generate flow table rules according to the global network topology view, the optimal path and the backup path, and send them to switches along the path.

Images