WO2020211833A1 - 一种基于机器学习的ap自适应优化选择方法 - Google Patents
一种基于机器学习的ap自适应优化选择方法 Download PDFInfo
- Publication number
- WO2020211833A1 WO2020211833A1 PCT/CN2020/085257 CN2020085257W WO2020211833A1 WO 2020211833 A1 WO2020211833 A1 WO 2020211833A1 CN 2020085257 W CN2020085257 W CN 2020085257W WO 2020211833 A1 WO2020211833 A1 WO 2020211833A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- tree
- node
- decision tree
- data set
- class
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Definitions
- the invention relates to the field of communication technology and car networking adaptive switching, in particular to an AP adaptive optimization selection method based on machine learning.
- WiFi wireless local area network
- WiFiAPs Access Points
- WiFi connections are still not satisfactory: According to a measurement study of more than 5 million users using WiFi networks in urban areas, as many as 45% of mobile devices cannot establish a WiFi connection with the corresponding AP. 15% (5%) of WiFi connections cost more than 5 seconds (10 seconds) to establish a connection.
- Previous measurement studies on WiFi networks have focused on general user experience metrics (for example, bandwidth and delay experienced in WiFi networks), and little attention has been paid to the performance of the WiFi connection establishment process.
- Collecting data from Android smartphones in a controlled environment it is found that the time cost of establishing a large number of connections is mainly caused by the loss of DHCP packets. In fact, the performance of the outdoor WiFi connection establishment process is still unknown, and a thorough investigation on a larger scale is lacking.
- connection establishment and adaptive handover process There are few researches on connection establishment and adaptive handover process at present. Most of the current research involving the connection establishment process is about WiFi handover mechanisms: that is, aimed at reducing handover delays.
- Various solutions have emerged in the prior art to alleviate these problems to receive information about the upcoming connection loss, such as the loss prediction of the device itself or the appropriate intervention of the player. But the key element of all anticipation strategies is the long-term prediction of appropriate channel conditions, whose time scale is much larger than small-scale fading. For prediction, most existing methods are based on specific channel models or extensive and detailed channel maps. Obviously, these two methods may not be enough to ensure data flow requirements.
- the main purpose of the present invention is to solve the problems in the prior art and provide a method for AP adaptive selection and optimization based on machine learning.
- the specific technical solutions are as follows:
- An AP adaptive optimization selection method based on machine learning includes the steps:
- Step 1 Collect data of connected devices in the current environment, establish a training data set, feature set, and determine a threshold;
- Step 2 Confirm whether it is a single-node tree according to the data set and ID3 algorithm
- the decision tree is a single-node tree, and the class C k is taken as the node's Class mark, return to the decision tree; if the feature set is an empty set, the decision tree is a single-node tree, and the class C k with the largest number of instances in the training data set is used as the class mark of the node, and the decision tree is returned; otherwise, press Algorithm ID3 calculates the information gain of each feature in the feature set to the training data set, and selects the feature Ag with the largest information gain; if the information gain of Ag is less than the threshold ⁇ , the decision tree is set as a single-node tree, and the number of instances in the training data set is maximized The class C k of the node is used as the class mark of the node, and the decision tree is returned;
- Step 3 If it is not a single-node tree, divide the subset to construct a sub-node spanning tree;
- Step 4 Call the above S2 and S3 recursively until a complete decision tree is generated to classify APs into a fast set and a slow set, and the AP with the fastest fast set is selected to establish a connection.
- the set of established characteristics includes but is not limited to the time of connection, signal strength, mobile device model, whether it is a public AP, whether it is encrypted, the number of connected devices, and the algorithm used includes but is not limited to Decision tree algorithms such as ID3 and C4.5.
- the steps of the ID3 algorithm to calculate the information gain are as follows:
- Step 2-1 calculate the empirical entropy H(D) of the data set D;
- Step 2-2 Calculate the empirical conditional entropy H(D
- Step 2-3 calculate the information gain g(D, A);
- the decision tree is composed of nodes and sub-nodes, and the decision tree is returned.
