Network optimization method and device and related equipment

Abstract

The invention provides a network optimization method and device and related equipment, which are applied to the technical field of communication, and the network optimization method comprises the following steps: obtaining signaling interaction data in a network system and traffic ticket information corresponding to the network system; determining a network to be optimized in the network system according to the signaling interaction data and the traffic ticket information; and carrying out optimization processing on the network to be optimized. Therefore, the embodiment does not need to independently detect the MDT parameter, so that the influence of poor network conditions on the network optimization efficiency is reduced, the network optimization efficiency is improved, and the network optimization effect is enhanced.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a network optimization method, apparatus and related equipment. Background Technology

[0002] With the development of communication technology, communication has become increasingly important in people's lives, and therefore, communication optimization has received more and more attention. However, current network optimization typically relies entirely on the access network equipment and terminals within the network system to collect Minimization Drive Testing (MDT) parameters based on the terminal's location. But when network conditions are poor, MDT parameters may be delayed or even fail to be received and processed correctly, resulting in low efficiency in network optimization. Therefore, current network optimization methods are ineffective. Summary of the Invention

[0003] This application provides a network optimization method, apparatus, and related equipment to address the problem of poor network optimization performance.

[0004] To solve the above problems, this application is implemented as follows: In a first aspect, embodiments of this application provide a network optimization method, including: Acquire signaling interaction data within the network system and traffic call detail records corresponding to the network system; Based on the signaling interaction data and the traffic call detail record information, determine the network within the network system that needs to be optimized; The network to be optimized is then optimized.

[0005] Secondly, embodiments of this application provide a network optimization apparatus, comprising: The acquisition module is used to acquire signaling interaction data within the network system and traffic call detail records corresponding to the network system; The first determining module is used to determine the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information; An optimization processing module is used to optimize the network to be optimized.

[0006] Thirdly, embodiments of this application also provide an electronic device, including: a memory, a processor, and a program stored in the memory and executable on the processor; the processor is configured to read the program in the memory to implement the steps in the method described in the first aspect above.

[0007] Fourthly, embodiments of this application also provide a readable storage medium for storing a program, which, when executed by a processor, implements the steps of the method described in the first aspect above.

[0008] Fifthly, embodiments of this application also provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the method described in the first aspect above.

[0009] In this embodiment, signaling interaction data and traffic call detail records (CDRs) corresponding to the network system are acquired; the network to be optimized within the network system is determined based on the signaling interaction data and the traffic CDRs; and the network to be optimized is then optimized. Thus, the network to be optimized can be determined based on the signaling interaction data and the traffic CDRs, which are information already generated within the network system. Therefore, the network to be optimized can be accurately determined and optimized based on the signaling interaction data and traffic CDRs. Compared to methods that optimize the network by separately detecting MDT parameters, this embodiment does not require separate detection of MDT parameters, thereby reducing the impact of poor network conditions on the efficiency of network optimization, improving the efficiency of network optimization, and enhancing the network optimization effect. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of the structure of a network system to which the embodiments of this application can be applied; Figure 2 This is a flowchart illustrating the network optimization method provided in an embodiment of this application; Figure 3 This is an application scenario diagram of the network optimization method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the network optimization device provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0013] The terms "first," "second," etc., used in the embodiments of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices. Additionally, the use of "and / or" in this application indicates at least one of the connected objects, such as A and / or B and / or C, representing seven possibilities: including A alone, B alone, C alone, and the presence of both A and B, both B and C, both A and C, and the presence of A, B, and C.

[0014] Please see Figure 1 , Figure 1 This is a structural diagram of a network system to which the embodiments of this application can be applied, such as... Figure 1 As shown, it includes a first electronic device 11, a second electronic device 12, and a third electronic device 13.

[0015] The first electronic device 11 and the second electronic device 12 can communicate with each other, the first electronic device 11 and the third electronic device 13 can communicate with each other, and the second electronic device 12 and the third electronic device 13 can also communicate with each other.

[0016] In practical applications, the first electronic device 11 may optionally be a mobile phone, tablet computer, laptop computer, personal digital assistant (PDA), mobile internet device (MID), wearable device, or vehicle-mounted device. For example, the first electronic device 11 may be a user equipment (UE); the second electronic device 12 may be a base station, access and mobility management function (AMF), relay, access point, or other network element, etc. The base station may be a radio access network (RAN) device; and the third electronic device 13 may be a core network access and mobility management function / user plane function (AMF / UPF) module, etc.

