Wireless communication network self-optimization anti-interference architecture system and construction method thereof

Abstract

The invention discloses a wireless communication network self-optimization anti-interference architecture system and a construction method thereof, specifically, a network self-learning module is internally provided with a confrontation attack model and a confrontation defense model, and the network self-learning module is nested with a network management and control module and a resource cooperation module; the adversarial attack model generates a large number of adversarial samples through the network, finds weak links of the network and initiates network adversarial attacks; the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model according to continuously initiated adversarial attacks, calls the network management and control module and the resource cooperation module, continuously adjusts algorithm learning parameters, updates a learning strategy, optimizes a network structure and resource configuration, and refines an anti-interference decision boundary; and the attack resisting model and the defense resisting model are alternately carried out to realize game playing and confrontation of the network self-learning module. According to the method, a new theory and a new implementation architecture are provided for self-optimization of the wireless communication network, and various emerging unknown interference threats can be effectively dealt with.

Description

Technical Field

[0001] The present invention belongs to the technical field of communication anti-interference, and in particular to a wireless communication network self-optimization anti-interference architecture system and a construction method thereof. Background Art

[0002] Based on the structural characteristics of wireless communication networks such as strong heterogeneity, high dynamics, and wide coverage, traditional anti-interference ideas need to overcome the uncertainty of the wireless environment and the unknown strategies of the interference party. With the surge in frequency-using equipment and the rapid improvement of the intelligence level of interference, the electromagnetic spectrum space is becoming increasingly complex and changeable, and new interference patterns are changing rapidly and are difficult to exhaust. In this context, "intelligent anti-interference" has become the mainstream direction of communication anti-interference research in the era of artificial intelligence. Researchers widely use a variety of AI technologies such as deep reinforcement learning, generative adversarial networks, adversarial machine learning, imitation learning, meta-learning, federated learning, and transfer learning to study environmental situation awareness, interference pattern recognition, anti-interference decision analysis, communication power control, dynamic spectrum access, motion trajectory planning, and other aspects of the OODA loop and anti-interference problems under strong adversarial conditions. The concept of intelligent anti-interference covers almost all communication anti-interference research enabled by artificial intelligence technology. However, the traditional passive anti-interference idea based on precise interference perception will face the dilemma of the "performance ceiling" that is difficult to defend against. Therefore, from the perspective of wireless communication networks, there is an urgent need to further study dynamically adaptive, intelligent and self-optimizing anti-interference technologies, and transform the anti-interference thinking of "the enemy changes and I change" into an anti-interference logic of active correction and self-optimization.

[0003] "Self-optimization" and "self-adaptation" are similar in their conceptual connotations, and a large number of self-optimization studies usually refer to research work in the field of self-adaptation. It mainly starts from system performance and environmental status, and optimizes parameters, configuration, control, etc. according to established strategies or learning capabilities. However, many researchers are more concerned with the generation mechanism and strengthening process of learning ability, strengthening strategic capabilities in self-learning and self-game, and optimizing system performance. Empowering adversarial learning and intelligent decision-making capabilities based on AI technology naturally belongs to the conceptual category of "intelligent anti-interference", but because of its characteristics of self-change, game evolution, and elastic adaptability, it has become a distinctive research direction.

[0004] At present, the concept of self-optimization in the field of wireless communications has not been authoritatively recognized, and there has been no research on establishing a self-optimizing and anti-interference architecture for wireless communication networks. Summary of the invention

[0005] The purpose of the present invention is to provide a wireless communication network self-optimization and anti-interference architecture system and a construction method thereof. Starting from the full cycle and multiple links of network construction, configuration, and update, a self-optimization and anti-interference architecture is established around three aspects: network management and control, resource collaboration, and learning model, thereby realizing self-optimization and anti-interference of the wireless communication network.

