Communication network simulation methods, systems, devices, media, and products

Abstract

The present disclosure provides a communication network simulation method, system, device, medium and product, wherein the method comprises: obtaining a target simulation intention and target of a user; selecting a plurality of agent groups according to the target simulation intention and target, determining a connection relationship of each agent group, and configuring a simulation running parameter of each agent group; in response to a simulation start instruction, combining and building each agent group according to the simulation running parameter and the connection relationship to obtain a communication simulation network; running the communication simulation network by calling a target tool in a tool library to obtain a simulation result, and updating tool attribute information of the tool and / or model parameters of the agent in the tool library according to a satisfaction score of the user on the simulation result.

Description

Technical Field

[0001] This disclosure relates to the technical field of artificial intelligence, and more specifically, to a communication network simulation method, system, device, medium, and product. Background Technology

[0002] Mobile communication networks consist of multiple systems, including terminals, wireless channels, wireless base stations, transmission networks, and core networks. The technologies involved include source-channel encoding / decoding, access control, mobility control, resource allocation, power control, IP routing, and quality of service assurance. In the field of communications, the operation of real communication systems is typically simulated using communication simulation systems. By establishing digital models, the various components of the communication system are abstracted and represented, and then simulated on a computer.

[0003] Most existing simulation technologies construct simulation environments based on network elements and communication function implementation methods required by communication theory and standard protocols. Existing simulation software typically divides communication systems into functional modules such as user distribution, terminal functions, channels, and base station functions, with each module designed and implemented according to standard protocol requirements. However, for communication system simulation, the usage threshold is relatively high, and the construction cost of simulation systems that incorporate user intent is high. Once a simulation system is built, the simulation purpose and the structure of the communication simulation network will not change, which increases the degree of limitation on the use of the simulation system. Summary of the Invention

[0004] This disclosure provides at least one communication network simulation method, system, device, medium, and product.

[0005] In a first aspect, embodiments of this disclosure provide a communication network simulation method, including:

[0006] Obtain the user's target simulation intent and objective;

[0007] Based on the target simulation intent and target, select multiple intelligent agent groups, determine the connection relationship of each intelligent agent group, and configure the simulation operation parameters of each intelligent agent group;

[0008] In response to the simulation start command, the intelligent agent groups are combined and constructed according to the simulation running parameters and the connection relationship to obtain a communication simulation network;

[0009] The communication simulation network is run by calling the target tool in the tool library to obtain simulation results, and the tool attribute information of the tool in the tool library and / or the model parameters of the agent are updated according to the user's satisfaction rating of the simulation results.

[0010] In one optional implementation, the plurality of agent groups includes: a master agent group and a backup agent group, and the method further includes:

[0011] If the simulation results of the main agent group and the backup agent group fail the rationality assessment, the initial network parameters of each agent group are adjusted, and the communication simulation network built by the multiple agent groups is rerun until a simulation result that passes the rationality assessment is obtained; wherein, the initial network parameters include: simulation running parameters and network model parameters.

[0012] In one optional implementation, the method further includes:

[0013] If the rationality assessment of the simulation results of the main intelligent agent group fails, but the rationality assessment of the simulation results of the backup intelligent agent group passes, obtain the user's satisfaction rating for the simulation results of the backup intelligent agent group.

[0014] If, based on the satisfaction rating, it is determined that the user is not satisfied with the simulation results of the backup agent group, then the target simulation intent and objective are adjusted, and the communication simulation network corresponding to each agent group is rerun based on the adjusted target simulation intent and objective.

[0015] In one optional implementation, the method further includes:

[0016] If the rationality assessment of the main intelligent agent group is passed, obtain the user's satisfaction score for the simulation results of the main intelligent agent group;

[0017] If, based on the satisfaction score, it is determined that the user is not satisfied with the simulation results of the main agent group, then the target simulation intent and objective are adjusted, and the communication simulation network corresponding to each agent group is rerun based on the adjusted target simulation intent and objective.

[0018] In one optional implementation, obtaining the user's target simulation intent and target includes:

[0019] Obtain the user's original target simulation intent and target input through a multimodal large language model;

[0020] The original target simulation intent and target are translated by combining simulation knowledge information in the knowledge base to obtain machine-recognizable target simulation intent and target, and the target simulation intent and target are stored in the memory bank.

[0021] In one optional implementation, the step of selecting multiple agent groups based on the target simulation intent and the target includes:

[0022] The main agent group is determined by combining the simulation knowledge in the knowledge base and the simulation scores of each agent in the memory bank;

[0023] Based on the satisfaction scores of the historically running agent groups in the memory bank, a backup agent group is determined, and the main agent group and the backup agent group are determined as the plurality of agent groups.

[0024] In one optional implementation, the step of determining the master agent group by combining simulation knowledge in the knowledge base and simulation scores of each agent in the memory bank includes:

[0025] The type of intelligent agent that matches the target simulation intent and target is determined by combining the simulation knowledge information in the knowledge base;

[0026] Based on the memory bank, the target agent with the highest score under the same agent type is determined, and the main agent group is determined based on the target agent.

[0027] In one optional implementation, determining the candidate agent group based on the satisfaction scores of the agent groups that have been running in the memory bank includes:

[0028] The candidate agent group is determined based on the historical operation of the agent group with the highest satisfaction score in the memory bank.

[0029] In one optional implementation, configuring the simulation operation parameters for each of the agent groups includes:

[0030] The simulation operation parameters are configured based on parameter configuration information and simulation knowledge in the knowledge base; wherein, the simulation operation parameters include at least one of the following: agent permissions and attribute parameters, interface attribute parameters, and the parameter configuration information is determined according to the target simulation intent and target.

[0031] In one optional implementation, the step of running the communication simulation network by calling the target tool in the tool library to obtain simulation results includes:

[0032] Based on the first type of attribute information of each tool in the tool library, candidate tools that match the target simulation intent and target are determined;

[0033] Based on the calling priority of the candidate tools, a target tool for running the communication simulation network is determined, and the communication simulation network is run through the target tool to obtain simulation results; wherein, the calling priority is determined by the tool attribute information of each tool.

[0034] In one optional implementation, updating the tool attribute information of the tools in the tool library and / or the model parameters of the agent based on the user's satisfaction rating of the simulation results includes:

[0035] Periodically query the memory bank for historical satisfaction scores of the communication simulation network that ran at a historical time.

[0036] If it is determined that the historical satisfaction score does not meet the satisfaction requirements, the tool attribute information of the tools in the tool library and / or the model parameters of the agent are updated.

[0037] In one optional implementation, triggering the update of tool attribute information of tools in the tool library includes:

[0038] The thought chain is obtained by retrieving the relationship between the target simulation intent and objective of the communication simulation network running at historical time points, the simulation process parameters of the communication simulation network, and the simulation resource information of the communication simulation network from the memory bank; wherein, the simulation resource information includes information related to the communication simulation network in the knowledge base and the tool library;

[0039] Runtime prompts for each simulation process in constructing this communication simulation network;

[0040] Based on the thought chain and the running prompt information, update the tool attribute information of the tools in the tool library.

[0041] In one optional implementation, updating the tool attribute information of the tools in the tool library based on the thought chain and the running prompt information includes:

[0042] The communication simulation network is run according to the aforementioned thought chain, and the tool calling process of each agent in the communication simulation network is adjusted through the running prompt information during the operation.

[0043] Record tool call information during the tool call process, and update the tool attribute information of the tools in the tool library based on the tool call information.

[0044] In one optional implementation, the triggering of updating the agent's model parameters includes:

[0045] Read the satisfaction score of each agent from the memory bank, and construct positive and negative samples based on the satisfaction score of the agent;

[0046] The target parameters corresponding to the positive or negative samples are used as the initial parameters of the agent; wherein, the target parameters include: simulation running parameters and model parameters;

[0047] The agent with the initial parameters is trained using the positive and negative samples until an agent that meets the satisfaction requirements is obtained.

[0048] Secondly, embodiments of this disclosure provide a communication network simulation system, including:

[0049] An intent processing entity is used to obtain the user's target simulation intent and target, select multiple intelligent agent groups based on the target simulation intent and target, and determine the connection relationship of each intelligent agent group.

[0050] The intelligent agent control entity is used to configure the simulation operation parameters of each intelligent agent group and update the tool attribute information of the tools in the tool library according to the user's satisfaction rating of the simulation results.

[0051] The simulation agent is used to respond to the simulation start command, combine and build the various agent groups according to the simulation running parameters and the connection relationship to obtain a communication simulation network, run the communication simulation network by calling the target tool in the tool library, obtain simulation results, and update the model parameters of the agent according to the user's satisfaction rating of the simulation results.