[0017] The network optimization method provided in the embodiments of this application will be described below.

[0018] See Figure 2 , Figure 2 This is a flowchart illustrating the network optimization method provided in the embodiments of this application. Figure 2 The network optimization method shown can be executed by a first electronic device, a second electronic device, or a third electronic device; the specific executing entity is not limited here. For ease of explanation, in this embodiment of the invention, the first electronic device is the vehicle terminal (UE), the second electronic device is the base station (RAN), and the third electronic device is the core network AMF / UPF.

[0019] like Figure 2 As shown, network optimization methods may include the following steps: Step 201: Obtain signaling interaction data within the network system and traffic call detail records corresponding to the network system.

[0020] The specific types of signaling interaction data and traffic call detail records are not limited here. Optionally, signaling interaction data may include at least one of the following: signaling interaction data between the vehicle terminal (UE) and the base station (RAN), between the vehicle terminal (UE) and the core network (AMF / UPF), and between the core network (AMF / UPF) and the base station (RAN). For example, signaling interaction data may include at least one of the following: Tracking Area Code (TAC), Evolved UMTSTerrestrial Radio Access Network (E-UTRAN) Cell Global Identifier (ECGI), Base Station Identifier (ID), Packet Data Network Gateway (PGW) ID, Serving Gateway (SGW) ID, Session Start Time, End Time, Duration, Location Update Handover Base Station Event, Received Signal Strength Indication (RSSI), and Signal-to-Noise Ratio (SNR). The above TAC can also be referred to as the TAC Tracking Area Code.

[0021] Among them, traffic call detail records (CDRs) can be referred to as service traffic data, and traffic CDRs can include at least one of the following: the user's Subscription Permanent Identifier (SUPI), Data Network Name (DNN), Access Network Cell Global Identifier (ECGI), Single Network Slice Selection Assistance Information (S-NSSAI), Service ID, uplink and downlink traffic under different Quality of Service (QoS) (QoS Flow ID, QFI), uplink and downlink traffic of vehicle SIM card services according to the time period specified in the service call detail records, accessed DNN, slice and service destination corresponding to ServiceID, and vehicle service data stream.

[0022] Step 202: Based on the signaling interaction data and the traffic call detail record information, determine the network to be optimized within the network system.

[0023] The specific type of network to be optimized is not limited here. Optionally, the network to be optimized may include: a transmission network between a first electronic device and a second electronic device, a transmission network between a first electronic device and a third electronic device, and a transmission network between a second electronic device and a third electronic device. Alternatively, the network to be optimized may also include at least one of the first electronic device, the second electronic device, and the third electronic device.

[0024] It should be noted that the specific method for determining the network to be optimized within the network system based on signaling interaction data and traffic call detail records is not limited here. Optionally, the network in an abnormal state in the network system can be determined based on signaling interaction data and traffic call detail records, and the network in an abnormal state can be determined as the network to be optimized. The aforementioned network in an abnormal state can be understood as: a network whose signaling interaction data exceeds a first preset value and / or whose traffic call detail records exceed a second preset threshold, and thus can be determined as being in an abnormal state.

[0025] Step 203: Optimize the network to be optimized.

[0026] The specific methods for optimizing the network to be optimized are not limited here. Optionally, the network to be optimized can be restarted, the links can be rebuilt, etc., to improve the communication performance of the network.

[0027] In this embodiment, steps 201 to 203 are used to obtain signaling interaction data and traffic call detail records (CDRs) corresponding to the network system. The network to be optimized within the network system is determined based on the signaling interaction data and the CDRs. The network to be optimized is then optimized. Thus, the network to be optimized can be determined based on the signaling interaction data and the corresponding CDRs. Since the signaling interaction data and CDRs are information already generated within the network system, the network to be optimized can be accurately determined and optimized based on them. Compared to methods that optimize the network by separately detecting MDT parameters, this embodiment does not require separate detection of MDT parameters, thereby reducing the impact of poor network conditions on the efficiency of network optimization, improving the efficiency of network optimization, and enhancing the network optimization effect.

