Communication system based on neural network model and configuration method thereof

By obtaining child node feature information and dynamically configuring neural network models, the problems of new child node optimization and model sharing in mobile communication systems are solved, and efficient model updates and accuracy improvements are achieved, especially performance improvements in long-term predictions .

CN115004649BActive Publication Date: 2024-11-29NTT DOCOMO INC
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202080093993.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-11-10
Publication Date
2024-11-29
Estimated Expiration
2040-11-10

AI Technical Summary

Technical Problem

In mobile communication systems, newly added child nodes cannot achieve targeted optimization configuration, and it is difficult to achieve optimal processing results during operation. The training data of a single child node is limited, so it is impossible to achieve optimal model optimization and model Sharing, using data from recent moments alone cannot improve the accuracy of neural network models.

Method used

By obtaining the feature information of multiple child nodes, dynamically configure the child node neural network model, use the main node for centralized or distributed updates, considering the characteristics of the child nodes such as height, antenna configuration, coverage area, etc., the neural network model index and model are used to consider the characteristics of the child nodes. Weight and semantic representation, introduce attention mechanism, use lightweight recurrent neural networks and gated recurrent units for training data representation and loss function construction.

Benefits of technology

Realize dynamic configuration and update of the new child node neural network model, improve training efficiency and model accuracy, especially in long-term prediction situations, make full use of the characteristics and training data of child nodes, and expand the availability The training data of the model improves the accuracy of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115004649B_ABST
    Figure CN115004649B_ABST
Patent Text Reader

Abstract

The present invention relates to a communication system based on a neural network model and a configuration method thereof. The communication system comprises at least one main node and a plurality of sub-nodes connected to the main node for communication, and each of the plurality of sub-nodes is configured with a sub-node neural network model, and the communication system configuration method comprises: obtaining feature information of the plurality of sub-nodes; and dynamically configuring the sub-node neural network model based on the acquired feature information.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the fields of mobile communication technology and artificial intelligence technology, and more specifically, to a communication system based on a neural network model and a configuration method thereof. Background Art

[0002] In traditional mobile communication systems, network deployment and operation and maintenance are mainly completed manually, which consumes a lot of human resources and increases operating costs, and the network optimization is not ideal. With the commercialization of the fifth generation of mobile communication technology, the communication system is developing in the direction of network diversification, broadband, integration, and intelligence, so complex tasks such as network optimization, large input data set processing, network recommendation or network element configuration are becoming greater challenges. At the same time, due to the breakthrough progress in big data technology, computing power, and various algorithms and network frameworks in recent years, artificial intelligence technology has also shown explosive growth. At present, artificial intelligence technology is increasingly combined with mobile communication technology. Mobile communication technology provides artificial intelligence technology with the big data throughput and low-latency transmission required for many intelligent application scenarios, while artificial intelligence technology also provides powerful solutions to solve various complex problems in mobile communication technology.

[0003] In a communication system consisting of at least one main node and multiple sub-nodes that are communicatively connected to the main node, a neural network model is configured in the main node and the multiple sub-nodes to perform complex tasks such as network optimization, large input data set processing, network recommendation or network element configuration. When a new sub-node is added to the communication system, the neural network model of the newly added sub-node needs to be initialized. If only the predetermined default settings are used, targeted optimization configuration cannot be achieved. At the same time, during the operation of the communication system, if the configured neural network model is not updated for a specific task, it is difficult to achieve the best processing effect. Furthermore, in the training process for a specific task, if only the local data of a single sub-node is used to perform training, the best model optimization and model sharing between the same or similar sub-nodes cannot be achieved due to limited training data. In addition, if only the data at the most recent moment is used for training, historical training data cannot be used to improve the accuracy of neural network model processing. Summary of the invention

[0004] The present disclosure is proposed in view of the above problems. The present disclosure provides a communication system based on a neural network model and a configuration method thereof.

[0005] According to one aspect of the present disclosure, a communication system configuration method based on a neural network model is provided, wherein the communication system includes at least one main node and multiple sub-nodes communicatively connected to the main node, and each of the multiple sub-nodes is configured with a sub-node neural network model, and the communication system configuration method includes: acquiring feature information of the multiple sub-nodes; and dynamically configuring the sub-node neural network model based on the acquired feature information.

[0006] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the acquiring of the characteristic information of the plurality of sub-nodes includes: receiving the characteristic information transmitted from one of the plurality of sub-nodes.