[0006] The technical solution to achieve the purpose of the present invention is: a method for constructing a self-optimizing anti-interference architecture system for a wireless communication network, comprising the following steps:

[0007] Step 1: Establish a network self-learning module, set an adversarial attack model and an adversarial defense model in the network self-learning module, and the network self-learning module embeds a network control module and a resource collaboration module;

[0008] Step 2: Run the network self-learning module, and alternate between the anti-attack model and the anti-defense model to achieve the game and confrontation of the network self-learning module:

[0009] The adversarial attack model generates a large number of adversarial samples through the network itself, finds the weak links of the network, and launches network adversarial attacks to counter the learning evolution of the defense model;

[0010] Based on the adversarial attacks continuously launched, the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model, calls the network control module and resource collaboration module, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary;

[0011] Step 3: In the face of ever-changing interference situations, repeat step 2 to establish a self-optimizing anti-interference architecture for the wireless communication network.

[0012] Furthermore, the network self-learning module in step 1 is specifically as follows:

[0013] The adversarial attack model is used to supplement the interference knowledge space, and the adversarial defense model is used to refine the anti-interference knowledge space;

[0014] Through the self-game process of alternating operation of the adversarial attack model and the adversarial defense model, the network control module is driven to update the data model, and the resource collaboration module is driven to update the strategy model.

[0015] Furthermore, in step 2, the adversarial defense model calls the network control module as follows:

[0016] In response to interference attacks generated by the adversarial attack model with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model, and the networking algorithm component is run to adjust the network topology update strategy to reduce the impact of interference and maintain network performance; the network topology update strategy components include subnet segmentation, link reconstruction, and network reconstruction, and the networking algorithm components include distributed dynamic networking and traction collaborative networking.

[0017] Furthermore, in step 2, the adversarial defense model calls the resource collaboration module as follows:

[0018] In response to the interference attacks generated by the adversarial attack model with the purpose of blocking key nodes and key links, the resource collaboration module is run in the adversarial defense model, and the resource scheduling algorithm component is run to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance; the network resource configuration strategy components include multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, and the resource scheduling algorithm components include distributed resource configuration and multi-user resource collaboration.

[0019] Furthermore, the network control module adopts a networking algorithm based on alliance formation game:

[0020] Each node first dynamically adjusts the set of neighboring nodes it interacts with by controlling the transmission power. Then, by setting the coalition game utility, it characterizes the utility mapping relationship from individual to local and from local to global. It forms an alliance between network nodes and divides the subnet according to the node alliance.

[0021] Execute the link reconstruction strategy, and each node dynamically adjusts the link of the local subnet structure to achieve distributed dynamic networking;

[0022] When interference destroys the local network structure, the network reconstruction strategy is executed to accelerate the traction collaborative networking process and collaborative efficiency through the traction effect of the cluster head node on other nodes.

[0023] Furthermore, the resource collaboration module adopts a resource scheduling algorithm based on asymmetric matching game:

[0024] Execute multi-domain resource matching strategies to sell redundant resources of each node to nodes that are severely disturbed or have insufficient resources;

[0025] Execute the neighboring node assistance strategy, and use the transaction price and mutual assistance score to enable each node to actively collaborate on resources based on limited interactions;

[0026] Execute pre-configuration and reconfiguration strategies, and perform distributed reconfiguration of resources at the network level based on environmental situation analysis and the different requirements of each node.

[0027] A wireless communication network self-optimization anti-interference architecture system, comprising a self-learning module, wherein the network self-learning module is provided with an adversarial attack model and an adversarial defense model, and the network self-learning module is nested with a network control module and a resource collaboration module;

[0028] The adversarial attack model generates a large number of adversarial samples through the network itself, finds the weak links of the network, and launches network adversarial attacks to counter the learning evolution of the defense model;

[0029] Based on the adversarial attacks continuously launched, the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model, calls the network control module and resource collaboration module, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary;

[0030] The adversarial attack model and the adversarial defense model are carried out alternately to realize the game and confrontation of the network self-learning module.

[0031] Furthermore, the adversarial defense model calls the network control module as follows:

[0032] In response to interference attacks generated by the adversarial attack model with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model, and the networking algorithm component is run to adjust the network topology update strategy to reduce the impact of interference and maintain network performance; the network topology update strategy components include subnet segmentation, link reconstruction, and network reconstruction, and the networking algorithm components include distributed dynamic networking and traction collaborative networking.

