Network connection optimization method based on collective base station switching vehicle state guessing algorithm

Abstract

The invention discloses a network connection optimization method based on a collective base station switching vehicle state guessing algorithm, and the method comprises the steps: S1, extracting related historical data of a user before the user initiates a connection request; s2, identifying a terminal group in the same vehicle through an AI model based on user related historical data, and judging the type of the vehicle; s3, predicting a plurality of base station nodes to which the user terminal is about to access, and performing resource allocation preparation in advance; and S4, when a user initiates a connection request, judging whether the user terminal successfully accesses the network from the predicted base station nodes. In the base station management, the switching data of the terminals are acquired in real time, the traffic tools are identified and collective switching behaviors are predicted through data collection and analysis, and resource allocation preparation is carried out in advance, so that delay reduction and switching efficiency improvement are facilitated; therefore, the network connection experience of the mobile terminal in the public transport means is further improved in a new communication environment.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communication algorithm optimization, and in particular to a network connection optimization method based on a collective base station switching vehicle state guessing algorithm. Background Art

[0002] In the old communication network, the base station switching management usually focuses on the connection stability of a single terminal, but in the scenario of large-scale mobile terminals moving across provinces, such as collective switching behaviors in public transportation, network delay problems are usually caused by the forced change of IP addresses. The prior art lacks effective detection and analysis methods for such collective switching behaviors. The current base station switching mechanism is difficult to identify the correlation between terminals. In particular, in the existing communication network, high-speed mobile scenarios (such as buses or subways) appear more and more frequently, and frequent base station switching will further affect the continuity of terminal connections. In addition, the prior art lacks a collective behavior recognition model based on switching frequency, success rate and geographic location data, and it is impossible to determine in real time whether multiple user terminals are in the same vehicle, and thus it is impossible to optimize resource allocation for this collective switching behavior. Therefore, it is necessary to study and design an optimization method that can improve and optimize the network connection effect of user terminals for this scenario. Summary of the invention

[0003] In view of the above problems, the present invention provides a network connection optimization method based on a collective base station switching vehicle state guessing algorithm, comprising the following steps:

[0004] S1, before the user initiates a connection request, relevant historical data of the user is extracted from historical data in the new communication network as the data basis of the guessing algorithm;

[0005] S2, using AI models based on the data features of user-related historical data, identifies the terminal groups in the same vehicle and determines the type of vehicle to which the terminal groups belong;

[0006] S3, predicting several base station nodes that the user terminals belonging to the terminal group are about to access through the prediction model, and making resource allocation preparations in advance based on these predicted base station nodes to provide support for the SDN networking process;

[0007] S4, when the user initiates a connection request, it is determined whether the user terminal successfully accesses the network from these predicted base station nodes;

[0008] If the user terminal successfully accesses these predicted base station nodes, positive feedback is formed to correct the prediction content in step S3;

[0009] Otherwise, return to step S3 to perform base station node prediction and resource allocation preparation again.

[0010] As a further illustration of the present invention, in step S1, the user's access behavior, network switching records and historical data of base station access points are extracted from historical data through regression analysis combined with historical trajectory playback.

[0011] Furthermore, in step S2, the AI ​​model uses cluster analysis to group and classify users' access patterns and mobile paths.

[0012] Furthermore, the AI ​​model determines whether the user terminal belongs to a terminal group in the same vehicle based on data such as the switching frequency, success rate, and geographic coordinates of the base station. It performs cluster analysis through the batch joining or exiting of certain nodes by these user terminals, and infers their traffic pattern and group type based on their number and movement speed.

[0013] Furthermore, the prediction model is a state space model, which predicts the user terminal through a Markov chain to obtain a base station node to which the user terminal may be connected in the future.

[0014] Furthermore, when the user initiates a connection request, the user terminal performs dynamic network negotiation with the control system.

[0015] Furthermore, during the user terminal network connection optimization process, the network status is monitored in real time and the bandwidth is adjusted dynamically.

