CN-121985378-A - WIFI6 self-adaptive load balancing method and system under multiple scenes

CN121985378ACN 121985378 ACN121985378 ACN 121985378ACN-121985378-A

Abstract

The invention provides a WIFI6 self-adaptive load balancing method and system under multiple scenes, and relates to the technical field of wireless communication networks, wherein the method comprises the steps of obtaining a network scene data sequence and a network running state data sequence of a target wireless network in a preset acquisition window; the method comprises the steps of calling a load balancing scene prototype library to identify a scene prototype, determining a target load balancing scene prototype, carrying out load state analysis, determining a load state feature set of a plurality of Access Points (AP) and an access state feature set of a plurality of terminal (STA), carrying out balancing action analysis, obtaining a target balancing action, and sending the target balancing action to a controller to carry out WIFI6 load balancing control. The invention solves the technical problems that the prior art is mainly aimed at a certain specific scene, can effectively work in the specific scene, but cannot adapt to more diversified environments, so that the resource waste or network congestion is caused, and the overall performance is reduced.

Inventors

XIE TIEMIN
ZHOU HE
JIN GUOHUA
ZHANG CHEN

Assignees

南京大洋通信系统有限公司

Dates

Publication Date: 20260505
Application Date: 20260327

Claims (10)

1. The WIFI6 self-adaptive load balancing method under multiple scenes is characterized by comprising the following steps: Acquiring a network scene data sequence and a network running state data sequence of a target wireless network in a preset acquisition window, wherein the target wireless network comprises a plurality of Access Points (AP), a plurality of terminal STAs and a controller; invoking a load balancing scene prototype library to perform scene prototype recognition on the network scene data sequence, and determining a target load balancing scene prototype, wherein each load balancing scene prototype in the load balancing scene prototype library is constructed based on a deep learning network algorithm; Load state analysis is carried out based on the network operation state data sequence, and a load state feature set of a plurality of Access Points (AP) and an access state feature set of a plurality of terminals (STA) are determined; and carrying out balance action analysis on the load state feature set and the access state feature set by using the load balance scene prototype library to obtain a target balance action, and sending the target balance action to a controller to carry out WIFI6 load balance control.
2. The WIFI6 adaptive load balancing method according to claim 1, wherein each of the plurality of access points AP operates in at least one of a 2.4GHz band, a 5GHz band, and a 6GHz band.
3. The WIFI6 adaptive load balancing method under multiple scenarios of claim 1, wherein invoking a load balancing scenario prototype library to perform scenario prototype recognition on the network scenario data sequence, determining a target load balancing scenario prototype, comprises: acquiring a plurality of sample network scene data sequences, a plurality of sample network running state data sequences, a plurality of corresponding sample equalization actions and a plurality of sample equalization effects; performing scene feature recognition based on the plurality of sample network scene data sequences, and determining a plurality of sample network scene features; Performing multidimensional clustering on the plurality of sample network scene features from user mobility distribution, service type proportion, AP overlapping density, uplink and downlink flow modes, low-power consumption periodic wake-up rhythm and roaming frequency, and determining a plurality of clustered sample network scene feature sets; According to the plurality of clustered sample network scene feature sets, mapping and dividing a plurality of sample network running state data sequences, a plurality of corresponding sample equalization actions and a plurality of sample equalization effects to obtain a plurality of clustered sample network running state data sequence sets, a plurality of clustered sample equalization action sets and a plurality of clustered sample equalization effect sets; and respectively constructing a load balancing scene prototype according to the plurality of clustering sample network scene feature sets, the plurality of clustering sample network running state data sequence sets, the plurality of clustering sample balancing action sets and the plurality of clustering sample balancing effect sets to obtain a load balancing scene prototype library.
4. The WIFI6 adaptive load balancing method under multiple scenarios according to claim 3, wherein performing load balancing scenario prototype construction according to a plurality of clustered sample network scenario feature sets, a plurality of clustered sample network running state data sequence sets, a plurality of clustered sample balancing action sets, and a plurality of clustered sample balancing effect sets, respectively, to obtain a load balancing scenario prototype library, includes: traversing the plurality of clustered sample network scene feature sets to execute balanced drift search and determine a plurality of central clustered sample network scene features; Dividing the plurality of clustered sample equalizing effect sets according to a preset sample equalizing effect threshold value to obtain a plurality of clustered positive sample equalizing effect sets and a plurality of clustered negative sample equalizing effect sets; Training by combining the plurality of clustered positive sample equalization effect sets, the plurality of clustered negative sample equalization effect sets and the plurality of clustered sample network running state data sequence sets to obtain a plurality of initial load equalization scene prototypes; And carrying out association identification on a plurality of initial load balancing scene prototypes by utilizing the network scene characteristics of the plurality of central clustering samples, and constructing the load balancing scene prototypes.