[0052] In one optional implementation, the intent processing entity includes: a knowledge base query instance module, a user intent translation module, and a simulator orchestration module;

[0053] The knowledge base query instance module is used to query simulation knowledge information from the knowledge base;

[0054] The user intent translation module is used to obtain the original target simulation intent and target input by the user through a multimodal large language model; and to translate the original target simulation intent and target by combining the simulation knowledge information to obtain a machine-recognizable target simulation intent and target.

[0055] The simulator orchestration module is used to select multiple intelligent agent groups according to the target simulation intent and the target, and to determine the connection relationship of each intelligent agent group.

[0056] In one optional implementation, the intent processing entity includes: a simulation result processing module and an intent alignment instance module;

[0057] The simulation result processing module is used to perform statistical analysis on the simulation results using a knowledge base and a tool library to obtain statistical analysis results.

[0058] The intent alignment instance module is used to push the statistical analysis results to the user and obtain the user's satisfaction score for the statistical analysis results, thereby obtaining the user's satisfaction score for the simulation results of the main intelligent agent group.

[0059] In one optional implementation, the intent processing entity includes:

[0060] The system update determination module is used to periodically query the historical satisfaction scores of the communication simulation network that ran at a historical time in the memory bank; and if it is determined that the historical satisfaction score does not meet the satisfaction requirements, it triggers the update of the tool attribute information of the tools in the tool library and / or the model parameters of the agent.

[0061] In one optional implementation, the intelligent agent control entity includes: a simulator permission and attribute setting module, an interface attribute configuration module, and a communication simulation network operation control and monitoring configuration module;

[0062] The simulator permission and attribute setting module is used to set the permission parameters and attribute parameters for each of the intelligent agents;

[0063] The interface attribute configuration module is used to configure the interface attribute parameters of each of the intelligent agents;

[0064] The communication simulation network operation control and monitoring configuration module is used to configure the control parameters and monitoring parameters of the communication network simulation system.

[0065] In one optional implementation, the communication network simulation system further includes: a knowledge base, a memory, and an agent toolkit;

[0066] The knowledge base contains simulation knowledge information required for each simulation stage of the communication simulation network.

[0067] The memory is used to store at least one of the following: target simulation intent and target, agent permission parameters and attribute parameters, interface attribute parameters, control parameters and monitoring parameters of the communication network simulation system, simulation results and their satisfaction scores;

[0068] The agent toolkit contains the tools for running each agent and their tool attribute information.

[0069] In one optional implementation, the intelligent agent control entity includes: a simulator update control module, a simulation effect analysis and evaluation module, and a tool update module;

[0070] The simulator update control module is used to update the model parameters of each agent in the simulated intelligent agent;

[0071] The tool update module is used to update the tool attribute information of the tools in the tool library;

[0072] The simulation effect analysis and evaluation module is used to evaluate the rationality of the simulation results and obtain the rationality evaluation results.

[0073] Thirdly, embodiments of this disclosure also provide an electronic device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the steps of the first aspect above, or any possible implementation of the first aspect, are performed.

[0074] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the first aspect or any possible implementation thereof.

[0075] Fifthly, embodiments of this disclosure also provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the first aspect described above, or any possible implementation of the first aspect.

[0076] This disclosure provides a communication network simulation method, system, device, medium, and product. In embodiments of this disclosure, firstly, the user's target simulation intent and objective are obtained; then, multiple agent groups are selected based on the target simulation intent and objective, and the connection relationships of each agent group are determined, as well as the simulation operation parameters of each agent group are configured; next, in response to a simulation start command, the agent groups are combined and constructed according to the simulation operation parameters and the connection relationships to obtain a communication simulation network; then, the communication simulation network is run by calling a target tool from a tool library to obtain simulation results, and the tool attribute information of the tools in the tool library and / or the model parameters of the agents are updated based on the user's satisfaction rating of the simulation results.

[0077] In the above embodiments, by selecting multiple intelligent agent groups based on the target simulation intent and the target, and then constructing and running a communication simulation network based on these groups, it is possible to flexibly combine intelligent agents and provide services according to user needs. This improves the flexibility of constructing the communication simulation network and narrows the scope of its application limitations. This processing method enables automatic construction of the communication simulation network, thereby saving the manual construction costs. Simultaneously, the technical solution of this disclosure can achieve self-learning and self-updating of tool attribute information and / or intelligent agent model parameters, reducing labor costs while improving the continuous accuracy and cross-scenario versatility of the simulation network.

[0078] To make the above-mentioned objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0079] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0080] Figure 1 A flowchart of a communication network simulation method provided by an embodiment of this disclosure is shown;

[0081] Figure 2 A schematic diagram of the structure of a communication network simulation system provided in an embodiment of this disclosure is shown;

[0082] Figure 3 This diagram illustrates a functional module schematic of an intent processing entity provided in an embodiment of the present disclosure;

[0083] Figure 4 This diagram illustrates the functional modules of an intelligent agent control entity provided in an embodiment of the present disclosure.

[0084] Figure 5 A schematic diagram of the intent transmission phase of the communication network simulation system provided in this embodiment of the present disclosure is shown;

[0085] Figure 6 A schematic diagram of the simulation network preparation stage and the operation stage of the communication network simulation system provided in the embodiments of this disclosure is shown;

[0086] Figure 7 A schematic diagram of the iterative optimization stage of the communication network simulation system provided in this embodiment of the present disclosure is shown;

[0087] Figure 8 A schematic diagram of an electronic device provided in an embodiment of the present disclosure is shown. Detailed Implementation

[0088] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0089] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0090] In this document, the term "and / or" merely describes a relationship, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0091] As described above, most existing simulation technologies construct simulation environments based on network elements and communication function implementation methods required by communication theory and standard protocols. Existing simulation software typically divides communication systems into functional modules such as user distribution, terminal functions, channels, and base station functions, with each module designed and implemented according to standard protocol requirements. However, for communication system simulation, the usage threshold is relatively high, and the construction cost of simulation systems that incorporate user intent is high. Once a simulation system is built, the simulation purpose and the structure of the communication simulation network will not change, which increases the degree of limitation on the use of the simulation system.

[0092] Based on the above research, this disclosure provides a communication network simulation method, system, device, medium, and product. In the embodiments of this disclosure, firstly, the user's target simulation intent and objective are obtained; then, multiple intelligent agent groups are selected according to the target simulation intent and objective, and the connection relationship of each intelligent agent group is determined, as well as the simulation operation parameters of each intelligent agent group are configured; next, in response to a simulation start command, the intelligent agent groups are combined and constructed according to the simulation operation parameters and the connection relationship to obtain a communication simulation network; finally, the communication simulation network is run by calling a target tool in a tool library to obtain simulation results.

[0093] In the above embodiments, by selecting multiple intelligent agent groups based on the target simulation intent and the target, and then constructing and running a communication simulation network based on these groups, it is possible to flexibly combine intelligent agents and provide services according to user needs. This improves the flexibility of building the communication simulation network and narrows the scope of its application limitations. This processing method enables the automatic construction of the communication simulation network, thereby saving the manual construction costs associated with building such networks.

[0094] To facilitate understanding of this embodiment, a detailed description of the communication network simulation method disclosed in this disclosure will be provided first. The execution subject of the communication network simulation method provided in this disclosure is generally an electronic device with a certain computing capability. In some possible implementations, this communication network simulation method can be implemented by a processor calling computer-readable instructions stored in memory.

[0095] See Figure 1 The diagram shows a flowchart of a communication network simulation method provided in this embodiment. This method is applied to a communication network simulation system and includes steps S101 to S104, wherein:

[0096] S101: Obtain the user's target simulation intent and objective.

[0097] S102: Select multiple intelligent agent groups according to the target simulation intent and target, determine the connection relationship of each intelligent agent group, and configure the simulation operation parameters of each intelligent agent group.

[0098] Here, multiple agent groups include a master agent group and a backup agent group. The simulation operation parameters include the agent permissions and attribute parameters of each agent in each agent group, as well as the interface attribute parameters.

[0099] S103: In response to the simulation start command, the intelligent agent groups are combined and built according to the simulation running parameters and the connection relationship to obtain a communication simulation network.

[0100] In this embodiment of the disclosure, after detecting the simulation start command, each agent can be configured according to the simulation running parameters, and the agents in the agent group can be combined and connected according to the connection relationship to obtain a communication simulation network.

[0101] S104: Run the communication simulation network by calling the target tool in the tool library, obtain the simulation results, and update the tool attribute information of the tool in the tool library and / or the model parameters of the agent according to the user's satisfaction rating of the simulation results.