- the number of failed connection attempts is less than 3.6%. 80% of the time cost is only 3 seconds, compared with more than 30 seconds using the baseline algorithm, that is, 80% reduces the connection time cost by 10 times.
- the algorithm of the present invention takes into account the possibility of connection failure events, even if the measured signal strength is the highest on the mobile device.
- the model of the present invention can predict these connection failure events with higher accuracy and prevent mobile devices from connecting to the SLOW set AP.
- Figure 1 is a schematic flow chart of the method of the present invention.
- Figure 2 shows the relative information gain of the feature set of the embodiment.
- Figure 3 is a decision tree model generated by the embodiment.
- An AP adaptive optimization selection method based on machine learning includes the steps:
- Step 1 Collect data of connected devices in the current environment, establish a training data set, a feature set, and determine a threshold.
- the established feature set includes but is not limited to the connection time, signal strength, mobile device model, whether it is a public AP, whether it is encrypted, the number of connected devices, and the algorithm used includes but not limited to ID3, C4 .5 and other decision tree algorithms.
- training data set D training data set D
- feature set A time when connected, signal strength, mobile device model, whether it is a public AP, whether it is encrypted, the number of connected devices, etc.
- threshold ⁇ threshold ⁇
- output decision tree T.
- the present invention uses coordinate axis visualization to display the connection time cost difference of each function. Since there are thousands of different mobile device models and AP models, the embodiment omits the coordinate axis visualization results of the functional mobile device models and AP models. In the embodiment, the relative information gain of some features is representatively selected, see FIG. 2.
- Step 2 Confirm whether it is a single-node tree according to the data set and ID3 algorithm.
- the decision tree is a single-node tree, and the class C k is taken as the node's Class mark, return to the decision tree; if the feature set is an empty set, the decision tree is a single-node tree, and the class C k with the largest number of instances in the training data set is used as the class mark of the node, and the decision tree is returned; otherwise, press Algorithm ID3 calculates the information gain of each feature in the feature set to the training data set, and selects the feature Ag with the largest information gain; if the information gain of Ag is less than the threshold ⁇ , the decision tree is set as a single-node tree, and the number of instances in the training data set is maximized The class C k is used as the class label of the node, and the decision tree is returned.
- step 2 the steps of the ID3 algorithm to calculate the information gain are as follows:
- Step 2-1 calculate the empirical entropy H(D) of the data set D;
- Step 2-2 Calculate the empirical conditional entropy H(D
- Step 2-3 calculate the information gain g(D, A);
- Step 3 If it is not a single-node tree, divide the subset to construct a sub-node spanning tree.
- Step 4 Call the above S2 and S3 recursively until a complete decision tree is generated to classify APs into a fast set and a slow set, and the AP with the fastest fast set is selected to establish a connection.
- the number of failed connection attempts is less than 3.6%. 80% of the time cost is only 3 seconds, compared with more than 30 seconds using the baseline algorithm, that is, 80% reduces the connection time cost by 10 times.
- the algorithm of the present invention takes into account the possibility of connection failure events, even if the measured signal strength is the highest on the mobile device.