[0016] Beneficial effects of the present invention:

[0017] The present invention acquires the switching data of these terminals in real time during base station management, and establishes a big data analysis algorithm through data collection and analysis, thereby effectively identifying terminals belonging to the same vehicle. Based on parameters such as switching frequency, switching success rate, and base station geographic coordinates, an AI model is constructed in a new communication network environment, which can predict collective switching behavior in advance, identify base stations to be switched in real time, and prepare resource allocation in advance, which helps to reduce delays. It can significantly reduce network delays of large-scale mobile terminals across provinces and cities in high-speed mobile scenarios, improve switching efficiency, and further improve the network connection experience of mobile terminals in public transportation vehicles in the new communication environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of a network connection optimization method based on a collective base station switching vehicle state guessing algorithm of the present invention. DETAILED DESCRIPTION

[0019] The following is a detailed description of the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0020] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside", "first", "second", etc., indicating directions or positions or sequential relationships, are based on the directions or positions or sequential relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0021] The invention discloses a network connection optimization method based on a collective base station switching vehicle state guessing algorithm, comprising the following steps:

[0022] S1, before the user initiates a connection request, the user's relevant historical data is extracted from the historical data in the new communication network as the data basis for the guessing algorithm. Specifically, the user's access behavior, network switching records, and historical data of base station access points can be extracted from the historical data through regression analysis combined with historical trajectory playback technology. The regression model is optimized by the least squares method to minimize the error, thereby improving the prediction accuracy.

[0023] S2, based on the data features of the user-related historical data, the AI ​​model identifies the terminal group in the same means of transport and determines the type of transport to which the terminal group belongs. This can be achieved using cluster analysis technology, which is a technology that divides data into different groups (clusters) to identify a collection of similar data points. In the present invention, cluster analysis is used to classify user behaviors under different modes of transport in order to select a more appropriate network optimization solution. Specifically, the AI ​​model identifies the terminal group in the same means of transport based on data such as switching frequency, success rate, and base station geographic coordinates, and performs cluster analysis by batch joining or exiting certain nodes of these terminals, and infers their traffic mode and group type by combining their number and moving speed. For example, in a certain period of time, the AI ​​model performs cluster analysis on the user-related historical data, and the results of the data analysis of the switching frequency, success rate, and base station geographic coordinates of 30 user terminals are basically the same. It can be inferred that these 30 user terminals are in the same means of transport, and then combined with the unit time change of these data, the average moving speed is calculated to be 30km / h, and it can be inferred that the type of transport to which the 30 user terminal group belongs is a bus. Based on the same algorithm, it is also possible to analyze and determine other types of transportation (such as subways, high-speed railways, etc.) to which other terminal groups belong based on the data characteristics of other types of terminal groups.

[0024] In a specific implementation, based on the Markov (MDP) reward process, a finite number of state sets S in the MDP are defined, each state represents a certain feature set of the user terminal, such as position, speed, direction, historical access base station information, and other environmental features. These states can represent the current transportation state of the user terminal. Based on historical data, the probability P of the user transferring between different states is calculated.

[0025] P is the state transition probability matrix, where s′ represents the state at the next moment and s represents the current state.

[0026] p(St+1=s′|st=s)

[0027] For example, the probability of moving from a certain position (state s) to another position (s'), combined with the association between the position and the type of transportation (such as bus, subway). Defining the reward value for each state adds a reward function R and a decay coefficient γ, using<S,R,P,γ> express.

[0028] R: represents the expected reward that the state St at a certain moment in the S state can obtain at the next moment (t+1).

[0029] Rs=E[Rt+1|St=s]

[0030] Gt: The harvest Gt is the decaying sum of all rewards from time t onwards in a Markov reward chain.

[0031] Gt=R t+1 +γR t+2 +γ 2 R t+3 +......+γ T+t+1 R T

[0032] γ: Discount factor, in order to avoid state cycles and the system's predictions for the future are not necessarily accurate, so a discount is required.

[0033] V(s): State value function: represents the expectation of receiving Gt from the Markov chain starting from this state.

[0034] v(s)=E[Gt|St=s]

[0035] According to the characteristics of state transition in MDP, such as whether the movement trajectory of the user terminal conforms to the rules of specific transportation tools (such as fixed subway lines, bus stops, etc.), by maximizing the difference in state transition probability and state value function weights, it is judged that the terminal group belongs to a specific type of transportation tool.