[0033] Furthermore, the adversarial defense model calls the resource collaboration module as follows:

[0034] In response to the interference attacks generated by the adversarial attack model with the purpose of blocking key nodes and key links, the resource collaboration module is run in the adversarial defense model, and the resource scheduling algorithm component is run to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance; the network resource configuration strategy components include multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, and the resource scheduling algorithm components include distributed resource configuration and multi-user resource collaboration.

[0035] A mobile terminal comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for constructing a self-optimizing anti-interference architecture system for a wireless communication network is implemented.

[0036] Compared with the prior art, the present invention has the following significant advantages:

[0037] (1) The present invention proposes a self-optimizing anti-interference architecture from three aspects: learning model, network control, and resource collaboration, providing a new theory and implementation architecture for self-optimization of wireless communication networks;

[0038] (2) The present invention strives to improve the coordination and flexibility of the network structure, enhance the intelligence and timeliness of resource allocation, and promote the self-game and self-evolution of the learning model by tapping into the inherent structural anti-interference ability of the communication network itself;

[0039] (3) A self-optimizing anti-interference communication network has been realized, which can actively correct weak links, dynamically adapt to complex electromagnetic environments, and effectively respond to various unknown interference threats that emerge in an endless stream. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0041] Figure 1 A diagram of the self-optimization and anti-interference mechanism of a wireless communication network provided in an embodiment of the present invention.

[0042] Figure 2 Schematic diagram of a self-optimizing anti-interference architecture system for wireless communication networks.

[0043] Figure 3 Schematic diagram of key technical architecture. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0045] The present invention provides a method for constructing a self-optimizing anti-interference architecture of a wireless communication network, comprising the following steps:

[0046] Step 1: Establish a network self-learning module, set an adversarial attack model and an adversarial defense model in the network self-learning module, and the network self-learning module embeds a network control module and a resource collaboration module;

[0047] Step 2: Run the network self-learning module, and alternate between the anti-attack model and the anti-defense model to achieve the game and confrontation of the network self-learning module:

[0048] The adversarial attack model generates a large number of adversarial samples through the network itself, finds the weak links of the network, and launches network adversarial attacks to counter the learning evolution of the defense model;

[0049] Based on the adversarial attacks continuously launched, the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model, calls the network control module and resource collaboration module, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary;

[0050] Step 3: In the face of ever-changing interference situations, repeat step 2 to establish a self-optimizing anti-interference architecture for the wireless communication network.

[0051] Through the self-optimizing anti-interference architecture, the network can self-discover weak links, self-update learning parameters, self-adapt to the environment, actively evolve, and generate self-optimizing anti-interference capabilities.

[0052] As a specific example, the network self-learning module in step 1 is as follows:

[0053] The adversarial attack model is used to supplement the interference knowledge space, and the adversarial defense model is used to refine the anti-interference knowledge space;

[0054] Through the self-game process of alternating operation of the adversarial attack model and the adversarial defense model, the network control module is driven to update the data model, and the resource collaboration module is driven to update the strategy model.

[0055] As a specific example, in step 2, the adversarial defense model calls the network control module as follows:

[0056] In response to interference attacks generated by the adversarial attack model with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model, and the networking algorithm component is run to adjust the network topology update strategy to reduce the impact of interference and maintain network performance; the network topology update strategy components include subnet segmentation, link reconstruction, and network reconstruction, and the networking algorithm components include distributed dynamic networking and traction collaborative networking.

[0057] As a specific example, in step 2, the adversarial defense model calls the resource collaboration module as follows:

[0058] In response to the interference attacks generated by the adversarial attack model with the purpose of blocking key nodes and key links, the resource collaboration module is run in the adversarial defense model, and the resource scheduling algorithm component is run to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance; the network resource configuration strategy components include multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, and the resource scheduling algorithm components include distributed resource configuration and multi-user resource collaboration.