[0028] It should be noted that the embodiments of this application represent a "zero-perception, zero-overhead" method for vehicle-to-everything (V2X) network quality detection and optimization. It completely eliminates reliance on the RPC signaling interaction between base stations and terminals in traditional MDT (Minimum Drive Testing), thus solving the problem of MDT report delays or loss due to network congestion in related technologies. Simultaneously, by integrating signaling call detail records (CDRs), passive detection data (such as Signal-to-Interference plus Noise Ratio (SINR), Reference Signal-Receiving Power (RSRP), and Reference Signal-Receiving Quality (RSRQ)), and multimodal data (signaling plane, control plane, and traffic), and combining machine learning (such as Long Short-Term Memory (LSTM), attention mechanisms, and Actor-Critic algorithms, as detailed in later sections), more accurate network quality assessment and dynamic optimization are achieved.

[0029] Its advantages are: minimal dependence on and impact on terminal networks: This application obtains the network status between terminals, base stations and core networks by parsing signaling call detail records and optimizes them without relying on traditional MDT reports, thus minimizing dependence on terminal networks. At the same time, because it does not introduce MDT reports, it avoids additional network overhead caused by the terminal network in transmitting MDT requests and reports, and avoids the adverse effects of network optimization on network quality.

[0030] Widely applicable: This application collects and parses signaling call details through a backend application system, without the need for core network (AMF, UPF) and base station (RAN) cooperation, and can realize the network optimization of 5G vehicle-to-everything (V2X) access network. It can be widely used in all 5G access network equipment, avoiding dependence on a single equipment manufacturer.

[0031] As an optional implementation, determining the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information includes: Obtain network communication event information within the network system, wherein the network communication event information is used to represent the corresponding communication quality information within the network system during network communication; Calculate network quality parameters based on the signaling interaction data, the traffic call detail records, and the network communication event information; Based on the network quality parameters, determine the networks within the network system that need to be optimized.

[0032] The network communication event information may include at least one of the following: QoS of various network communication protocols, network latency, DL / UL transmission rate, etc.

[0033] The network quality parameters may include at least one of the following: Signal to Interference plus Noise Ratio (SINR), Reference Signal Receiving Power (RSRP), and Reference Signal Receiving Quality (RSRQ).

[0034] Where SINR = 10 * lg(effective signal power / (noise power + interference information power)), RSRP = 10 * lg(signal power / 1 milliwatt), and RSRQ = 10 * lg((number of physical resource blocks (PRB) * signal power / total E-UTRA carrier power)).

[0035] The specific method for calculating network quality parameters based on the signaling interaction data, traffic detail records (CDRs), and network communication event information is not limited here. Optionally, the signaling interaction data, traffic detail records (CDRs), and network communication event information can be obtained, and a weighted sum of these three information types can be calculated. This weighted sum can then be used as the network quality parameter. It should be noted that when the signaling interaction data, traffic detail records, and network communication event information all include multiple types of information, a first weighted sum of the multiple types of information included in the signaling interaction data, a second weighted sum of the multiple types of information included in the traffic detail records, and a third weighted sum of the multiple types of information included in the network communication event information can be calculated separately. Then, a weighted sum of the first, second, and third weighted sums can be calculated, and the final weighted sum can be used as the network quality parameter.

[0036] In this embodiment of the application, network quality parameters are calculated based on the signaling interaction data, the traffic call detail record information, and the network communication event information. Based on the network quality parameters, the network to be optimized within the network system is determined, which can make the determination of the network to be optimized more accurate.

[0037] It should be noted that the network quality parameters of each network in the network system can be calculated based on signaling interaction data, traffic bill information, and network communication event information, and then it can be determined whether the network needs to be optimized according to the network quality parameters of each network. For example, when SINR > 20dB, RSRP > -80dBm, and RSRQ > -10dB, it indicates that the current network quality is excellent; correspondingly, 10dB < SIN < 20dB, -90dBm < RSRP < -80dBm, and -15dB < RSRQ < -10dB indicate that the network quality is good; 0dB < SINR < 10dB, -100dBm < RSRP < -90dBm, and -20dB < RSRQ < -15dB indicate that the network quality is very poor and needs to be optimized; SINR < 0dB, RSRP < -100dBm, and RSRQ < -20dB indicate that the network quality is completely unavailable and a cell switch needs to be performed to achieve the effect of optimizing the network.