[0036] S3, predicting several base station nodes that the user terminals belonging to the terminal group will access through the prediction model, and making resource allocation preparations in advance based on these predicted base station nodes to provide support for the SDN networking process. The changes of the base station can be predicted based on the data characteristics of the user terminal. Specifically, by combining the above algorithm, the system constructs a prediction model based on the state space model. When the data characteristics of a terminal group and its changes are analyzed and identified to be consistent with the known transportation routes, the subsequent changes of these data can be predicted, such as predicting the next two base station nodes that the user terminal will access in advance, providing support for the SDN networking process and reducing network delays caused by inter-provincial and inter-city movement.

[0037] Specifically, the prediction model is a state space model, which predicts the user terminal through a Markov chain to obtain the base station node that the user terminal may connect to in the future. The state space model is a mathematical framework that can describe the changes in time series data. In the present invention, it is used to describe the user's network switching process to predict the future network state and optimization plan. The Markov chain is a random process model, and its next state depends only on the current state. In the present invention, the Markov chain is used to predict the base station node that the user terminal may connect to in the future, thereby optimizing the base station switching process.

[0038] As described above, the future state of the user terminal, i.e., the sequence of base station nodes accessed by the user terminal, is predicted by using the state transition model of MDP. At this time, the state definition is extended to include base station node connection information. Based on the user's current state (such as the base station, location, and direction of movement) and the state transition probability, the set of base station nodes that the user may access in the next several time steps is predicted.

[0039] Based on the predicted future base station node set, the Bellman Equation is introduced

[0040] v(s)=E[Gt|St=s]

[0041] =E[R t+1 +γR t+2 +γ2R t+3 +.......|St=s]

[0042] =E[R t+1 +γγ( t+2 +γR t+3 +......)|St=s]

[0043] =E[R t+1 +γv(S t+1 )|St=s]

[0044] =E[R t+|1St=s]+γγE[v( t+1 )|St=s]

[0045] V(s)=R(s)+γs′∈S∑P(s′∣s)V(s′)

[0046] Through the Bellman equation, the value function v(s) consists of two parts: one is the expectation of the current reward, that is, R(s), and the other is the state expectation at the next moment, that is, the sum of the expected reward and the product of the corresponding state transition probability for the possible state at the next moment, and finally a discount. For example, when the user terminal successfully accesses a base station with high-quality resources, the reward is higher; when accessing a base station with congested or low-quality resources, the reward is lower. By optimizing the reward function, the selection of the state transition strategy is guided.

[0047] The SDN network can allocate sufficient network resources (such as bandwidth and transmission paths) to these nodes in advance to ensure the access quality of the terminal group. Through the prediction results of MDP, resource allocation preparation is carried out in advance to provide support for the SDN networking process; the resource scheduling and path planning between base station nodes in the SDN networking process are optimized, and several base station nodes that user terminals belonging to the terminal group are about to access are predicted. Based on these predicted base station nodes, resource allocation preparation is carried out in advance to provide support for the SDN networking process.

[0048] S4, when the user initiates a connection request, it is determined whether the user terminal successfully accesses the network from these predicted base station nodes;

[0049] If the user terminal successfully accesses these predicted base station nodes, positive feedback is formed to correct the prediction content in step S3 and improve the algorithm accuracy; otherwise, the process returns to step S3 to re-predict the base station nodes and prepare for resource allocation.

[0050] Specifically, when a user initiates a connection request, the user terminal conducts dynamic network negotiation with the control system. The control system caches the negotiated network connection information into the system and records it as user node access information. This information includes the base station node relationship, sequence, cut-in speed, and the status of the user terminal in the heterogeneous network. The system also records the current network status (such as bandwidth, user location information) and negotiation results (such as maximum allowed bandwidth, delay requirements) in the new communication network, and dynamically adjusts resources through adaptive bandwidth control, laying the foundation for the upper-layer predicted SDN networking. Adaptive bandwidth control is a technology that dynamically adjusts bandwidth allocation according to current network conditions to ensure the service quality of high-priority applications. In this system, it helps to adjust bandwidth in real time when user network demand fluctuates to ensure the smoothness of the interactive experience.