[0102] Here, the tool that matches the communication simulation network can be determined by the tool attribute information of each tool in the tool library. Then, the communication simulation network can be run by calling the tool to obtain the simulation results.

[0103] In this embodiment of the disclosure, firstly, the user's target simulation intent and objective are obtained; then, multiple intelligent agent groups are selected according to the target simulation intent and objective, and the connection relationship of each intelligent agent group is determined, as well as the simulation operation parameters of each intelligent agent group are configured; next, in response to the simulation start command, each intelligent agent group is combined and built according to the simulation operation parameters and the connection relationship to obtain a communication simulation network; then, the communication simulation network is run by calling the target tool in the tool library to obtain the simulation results.

[0104] In the above embodiments, by selecting multiple intelligent agent groups based on the target simulation intent and the target, and then constructing and running a communication simulation network based on these groups, it is possible to flexibly combine intelligent agents and provide services according to user needs. This improves the flexibility of constructing the communication simulation network and narrows the scope of its application limitations. This processing method enables automatic construction of the communication simulation network, thereby saving the manual construction costs. Simultaneously, the technical solution of this disclosure can achieve self-learning and self-updating of tool attribute information and / or intelligent agent model parameters, reducing labor costs while improving the continuous accuracy and cross-scenario versatility of the simulation network.

[0105] In this embodiment, the communication network simulation system comprises an intent processing entity, a simulation agent, an agent control entity, and a simulation basic resource library. The simulation basic resource library includes a knowledge base, a memory library, and an agent tool library (hereinafter referred to as the tool library). The communication network simulation method described above will be introduced below in conjunction with the specific structure of the communication network simulation system.

[0106] In an optional implementation, step S101 above, which obtains the user's target simulation intent and target, specifically includes the following steps:

[0107] First, obtain the user's original target simulation intent and target input through a multimodal large language model;

[0108] Secondly, the original target simulation intent and target are translated by combining the simulation knowledge information in the knowledge base to obtain a machine-recognizable target simulation intent and target, and the target simulation intent and target are stored in the memory bank.

[0109] In this embodiment of the disclosure, the user can input the original simulation intent and target using a multimodal large language model (MM-LLM). After obtaining the user's input of the original simulation intent and target, the intent processing entity can call upon simulation knowledge information in the knowledge base, and then translate the user's input of the original simulation intent and target into a language that the communication system can understand, i.e., the target simulation intent and target. Afterwards, the intent processing entity can store the translated target simulation intent and target in a memory bank and simultaneously pass it to the intelligent agent control entity for further processing.

[0110] In an optional implementation, step S102 above selects multiple agent groups based on the target simulation intent and the target, specifically including the following steps:

[0111] First, the main agent group is determined by combining the simulation knowledge in the knowledge base and the simulation scores of each agent in the memory bank;

[0112] Secondly, based on the satisfaction scores of the agent groups that have been running in the memory bank in the past, a backup agent group is determined, and the main agent group and the backup agent group are determined as the multiple agent groups.

[0113] In this embodiment of the disclosure, the multiple intelligent agent groups include a primary intelligent agent group and a backup intelligent agent group. Here, the intent processing entity can query the knowledge base and memory, and combine the target simulation intent and the target to complete the selection of the primary intelligent agent group and the backup intelligent agent group, as well as the confirmation of the connection relationship.

[0114] The knowledge base contains simulation knowledge information required for each simulation stage of the communication simulation network; the memory is used to store at least one of the following: target simulation intent and target, agent permission parameters and attribute parameters, interface attribute parameters, control parameters and monitoring parameters of the communication network simulation system, simulation results and their satisfaction scores.

[0115] Therefore, the primary agent group can be determined by combining the knowledge base and the memory; this primary agent group has not been operating as a communication simulation network at any historical time. The backup agent group is a set of agent groups that meet the simulation requirements and has been operating as a communication simulation network at any historical time.

[0116] As described above, the simulation agent combination scheme, which combines the simulation intent and target selection, is the lowest cost simulation agent combination that can meet the requirements, and provides simulation services on demand.

[0117] For example, if the user's input intent is to simulate the performance indicators of a wireless network that closely resemble reality, after reviewing the historical records in the knowledge base and memory, the intent processing entity selects the following agent group: geographic environment simulation agent + user trajectory simulation agent + user internet access service simulation agent + wireless channel simulation agent + wireless network performance simulation agent. Combining real physical space characteristics, the user's internet access behavior and channel simulation are completed, and then the network performance indicators that closely resemble reality are output.

[0118] If the user's input intent is to simulate wireless network performance indicators, the intent processing entity adjustment agent group is: user distribution agent (only user grid location distribution, no continuous latitude and longitude trajectory) + user internet access service simulation agent + wireless network performance simulation agent (calling theoretical channel model) to simulate network performance indicators. The simulation results conform to the operation law of communication system but cannot closely match reality.

[0119] If the user's input intent is theoretical wireless network performance simulation, the intent processing entity will adjust the agent group to: wireless network performance simulation agent (calling the theoretical channel model and default FTP service), and the user will select the wireless user theoretical distribution model (good, medium and bad percentage) and FTP service transmission data traffic for simulation.

[0120] In this embodiment of the disclosure, the above steps combine simulation knowledge in the knowledge base and simulation scores of each agent in the memory bank to determine the master agent group, specifically including the following steps:

[0121] Step S11: Determine the type of intelligent agent that matches the target simulation intent and target by combining the simulation knowledge information in the knowledge base;

[0122] Step S12: Based on the memory bank, determine the target agent with the highest score under the same agent type, and determine the main agent group based on the target agent.

[0123] In this embodiment of the disclosure, the agent with the highest score of the same type in the memory bank can be selected first to build the communication simulation network.

[0124] In practice, simulation knowledge information can be queried in the knowledge base through the knowledge base query instance in the intent processing entity. Based on this simulation knowledge information, the agent or agent type that matches the target simulation intent and target can be determined. For each agent type, there can be multiple agents. In this case, it is necessary to select the target agent from the agents belonging to each agent type by combining the scores of each agent, and then determine the target agent as the main agent group.

[0125] Each agent's score is stored in a memory bank, and this score can be retrieved from the memory bank. Then, from multiple agents of the same agent type, the agent with the highest score is selected as the target agent, thus obtaining the master agent group. Next, the connection relationships between the multiple target agents can be determined using simulation knowledge information, thereby obtaining the connection relationships of the master agent group.

[0126] By using the above processing method, a minimum number of intelligent agents can be selected to build a communication simulation network while meeting the simulation intent and objectives, thereby reducing the construction cost of the communication simulation network.

[0127] In this embodiment of the disclosure, the above steps determine the candidate agent group based on the satisfaction scores of the agent groups that have been running in the memory bank, specifically including the following steps:

[0128] The candidate agent group is determined based on the historical operation of the agent group with the highest satisfaction score in the memory bank.

[0129] In this embodiment of the disclosure, satisfaction scores of agent groups that have been running at historical times can be stored in a memory bank. Here, the memory bank can be used to filter out running agent groups that match the target simulation intent and target, and the satisfaction scores of the filtered running agent groups can be determined. Then, a backup agent group can be determined based on the agent group with the highest satisfaction score.

[0130] In practice, the agent group with the highest satisfaction score is directly selected as the backup agent group.

[0131] In addition, the agents in the agent group with the highest satisfaction rating (denoted as agent group A) can be replaced, and the replaced agents can be identified as the backup agent group. For example, regarding the type of agents in the agent group with the highest satisfaction rating (denoted as agent group A), the agent with the higher rating among agents of the same type can be identified from the memory bank, and then the corresponding agent in agent group A can be replaced with the agent with the higher rating, thus obtaining the backup agent group.

[0132] In this case, the connection relationship between the agents in agent group A is the connection relationship of the backup agent group.

[0133] This processing method allows for the rapid selection of agent groups that meet the satisfaction requirements. Furthermore, by replacing and adjusting the agents in the backup agent groups, the backup agent groups can be further optimized, resulting in a communication simulation network with better simulation effects.

[0134] In this embodiment of the disclosure, after selecting multiple intelligent agent groups and determining the connection relationship of each intelligent agent group, the intent processing entity can combine the target simulation intent and the target to complete the processing of information such as intelligent agent parameter configuration, business process, network operation mode and simulation target, and transmit the processed full information to the memory and intelligent agent control entity, and transmit the connection relationship of each intelligent agent group to the simulated intelligent agent.