[0038] It should be noted that when the signaling interaction data and traffic bill information are obtained, at least one of the signaling interaction data and traffic bill information can also be corrected, and the correction can include at least one of the following: data cleaning, data formatting, time series filling, data aggregation, etc.

[0039] Among them, the above data cleaning can be understood as: removing invalid data (including outliers and duplicates); the above data formatting can also be referred to as data standardization processing, and data standardization processing can include: unifying time formats, geographical location formats, etc.; the above time series filling can be understood as filling in the missing data through time series prediction; the above data aggregation can be understood as: aggregating the data of the same vehicle terminal or area.

[0040] As an optional implementation manner, before determining the network to be optimized in the network system according to the signaling interaction data and the traffic bill information, the method further includes: Determining the target location in the signaling interaction data and the traffic bill information, where the information at the target location is missing; Inputting at least part of the information in the signaling interaction data and at least part of the information in the traffic bill information into a pre-constructed interaction layer respectively to output prediction information; Determining the prediction information as the missing information and filling the missing information into the corresponding target locations in the signaling interaction data and the traffic bill information.

[0041] The target location can be understood as a location present in both signaling interaction data and traffic call detail record (CDR) information, meaning that information is missing in both. Therefore, prediction information can be output by inputting at least a portion of the information from the signaling interaction data and at least a portion of the information from the traffic CDR information into a pre-built interaction layer. The locations corresponding to at least a portion of the information in the signaling interaction data and at least a portion of the information in the traffic CDR information can be understood as having a correlation with the target location. For example, the location corresponding to at least a portion of the information in the signaling interaction data can be understood as the context location of the target location in the signaling interaction data, and the location corresponding to at least a portion of the information in the traffic CDR information can be understood as the context location of the target location in the traffic CDR information. This improves the accuracy of the output prediction information.

[0042] In this embodiment, predicted information can be output through the interaction layer, and the predicted information can be identified as missing information to supplement the missing information in the signaling interaction data and traffic call detail records. This improves the efficiency and accuracy of supplementing the missing information in the signaling interaction data and traffic call detail records.

[0043] It should be noted that the specific type of the above interaction layer is not limited here. Optionally, the interaction layer can adopt a Long Short-Term Memory (LSTM) network model, and the LSTM model can be based on TensorFlow. The Lite architecture allows it to run on various edge nodes except for low-power UPF, resulting in good general applicability. As an optional implementation, the interaction layer includes a forget gate, an input gate, and an output gate. The forget gate and the input gate are connected to the output gate, and the output gate is used to determine at least a portion of the information output by the forget gate and the input gate as the predicted information, and outputs the predicted information. Thus, when the LSTM model includes the forget gate, input gate, and output gate, it can have three gating mechanisms. These three gating mechanisms can interact in a special way, selectively remembering or forgetting information, thereby more effectively capturing data indicators in long-term series. Based on a series of preceding data indicators (i.e., signaling interaction data and traffic call detail records), it can predict and fill in missing data relatively accurately. Since signaling interaction data and traffic call detail records are usually short-term small datasets, and the LSTM model has higher prediction accuracy and better prediction effect for short-term small datasets, it can enhance the prediction effect for missing data.

[0044] It's important to note that the Forget Gate determines which information should be retained and which should be discarded. It takes the current input and the hidden state from the previous time step (which can be understood as missing information) and normalizes it using the Sigmoid activation function. A result close to 0 indicates that the state information will be forgotten, while a result close to 1 indicates that the state information will be retained. The formula is as follows: W f and b f These are the weight matrix and bias vector of the forget gate, h. t-1 ,x t These are the current input and the hidden state at the previous time step, respectively, with σ being a fixed parameter.

[0045] The input gate controls how much of the current input information will be updated into the state. An update ratio is calculated using the sigmoid function, the current input is transformed using a Tanh activation function, and then the transformed input is multiplied by the update ratio to obtain the information that needs to be updated into the state. The formula for calculating the input gate is as follows: W i W C Both are the weight matrices of the input gate, b i and b C is the input gate bias vector, and tanh represents the Tanh activation function.