- the model of the present invention can predict these connection failure events with higher accuracy, and prevent the mobile device from connecting to the SLOW set AP.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
一种基于机器学习的AP自适应优化选择方法，方法应用于移动设备与AP建立WIFI连接及车联网自适应网络切换的过程中，方法包括：收集当前环境中的连接设备数据，建立训练数据集，特征集，确定阈值；根据数据集及ID3算法确认是否为单结树；若非单结树，则分割子集构建子结点生成树；递归调用，直至生成完整的决策树，以将AP分类为FAST集和SLOW集，选择FAST集中最快的AP建立连接。根据机器学习模型对AP接入点进行选择以缩短连接时间，减少WIFI连接设置时间成本。
Description
本发明涉及通信技术和车联网自适应切换领域，尤其涉及一种基于机器学习的AP自适应优化选择方法。
近年来，由于智能设备的爆炸性增长，无线数据流量呈指数上升趋势。在这些无线网络中，802.11无线局域网(WiFi)已成为当今无线业务的主要部分。在过去十年中，已部署超过10亿个WiFiAP(接入点)以提供无线连接。即使用户使用支持3G/4G蜂窝网络的智能设备，也可以使用今天随处可见的WiFi热点。
然而，WiFi网络中的网络性能和用户体验仍然不尽如人意：根据对城市地区使用WiFi网络的500多万用户的测量研究，多达45％的移动设备无法与相应的AP建立WiFi连接，成功的WiFi连接的15％(5％)连接建立时间成本超过5秒(10秒)。以前关于WiFi网络的测量研究已经集中在一般用户体验度量(例如，WiFi网络中经历的带宽和延迟)上，很少关注WiFi连接建立过程的性能。从受控环境中的安卓智能手机中收集数据，发现大量连接建立时间成本的损失主要是由于DHCP数据包丢失造成的。实际上，室外WiFi连接建立过程的性能仍然未知，并且缺乏更大规模的彻底调查。有很多研究专注于WiFi性能测量,旨在估计某些AP-Client链路的可用吞吐量并探讨AP方面的延迟。但是，基于现阶段的研究，迫切需要注意连接建立时间成本度量，因为高连接失败率已经影响了用户体验。
同时在车联网领域同时存在网络连接切换的问题，现有技术下，移动终端设备在网络内移动时，移动设备将连接到各种AP，切换过程中导致服务质量的显著波动和可能的长连接中断，其中信号功率不足以支持数据速率：这些通常包括各种生活场景如电梯和楼梯，尤其体现在车联网领域中。当用户到达网络盲点时，连接中断。显然，移动终端设备上的数据流将特别受到临时连接损失的影响，并且被用户视为主要问题。
目前关于连接建立及自适应切换过程的研究很少。涉及连接建立过程的大多数当前研究都是关于WiFi切换机制：即旨在减少切换延迟。现有技术中出现了各种解决方案来缓解这些问题，以接收关于即将到来的连接丢失的信息，比如设备本身的丢失预测或对播放器的适当干预。但所有预期策略的关键要素是适当的信道条件的长期预测，其时间尺度远大于小规模衰落。对于预测，大多数现有方法基于特定的信道模型或广泛而详细的信道地图。显然，这两种方法可能不足以确保数据流需求。
发明内容
本发明的主要目的在于解决现有技术中存在的问题，提供一种基于机器学习的AP自适应选择优化方法，具体技术方案如下：
一种基于机器学习的AP自适应优化选择方法，所述方法包括步骤：
步骤1，收集当前环境中的连接设备数据，建立训练数据集，特征集，确定阈值；
步骤2，根据数据集及ID3算法确认是否为单结树；
设有k个类C
k(k＝1,2,3，K)，若训练数据集中所有实例属于同一类C
k，则决策树为单结点树，并将类C
k作为该结点的类标记，返回决策树；若特征集为空集，则决策树为单结点树，并将训练数据集中实例数最大的类C
k作为该结点的类标记，返回决策树；否则，按算法ID3计算特征集中各特征对训练数据集的信息增益，选择信息增益最大的特征Ag；如果Ag的信息增益小于阈值ε，则置决策树为单结点树，并将训练数据集中实例数最大的类C
k作为该节点的类标记，返回决策树；
步骤3，若非单结树，则分割子集构建子结点生成树；
步骤4，递归调用以上S2、S3，直至生成完整的决策树，以将AP分类为fast集和slow集，选择fast集最快的AP建立连接。
进一步地，所述步骤1中，所述建立的特征集中，包括但不仅限于连接时的时间、信号强度、移动设备型号、是否为公共AP、是否加密、连接设备数量，使用算法包括但不限于ID3、C4.5等决策树算法。
进一步地，所述步骤2中，所述ID3算法计算信息增益的步骤如下：
步骤2-1，计算数据集D的经验熵H(D)；
步骤2-2，计算特征A对数据集D的经验条件熵H(D|A)；
步骤2-3，计算信息增益g(D,A)；
g(D,A)＝H(D)-H(D|A)
进一步地，所述步骤3中，具体地，在确认是否为单结树之后对Ag的每一可能值ai，依Ag＝ai将训练数据集分割为若干非空子集Di，将Di中实例数最大的类作为标记，构建子结点，由结点及子结点构成决策树，返回该决策树。