5. The WIFI6 adaptive load balancing method under multiple scenarios of claim 4, wherein training is performed in combination with the multiple clustered positive sample balancing effect sets and multiple clustered negative sample balancing effect sets, and multiple clustered sample network running state data sequence sets, to obtain multiple initial load balancing scenario prototypes, comprising: Performing action rewarding identification and action punishment identification on the plurality of clustered sample equalization action sets based on the plurality of clustered positive sample equalization effect sets and the plurality of clustered negative sample equalization effect sets to obtain a plurality of identified clustered sample equalization action sets; Traversing the running state data sequence sets of the plurality of clustered sample networks to analyze states of access points and terminals, and obtaining a plurality of clustered sample access state feature set sets and a plurality of clustered load state feature set sets; And performing positive and negative sample training on a framework constructed based on a deep learning network algorithm according to the action rewarding mark and the action punishment mark by utilizing a plurality of clustering sample access state feature set, a plurality of clustering load state feature set and a plurality of identification clustering sample balance action set until training converges, and obtaining a plurality of initial load balance scene prototypes after training is completed.
6. The WIFI6 adaptive load balancing method according to claim 5, wherein determining the load status feature group of the plurality of access points AP and the access status feature group of the plurality of terminal STAs based on the network operation status data sequence performs load status analysis, includes: Based on the multiple Access Points (AP) and the multiple terminal STAs, extracting data of the network operation state data sequence to obtain multiple Access Points (AP) data sequences and multiple terminal STA data sequences; extracting a first access point AP data sequence from the plurality of access point AP data sequences; Performing feature trend superposition on the first access point AP data sequence, determining a first load state feature, and adding the first load state feature into a load state feature group; And traversing the data sequences of the plurality of terminal STAs to perform feature trend superposition, and determining an access state feature set.
7. The WIFI6 adaptive load balancing method under multiple scenarios of claim 6, wherein performing feature trend stacking on the first access point AP data sequence to determine a first load status feature includes: Extracting multi-scale trend features of the AP data sequence of the first access point to obtain a first multi-scale trend feature set, wherein each multi-scale trend feature comprises a change rate of the associated user number, frequency band load distribution, BSS Coloring conflict degrees and neighbor AP overlapping degrees; Performing repeated random extraction without returning on the first multi-scale trend feature set, and extracting two multi-scale trend features each time to obtain a first multi-scale trend feature extraction combination set; traversing the first multi-scale trend feature extraction combination set to carry out overall analysis, and determining a first target multi-scale trend feature; And extracting first access point AP data positioned at the last position in the first access point AP data sequence, executing data coding, and splicing and fusing the first access point AP data with the first target multiscale trend characteristic to obtain a first load state characteristic.
8. The WIFI6 adaptive load balancing method under multiple scenarios of claim 7, wherein traversing the first multi-scale trend feature extraction combined set for overall resolution, determining a first target multi-scale trend feature, comprises: Traversing each first multi-scale trend feature extraction combination in the first multi-scale trend feature extraction combination set, and performing independent deviation analysis on the change rate of the associated user number, the frequency band load distribution, BSS Coloring conflict degrees and the overlapping degree of neighbor APs to construct a first combination deviation group set; Respectively carrying out normalization processing on each combination deviation group in the first combination deviation group set, and matrixing processing results to construct a first combination interaction matrix set; Performing map convolution operation on two corresponding multi-scale trend features in the first multi-scale trend feature extraction combined set by using the first combined interaction matrix set to obtain the first multi-scale interaction trend feature extraction combined set; and calculating feature average values in the first multi-scale interaction trend feature extraction combination set to obtain a first target multi-scale trend feature.
9. The WIFI6 adaptive load balancing method according to claim 8, wherein each first combination bias group in the first combination bias group set includes a variation rate bias of the number of associated users, a bias of frequency band load distribution, a BSS Coloring collision degree bias, and a bias of overlapping degrees of neighboring APs.
10. A WIFI6 adaptive load balancing system in multiple scenarios, configured to implement a WIFI6 adaptive load balancing method in multiple scenarios according to any one of claims 1-9, the system comprising: The system comprises a data sequence acquisition module, a network operation state acquisition module and a network operation state acquisition module, wherein the data sequence acquisition module is used for acquiring a network scene data sequence and a network operation state data sequence of a target wireless network in a preset acquisition window, and the target wireless network comprises a plurality of Access Points (AP), a plurality of terminal (STA) and a controller; The scene prototype recognition module is used for calling a load balancing scene prototype library to perform scene prototype recognition on the network scene data sequence and determining a target load balancing scene prototype, wherein each load balancing scene prototype in the load balancing scene prototype library is constructed based on a deep learning network algorithm; the load state analysis module is used for carrying out load state analysis based on the network operation state data sequence and determining load state characteristic groups of a plurality of Access Points (AP) and access state characteristic groups of a plurality of terminals (STA); And the load balancing control module is used for carrying out balancing action analysis on the load state characteristic group and the access state characteristic group by using the load balancing scene prototype library to obtain a target balancing action, and sending the target balancing action to the controller for WIFI6 load balancing control.