[0046] The state is updated based on the calculation results of the forget gate and the input gate. t =f t ⊙C t-1 +i t ⊙C t ', where "⊙" represents element-wise multiplication.

[0047] The output gate determines which information from the state will be output as the hidden state at the current time step (i.e., it can be understood as determining at least a portion of the information output from the forget gate and input gate as the predicted information). It calculates an output ratio using the sigmoid function, then multiplies it by the state after processing with the Tanh activation function to obtain the hidden state at the current time step. The formula is: W o b o These are the weight matrix and bias vector of the output gate, respectively.

[0048] It should be noted that, optionally, determining the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information includes: Based on the signaling interaction data and the traffic call detail record information, target processing is performed to determine the network to be optimized within the network system. The target processing includes at least one of the following: Network performance testing, abnormal behavior analysis, correlation mining, anomaly prediction and early warning, and multimodal data fusion analysis.

[0049] Among them, network performance testing includes evaluating indicators such as latency, bandwidth utilization, and packet loss rate; abnormal behavior analysis includes locating problems such as network outages, excessive latency, and signaling anomalies; correlation mining includes exploring the correlation between network performance and vehicle service behavior; anomaly prediction and early warning includes early detection and warning of potential network anomalies; and multimodal data fusion analysis includes integrating network data from different sources (signaling plane, control plane, and service traffic data) for multi-dimensional comprehensive analysis to uncover deeper correlations between network performance and anomalies.

[0050] It should be noted that, in the embodiments of this application, a multi-dimensional fusion model based on an attention mechanism can be used to dynamically weight data from different sources (signaling call detail records, network latency, DL / UL transmission rates, and SINR / RSRP / RSRQ metrics calculated by passive probing), automatically learning the importance of each source data under different network conditions. The multi-dimensional fusion model overcomes the information limitations of a single metric and avoids the problem of traditional weighted averaging's inability to dynamically adjust the contribution of each dimension metric. It can flexibly adjust the network quality contribution of different dimensions, features, and time periods according to specific metric conditions.

[0051] It should be noted that the multi-dimensional fusion attention mechanism allows the querying of key-value pairs from another dimension while calculating a metric for one dimension. When measuring latency impact, it's necessary to "query" QoS-related metrics, expressed as: Attention(Q,K,V) = softmax( V, where softmax is the normalization function, significantly outperforming traditional concatenation or average pooling methods, Q is the query matrix (i.e., latency index matrix), K is the QoS key matrix, V is the QoS value matrix, and d K It represents the dimension of the key vector, which can handle the problem of scale inconsistency between different dimensions.

[0052] As an optional implementation, determining the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information includes: The signaling interaction data and the traffic call detail record information are processed using a policy optimization-effect evaluation learning algorithm to determine the network to be optimized within the network system.

[0053] In this embodiment, signaling interaction data and traffic call detail record information are processed using a policy optimization-effect evaluation learning algorithm to identify networks to be optimized within the network system. This improves the accuracy of identifying networks to be optimized. It should be noted that, optionally, the policy optimization-effect evaluation learning algorithm can be applied to an artificial intelligence (AI) model. By combining it with an AI model, the accuracy of identifying networks to be optimized can be further improved.

[0054] As an optional implementation, the step of processing the signaling interaction data and the traffic call detail record information according to the policy optimization-effect evaluation learning algorithm to determine the network to be optimized within the network system includes: The signaling interaction data and the traffic call detail record information are input into the policy optimization network for processing to obtain the network optimization policy. The policy optimization network uses a policy optimization learning algorithm and is a pre-set network used to generate the network optimization policy. The network optimization strategy is input into the effect evaluation network for time error processing to obtain the time error value. The effect evaluation network uses an effect evaluation learning algorithm and is a pre-set network used to output the time error value. If the time error value is within a preset range, the network to be optimized corresponding to the network optimization strategy is determined.

[0055] Optionally, the policy optimization network and the effect evaluation network can be integrated into different networks in the same model, which can improve the integration of the policy optimization network and the effect evaluation network and improve the efficiency of determining the network to be optimized; alternatively, the policy optimization network and the effect evaluation network can be two separate networks, that is, the policy optimization network and the effect evaluation network do not need to be integrated into the same model.