与现有技术相比，本发明在应用过程中，连接尝试失败的次数少于3.6％。80％的时间成本仅为3秒，相比之下使用基线算法超过30秒，即80％的减少10倍连接时间成本。本发明的算法考虑了连接失败事件的可能性很大，即使测量的信号强度在移动设备上也是最高的。本发明的模型可以更高精度地预测这些连接故障事件，并避免移动设备连接到SLOW集AP。
图1为本发明所述方法的流程示意图。
图2为实施例特征集的相对信息增益。
图3为实施例生成的决策树模型。
下面结合说明书附图对本发明的技术方案做进一步的详细说明。
一种基于机器学习的AP自适应优化选择方法，所述方法包括步骤：
步骤1，收集当前环境中的连接设备数据，建立训练数据集，特征集，确定阈值。
所述步骤1中，所述建立的特征集中，包括但不仅限于连接时的时间、信号强度、移动设备型号、是否为公共AP、是否加密、连接设备数量，使用算法包括但不限于ID3、C4.5等决策树算法。
具体的，输入：训练数据集D，特征集A(连接时的时间、信号强度、移动设备型号、是否为公共AP、是否加密、连接设备数量等)，阈值ε；输出：决策树T。建立数据集时，要了解每个功能如何影响连接时间成本，本发明使用坐标轴可视化来显示每个功能的连接时间成本差异。由于存在数千种不同的移动设备模型和AP模型，实施例省略了功能移动设备模型和AP模型的坐标轴可视化结果。在实施例中，有代表性的选择了一些特征的相对信息增益，参阅图2。
步骤2，根据数据集及ID3算法确认是否为单结树。
设有k个类C
k(k＝1,2,3，K)，若训练数据集中所有实例属于同一类C
k，则决策树为单结点树，并将类C
k作为该结点的类标记，返回决策树；若特征集为空集，则决策树为单结点树，并将训练数据集中实例数最大的类C
k作为该结点的类标记，返回决策树；否则，按算法ID3计算特征集中各特征对训练数据集的信息增益，选择信息增益最大的特征Ag；如果Ag的信息增益小于阈值ε，则置决策树为单结点树，并将训练数据集中实例数最大的类C
k作为该节点的类标记，返回决策树。
所述步骤2中，所述ID3算法计算信息增益的步骤如下：
步骤2-1，计算数据集D的经验熵H(D)；
步骤2-2，计算特征A对数据集D的经验条件熵H(D|A)；
步骤2-3，计算信息增益g(D,A)；
g(D,A)＝H(D)-H(D|A)
步骤3，若非单结树，则分割子集构建子结点生成树。
所述步骤3中，具体地，在确认是否为单结树之后对Ag的每一可能值ai，依Ag＝ai将训练数据集分割为若干非空子集Di，将Di中实例数最大的类作为标记，构建子结点，由结点及子结点构成决策树，返回该决策树。
步骤4，递归调用以上S2、S3，直至生成完整的决策树，以将AP分类为fast集和slow集，选择fast集最快的AP建立连接。
对第i个子结点，以Di为训练集，以A-{Ag}为特征集，递归地调用，得到子树Ti，返回Ti。实施例所生成的决策树模型参阅图3。
与现有技术相比，本发明在应用过程中，连接尝试失败的次数少于3.6％。80％的时间成本仅为3秒，相比之下使用基线算法超过30秒，即80％的减少10倍连接时间成本。本发明的算法考虑了连接失败事件的可能性很大，即使测量的信号强度在移动设备上也是最高的。本发明的模型可以更高精度地预测这些连接故障事件，并避免移动设备连接到SLOW 集AP。
以上所述仅为本发明的较佳实施方式，本发明的保护范围并不以上述实施方式为限，但凡本领域普通技术人员根据本发明所揭示内容所作的等效修饰或变化，皆应纳入权利要求书中记载的保护范围内。
Claims (4)
- 一种基于机器学习的AP自适应优化选择方法，其特征在于：所述方法包括步骤：步骤1，收集当前环境中的连接设备数据，建立训练数据集，特征集，确定阈值；步骤2，根据数据集及ID3算法确认是否为单结树；设有k个类C k(k＝1,2,3，K)，若训练数据集中所有实例属于同一类C k，则决策树为单结点树，并将类C k作为该结点的类标记，返回决策树；若特征集为空集，则决策树为单结点树，并将训练数据集中实例数最大的类C k作为该结点的类标记，返回决策树；否则，按算法ID3计算特征集中各特征对训练数据集的信息增益，选择信息增益最大的特征Ag；如果Ag的信息增益小于阈值ε，则置决策树为单结点树，并将训练数据集中实例数最大的类C k作为该节点的类标记，返回决策树；步骤3，若非单结树，则分割子集构建子结点生成树；步骤4，递归调用以上S2、S3，直至生成完整的决策树，以将AP分类为fast集和slow集，选择fast集最快的AP建立连接。
- 根据权利要求1所述的一种基于机器学习的AP自适应优化选择方法，其特征在于：所述步骤1中，所述建立的特征集中，包括但不仅限于连接时的时间、信号强度、移动设备型号、是否为公共AP、是否加密、连接设备数量，使用算法包括但不限于ID3、C4.5等决策树算法。
- 根据权利要求1所述的一种基于机器学习的AP自适应优化选择方法，其特征在于：所述步骤2中，所述ID3算法计算信息增益的步骤如下：步骤2-1，计算数据集D的经验熵H(D)；
步骤2-2，计算特征A对数据集D的经验条件熵H(D|A)；