[0007] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the acquiring of characteristic information of the multiple subnodes includes: receiving initial information transmitted from one of the multiple subnodes; and predicting the characteristic information of the one subnode based on the initial information.

[0008] In addition, according to a communication system configuration method according to one aspect of the present disclosure, dynamically configuring the sub-node neural network model based on the acquired feature information includes: selecting a neural network model from a plurality of predetermined neural network models based on the feature information; and configuring the sub-node neural network model of the sub-node using the selected one neural network model.

[0009] In addition, according to an aspect of the present disclosure, a communication system configuration method, wherein the dynamically configuring the sub-node neural network model based on the acquired feature information includes: selecting a matching sub-node that matches the one sub-node from the multiple sub-nodes based on the feature information; receiving the sub-node neural network model of the matching sub-node from the matching sub-node; and configuring the sub-node neural network model of the one sub-node using the sub-node neural network model of the matching sub-node.

[0010] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the one child node is a child node newly added to the communication system.

[0011] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the acquiring of the characteristic information of the plurality of sub-nodes includes: receiving the characteristic information transmitted from each of the plurality of sub-nodes.

[0012] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the dynamic configuration of the sub-node neural network model based on the acquired feature information includes: dividing the multiple sub-nodes into multiple categories based on the feature information; performing training of the sub-node neural network model for the multiple categories using the feature information to obtain an updated sub-node neural network model; and updating the sub-node neural network models of the multiple sub-nodes using the sub-node neural network model.

[0013] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the dynamic configuration of the sub-node neural network model based on the acquired feature information includes: dividing the multiple sub-nodes into multiple categories based on the feature information; notifying the sub-nodes of the same category of the feature information of the sub-nodes belonging to the same category in the multiple categories; and the sub-nodes of the same category performing training using the feature information of the sub-nodes of the same category to update the sub-node neural network model of the sub-nodes of the same category.

[0014] In addition, according to a communication system configuration method according to one aspect of the present disclosure, the characteristic information includes the height of the subnode, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, and historical configuration information.

[0015] In addition, according to an aspect of the present disclosure, a communication system configuration method, wherein the configuring the sub-node neural network model includes one of the following: establishing an index of multiple neural network models, and using the index to indicate that the sub-node neural network model is one of the multiple neural network models; using the model weight of the neural network model to indicate the sub-node neural network model; using the change in the model weight of the neural network model to indicate the sub-node neural network model; and using the semantic representation of the neural network model to indicate the sub-node neural network model.

[0016] In addition, according to a communication system configuration method in one aspect of the present disclosure, the characteristic information is a historical optimal beam set of a user equipment corresponding to the subnode, and the historical optimal beam set includes: a difference sequence between multiple optimal beams at multiple consecutive time points and an optimal beam at a recent time point; or a difference sequence between optimal beams at two adjacent time points in multiple consecutive time points;

[0017] In addition, according to a communication system configuration method according to one aspect of the present disclosure, using the feature information to update the sub-node neural network model includes: determining the weight of each historical optimal beam using the number of occurrences of each historical optimal beam in the historical optimal beam set; and constructing a weighted loss function based on the weight of each historical optimal beam and the historical optimal beam set to perform training to update the sub-node neural network model.

[0018] In addition, according to an aspect of the present disclosure, a communication system configuration method, wherein updating the sub-node neural network model using the feature information includes: configuring an attention layer in the sub-node neural network model, and performing training using the sub-node neural network model including the attention layer to update the sub-node neural network model.

[0019] According to another aspect of the present disclosure, there is provided a communication system based on a neural network model, comprising: at least one main node; a plurality of sub-nodes, which are communicatively connected to the main node, and each of the plurality of sub-nodes is configured with a sub-node neural network model, wherein the at least one main node obtains feature information of the plurality of sub-nodes; and based on the obtained feature information, the sub-node neural network model is dynamically configured.

[0020] In addition, according to a communication system in another aspect of the present disclosure, the at least one master node receives the feature information transmitted from one of the plurality of subnodes.

[0021] In addition, according to another aspect of the communication system of the present disclosure, the at least one master node receives initial information transmitted from one of the multiple sub-nodes; and predicts the characteristic information of the one sub-node based on the initial information.

[0022] In addition, according to a communication system in another aspect of the present disclosure, the at least one main node selects a neural network model from a plurality of predetermined neural network models based on the feature information; and configures the sub-node neural network model of the sub-node using the selected one neural network model.