[0059] As a specific example, the network control module adopts a networking algorithm based on alliance formation game:

[0060] Each node first dynamically adjusts the set of neighboring nodes it interacts with by controlling the transmission power. Then, by setting the coalition game utility, it characterizes the utility mapping relationship from individual to local and from local to global. It forms an alliance between network nodes and divides the subnet according to the node alliance.

[0061] Execute the link reconstruction strategy. Each node dynamically adjusts the link of the local subnet structure, that is, realizes distributed dynamic networking, improves the performance of the entire network topology, and thus improves the coordination and flexibility of the link structure;

[0062] When interference destroys the local network structure, the network reconstruction strategy is executed. Through the traction effect of the cluster head node on other nodes, the traction collaborative networking process and collaborative efficiency are accelerated to achieve efficient collaboration and rapid reconstruction.

[0063] As a specific example, the resource collaboration module adopts a resource scheduling algorithm based on an asymmetric matching game:

[0064] Execute multi-domain resource matching strategies to sell redundant resources of each node to nodes that are severely disturbed or have insufficient resources;

[0065] Execute the neighboring node assistance strategy, and use the transaction price and mutual assistance score to enable each node to actively collaborate on resources based on limited interactions;

[0066] Execute pre-configuration and reconfiguration strategies, and perform flexible and efficient distributed reconfiguration of resources at the network level based on environmental situation analysis and the different needs of each node;

[0067] The above process realizes distributed resource configuration.

[0068] When faced with a serious resource gap, multiple users can also adopt a multi-user resource collaboration algorithm based on non-orthogonal technology to further improve resource utilization efficiency and enhance the intelligence and timeliness of multi-domain anti-interference resource configuration.

[0069] The present invention also provides a wireless communication network self-optimization anti-interference architecture system, including a self-learning module, wherein the network self-learning module is provided with a counter-attack model and a counter-defense model, and the network self-learning module is nested with a network control module and a resource collaboration module;

[0070] The adversarial attack model generates a large number of adversarial samples through the network itself, finds the weak links of the network, and launches network adversarial attacks to counter the learning evolution of the defense model;

[0071] Based on the adversarial attacks continuously launched, the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model, calls the network control module and resource collaboration module, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary;

[0072] The adversarial attack model and the adversarial defense model are carried out alternately to realize the game and confrontation of the network self-learning module.

[0073] As a specific example, the adversarial defense model calls the network control module as follows:

[0074] In response to interference attacks generated by the adversarial attack model with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model, and the networking algorithm component is run to adjust the network topology update strategy to reduce the impact of interference and maintain network performance; the network topology update strategy components include subnet segmentation, link reconstruction, and network reconstruction, and the networking algorithm components include distributed dynamic networking and traction collaborative networking.

[0075] As a specific example, the adversarial defense model calls the resource collaboration module as follows:

[0076] In response to the interference attacks generated by the adversarial attack model with the purpose of blocking key nodes and key links, the resource collaboration module is run in the adversarial defense model, and the resource scheduling algorithm component is run to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance; the network resource configuration strategy components include multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, and the resource scheduling algorithm components include distributed resource configuration and multi-user resource collaboration.

[0077] The present invention also provides a mobile terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for constructing the self-optimizing anti-interference architecture system of the wireless communication network when executing the program.

[0078] The present invention is further described in detail below with reference to the accompanying drawings and specific implementations.

[0079] Example 1

[0080] This embodiment provides a method for constructing a wireless communication network self-optimization anti-interference architecture system, the steps of which are as follows:

[0081] Step 1: Run the network self-learning module;

[0082] The adversarial attack model generates a large number of adversarial samples through the network, finds the weak links of the network, and launches network adversarial attacks. According to the adversarial attacks launched continuously, the adversarial defense model calls the network control module and resource collaboration module described in steps 2 and 3, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary. The adversarial attack model and the adversarial defense model are carried out alternately to realize the game and confrontation of the network self-learning module.