[0051] In a preferred embodiment, during the user terminal network connection optimization process, the network status is monitored in real time and the bandwidth is adjusted dynamically. The bandwidth and delay fluctuations can be smoothed by Kalman filtering technology and compared with the last negotiation data. Kalman filtering is a recursive algorithm used to estimate the true state of the system through noise measurement. In the present invention, Kalman filtering is used to smooth the fluctuations of bandwidth and delay to make network status monitoring more accurate and stable. When changes in collective user behavior are detected, the system selects a prediction scheme that meets the current needs from the cached data to predict the upcoming collective base station switching node to ensure the stability and efficiency of the connection in the new communication network.

[0052] Through the above-mentioned network connection optimization method based on the collective base station switching vehicle state guessing algorithm of the present invention, the present invention obtains the switching data of these terminals in real time in base station management, and establishes a big data analysis algorithm through data collection and analysis, so as to effectively identify the terminals belonging to the same vehicle; based on the switching frequency, switching success rate, base station geographic coordinates and other parameters, an AI model is constructed in a new communication network environment, which can predict the collective switching behavior in advance, identify the base station to be switched in real time, and prepare resource allocation in advance, which helps to reduce delays, and can significantly reduce the network delay of large-scale mobile terminals across provinces and cities in high-speed mobile scenarios, improve switching efficiency, and further improve the network connection experience of mobile terminals in public transportation vehicles under the new communication environment.

[0053] The above only describes the preferred embodiments of the present invention, but it cannot be understood as limiting the claims. The present invention is not limited to the above embodiments, and its specific structure is allowed to be changed. In short, all changes made within the protection scope of the independent claims of the present invention are within the protection scope of the present invention.

Claims

1. A network connection optimization method based on a collective base station switching vehicle state guessing algorithm, characterized in that: The following steps are involved: S1, before the user initiates a connection request, relevant historical data of the user is extracted from historical data in the new communication network as the data basis of the guessing algorithm; S2, using AI models based on the data features of user-related historical data, identifies the terminal groups in the same vehicle and determines the type of vehicle to which the terminal groups belong; S3, predicting several base station nodes that the user terminals belonging to the terminal group are about to access through the prediction model, and making resource allocation preparations in advance based on these predicted base station nodes to provide support for the SDN networking process; S4, when the user initiates a connection request, it is determined whether the user terminal successfully accesses the network from these predicted base station nodes; If the user terminal successfully accesses these predicted base station nodes, positive feedback is formed to correct the prediction content in step S3; Otherwise, return to step S3 to perform base station node prediction and resource allocation preparation again.

2. The network connection optimization method based on the collective base station switching vehicle state guessing algorithm according to claim 1 is characterized by: In step S1, the user's access behavior, network switching records and historical data of base station access points are extracted from historical data through regression analysis combined with historical trajectory playback.

3. The network connection optimization method based on the collective base station switching vehicle state guessing algorithm according to claim 1 is characterized by: In step S2, the AI ​​model uses cluster analysis to group and classify users' access patterns and mobile paths.

4. The network connection optimization method based on the collective base station switching vehicle state guessing algorithm according to claim 3 is characterized by: The AI ​​model determines whether the user terminal belongs to a terminal group in the same vehicle based on data such as the switching frequency, success rate, and base station geographic coordinates of the user terminal. It performs cluster analysis through the batch joining or exiting of certain nodes by these user terminals, and infers their traffic pattern and group type based on their number and movement speed.

5. The network connection optimization method based on the collective base station switching vehicle state guessing algorithm according to claim 1 is characterized by: The prediction model is a state space model, which predicts the user terminal through a Markov chain to obtain a base station node to which the user terminal may be connected in the future.

6. The network connection optimization method based on collective base station switching vehicle state guessing algorithm according to claim 1, characterized in that: When a user initiates a connection request, the user terminal performs dynamic network negotiation with the control system.

7. The network connection optimization method based on collective base station switching vehicle state guessing algorithm according to claim 1, characterized in that: During the user terminal network connection optimization process, the network status is monitored in real time and the bandwidth is adjusted dynamically.

Images