[0023] In addition, according to another aspect of the present disclosure, a communication system is provided, wherein the at least one main node selects a matching subnode that matches the one subnode from the multiple subnodes based on the feature information; receives a subnode neural network model of the matching subnode from the matching subnode; and configures the subnode neural network model of the one subnode using the subnode neural network model of the matching subnode.

[0024] In addition, according to a communication system according to another aspect of the present disclosure, the one child node is a child node newly added to the communication system.

[0025] In addition, according to a communication system in another aspect of the present disclosure, the at least one master node receives the feature information transmitted from each of the plurality of child nodes.

[0026] In addition, according to another aspect of the present disclosure, in a communication system, the at least one main node divides the multiple sub-nodes into multiple categories based on the feature information; performs training of the sub-node neural network models for the multiple categories using the feature information to obtain updated sub-node neural network models; and updates the sub-node neural network models of the multiple sub-nodes using the sub-node neural network model.

[0027] In addition, according to another aspect of the present disclosure, a communication system is provided, wherein the at least one main node divides the multiple child nodes into multiple categories based on the feature information; the feature information of the child nodes belonging to the same category in the multiple categories is notified to the child nodes of the same category according to the multiple categories; and the child nodes of the same category perform training using the feature information of the child nodes of the same category to update the child node neural network model of the child nodes of the same category.

[0028] In addition, according to a communication system in another aspect of the present disclosure, the characteristic information includes the height, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, and historical configuration information of the subnode. In addition, according to a communication system in another aspect of the present disclosure, the configuration of the subnode neural network model includes one of the following: establishing an index of multiple neural network models, using the index to indicate that the subnode neural network model is one of the multiple neural network models; using the model weight of the neural network model to indicate the subnode neural network model; using the model weight change of the neural network model to indicate the subnode neural network model; and using the semantic representation of the neural network model to indicate the subnode neural network model.

[0029] In addition, according to another aspect of the present disclosure, in a communication system, the characteristic information is a historical optimal beam set of a user equipment corresponding to the subnode, and the historical optimal beam set includes: a difference sequence between multiple optimal beams at multiple consecutive time points and an optimal beam at a recent time point; or a difference sequence between optimal beams at two adjacent time points in multiple consecutive time points;

[0030] In addition, according to a communication system in accordance with another aspect of the present disclosure, the at least one master node or the sub-node determines the weight of each historical optimal beam by using the number of occurrences of each historical optimal beam in the historical optimal beam set; and based on the weight of each historical optimal beam and the historical optimal beam set, a weighted loss function is constructed to perform training to update the sub-node neural network model.

[0031] In addition, according to another aspect of the present disclosure, in a communication system, the at least one main node or the sub-node configures an attention layer in the sub-node neural network model, and performs training using the sub-node neural network model including the attention layer to update the sub-node neural network model.

[0032] As will be described in detail below, according to the communication system based on the neural network model and the configuration method thereof disclosed in the present invention, the dynamic configuration of the neural network model of the new sub-node in the communication system is realized, and the online data is fully utilized during the operation, and the master node realizes the centralized update at the master node or the distributed update at each sub-node. In the process of dynamic configuration and update, the full sharing and utilization of the training data and the neural network model between the same or similar sub-nodes is considered, which improves the efficiency of training and the accuracy of the obtained model. In addition, in the configuration process of the neural network model, various characteristics of the sub-nodes are fully considered, such as the height of the sub-node, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, and historical configuration information, and the neural network model is represented in different ways such as the neural network model index, the model weight of the neural network model, the model weight change of the neural network model, and the semantic representation of the neural network model, which further improves the efficiency of training and the accuracy of the obtained model. Furthermore, when a neural network model is used for a specific task such as configuring an optimal beam candidate set for a user device, a lightweight recurrent neural network (RNN) is adopted, and a gated recurrent unit (GRU) module is used to capture the long-term dependency information of the input sequence, an appropriate training data representation method is selected, and the construction of the loss function is improved in a targeted manner. At the same time, an attention mechanism is introduced into the neural network model to effectively extract valuable information from the input sequence, thereby effectively improving the accuracy of the prediction of the optimal beam candidate set, especially improving the accuracy in the long-term prediction case.