[0135] In an optional implementation, the above steps configure the simulation operation parameters for each of the intelligent agent groups, specifically including the following steps:

[0136] The simulation operation parameters are configured based on parameter configuration information and simulation knowledge in the knowledge base; wherein, the simulation operation parameters include at least one of the following: agent permissions and attribute parameters, interface attribute parameters, and the parameter configuration information is determined according to the target simulation intent and target.

[0137] Here, the parameter configuration information is denoted as the parameter configuration, business process, network operation mode, and simulation target of the intelligent agent described above. After receiving the parameter configuration information, the intelligent agent control entity can combine the received parameter configuration information with the corresponding description in the knowledge base to complete the intelligent agent permission and attribute configuration, interface attribute configuration, and then send the configured simulation operation parameters to the simulation intelligent agent.

[0138] Next, the intelligent agent control entity can configure the operation control and monitoring parameters of the communication simulation network. After completing this configuration process, all configured parameters can be checked. This check can identify incorrectly configured parameters and parameters that have not been configured. After completing the parameter configuration, all configured parameters can be stored in the memory.

[0139] After completing the network preparation phase described above, the network operation phase can begin. During network operation, the agent control entity generates a simulation start command and transmits this command to the simulated agent to begin agent assembly and construction, thus obtaining the communication simulation network. Simultaneously, monitoring of the agent's operational status begins. The simulated agent can combine information received from the intent processing entity and the agent control entity to initiate the communication simulation network and continuously invoke tools from the tool library.

[0140] In an optional implementation, the above steps involve running the communication simulation network by calling the target tool in the tool library to obtain simulation results, specifically including the following steps:

[0141] Step S31: Based on the first type of attribute information of each tool in the tool library, determine the candidate tools that match the target simulation intent and the target;

[0142] Step S32: Based on the calling priority of the candidate tools, determine the target tool for running the communication simulation network, and run the communication simulation network through the target tool to obtain simulation results; wherein, the calling priority is determined by the tool attribute information of each tool.

[0143] In this implementation, the tool library consists of a set of tools that the simulator can call and tool attribute information. Multiple tools of the same type exist in the tool library, managed by a tool attribute information table. The tool attribute information table records the calling priority of each tool, information on the combined use of tools for various simulation tasks, the number of times they are called, and other information, determining the calling behavior during the simulation agent's runtime phase. The tool attribute information table format is as follows:

[0144]

[0145]

[0146] The meanings of each field in the tool attribute information table are as follows:

[0147] Tool Name: The tool represents information, such as a probability model;

[0148] Tool ID: Tool identification information, for example, ID is 1;

[0149] Tool types: Divided into three categories: inherent knowledge tools, AI atomic capabilities, and system-owned tools;

[0150] Tool tags: These represent the business scenarios in which the tool is applicable throughout the entire simulation process, such as physical environment feature analysis, channel simulation, and simulation result calculation.

[0151] Scope of application: This refers to the intent processing entities and specific simulated intelligent agents to which the tool is applicable. The tool can be general-purpose or specific to a particular intelligent agent.

[0152] Simulation task types: Simulation task types are organized according to simulation intent and objectives, such as user behavior simulation, wireless channel simulation, wireless network performance simulation, etc.

[0153] The fields of call priority, number of times used in combination, and number of times called are all recorded separately according to different task types.

[0154] Invocation Priority: Represents the priority of a tool in being invoked among all tools of the same type under a specific simulation task. The value ranges from [1, 2, 3, 4, 5] and only takes effect when multiple tools of the same type exist in the tool library. For example, if there are n channel models, and m channel models have been invoked under a specific task Ti, when a user wants to initiate a simulation task of type Ti, the channel model selection has a high probability of being chosen from the m channel models. The higher the priority, the higher the probability of selection. There is also a small probability of being chosen from the remaining (nm) channel models. The invocation priority is calculated by weighting the number of times it has been invoked under the same task, the number of times it has been used in conjunction with other tools, the average user satisfaction during invocation, and the probability of it remaining active.

[0155] Joint usage count: Based on simulation task statistics, calculate the call status and call count of other simulators under the same type of task, in the format {Simulator 1: call count; Simulator 2: call count; ...}.

[0156] Number of calls: Calculate the number of times the target simulator is called under the same type of task based on simulation task statistics.

[0157] Persistence Probability: This represents the probability that a system-exclusive tool will become a fixed tool. The default probability for inherent knowledge tools and AI atomic capabilities is always 1. The default probability for system-exclusive tools is 0.3, determined by the total number of calls, the number of task categories called, the number of simulated intelligent agents, and user satisfaction during the call. If the persistence probability remains below 0.1 for an extended period, the tool will be removed from the tool attribute information table.

[0158] During simulation, based on the specific simulation task, the intended entity or simulated intelligent agent invokes tools from the tool library by combining tool tags, applicable scope, simulation task type, and invocation priority fields. Unlike the general system tools of traditional AI agents, the tools in the tool library include the following three categories:

[0159] Inherent knowledge tools: These are classic tools derived from commonly used knowledge and experience in mathematics, communications, and computer science accumulated throughout the history of communication system development. They are used for the construction, operation, observation, and evaluation of networks. These include basic tools such as correlation calculation, time-domain and frequency-domain signal processing, and probability statistics, as well as classic network system and sociological models such as user internet access service models, routing models, wireless channel models, and resource allocation algorithms.

[0160] AI atomic capabilities: These include capabilities commonly used in the field of artificial intelligence such as data preprocessing and postprocessing, feature analysis and construction, and model parameter optimization. In the field of network intelligence, they have been proven in practice to possess universal atomic capabilities such as prediction, analysis, perception, and decision-making.

[0161] System-owned tools: Tools accumulated during the simulation system's operation and iterative optimization, including memory reading and processing tools, simulation result calculation tools, and agent parameter tuning tools accumulated during agent operation, as well as new tools of the same type formed by modifying parameters / formula coefficients of inherent knowledge tools and AI atomic capabilities during the optimization process.

[0162] In this embodiment of the disclosure, candidate tools that match the target simulation intent and target can be determined based on the first type of attribute information of each tool in the tool library. The first type of attribute information includes attribute information such as tool name, tool ID, tool type, tool tag, applicable scope, and simulation task type from the tool attribute information table.

[0163] After obtaining candidate tools, the target tool for running the communication simulation network can be determined based on the tool's invocation priority. For example, the candidate tool with the highest priority can be selected as the target tool for running the communication simulation network. Then, the communication simulation network can be run using the target tool to obtain simulation results.

[0164] As described above, the multiple agent groups include a primary agent group and a backup agent group. Therefore, after running the communication simulation network by calling the target tool in the tool library and obtaining the simulation results, the method further includes the following steps:

[0165] If the simulation results of the main agent group and the backup agent group fail the rationality assessment, the initial network parameters of each agent group are adjusted, and the communication simulation network built by the multiple agent groups is rerun until a simulation result that passes the rationality assessment is obtained; wherein, the initial network parameters include: simulation running parameters and network model parameters.

[0166] In this embodiment, the communication network simulation system first runs the communication simulation network built by the main intelligent agent group, denoted as communication simulation network A, in the manner described above. At this time, the intelligent agent control entity can evaluate the rationality of the simulation results of communication simulation network A from aspects such as result completeness and usability. If the rationality evaluation of the simulation results of communication simulation network A fails, the communication simulation network built by the backup intelligent agent group, denoted as communication simulation network B, is run in the manner described above. If the rationality evaluation of the simulation results of communication simulation network B also fails, the initial network parameters of each intelligent agent group are adjusted, and the communication simulation networks built by multiple intelligent agent groups are rerun until a simulation result that passes the rationality evaluation is obtained.

[0167] In this embodiment of the disclosure, if the rationality assessment of the main agent group passes, the user's satisfaction score for the simulation results of the main agent group is obtained; then, if it is determined based on the satisfaction score that the user is not satisfied with the simulation results of the main agent group, the target simulation intent and target are adjusted, and the communication simulation network corresponding to each agent group is rerun based on the adjusted target simulation intent and target.

[0168] Here, the intent processing entity retrieves the target simulation diagram and objective from the memory, calls the knowledge base and tool library to perform statistical analysis on the simulation results of the communication simulation network A, and obtains the statistical analysis results. These results are then presented to the user via MMLLM, while simultaneously obtaining user feedback on the simulation results. The intent processing entity translates the user feedback into a satisfaction score, stores it in the memory, and updates the agent's score in the memory based on the satisfaction score and the simulation results. If the user is satisfied with the simulation results, the simulation is completed; if the user is dissatisfied, the simulation intent and objective are adjusted based on the user feedback, and the intent transmission phase and network organization and operation phase are repeated until a satisfactory simulation result is obtained.