[0083] Step 2: In the face of the interference attack generated by the adversarial attack model described in step 1 with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model described in step 1, and each algorithm component is run to adjust the network topology update strategy to reduce the interference impact and maintain network performance;

[0084] Step 3. In the face of interference attacks generated by the adversarial attack model described in step 1 with the purpose of blocking key nodes and key links, run the resource collaboration module in the adversarial defense model described in step 1, and run each algorithm component to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance.

[0085] In the present invention, the self-optimization system can achieve internal "self-reflection" without relying entirely on external stimuli. Through self-discovery, self-renewal, and self-adaptation, it can explore environmental laws, rehearse confrontation situations, and plan in advance. It can thus get rid of the rigid logic of "the enemy changes and I change", and achieve "innate immunity" to interference attacks, with the advantages of game evolution and endogenous wisdom.

[0086] Furthermore, in step 1, a large number of adversarial samples can be generated through the adversarial attack model, and the decision boundary of the network learning model can be further refined through self-learning of the adversarial samples.

[0087] Furthermore, step 1 can adjust the learning parameters through the adversarial defense model, update the learning strategy to defend against the adversarial attacks that are constantly launched, and timely optimize the network structure and resource allocation.

[0088] Furthermore, step 2 can adopt a networking solution based on alliance formation game. Each node first dynamically adjusts the set of interactive neighbor nodes by controlling the transmission power. Then, by setting the alliance game utility, the utility mapping relationship from individual to local and local to overall is characterized. Through the distributed collaborative optimization of the local subnet structure, each node can achieve performance improvement of the topology of the entire network, thereby improving the coordination and flexibility of the link structure; at the same time, when interference destroys the local network structure, the cluster head node can accelerate the networking process and coordination efficiency through the traction effect on other nodes, and achieve efficient coordination and rapid reconstruction.

[0089] Furthermore, step 3 can adopt a resource scheduling solution based on asymmetric matching game. The redundant resources of each node are sold to nodes that are seriously disturbed or lack resources. With the help of transaction prices and mutual assistance scores, each node can actively collaborate on resources based on limited interactions, and flexibly and efficiently reconfigure resources in a distributed manner from the network level based on environmental situation analysis and the different needs of each node; when facing serious resource gaps, multi-users can also adopt a multi-user resource collaboration solution based on non-orthogonal technology, thereby further improving resource utilization efficiency and enhancing the intelligence and timeliness of multi-domain anti-interference resource configuration.

[0090] Example 2

[0091] This embodiment provides a wireless communication network self-optimization anti-interference architecture system, which can organize the three modules of learning model, network control, and resource collaboration to build a self-optimization anti-interference architecture, and provide a new theory and implementation architecture for wireless communication network self-optimization. The learning model enables adversarial learning through AI technology, actively guides network evolution to drive network control to adjust networking strategies and resource collaboration to adjust resource configuration, and finally feeds back to the network model to form a loop.

[0092] The wireless communication network self-optimization and anti-interference architecture system can classify each functional component that uses different technologies into three modules: learning model, network control, and resource collaboration. The endogenously driven and self-game network self-learning model is placed at the core of self-optimization and anti-interference. The networking mechanism and resource allocation are used as the structural basis of self-optimization and anti-interference, which are driven by the self-learning model and provide feedback to the self-learning model.

[0093] As a specific implementation method, Figure 1 As shown, the method for constructing a wireless communication network self-optimization anti-interference architecture system in this embodiment includes the following steps:

[0094] Step S1, driving the strategy update of the network self-learning module in an adversarial training learning manner:

[0095] Through the adversarial attack model, a large amount of adversarial data is generated by itself, and the attack network self-discovers weak links and updates the network data model; through adversarial learning of the adversarial defense model, the strategy model is self-updated, and the output execution actions enable the network to self-adapt to the complex electromagnetic environment in the presence of unknown interference.

[0096] In this implementation, during the adversarial attack, the algorithm actively introduces imperceptible adversarial noise to make the mapping of adversarial samples in the learning model produce the maximum deviation from the mapping of normal samples.

[0097] Step S2: Learning strategies drive network management and control, adjust networking strategies, and update network topology:

[0098] Topology control is performed through link reconstruction, subnet segmentation, network reconstruction, etc., mainly involving algorithms such as distributed dynamic networking and traction collaborative networking.