[0056] It should be noted that the policy optimization network and the performance evaluation network can be pre-trained networks. The training process can be described as follows: Sample data is input into the policy optimization network to be trained for processing to obtain the sample optimization policy. The sample optimization policy is then input into the performance evaluation network to be trained for evaluation. The performance evaluation network outputs the sample scoring error value. After multiple iterations of training with multiple sample data, when the difference between the sample scoring error value evaluated by the performance evaluation network after a certain round of training and the actual error value carried in the sample data input during that round of training is less than a preset difference, it can be determined that the policy optimization network and the performance evaluation network after that round of training have converged. The policy optimization network and the performance evaluation network after that round of training can be identified as the aforementioned policy optimization network and performance evaluation network, respectively.

[0057] In this embodiment, since the strategy optimization network is used to generate the network optimization strategy and the effect evaluation network is used to evaluate the network optimization strategy, the efficiency of the determined network optimization strategy is improved through division of labor and cooperation. At the same time, the accuracy of the network to be optimized corresponding to the determined network optimization strategy is also higher.

[0058] In this algorithm, policy optimization can be referred to as the Actor, and performance evaluation can be referred to as the Critic. Therefore, the policy optimization-performance evaluation learning algorithm can be called the Actor-Critic deep learning algorithm. The core idea is to combine the policy optimization network (Actor) and the performance evaluation network (Critic), where the Actor is responsible for generating the policy, and given a state s, outputs the probability distribution π of the action. θ (a|s); Critic is responsible for evaluating the value of the current policy. Given a state s, it outputs the state value V(s) to estimate the time difference (TD) error, thereby guiding the policy update.

[0059] The objective function for maximizing the cumulative reward in the policy optimization-performance evaluation learning algorithm is defined as follows: ,in Let be the advantage function, representing the relative value of the policy. The Actor uses the TD error adjustment policy calculated by the Critic to increase the probability of actions producing optimization effects. Where α is the learning rate.

[0060] It should be noted that: optionally, the policy optimization-performance evaluation learning algorithm is completed iteratively through the following process: Signaling call detail records (CDRs) are used to collect signaling interaction information between the vehicle terminal (UE) and the base station (RAN) and the core network, as well as network quality information such as QoS, network latency, and DL / UL transmission rate of the current bearer network communication protocol. These information are used as the feature vector of the network state at time t, denoted as [vector]. ; Use in the Actor module As input, the output network optimization scheme A is used, and the network state S at time t+1 is obtained based on action A. t+1 Similarly, the feature vector of the new state is obtained through signaling call detail records. ; Use them respectively in the Critic module , As input, the Q-value output V(S) is obtained. t ), V(S t+1 ); Calculate the TD error (network performance benefit) of action (network optimization scheme) A. ; Using the mean squared error loss function This serves as the gradient update for the Critic module; Following the algorithm described above, the Actor is used as the input state, and the output is a Gaussian distribution to handle continuous actions. The Critic network outputs a state value estimate V(S). t The actor selects action a based on the current strategy. t Examine the efficiency and quality of signaling call detail records to obtain the new state S. t+1 Critic calculates the TD error δ t Update the Critic network parameters. The Actor updates the parameters based on δ. t Adjust the strategy parameters to increase the probability of high-value actions. Repeat this iterative process until the strategy converges to obtain the best real-time optimization solution.

[0061] It should be noted that, optionally, network quality analysis results can also be presented in a visually intuitive format to support network strategies, path optimization, and management. Specifically, this may include at least one of the following: (1) Real-time global network quality monitoring view: The network quality of each cell is calculated based on ECGI. The data is fitted according to the region (province, city) of the cell to provide a real-time updated visual global (national) network quality monitoring view, which displays key network performance indicators and abnormal status.

[0062] (2) Dynamic network performance trend analysis: By analyzing historical data, network performance trend charts are generated to predict future load and generate optimization suggestions.

[0063] (3) Multi-scenario decision-making and scheduling simulation: Construct simulation scenarios and analyze the impact of different network routing or traffic scheduling strategies on network performance through virtual environments.

[0064] (4) Intelligent recommendation for routing strategy decision support: Based on big data analysis results, recommend efficient routing strategies, optimize V2X data transmission, quickly generate recovery plan suggestions in abnormal situations, and automatically generate network optimization and resource scheduling suggestions.