[0169] In this embodiment of the disclosure, if the rationality assessment of the simulation results of the main intelligent agent group fails, but the rationality assessment of the simulation results of the backup intelligent agent group passes, a user satisfaction score for the simulation results of the backup intelligent agent group is obtained; wherein, if it is determined based on the satisfaction score that the user is not satisfied with the simulation results of the backup intelligent agent group, the target simulation intent and target are adjusted, and the communication simulation network corresponding to each intelligent agent group is rerun based on the adjusted target simulation intent and target.

[0170] Here, if the rationality assessment of the simulation results of the main agent group fails, while the rationality assessment of the simulation results of the backup agent group passes, the intent processing entity retrieves the target simulation diagram and objective of this simulation from the memory, calls the knowledge base and tool library to perform statistical analysis on the simulation results of the communication simulation network B, and obtains the statistical analysis results; then, it presents these results to the user through MMLLM, while simultaneously obtaining user feedback on the simulation results. The intent processing entity translates the user's feedback into a satisfaction score, stores it in the memory, and updates the agent's score in the memory based on the satisfaction score and the simulation results. If the user is satisfied with the simulation results, the simulation is completed; if the user is not satisfied, the simulation intent and objective are adjusted based on the user feedback, and the intent transmission phase and network organization and operation phase are repeated until a simulation result satisfactory to the user is obtained.

[0171] In an optional implementation, updating the tool attribute information of the tools in the tool library and / or the model parameters of the agent based on the user's satisfaction rating of the simulation results includes the following steps:

[0172] S105: Periodically query the memory bank for the historical satisfaction scores of the communication simulation network that was running at a historical time.

[0173] S106: If it is determined that the historical satisfaction score does not meet the satisfaction requirements, trigger the updating of the tool attribute information of the tools in the tool library and / or the model parameters of the agent.

[0174] In this embodiment of the disclosure, the intent processing entity can periodically query the satisfaction ratings of recently run communication simulation networks from the memory. If the satisfaction ratings indicate that the user's satisfaction with the communication network simulation system is poor, a simulation system update start command is sent to the agent control entity. This command then triggers updates to the tool attribute information of tools in the tool library and the model parameters of the agent.

[0175] In this embodiment of the disclosure, the above steps trigger the updating of tool attribute information of tools in the tool library, specifically including the following steps:

[0176] Step S41: Read the relationship between the target simulation intent and objective of the communication simulation network running at a historical time in the memory bank, the simulation process parameters of the communication simulation network, and the simulation resource information of the communication simulation network to obtain the thought chain; wherein, the simulation resource information includes information related to the communication simulation network in the knowledge base and the tool library;

[0177] Step S42: Construct the running prompt information for each simulation process of the communication simulation network;

[0178] Step S43: Based on the thought chain and the running prompt information, update the tool attribute information of the tools in the tool library.

[0179] In this embodiment, the intelligent agent control entity reads simulation records and user satisfaction data from the memory bank to establish the relationships between the target simulation intent and target graph, simulation process parameters, knowledge base, and tool library, thus beginning the construction of a thought chain. Next, starting with the target simulation intent and target graph, it constructs step-by-step prompts for knowledge filtering, simulation parameter configuration, simulation execution, and tool invocation (i.e., execution prompts for each simulation execution process). Finally, based on this thought chain and execution prompts, the tool attribute information of the tools in the tool library can be updated.

[0180] In this embodiment of the disclosure, the above steps, based on the thought chain and the running prompt information, update the tool attribute information of the tools in the tool library, specifically including:

[0181] The communication simulation network is run according to the aforementioned thought chain, and the tool calling process of each agent in the communication simulation network is adjusted through the running prompt information during the operation.

[0182] Record tool call information during the tool call process, and update the tool attribute information of the tools in the tool library based on the tool call information.

[0183] In this embodiment, the communication simulation network can be run according to a thought chain, and the tool invocation process of each agent in the communication simulation network can be adjusted through running prompts during each simulation run. After the simulation run is completed, the tool invocation information of each simulation run is recorded, and the tool attribute information in the tool attribute information table is updated. Simultaneously, new system-owned tools can be generated by adjusting the process based on the tool attribute information. After the agent tool library update is completed, the agent control entity notifies the simulation agent to start the update process.

[0184] For example, suppose the simulation intention and goal is to simulate the performance indicators of a wireless network that are close to reality. However, if the actual wireless channel simulation uses a theoretical channel model, it will result in poor user satisfaction.

[0185] At this point, the input running prompt message can be described as follows:

[0186] (1) To output real wireless network performance indicators, a combination of real user internet behavior simulation, real wireless channel simulation, and real wireless performance indicator simulation agents is needed.

[0187] (2) A realistic wireless channel simulation requires an accurate channel model and cannot use a theoretical channel model.

[0188] (3) An accurate channel model can be obtained by using the regression model (hereinafter referred to as Model A) based on the spatial characteristics of the simulated agent in the geographic environment in the tool library to simulate the real channel.

[0189] By increasing the number of times model A is called during the simulation process through multiple rounds of simulation, and by improving the user satisfaction effect obtained from the simulation results of intent processing entity analysis, the calling priority of model A in the tool attribute information table is increased.

[0190] In this embodiment of the disclosure, the above steps trigger the updating of the agent's model parameters, specifically including:

[0191] Read the satisfaction score of each agent from the memory bank, and construct positive and negative samples based on the satisfaction score of the agent;

[0192] The target parameters corresponding to the positive or negative samples are used as the initial parameters of the agent; wherein, the target parameters include: simulation running parameters and model parameters;

[0193] The agent with the initial parameters is trained using the positive and negative samples until an agent that meets the satisfaction requirements is obtained.

[0194] In this embodiment of the disclosure, the simulated agent can read simulation records and satisfaction scores from the memory bank. For each agent, the user's satisfied and dissatisfied simulator results can be constructed into positive and negative sample labels, respectively.

[0195] Here, the target parameters corresponding to either positive or negative samples can be used as the initial parameters for the agent to begin parameter fine-tuning. The loss function is MSE, and the goal is to make the simulation results output by a single simulated agent as close as possible to the original results that satisfy the user. After the agent completes parameter fine-tuning based on positive samples, it is fine-tuned again using negative samples until the simulation results are as close as possible to the original results that satisfy the user, thus obtaining an agent that can adapt to multiple sets of parameter inputs.

[0196] See Figure 2 The figure shows a schematic diagram of a communication network simulation system provided in an embodiment of this disclosure. As shown, the communication network simulation system includes: an intent processing entity, an intelligent agent control entity, a simulated intelligent agent, a global memory, a global knowledge base, and an intelligent agent tool library. The global memory is the memory library in the above embodiment, the global knowledge base is the knowledge base in the above embodiment, and the intelligent agent tool library is the tool library in the above embodiment.

[0197] An intent processing entity is used to obtain the user's target simulation intent and goal, select multiple intelligent agent groups based on the target simulation intent and goal, and determine the connection relationship of each intelligent agent group.

[0198] In this embodiment of the disclosure, the intent processing entity is responsible for aligning the simulation intent and the target, as well as collecting user feedback. For example... Figure 3 As shown, the internal structure can be composed of modules such as user intent translation, simulator orchestration, simulation result processing, system update determination, tool call instance, knowledge base query instance, intent memory instance, and intent alignment instance.

[0199] The intelligent agent control entity is used to configure the simulation operation parameters of each intelligent agent group and update the tool attribute information of the tools in the tool library based on the user's satisfaction rating of the simulation results.

[0200] The intelligent agent control entity is responsible for controlling the operation of the simulated intelligent agent and updating the simulation system. For example... Figure 4As shown, it specifically includes modules such as simulator update control, simulation effect analysis and evaluation, global tool library update, simulator permission and attribute settings, interface attribute settings, and simulation network operation control and monitoring.

[0201] The simulation agent is used to respond to the simulation start command, combine and build the various agent groups according to the simulation running parameters and the connection relationship to obtain a communication simulation network, run the communication simulation network by calling the target tool in the tool library, obtain simulation results, and update the model parameters of the agent according to the user's satisfaction rating of the simulation results.

[0202] Simulated intelligent agents are a collection of simulators based on large model intelligent agents. Each intelligent agent is responsible for simulating different elements in the communication system, including physical space environment, user Internet behavior, signal transmission, network element functions and other types of models.

[0203] The knowledge base contains simulation knowledge information required for each simulation stage of the communication simulation network.

[0204] The global knowledge base is composed of theoretical knowledge such as the business principles corresponding to the simulated intelligent agent and the communication and interaction methods of various elements in the network system, as well as empirical knowledge such as the manuals of real equipment in the real network. It is the source of knowledge for the intent processing entity, the intelligent agent control entity to perform intent alignment, the simulation intelligent agent control orchestration and the simulation operation process during the simulation phase, and provides feature vector input for simulation parameter configuration during the simulation system update and iteration phase.