[0033] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the technology as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The above and other purposes, features and advantages of the present disclosure will become more apparent by describing the embodiments of the present disclosure in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0035] Figure 1 is a schematic diagram outlining a communication system according to an embodiment of the present disclosure;

[0036] Figure 2 is a flowchart outlining a method for configuring a communication system according to an embodiment of the present disclosure;

[0037] Figure 3 is a schematic diagram illustrating one configuration example of a communication system according to an embodiment of the present disclosure;

[0038] Figure 4 is a flowchart illustrating an example of a communication system configuration method according to an embodiment of the present disclosure;

[0039] Figure 5 is a flowchart illustrating an example of a communication system configuration method according to an embodiment of the present disclosure;

[0040] Figure 6 is a schematic diagram illustrating one configuration example of a communication system according to an embodiment of the present disclosure;

[0041] Figure 7 is a flowchart illustrating an example of a communication system configuration method according to an embodiment of the present disclosure;

[0042] Figure 8 is a flowchart illustrating an example of a communication system configuration method according to an embodiment of the present disclosure;

[0043] Fig. 9 is a schematic diagram illustrating one configuration example of a communication system according to an embodiment of the present disclosure;

[0044] Fig.10 is a flowchart illustrating an example of a communication system configuration method according to an embodiment of the present disclosure;

[0045] Fig.11 is a flowchart illustrating an example of a communication system configuration method according to an embodiment of the present disclosure;

[0046] Fig.12 is a schematic diagram illustrating a communication system according to an embodiment of the present disclosure performing an optimal beam scanning task;

[0047] Fig.13is a schematic diagram illustrating the training and prediction process of a neural network model configured in a communication system according to an embodiment of the present disclosure;

[0048] Fig.14 is a schematic diagram illustrating a neural network model configured in a communication system according to an embodiment of the present disclosure; and

[0049] Fig.15 is a block diagram illustrating an example of a hardware configuration of a sub-node and a user equipment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present disclosure more obvious, the exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments of the present disclosure, and it should be understood that the present disclosure is not limited to the exemplary embodiments described here.

[0051] The solution provided by the present disclosure involves the combination of mobile communication technology and artificial intelligence technology, which is specifically illustrated by the following embodiments.

[0052] Figure 1 is a schematic diagram outlining a communication system according to an embodiment of the present disclosure.

[0053] like Figure 1 As shown, the communication system 1 according to the embodiment of the present disclosure includes at least one master node 10 and a plurality of sub-nodes 11, 12, 13 and 14 that are communicatively connected to the master node. The master node 10 serves as a central control unit to implement configuration, scheduling and management of the plurality of sub-nodes 11, 12, 13 and 14 and corresponding resources. Each of the plurality of sub-nodes 11, 12, 13 and 14 is configured with a sub-node neural network model 111, 121, 131 and 141.

[0054] In one embodiment of the present disclosure, the master node 10 is, for example, a central unit (CU) of a communication network, and the subnodes 11, 12, 13, and 14 are, for example, distributed units (DU) of the communication network. In another embodiment of the present disclosure, the master node 10 is, for example, a cloud server, and the subnodes 11, 12, 13, and 14 are, for example, multi-access edge computing (MEC) servers. It is easy to understand that the number and type of master nodes and subnodes, and the number and type of subnode neural networks are non-limiting.

[0055] Figure 2 1 is a flowchart summarizing a method for configuring a communication system according to an embodiment of the present disclosure. Figure 1 In the communication system 1 shown in FIG. Figure 2 The communication system configuration method according to an embodiment of the present disclosure is shown.

[0056] Specifically, in step S201, feature information of multiple child nodes is obtained.

[0057] As will be described in detail with reference to the accompanying drawings, in an embodiment of the present disclosure, the characteristic information of the plurality of subnodes may be the height of the subnode, antenna configuration, size of the coverage area, service type, service volume, user distribution, environmental information, etc. In addition, in an embodiment of the present disclosure, the characteristic information of the plurality of subnodes may also include historical information acquired by the subnode neural network of the plurality of subnodes in the process of performing a specific task. In an embodiment of the present disclosure, the characteristic information of the plurality of subnodes may be the subnodes reporting to the main node, or the main node predicting the relevant characteristic information of the subnodes based on the acquired initial information of the subnodes.

[0058] In step S202, the sub-node neural network model is dynamically configured based on the acquired feature information.

[0059] As will be described in detail below with reference to the accompanying drawings, in an embodiment of the present disclosure, dynamically configuring the subnode neural network model may be to initialize the subnode neural network model of the new subnode when the new subnode joins the communication network. In an embodiment of the present disclosure, dynamically configuring the subnode neural network model may also be to train and update the subnode neural network model of each subnode using real-time generated online data as training data during the operation of the communication network.