步骤2-3，计算信息增益g(D,A)；g(D,A)＝H(D)-H(D|A) - 根据权利要求1所述的一种基于机器学习的AP自适应优化选择方法，其特征在于：所述步骤3中，具体地，在确认是否为单结树之后对Ag的每一可能值ai，依Ag＝ai将训练数据集分割为若干非空子集Di，将Di中实例数最大的类作为标记，构建子结点，由结点及子结点构成决策树，返回该决策树。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910314113.1A CN110009061B (zh) | 2019-04-18 | 2019-04-18 | 一种基于机器学习的ap自适应优化选择方法 |
| CN201910314113.1 | 2019-04-18 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020211833A1 true WO2020211833A1 (zh) | 2020-10-22 |
Family
ID=67172902
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2020/085257 WO2020211833A1 (zh) | 2019-04-18 | 2020-04-17 | 一种基于机器学习的ap自适应优化选择方法 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110009061B (zh) |
| WO (1) | WO2020211833A1 (zh) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110009061B (zh) * | 2019-04-18 | 2022-08-05 | 南京邮电大学 | 一种基于机器学习的ap自适应优化选择方法 |
| CN111008646A (zh) * | 2019-11-06 | 2020-04-14 | 国网浙江武义县供电有限公司 | 基于设备使用情况的人机关系验证方法和装置 |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105335752A (zh) * | 2015-09-18 | 2016-02-17 | 国网山东省电力公司菏泽供电公司 | 一种基于主成分分析多变量决策树的接线方式识别方法 |
| CN106257872A (zh) * | 2015-06-22 | 2016-12-28 | 埃森哲环球服务有限公司 | Wi‑fi接入点性能管理 |
| CN106792992A (zh) * | 2016-12-12 | 2017-05-31 | 上海掌门科技有限公司 | 一种用于提供无线接入点信息的方法与设备 |
| US9674094B1 (en) * | 2014-03-13 | 2017-06-06 | Marvell International Ltd. | Method and apparatus for controlling transmission of packets over a link aggregation group |
| CN109068350A (zh) * | 2018-08-15 | 2018-12-21 | 西安电子科技大学 | 一种无线异构网络的终端自主选网系统及方法 |
| CN110009061A (zh) * | 2019-04-18 | 2019-07-12 | 南京邮电大学 | 一种基于机器学习的ap自适应优化选择方法 |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6226701B2 (ja) * | 2013-11-05 | 2017-11-08 | キヤノン株式会社 | データ処理方法及び装置、データ識別方法及び装置、プログラム |
| CN106096748A (zh) * | 2016-04-28 | 2016-11-09 | 武汉宝钢华中贸易有限公司 | 基于聚类分析和决策树算法的装车工时预测模型 |
| CN105978760B (zh) * | 2016-06-27 | 2019-07-19 | 桂林电子科技大学 | 一种基于C4.5决策树的VoIP流量在线识别方法 |
-
2019
- 2019-04-18 CN CN201910314113.1A patent/CN110009061B/zh active Active
-
2020
- 2020-04-17 WO PCT/CN2020/085257 patent/WO2020211833A1/zh active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9674094B1 (en) * | 2014-03-13 | 2017-06-06 | Marvell International Ltd. | Method and apparatus for controlling transmission of packets over a link aggregation group |