[0065] (5) Interactive visualization analysis: Supports users to dynamically adjust the visualization dimensions and scope to deeply explore the value of data.

[0066] (6) Customized business view: Customize personalized visualization views for different user roles (such as operations, maintenance, and management).

[0067] It should be noted that, based on the above embodiments and the concept of privacy-preserving computation, a federated learning framework is adopted, focusing on the collaborative optimization needs across regions and operators. Model training can be performed at the edge, aggregating network-wide optimization strategies while protecting user privacy and reducing the computational burden on the central node. Lightweight AI algorithms (i.e., policy optimization-effect evaluation learning algorithms) more suitable for edge computing are selected to reduce the demand for computing resources and improve the system's deployability and real-time performance. A dynamic model update mechanism is designed to allow the AI ​​model (i.e., the model applied in the interaction layer) to quickly adjust according to new data and environmental changes during operation, improving the model's adaptability and stability. Edge computing technology can also be used to offload some AI computing tasks to the network edge, reducing the computational burden on central nodes and improving the system's response speed and efficiency.

[0068] To illustrate the above embodiments more clearly, a specific embodiment is given below as an example. For details, please refer to [link / reference needed]. Figure 3 , Figure 3 The diagram shown is an application scenario illustration of an embodiment of this application. For details, please refer to... Figure 3 As shown, Figure 3As shown, the embodiments of this application can be applied to electronic devices, which can be applied in the application scenario of vehicle-to-everything (V2X) communication. Therefore, this electronic device can be called a V2X electronic device. The V2X electronic device may include a V2X data source acquisition module, a V2X data processing and analysis module, and a V2X network protection application module. The V2X data source acquisition module can be used to collect data, such as signaling interaction data within the network system and the corresponding traffic call detail records (CDRs). The V2X data processing and analysis module can analyze the data collected by the V2X data source acquisition module to determine the network to be optimized. This includes multiple functions such as data preprocessing, network quality and anomaly analysis and detection services, policy generation and model training analysis engine, and data security. The V2X network protection application module may include a data visualization and V2X network policy selection support module. For specific functions, please refer to [link to relevant documentation]. Figure 3 As mentioned above, the specifics will not be repeated here.

[0069] See Figure 4 , Figure 4 This is a structural diagram of the network optimization device provided in the embodiments of this application. The network optimization device 400 includes: The acquisition module 401 is used to acquire signaling interaction data within the network system and traffic call detail records corresponding to the network system; The first determining module 402 is used to determine the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information; The optimization processing module 403 is used to perform optimization processing on the network to be optimized.

[0070] As an optional implementation, the first determining module 402 includes: The acquisition submodule is used to acquire network communication event information within the network system. The network communication event information is used to represent the communication quality information within the network system during network communication. The first calculation submodule is used to calculate network quality parameters based on the signaling interaction data, the traffic call detail record information and the network communication event information; The first determining submodule is used to determine the network to be optimized within the network system based on the network quality parameters.

[0071] As an optional implementation, the network optimization device 400 further includes: The second determining module is used to determine the target location in the signaling interaction data and the traffic call detail information, where the target location information is missing. The output module is used to input at least a portion of the information in the signaling interaction data and at least a portion of the information in the traffic call detail record information into a pre-built interaction layer to output prediction information; The third determining module is used to determine the predicted information as the missing information and fill the missing information into the target position in the signaling interaction data and the traffic call detail record information.

[0072] As an optional implementation, the interaction layer includes a forget gate, an input gate, and an output gate. The forget gate and the input gate are respectively connected to the output gate. The output gate is used to determine at least a portion of the information output by the forget gate and the input gate as the predicted information, and output the predicted information.

[0073] As an optional implementation, the first determining module 402 is further configured to process the signaling interaction data and the traffic call detail record information according to the strategy optimization-effect evaluation learning algorithm to determine the network to be optimized within the network system.