[0060] Below, we will refer to Figures 3 to 11 A specific example of a communication system and a configuration method thereof according to an embodiment of the present disclosure is described in detail.

[0061] Figure 3 1 is a schematic diagram illustrating a configuration example of a communication system according to an embodiment of the present disclosure. Figure 3 As shown, a new subnode 14 is added to the communication system 1 , and the master node 10 performs initialization configuration on the new subnode 14 . Figure 4 and Figure 5 is with Figure 3 An example flow chart of a communication system configuration method corresponding to a scenario, and Figure 4 and Figure 5 Two different ways of acquiring feature information are shown respectively.

[0062] like Figure 4 As shown, an example of a communication system configuration method according to an embodiment of the present disclosure includes the following steps.

[0063] In step S401, the characteristic information transmitted from one of the plurality of child nodes is received. Figure 3 As shown, the master node 10 receives the characteristic information P transmitted from the newly added child node 14 i,4 , feature information Pi,4 It may be the height of the subnode 14, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, etc.

[0064] In step S402, based on the feature information, a neural network model is selected from a plurality of predetermined neural network models. The master node 10 is pre-set with a plurality of neural network models for different sub-node types and task types. The master node 10 selects a neural network model based on the feature information P transmitted from the newly added sub-node 14. i,4 , selecting a neural network model from a plurality of predetermined neural network models.

[0065] In step S403, the subnode neural network model of the subnode is configured using the selected one neural network model. The master node 10 selects a neural network model from a plurality of predetermined neural network models through the signaling information P i+1,4 Send to the child node 14, so as to configure the child node neural network model 114 of the child node 14.

[0066] In an embodiment of the present disclosure, the main node can represent the neural network model in a variety of different ways. For example, an index of multiple neural network models can be established, and the index can be used to indicate that the sub-node neural network model is one of the multiple neural network models. The model weight of the neural network model can be used to indicate the sub-node neural network model. The change in the model weight of the neural network model can be used to indicate the sub-node neural network model. In addition, the semantic representation of the neural network model can also be used to indicate the sub-node neural network model, for example, using the topological architecture diagram of the neural network as the semantic representation of the neural network model. It is easy to understand that the above representation of the neural network model is only schematic, and the representation of the neural network model in the communication system configuration method according to the embodiment of the present disclosure is not limited to this.

[0067] like Figure 5 As shown, an example of a communication system configuration method according to an embodiment of the present disclosure includes the following steps.

[0068] In step S501, initial information transmitted from one of the plurality of sub-nodes is received. Figure 3 As shown, the master node 10 receives the initial information P transmitted from the newly added child node 14 i,4 It should be noted that the initial information transmitted by the child node 14 may be different from the reference Figure 4 In addition, in the embodiment of the present disclosure, step S501 is optional, and the newly added child node 14 does not need to report the initial information.

[0069] In step S502, based on the initial information, the feature information of the one child node is predicted. Figure 4 The example described in Figure 5 In the flowchart shown, the main node 10 predicts the feature information of the newly added child node 14.

[0070] In step S503, based on the feature information, a neural network model is selected from a plurality of predetermined neural network models. The master node 10 is pre-set with a plurality of neural network models for different sub-node types and task types. The master node 10 selects a neural network model from a plurality of predetermined neural network models based on the predicted feature information of the newly added sub-node 14.

[0071] In step S504, the subnode neural network model of the subnode is configured using the selected one neural network model. Figure 4 The configuration steps described above are the same. The master node 10 selects a neural network model from multiple predetermined neural network models through the signaling information P i+1,4 Send to the child node 14, so as to configure the child node neural network model 114 of the child node 14.

[0072] Figure 6 is a schematic diagram illustrating a configuration example of a communication system according to an embodiment of the present disclosure. Figure 3 In the example shown, a new subnode 14 is added to the communication system 1, and the master node 10 performs initialization configuration on the new subnode 14. Figure 3 Unlike the example shown, Figure 6 In the configuration example shown, the master node selects a neural network model from a child node that matches the newly joined child node similarly, rather than selecting from multiple neural network models that are set. Figure 7 and Figure 8 is with Figure 6 An example flow chart of a communication system configuration method corresponding to a scenario, and Figure 7 and Figure 8 Two different ways of acquiring feature information are shown respectively.

[0073] like Figure 7 As shown, an example of a communication system configuration method according to an embodiment of the present disclosure includes the following steps.