| CN106257872A (zh) * | 2015-06-22 | 2016-12-28 | 埃森哲环球服务有限公司 | Wi‑fi接入点性能管理 |
| CN105335752A (zh) * | 2015-09-18 | 2016-02-17 | 国网山东省电力公司菏泽供电公司 | 一种基于主成分分析多变量决策树的接线方式识别方法 |
| CN106792992A (zh) * | 2016-12-12 | 2017-05-31 | 上海掌门科技有限公司 | 一种用于提供无线接入点信息的方法与设备 |
| CN109068350A (zh) * | 2018-08-15 | 2018-12-21 | 西安电子科技大学 | 一种无线异构网络的终端自主选网系统及方法 |
| CN110009061A (zh) * | 2019-04-18 | 2019-07-12 | 南京邮电大学 | 一种基于机器学习的ap自适应优化选择方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110009061B (zh) | 2022-08-05 |
| CN110009061A (zh) | 2019-07-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| El Helou et al. | A network-assisted approach for RAT selection in heterogeneous cellular networks | |
| CN113261244B (zh) | 联合mec主机和upf选择的网络节点 | |
| CN111614563A (zh) | 一种用户面路径的选择方法及装置 | |
| CN106257942A (zh) | 在网络环境中使用用户设备组协助服务移出的系统和方法 | |
| CN103959850A (zh) | 用于无线宽域网和无线局域网的互通的测量 | |
| WO2020211833A1 (zh) | 一种基于机器学习的ap自适应优化选择方法 | |
| US11855846B2 (en) | Communication method and communication apparatus | |
| CN103918328A (zh) | 寻呼期间小区数量的减少 | |
| WO2019129300A1 (zh) | 缓存决策方法及装置 | |
| EP3682663B1 (en) | Global optimization process for link associations | |
| CN101646157B (zh) | 池信息管理方法及其设备 | |
| CN105532012A (zh) | 视频分发方法、设备和系统 | |
| US11895543B2 (en) | MRO for 5G networks | |
| WO2012095057A2 (zh) | 拥塞控制方法及设备、系统 | |
| CN111669785B (zh) | 一种基于校园网的异构无线网络的移动数据分流方法 | |
| WO2014181536A1 (ja) | 通信システム、基地局、及び通信方法 | |
| KR101086033B1 (ko) | 네트워크 관리 장치 및 그의 이동국 핸드오버 제어 방법 | |
| Mazza et al. | A user-satisfaction based offloading technique for smart city applications | |
| CN105430700A (zh) | 基于多虚拟接入点关联接入的无线局域网移动性管理方法 | |
| LU102762A1 (en) | Machine learning-based and adaptive optimization method of ap selection | |
| CN101621783A (zh) | 避免冗余位置更新的方法、系统、用户设备及网络 | |
| KR101686186B1 (ko) | 핸드오버 대기시간 감소를 위한 핸드오버 수행방법 및 모바일 무선 네트워크 시스템 | |
| WO2019134174A1 (zh) | 一种业务传输方法及装置、计算机存储介质 | |
| CN104254091A (zh) | 一种空口信道状态参数上报方法和设备 | |
| CN111095990A (zh) | 选择性改变无线接入点的激活状态 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20791293 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 20791293 Country of ref document: EP Kind code of ref document: A1 |