Description

WIFI6 self-adaptive load balancing method and system under multiple scenes Technical Field The invention relates to the technical field of wireless communication networks, in particular to a WIFI6 self-adaptive load balancing method and system under multiple scenes. Background WIFI6 makes a breakthrough in terms of improving throughput, reducing delay and increasing network capacity, and particularly in a high-density environment, WIFI6 can provide more efficient wireless access for more terminals, however, as wireless technology is continuously developed, network environments become more and more complex, scene types become more diversified, and different load balancing strategies are required to optimize the performance of the network for different scenes. Most of the prior art is directed to a certain specific type of scenario, such as a high-density conference, a low-power consumption internet of things device, and the like, and each scenario has specific characteristics and requirements, so that although the prior art can effectively work in the specific scenario, the prior art cannot adapt to more diversified environments, for example, a load balancing strategy for the high-density conference scenario is not suitable for being used in an environment in which the low-power consumption IoT device is concentrated, and vice versa. This results in that the load balancing effect of the network is greatly compromised in the case of multiple alternate scenarios, resulting in resource waste or network congestion, and reduced overall performance. Disclosure of Invention The application provides a WIFI6 self-adaptive load balancing method and system under multiple scenes, and aims to solve the technical problems that the prior art is mainly aimed at a certain type of specific scenes, can effectively work under the specific scenes, cannot adapt to more diversified environments, causes resource waste or network congestion and reduces overall performance. The application discloses a first aspect of WIFI6 self-adaptive load balancing method under multiple scenes, which comprises the steps of obtaining a network scene data sequence and a network operation state data sequence of a target wireless network in a preset acquisition window, wherein the target wireless network comprises a plurality of Access Points (AP), a plurality of terminal STAs and a controller, invoking a load balancing scene prototype library to conduct scene prototype recognition on the network scene data sequence to determine a target load balancing scene prototype, constructing each load balancing scene prototype in the load balancing scene prototype library based on a deep learning network algorithm, conducting load state analysis based on the network operation state data sequence to determine a load state feature set of the plurality of Access Points (AP) and an access state feature set of the plurality of terminal STAs, conducting balance action analysis on the load state feature set and the access state feature set by utilizing the load balancing scene prototype library to obtain a target balance action, and sending the target balance action to the controller to conduct WIFI6 load balancing control. The application discloses a second aspect of the WIFI6 self-adaptive load balancing system under multiple scenes, which is used for the WIFI6 self-adaptive load balancing method under the multiple scenes, and comprises a data sequence acquisition module, a load balancing control module and a scene prototype recognition module, wherein the data sequence acquisition module is used for acquiring a network scene data sequence and a network operation state data sequence of a target wireless network in a preset acquisition window, the target wireless network comprises a plurality of Access Points (AP), a plurality of terminal STAs and a controller, the scene prototype recognition module is used for calling a load balancing scene prototype library to recognize scene prototypes of the network scene data sequence and determining a target load balancing scene prototype, each load balancing scene prototype in the load balancing scene prototype library is constructed based on a deep learning network algorithm, the load state analysis module is used for carrying out load state analysis based on the network operation state data sequence and determining a load state feature set of the Access Points (AP) and an access state feature set of the terminal STAs, and the load balancing control module is used for carrying out analysis on the load state feature set and the access state feature set by the load balancing scene prototype library and sending the target balancing action feature set to the WIFI 6. The one or more technical schemes provided by the application have at least the following beneficial effects: The method comprises the steps of periodically obtaining a data sequence of a target wireless network, monitoring the change of the network environment in rea