[0099] In this embodiment, the alliance formation game is selected to efficiently model the subnet segmentation and fusion problems in network topology control. By setting different alliance formation criteria and optimization principles, different alliance groups can be flexibly formed, and the subnet structure can be adjusted in time according to the dynamic changes in the environmental situation, thereby updating the network topology in time.

[0100] Step S3, learning strategy drives resource collaboration, schedules multi-node resources, and updates resource configuration:

[0101] Through the collaborative scheduling of multi-domain resources between neighboring nodes, it mainly involves strategy algorithms such as multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, as well as scheduling algorithms such as distributed resource configuration and multi-user resource collaboration.

[0102] In this embodiment, the selection matching game quickly and effectively allocates the required resources to the nodes in the wireless communication network, while maintaining the simplicity and reliability of the decision-making process. The bilateral decision makers in the matching game can be divided into applicants and recipients. Applicants initiate matching applications to recipients in turn according to their respective preference lists. The recipients sort the applicants according to their own preference lists, and accept the more favorable options and reject other options. After the two-way selection of applicants and recipients, a stable bilateral matching result can be finally obtained.

[0103] Step S4, taking the new network state after network control and resource collaboration as the new input of the learning module, continuing to learn and dynamically updating the learning strategy.

[0104] The basic process of network self-evolution is as follows:

[0105] The network self-learning module embeds the network control module and the resource collaboration module. The network self-learning module supplements the interference knowledge space through the network adversarial attack model, refines the anti-interference knowledge space through the adversarial defense model, and drives the network control module to update the data model and the resource collaboration module to update the strategy model through the self-game process of alternating operation of the adversarial model and the adversarial defense model.

[0106] (1) First, the strategy update of the network self-learning module is driven by adversarial training learning.

[0107] (2) Apply the learning strategy in step (1) to drive network control, adjust the networking strategy, and update the network topology.

[0108] (3) Apply the learning strategy in step (1) to drive resource collaboration, schedule multi-node resources, and update resource configuration.

[0109] (4) The new network status after network control and resource collaboration is used as the new input of the learning module to continue learning and dynamically update the learning strategy.

[0110] Figure 1 A diagram of the self-optimization anti-interference mechanism of wireless communication networks is given; Figure 2 Self-optimizing anti-interference architecture for wireless communication networks; Figure 3 It is the key technical architecture.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0112] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the creative concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for constructing a self-optimizing anti-interference architecture system for a wireless communication network, characterized in that: The following steps are involved: Step 1: Establish a network self-learning module, set an adversarial attack model and an adversarial defense model in the network self-learning module, and the network self-learning module embeds a network control module and a resource collaboration module; Step 2: Run the network self-learning module, and alternate between the anti-attack model and the anti-defense model to achieve the game and confrontation of the network self-learning module: The adversarial attack model generates a large number of adversarial samples through the network itself, finds the weak links of the network, and launches network adversarial attacks to counter the learning evolution of the defense model; Based on the adversarial attacks continuously launched, the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model, calls the network control module and resource collaboration module, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary; Step 3: In the face of ever-changing interference situations, repeat step 2 to establish a self-optimizing anti-interference architecture for the wireless communication network.

2. The method for constructing a wireless communication network self-optimizing anti-interference architecture system according to claim 1, characterized in that: The network self-learning module described in step 1 is as follows: The adversarial attack model is used to supplement the interference knowledge space, and the adversarial defense model is used to refine the anti-interference knowledge space; Through the self-game process of alternating operation of the adversarial attack model and the adversarial defense model, the network control module is driven to update the data model, and the resource collaboration module is driven to update the strategy model.

3. The method for constructing a wireless communication network self-optimizing anti-interference architecture system according to claim 1, characterized in that: In step 2, the adversarial defense model calls the network control module as follows: In response to interference attacks generated by the adversarial attack model with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model, and the networking algorithm component is run to adjust the network topology update strategy to reduce the impact of interference and maintain network performance; the network topology update strategy components include subnet segmentation, link reconstruction, and network reconstruction, and the networking algorithm components include distributed dynamic networking and traction collaborative networking.