[0205] The memory is used to store at least one of the following: target simulation intent and target, agent permission parameters and attribute parameters, interface attribute parameters, control parameters and monitoring parameters of the communication network simulation system, simulation results and their satisfaction scores.

[0206] The global memory bank includes short-term memory and long-term memory. The short-term memory is implemented in the same way as the temporary memory in AI agent technology. The long-term memory consists of the agent simulation process, agent connection topology, and user feedback for each historical simulation task, and serves as the information foundation for updating the agent toolkit. The memory bank contains a simulation agent rating table, which is updated by the intent processing entity based on user satisfaction with previous simulation tasks and the reasonableness of the original results of the participating simulation agents.

[0207] The agent toolkit contains the tools for running each agent and their attribute information.

[0208] The intelligent agent tool library consists of a set of tools that the simulator can call, along with tool attribute information. Multiple tools of the same type exist in the library, managed by a tool attribute information table. This table records the calling priority of each tool, information on the combined use of tools for various simulation tasks, the number of times each tool is called, and other information that determines the calling behavior during the simulation agent's runtime phase.

[0209] This disclosure proposes a communication network simulation system based on large-scale agent modeling technology. This system supports various user simulation intent inputs, including natural language, speech, and images. It allows for the selection of simulation agents and the construction and operation of the simulation network according to user needs. A knowledge base ensures accurate translation of user intents into the network simulation target and facilitates collaborative simulation by the simulation agents. Furthermore, a long-term memory is built based on user feedback of the simulation results, which is used for optimizing and updating a general toolkit for the simulation agents. This achieves self-iterative, high-precision, and transferable simulation systems.

[0210] As described above, the intent processing entity includes: a knowledge base query instance module, a user intent translation module, and a simulator orchestration module.

[0211] The knowledge base query instance module is used to query simulation knowledge information from the knowledge base;

[0212] The user intent translation module is used to obtain the original target simulation intent and target input by the user through a multimodal large language model; and to translate the original target simulation intent and target by combining the simulation knowledge information to obtain a machine-recognizable target simulation intent and target.

[0213] The simulator orchestration module is used to select multiple intelligent agent groups according to the target simulation intent and the target, and to determine the connection relationship of each intelligent agent group.

[0214] In this embodiment, the user can input the original simulation intent and target using a Multimodal Large Language Model (MM-LLM). After obtaining the user's input, the intent processing entity retrieves simulation knowledge information from the knowledge base through the knowledge base query instance module. The user intent translation module translates the user's input into a language understandable by the communication system, namely the target simulation intent and target, using this simulation knowledge information. The intent processing entity then stores the translated target simulation intent and target in a memory bank and passes it to the agent control entity for further processing. Subsequently, the simulator orchestration module can select multiple agent groups based on the target simulation intent and target and determine the connection relationship between each agent group. The specific implementation process is as described above and will not be further elaborated here.

[0215] As described above, the intelligent agent control entity includes: a simulator permission and attribute setting module, an interface attribute configuration module, and a communication simulation network operation control and monitoring configuration module.

[0216] The simulator permission and attribute setting module is used to set the permission parameters and attribute parameters for each of the intelligent agents;

[0217] The interface attribute configuration module is used to configure the interface attribute parameters of each of the intelligent agents;

[0218] The communication simulation network operation control and monitoring configuration module is used to configure the control parameters and monitoring parameters of the communication network simulation system.

[0219] In this embodiment of the disclosure, after selecting multiple intelligent agent groups and determining the connection relationship of each intelligent agent group, the intent processing entity can combine the target simulation intent and the target to complete the processing of information such as intelligent agent parameter configuration, business process, network operation mode and simulation target, and transmit the processed full information to the memory and intelligent agent control entity, and transmit the connection relationship of each intelligent agent group to the simulated intelligent agent.

[0220] Here, the parameter configuration information is denoted as the parameter configuration, business process, network operation mode, and simulation target of the intelligent agent described above. After receiving this parameter configuration information, the intelligent agent control entity can use the simulator permission and attribute setting module and the interface attribute configuration module, combined with the received parameter configuration information and the corresponding description in the knowledge base, to sequentially complete the intelligent agent permission and attribute configuration and the interface attribute configuration, and then send the configured simulation operation parameters to the simulated intelligent agent. The communication simulation network operation control and monitoring configuration module in the intelligent agent control entity can complete the configuration of communication simulation network operation control and monitoring parameters.

[0221] In this embodiment of the disclosure, the intent processing entity includes: a simulation result processing module and an intent alignment instance module.

[0222] The simulation result processing module is used to perform statistical analysis on the simulation results using a knowledge base and a tool library to obtain statistical analysis results.

[0223] The intent alignment instance module is used to push the statistical analysis results to the user and obtain the user's satisfaction score for the statistical analysis results, thereby obtaining the user's satisfaction score for the simulation results of the main intelligent agent group.

[0224] In this embodiment, for each communication simulation network, the simulation results can be evaluated for their rationality using the simulation effect analysis and evaluation module within the intelligent agent control entity, resulting in a rationality evaluation result. If the rationality evaluation is passed, the simulation results for each communication simulation network built by the intelligent agent group can be statistically analyzed using the simulation result processing module, yielding a statistical analysis result. This statistical analysis result is then pushed to the user via the intent alignment instance module, allowing the user to obtain a satisfaction rating. If the user is satisfied with the simulation result, the simulation is completed. If the user is dissatisfied, the simulation intent and objective are adjusted based on user feedback, and the intent transmission phase and network organization and operation phase are repeated until a satisfactory simulation result is obtained.

[0225] In one alternative implementation, the entity to be processed includes:

[0226] The system update determination module is used to periodically query the historical satisfaction scores of the communication simulation network that ran at a historical time in the memory bank; and if it is determined that the historical satisfaction score does not meet the satisfaction requirements, it triggers the update of the tool attribute information of the tools in the tool library and / or the model parameters of the agent.

[0227] In this embodiment of the disclosure, the intent processing entity can periodically query the satisfaction ratings of recently run communication simulation networks from the memory. Specifically, if the system update determination module determines, based on the satisfaction ratings, that the user's satisfaction with the communication network simulation system is poor, it sends a simulation system update start command to the intelligent agent control entity. This command then triggers updates to the tool attribute information of tools in the tool library and the model parameters of the intelligent agent.

[0228] In one optional implementation, the intelligent agent control entity includes: a simulator update control module, a simulation effect analysis and evaluation module, and a tool update module.

[0229] The simulator update control module is used to update the model parameters of each agent in the simulated intelligent agent;

[0230] The tool update module is used to update the tool attribute information of the tools in the tool library;

[0231] In this embodiment of the disclosure, the intent processing entity can periodically query the satisfaction ratings of recently run communication simulation networks from the memory. Specifically, if the system update determination module determines, based on the satisfaction ratings, that the user's satisfaction with the communication network simulation system is poor, it sends a command to the intelligent agent control entity to initiate a simulation system update. This, in turn, triggers an update of the tool attribute information of tools in the tool library via the tool update module, and triggers an update of the intelligent agent's model parameters via the simulator update control module.

[0232] The following is combined with Figures 5 to 7 The above process will be described.

[0233] like Figure 5 The diagram shown illustrates the intent transmission phase. Figure 5 As shown, the specific process of this intent transmission phase is described below:

[0234] Multimodal large model: Simulation intent input.

[0235] In this embodiment of the disclosure, the user can use a multimodal large language model (MM-LLM) to input the original target simulation intent and target.

[0236] Intent processing entity: Simulated intent translation.

[0237] After obtaining the simulation intent and target input by the user, the intent processing entity can call the simulation knowledge information in the knowledge base, and then use the simulation knowledge information to translate the simulation intent and target input by the user into a language that the communication system can understand, namely the target simulation intent and target.

[0238] Simulated target memory. The intent processing entity can store the translated target simulation intent and target into a memory bank.

[0239] Intelligent agent control entity: Synchronization of simulation intent and goal. The intent processing entity can transmit the simulation intent and goal to the intelligent agent control entity for subsequent processing.

[0240] like Figure 6 The diagram illustrates the preparation and execution phases of the simulation network. The preparation phase establishes the connections and parameters of the simulated agents. The execution phase iterates through multiple rounds of simulation based on user feedback until a satisfactory simulation result is output.