[0074] As an optional implementation, the first determining module 402 includes: The second calculation submodule is used to input the signaling interaction data and the traffic call detail record information into the policy optimization network for processing to obtain the network optimization policy. The policy optimization network applies a policy optimization learning algorithm. The policy optimization network is a pre-set network used to generate the network optimization policy. The third calculation submodule is used to input the network optimization strategy into the effect evaluation network for time error processing to obtain the time error value. The effect evaluation network uses an effect evaluation learning algorithm and is a pre-set network used to output the time error value. The second determining submodule is used to determine the network to be optimized corresponding to the network optimization strategy when the time error value is within a preset range.

[0075] The network optimization device 400 can achieve the functions described in the embodiments of this application. Figure 2 The various processes in the method embodiments, and the ways to achieve the same beneficial effects, will not be repeated here to avoid repetition.

[0076] This application also provides an electronic device. Please refer to [link to relevant documentation]. Figure 5 The electronic device may include a processor 501, a memory 502, and a program 5021 stored in the memory 502 and executable on the processor 501. When the program 5021 is executed by the processor 501, it can achieve... Figure 2 Any steps in the corresponding method embodiments and the achievement of the same beneficial effects will not be repeated here.

[0077] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by hardware related to program instructions, and the program can be stored in a readable medium. This application also provides a readable storage medium storing a computer program, which, when executed by a processor, can implement the above-described methods. Figure 2 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0078] The storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0079] This application also provides a computer program product, including computer instructions, which, when executed by a processor, can perform the above-described functions. Figure 2 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0080] The above description represents the preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles described in this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A network optimization method, characterized in that, include: Acquire signaling interaction data within the network system and traffic call detail records corresponding to the network system; Based on the signaling interaction data and the traffic call detail record information, determine the network within the network system that needs to be optimized; The network to be optimized is then optimized.

2. The method according to claim 1, characterized in that, The step of determining the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information includes: Obtain network communication event information within the network system, wherein the network communication event information is used to represent communication quality information within the network system during network communication; Calculate network quality parameters based on the signaling interaction data, the traffic call detail records, and the network communication event information; Based on the network quality parameters, determine the networks within the network system that need to be optimized.

3. The method according to claim 1, characterized in that, Before determining the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information, the method further includes: The target location in the signaling interaction data and the traffic call detail record information is determined, but the information of the target location is missing. At least a portion of the information in the signaling interaction data and at least a portion of the information in the traffic call detail record information are respectively input into a pre-built interaction layer to output prediction information; The predicted information is identified as missing information, and the missing information is filled into the target position in the signaling interaction data and the traffic call detail record information.

4. The method according to claim 3, characterized in that, The interaction layer includes a forget gate, an input gate, and an output gate. The forget gate and the input gate are respectively connected to the output gate. The output gate is used to determine at least a portion of the information output by the forget gate and the input gate as the predicted information, and output the predicted information.

5. The method according to any one of claims 1 to 4, characterized in that, The step of determining the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information includes: The signaling interaction data and the traffic call detail record information are processed using a policy optimization-effect evaluation learning algorithm to determine the network to be optimized within the network system.

6. The method according to claim 5, characterized in that, The step of processing the signaling interaction data and the traffic call detail record information according to the policy optimization-effect evaluation learning algorithm to determine the network to be optimized within the network system includes: The signaling interaction data and the traffic call detail record information are input into the policy optimization network for processing to obtain the network optimization policy. The policy optimization network uses a policy optimization learning algorithm and is a pre-set network used to generate the network optimization policy. The network optimization strategy is input into the effect evaluation network for time error processing to obtain the time error value. The effect evaluation network uses an effect evaluation learning algorithm and is a pre-set network used to output the time error value. If the time error value is within a preset range, the network to be optimized corresponding to the network optimization strategy is determined.

7. A network optimization device, characterized in that, include: The acquisition module is used to acquire signaling interaction data within the network system and traffic call detail records corresponding to the network system; The first determining module is used to determine the network to be optimized within the network system based on the signaling interaction data and the traffic call detail record information; An optimization processing module is used to optimize the network to be optimized.

8. An electronic device, comprising: A memory, a processor, and a program stored in the memory and executable on the processor; characterized in that the processor is configured to read the program from the memory to implement the steps of the network optimization method as described in any one of claims 1 to 6.

9. A readable storage medium for storing a program, characterized in that, When the program is executed by the processor, it implements the steps in the network optimization method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the steps in the network optimization method as described in any one of claims 1 to 6.

Images