4. The method for constructing a wireless communication network self-optimizing anti-interference architecture system according to claim 1, characterized in that: In step 2, the adversarial defense model calls the resource collaboration module as follows: In response to the interference attacks generated by the adversarial attack model with the purpose of blocking key nodes and key links, the resource collaboration module is run in the adversarial defense model, and the resource scheduling algorithm component is run to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance; the network resource configuration strategy components include multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, and the resource scheduling algorithm components include distributed resource configuration and multi-user resource collaboration.

5. The method for constructing a wireless communication network self-optimizing anti-interference architecture system according to claim 3, characterized in that: The network control module adopts a networking algorithm based on alliance formation game: Each node first dynamically adjusts the set of neighboring nodes it interacts with by controlling the transmission power. Then, by setting the coalition game utility, it characterizes the utility mapping relationship from individual to local and from local to global. It forms an alliance between network nodes and divides the subnet according to the node alliance. Execute the link reconstruction strategy, and each node dynamically adjusts the link of the local subnet structure to achieve distributed dynamic networking; When interference destroys the local network structure, the network reconstruction strategy is executed to accelerate the traction collaborative networking process and collaborative efficiency through the traction effect of the cluster head node on other nodes.

6. The method for constructing a wireless communication network self-optimizing anti-interference architecture system according to claim 4, characterized in that: The resource collaboration module adopts a resource scheduling algorithm based on asymmetric matching game: Execute multi-domain resource matching strategies to sell redundant resources of each node to nodes that are severely disturbed or have insufficient resources; Execute the neighboring node assistance strategy, and use the transaction price and mutual assistance score to enable each node to actively collaborate on resources based on limited interactions; Execute pre-configuration and reconfiguration strategies, and perform distributed reconfiguration of resources at the network level based on environmental situation analysis and the different requirements of each node.

7. A wireless communication network self-optimization anti-interference architecture system, characterized in that: It includes a self-learning module, in which a counter-attack model and a counter-defense model are provided, and the network self-learning module embeds a network control module and a resource collaboration module; The adversarial attack model generates a large number of adversarial samples through the network itself, finds the weak links of the network, and launches network adversarial attacks to counter the learning evolution of the defense model; Based on the adversarial attacks continuously launched, the adversarial defense model inputs a large number of adversarial samples generated by the adversarial attack model, calls the network control module and resource collaboration module, continuously adjusts the algorithm learning parameters, updates the learning strategy, optimizes the network structure and resource allocation, and refines the anti-interference decision boundary; The adversarial attack model and the adversarial defense model are carried out alternately to realize the game and confrontation of the network self-learning module.

8. The wireless communication network self-optimization anti-interference architecture system according to claim 7, characterized in that: The adversarial defense model calls the network control module as follows: In response to interference attacks generated by the adversarial attack model with the purpose of destroying points, breaking links, and paralyzing the network, the network control module is run in the adversarial defense model, and the networking algorithm component is run to adjust the network topology update strategy to reduce the impact of interference and maintain network performance; the network topology update strategy components include subnet segmentation, link reconstruction, and network reconstruction, and the networking algorithm components include distributed dynamic networking and traction collaborative networking.

9. The wireless communication network self-optimization anti-interference architecture system according to claim 7, characterized in that: The adversarial defense model calls the resource collaboration module as follows: In response to the interference attacks generated by the adversarial attack model with the purpose of blocking key nodes and key links, the resource collaboration module is run in the adversarial defense model, and the resource scheduling algorithm component is run to update the network resource configuration strategy to enhance the resilience and anti-interference capabilities of key nodes and maintain network performance; the network resource configuration strategy components include multi-domain resource matching, neighboring node assistance, pre-configuration and reconfiguration, and the resource scheduling algorithm components include distributed resource configuration and multi-user resource collaboration.

10. A mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for constructing a wireless communication network self-optimizing anti-interference architecture system according to any one of claims 1 to 6 is implemented.

Images