[0241] like Figure 6 As shown, the specific process of the simulation network preparation phase is described below:

[0242] Intent processing entities: Simulated agents are flexibly orchestrated as needed. Here, intent processing entities can combine simulation knowledge in the knowledge base and simulation ratings of each agent in the memory bank to determine the master agent group and the connection relationships of the master agent group; secondly, based on the satisfaction ratings of agent groups that have been running in the memory bank in the past, reserve agent groups are determined and the connection relationships of the reserve agent groups are determined.

[0243] After selecting multiple agent groups and determining the connection relationships between each agent group, the intent processing entity can process information such as the target simulation intent, target completion agent parameter configuration, business process, network operation mode, and simulation target (i.e., parameter configuration information). It then transmits the processed full information to the memory and agent control entity, and transmits the connection relationships of each agent group to the simulation agent. The simulation agent then receives the simulation intent and target, connection relationships, and simulation operation parameters.

[0244] After receiving the parameter configuration information, the intelligent agent control entity sequentially completes the configuration of intelligent agent attributes and permissions, the configuration of intelligent agent interaction interface, and the setting of simulation network operation monitoring parameters, thereby completing the setting of simulation operation parameters and memorizing the simulation operation parameters, that is, storing the configured simulation operation parameters in the global memory.

[0245] like Figure 6 As shown, the specific process of this operational phase is described below:

[0246] The intelligent agent control entity triggers a simulation start command to the simulated intelligent agent, which then performs combined operation. Upon detecting the simulation start command, the simulated intelligent agent can configure each agent according to simulation parameters and connect the agents in the group according to their connection relationships to obtain a communication simulation network. The simulation results are obtained by running the communication simulation network using a target tool from the tool library.

[0247] Result storage: The simulation agent stores the simulation results and process parameters into a global memory.

[0248] Simulation run raw result output: The simulation agent outputs the simulation results to the agent control entity.

[0249] Simulation Result Reasonableness Analysis: The intelligent agent control entity evaluates the reasonableness of the results from aspects such as completeness and usability. If the result fails the reasonableness evaluation, a backup intelligent agent group is used to run the communication simulation network again. If it still fails the reasonableness evaluation, the initial network parameters are adjusted. If the reasonableness evaluation passes, the intent processing entity summarizes and generates the simulation results, thereby obtaining the statistical analysis results of the simulation results. This is presented to the user through MMLLM, and user responses and feedback are collected through a multimodal large model to obtain user feedback on the simulation results. If the user is satisfied with the simulation results, the simulation ends, completing this simulation. If the user is not satisfied with the simulation results, the intent processing entity adjusts the simulation intent and objective, and re-translates the adjusted simulation intent and objective, repeating the intent transmission phase and network organization and operation phase until a simulation result satisfactory to the user is obtained.

[0250] like Figure 7 The diagram shown illustrates the iterative optimization phase of a communication network simulation system. Figure 7 As shown, the specific process of this intent phase is described below:

[0251] The intent is to process entity queries for recent user satisfaction ratings. For example, it can query the satisfaction ratings of recently run communication simulation networks from the memory. If the satisfaction ratings indicate that users have poor satisfaction with the communication network simulation system, then an update of the simulation system is initiated by sending an update command to the intelligent agent control entity.

[0252] Upon receiving the command to start the simulation system update, the intelligent agent control entity initiates an update of the intelligent agent tool library. First, it reads simulation records and user satisfaction data to construct a thought chain connecting the target simulation intent and target graph, simulation process parameters, knowledge base, and tool library. Next, it constructs runtime prompts; for example, starting with the target simulation intent and target graph, it builds step-by-step prompts for knowledge filtering, simulation parameter configuration, simulation execution, and tool invocation. Finally, it updates the tool attribute information of the tools in the tool library.

[0253] After completing the update of the agent tool library, the agent control entity initiates the simulation agent update process.

[0254] The simulated agent retrieves simulation records and user feedback (i.e., satisfaction ratings) from the global memory.

[0255] Construct a dataset of positive and negative simulation results. Specifically, user-satisfied and unsatisfactory simulator results can be constructed as positive and negative samples, respectively.

[0256] Read the optimal initial simulator and tool parameters. Here, the target parameters corresponding to the positive or negative samples can be used as the initial parameters of the agent.

[0257] The simulation agent operates and undergoes iterative training. Here, the agent's parameters are fine-tuned initially, using the loss function MSE, with the goal of making the simulation results output by a single agent as close as possible to the user-satisfied original results. After the agent completes parameter fine-tuning based on positive samples, it is fine-tuned again using negative samples until the simulation results are as close as possible to the user-satisfied original results. This yields an agent capable of adapting to multiple sets of parameter inputs, thus completing the update of the simulation agent model parameters.

[0258] As described above, the technical solution disclosed herein forms a simulation system by combining multiple types of intelligent agents and management modules. It can flexibly combine intelligent agents according to the user's simulation intent and objectives to form a communication simulation network. Furthermore, it iteratively generates simulation results based on user feedback and regularly updates the simulation collaboration, thereby improving the accuracy and cross-scenario transferability of the simulation results. Specifically, the technical solution disclosed herein has the following advantages:

[0259] (1) It can achieve decoupling and independence of intelligent agents, and flexibly combine them according to user intentions, which effectively improves the flexibility and multi-scenario versatility of simulation networks compared with traditional solutions.

[0260] (2) Compared with the independent tools and memory interaction methods of traditional intelligent agents, the simulation system’s construction and operation efficiency and overall accuracy are improved by using a globally shared simulation basic resource library, including the design of memory library, knowledge base and tool library.

[0261] (3) An innovative design was made to the composition of the intelligent agent tool library. Compared with traditional intelligent agent tools, it introduces AI atomic capabilities, system-owned tools, and tool attribute information tables. While enriching the implementation methods of simulation intelligent agents and improving simulation accuracy, it effectively promotes the iterative optimization effect of the simulation system.

[0262] (4) Based on the user satisfaction in long-term memory, the multi-agent and tool library are synchronously iterated and updated. While ensuring the flexibility of the system, the simulation accuracy of the simulation system when migrating across scenarios is effectively improved, avoiding the manual cost of adapting to new scenarios and reducing the cost of migration and promotion.

[0263] Corresponding to Figure 1 In addition to the communication network simulation method described in this disclosure, an electronic device 800 is also provided in this embodiment, such as... Figure 8 The diagram shown is a structural schematic of an electronic device 800 provided in an embodiment of this disclosure, including:

[0264] The system includes a processor 81, a memory 82, and a bus 83. The memory 82 stores execution instructions and includes main memory 821 and external memory 822. The main memory 821, also called internal memory, temporarily stores the computational data in the processor 81, as well as data exchanged with external memory such as a hard disk. The processor 81 exchanges data with the external memory 822 through the main memory 821. When the electronic device 800 is running, the processor 81 communicates with the memory 82 through the bus 83, causing the processor 81 to execute the following instructions:

[0265] Obtain the user's target simulation intent and objective;

[0266] Based on the target simulation intent and target, select multiple intelligent agent groups, determine the connection relationship of each intelligent agent group, and configure the simulation operation parameters of each intelligent agent group;

[0267] In response to the simulation start command, the intelligent agent groups are combined and constructed according to the simulation running parameters and the connection relationship to obtain a communication simulation network;

[0268] The communication simulation network is run by calling the target tool in the tool library to obtain simulation results, and the tool attribute information of the tool in the tool library and / or the model parameters of the agent are updated according to the user's satisfaction rating of the simulation results.

[0269] This disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the communication network simulation method described in the above-described method embodiments. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0270] This disclosure also provides a computer program product carrying program code. The program code includes instructions that can be used to execute the steps of the communication network simulation method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0271] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0272] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this disclosure, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division; in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0273] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0274] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0275] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0276] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit it. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this disclosure. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the protection scope of the claims.

Claims

1. A communication network simulation method, characterized in that, include: Obtain the user's target simulation intent and objective; Based on the target simulation intent and target, select multiple intelligent agent groups, determine the connection relationship of each intelligent agent group, and configure the simulation operation parameters of each intelligent agent group; In response to the simulation start command, the intelligent agent groups are combined and constructed according to the simulation running parameters and the connection relationship to obtain a communication simulation network; The communication simulation network is run by calling the target tool in the tool library to obtain simulation results, and the tool attribute information of the tool in the tool library and / or the model parameters of the agent are updated according to the user's satisfaction rating of the simulation results.

2. The method according to claim 1, characterized in that, The plurality of agent groups includes: a master agent group and a backup agent group, and the method further includes: If the simulation results of the main agent group and the backup agent group fail the rationality assessment, the initial network parameters of each agent group are adjusted, and the communication simulation network built by the multiple agent groups is rerun until a simulation result that passes the rationality assessment is obtained; wherein, the initial network parameters include: simulation running parameters and network model parameters.

3. The method according to claim 2, characterized in that, The method further includes: If the rationality assessment of the simulation results of the main intelligent agent group fails, but the rationality assessment of the simulation results of the backup intelligent agent group passes, obtain the user's satisfaction rating for the simulation results of the backup intelligent agent group. If, based on the satisfaction rating, it is determined that the user is not satisfied with the simulation results of the backup agent group, then the target simulation intent and objective are adjusted, and the communication simulation network corresponding to each agent group is rerun based on the adjusted target simulation intent and objective.

4. The method according to claim 2, characterized in that, The method further includes: If the rationality assessment of the main intelligent agent group is passed, obtain the user's satisfaction score for the simulation results of the main intelligent agent group; If, based on the satisfaction score, it is determined that the user is not satisfied with the simulation results of the main agent group, then the target simulation intent and objective are adjusted, and the communication simulation network corresponding to each agent group is rerun based on the adjusted target simulation intent and objective.

5. The method according to claim 1, characterized in that, The acquisition of the user's target simulation intent and objective includes: Obtain the user's original target simulation intent and target input through a multimodal large language model; The original target simulation intent and target are translated by combining simulation knowledge information in the knowledge base to obtain machine-recognizable target simulation intent and target, and the target simulation intent and target are stored in the memory bank.

6. The method according to claim 1, characterized in that, The step of selecting multiple agent groups based on the target simulation intent and the target includes: The main agent group is determined by combining the simulation knowledge in the knowledge base and the simulation scores of each agent in the memory bank; Based on the satisfaction scores of the historically running agent groups in the memory bank, a backup agent group is determined, and the main agent group and the backup agent group are determined as the plurality of agent groups.

7. The method according to claim 6, characterized in that, The process of combining simulation knowledge in the knowledge base and simulation scores of each agent in the memory bank to determine the main agent group includes: The type of intelligent agent that matches the target simulation intent and target is determined by combining the simulation knowledge information in the knowledge base; Based on the memory bank, the target agent with the highest score under the same agent type is determined, and the main agent group is determined based on the target agent.

8. The method according to claim 6, characterized in that, The process of determining candidate agent groups based on satisfaction scores of historically running agent groups in the memory bank includes: The candidate agent group is determined based on the historical operation of the agent group with the highest satisfaction score in the memory bank.

9. The method according to claim 1, characterized in that, The configuration of simulation operation parameters for each of the intelligent agent groups includes: The simulation operation parameters are configured based on parameter configuration information and simulation knowledge in the knowledge base; wherein, the simulation operation parameters include at least one of the following: agent permissions and attribute parameters, interface attribute parameters, and the parameter configuration information is determined according to the target simulation intent and target.

10. The method according to claim 1, characterized in that, The process of running the communication simulation network by calling the target tool in the tool library to obtain simulation results includes: Based on the first type of attribute information of each tool in the tool library, candidate tools that match the target simulation intent and target are determined; Based on the calling priority of the candidate tools, a target tool for running the communication simulation network is determined, and the communication simulation network is run through the target tool to obtain simulation results; wherein, the calling priority is determined by the tool attribute information of each tool.

11. The method according to claim 1, characterized in that, The step of updating the tool attribute information of tools in the tool library and / or the model parameters of the agent based on the user's satisfaction rating of the simulation results includes: Periodically query the memory bank for historical satisfaction scores of the communication simulation network that ran at a historical time. If it is determined that the historical satisfaction score does not meet the satisfaction requirements, the tool attribute information of the tools in the tool library and / or the model parameters of the agent are updated.

12. The method according to claim 11, characterized in that, The triggering of updating the tool attribute information of the tools in the tool library includes: The thought chain is obtained by retrieving the relationship between the target simulation intent and objective of the communication simulation network running at historical time points, the simulation process parameters of the communication simulation network, and the simulation resource information of the communication simulation network from the memory bank; wherein, the simulation resource information includes information related to the communication simulation network in the knowledge base and the tool library; Runtime prompts for each simulation process in constructing this communication simulation network; Based on the thought chain and the running prompt information, update the tool attribute information of the tools in the tool library.

13. The method according to claim 12, characterized in that, The step of updating the tool attribute information of the tools in the tool library based on the thought chain and the running prompt information includes: The communication simulation network is run according to the aforementioned thought chain, and the tool calling process of each agent in the communication simulation network is adjusted through the running prompt information during the operation. Record tool call information during the tool call process, and update the tool attribute information of the tools in the tool library based on the tool call information.

14. The method according to claim 11, characterized in that, The model parameters that trigger the update of the agent include: Read the satisfaction score of each agent from the memory bank, and construct positive and negative samples based on the satisfaction score of the agent; The target parameters corresponding to the positive or negative samples are used as the initial parameters of the agent; wherein, the target parameters include: simulation running parameters and model parameters; The agent with the initial parameters is trained using the positive and negative samples until an agent that meets the satisfaction requirements is obtained.

15. A communication network simulation system, characterized in that, include: An intent processing entity is used to obtain the user's target simulation intent and target, select multiple intelligent agent groups based on the target simulation intent and target, and determine the connection relationship of each intelligent agent group. The intelligent agent control entity is used to configure the simulation operation parameters of each intelligent agent group and update the tool attribute information of the tools in the tool library according to the user's satisfaction rating of the simulation results. The simulation agent is used to respond to the simulation start command, combine and build the various agent groups according to the simulation running parameters and the connection relationship to obtain a communication simulation network, run the communication simulation network by calling the target tool in the tool library, obtain simulation results, and update the model parameters of the agent according to the user's satisfaction rating of the simulation results.

16. The system according to claim 15, characterized in that, The intent processing entity includes: a knowledge base query instance module, a user intent translation module, and a simulator orchestration module; The knowledge base query instance module is used to query simulation knowledge information from the knowledge base; The user intent translation module is used to obtain the original target simulation intent and target input by the user through a multimodal large language model; and to translate the original target simulation intent and target by combining the simulation knowledge information to obtain a machine-recognizable target simulation intent and target. The simulator orchestration module is used to select multiple intelligent agent groups according to the target simulation intent and the target, and to determine the connection relationship of each intelligent agent group.

17. The system according to claim 15, characterized in that, The intent processing entity includes: a simulation result processing module and an intent alignment instance module; The simulation result processing module is used to perform statistical analysis on the simulation results using a knowledge base and a tool library to obtain statistical analysis results. The intent alignment instance module is used to push the statistical analysis results to the user and obtain the user's satisfaction score for the statistical analysis results, thereby obtaining the user's satisfaction score for the simulation results of the main intelligent agent group.

18. The system according to claim 15, characterized in that, The intent processing entity includes: The system update determination module is used to periodically query the historical satisfaction scores of the communication simulation network that ran at a historical time in the memory bank; and if it is determined that the historical satisfaction score does not meet the satisfaction requirements, it triggers the update of the tool attribute information of the tools in the tool library and / or the model parameters of the agent.

19. The system according to claim 15, characterized in that, The intelligent agent control entity includes: a simulator permission and attribute setting module, an interface attribute configuration module, and a communication simulation network operation control and monitoring configuration module; The simulator permission and attribute setting module is used to set the permission parameters and attribute parameters for each of the intelligent agents; The interface attribute configuration module is used to configure the interface attribute parameters of each of the intelligent agents; The communication simulation network operation control and monitoring configuration module is used to configure the control parameters and monitoring parameters of the communication network simulation system.

20. The system according to claim 19, characterized in that, The communication network simulation system also includes: a knowledge base, a memory base, and an agent tool library; The knowledge base contains simulation knowledge information required for each simulation stage of the communication simulation network. The memory is used to store at least one of the following: target simulation intent and target, agent permission parameters and attribute parameters, interface attribute parameters, control parameters and monitoring parameters of the communication network simulation system, simulation results and their satisfaction scores; The agent toolkit contains the tools for running each agent and their tool attribute information.

21. The system according to claim 15, characterized in that, The intelligent agent control entity includes: a simulator update control module, a simulation effect analysis and evaluation module, and a tool update module; The simulator update control module is used to update the model parameters of each agent in the simulated intelligent agent; The tool update module is used to update the tool attribute information of the tools in the tool library; The simulation effect analysis and evaluation module is used to evaluate the rationality of the simulation results and obtain the rationality evaluation results.

22. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the communication network simulation method according to any one of claims 1 to 14.

23. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the communication network simulation method according to any one of claims 1 to 14.

24. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the communication network simulation method according to any one of claims 1 to 14.

Images