[0074] In step S701, the characteristic information transmitted from one of the plurality of child nodes is received. Figure 6 As shown, the master node 10 receives the characteristic information P transmitted from the newly added child node 14 i,4 , feature information P i,4It may be the height of the subnode 14, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, etc.

[0075] In step S702, based on the feature information, a matching sub-node that matches the one sub-node is selected from the plurality of sub-nodes. Figure 6 As shown, the master node 10 receives the feature information P transmitted from the newly added child node 14. i,4 , selecting a matching child node 11 that matches the newly added child node 14 from the existing multiple child nodes.

[0076] In step S703, the subnode neural network model of the matching subnode is received from the matching subnode. Figure 6 As shown, the master node 10 receives the signaling P from the matching child node 11. i,1 The transmitted sub-node neural network model 111.

[0077] In step S704, the sub-node neural network model of the matching sub-node is used to configure the sub-node neural network model of the one sub-node. Figure 6 As shown, the master node 10 will receive the signaling P from the matching child node 11. i,1 The transmitted child node neural network model 111 is transmitted through the signaling information P i+1,4 Send to the child node 14, so as to configure the child node neural network model 114 of the child node 14.

[0078] like Figure 8 As shown, an example of a communication system configuration method according to an embodiment of the present disclosure includes the following steps.

[0079] In step S801, initial information transmitted from one of the plurality of subnodes is received. Figure 6 As shown, the master node 10 receives the initial information P transmitted from the newly added child node 14 i,4 It should be noted that the initial information transmitted by the child node 14 may be different from the reference Figure 7 In addition, in the embodiment of the present disclosure, step S801 is optional, and the newly added child node 14 does not need to report the initial information.

[0080] In step S802, based on the initial information, the feature information of the one child node is predicted. Figure 7 The example described in Figure 8 In the flowchart shown, the main node 10 predicts the feature information of the newly added child node 14.

[0081] In step S803, based on the feature information, a matching sub-node matching the one sub-node is selected from the plurality of sub-nodes. Figure 6 As shown, the master node 10 receives the feature information P transmitted from the newly added child node 14. i,4 , selecting a matching child node 11 that matches the newly added child node 14 from the existing multiple child nodes.

[0082] In step S804, the subnode neural network model of the matching subnode is received from the matching subnode. Figure 6 As shown, the master node 10 receives the signaling P from the matching child node 11. i,1 The transmitted sub-node neural network model 111.

[0083] In step S805, the sub-node neural network model of the matching sub-node is used to configure the sub-node neural network model of the one sub-node. Figure 6 As shown, the master node 10 will receive the signaling P from the matching child node 11. i,1 The transmitted child node neural network model 111 is transmitted through the signaling information P i+1,4 Send to the child node 14, so as to configure the child node neural network model 114 of the child node 14.

[0084] Above, refer to Figures 3 to 8 It is described that when a new sub-node is added to a communication system, the new sub-node is not initialized and configured using a predetermined default setting, but a targeted optimization configuration is performed based on the characteristics of the new sub-node itself.

[0085] Fig. 9 1 is a schematic diagram illustrating a configuration example of a communication system according to an embodiment of the present disclosure. Fig. 9 As shown, in the communication system 1 , the master node 10 coordinates the training updates for the respective child nodes 11 , 12 , 13 and 14 . Fig.10 and Fig.11 is with Fig. 9 An example flow chart of a communication system configuration method corresponding to a scenario, and Fig.10 and Fig.11 Two different updating modes are shown respectively.

[0086] like Fig.10 As shown, an example of a communication system configuration method according to an embodiment of the present disclosure includes the following steps.

[0087] In step S1001, the feature information transmitted from each of the plurality of sub-nodes is received. Fig. 9As shown, the master node 10 receives the characteristic information P transmitted from the child nodes 11, 12, 13 and 14. i,1 , P i,2 , P i,3 and P i,4 More specifically, in the scenario of training and updating each of the child nodes 11, 12, 13, and 14, the feature information P transmitted from each of the child nodes 11, 12, 13, and 14 is i,1 , P i,2 , P i,3 and P i,4 It is the online data generated for a specific task by each sub-node 11, 12, 13 and 14. For example, in the embodiment further described below, the online data is the historical optimal beam set predicted by the sub-node for the user equipment.

[0088] In step S1002, the plurality of child nodes are divided into a plurality of categories based on the characteristic information. Fig. 9 As shown, the master node 10 is based on the feature information P transmitted from each of the child nodes 11, 12, 13 and 14. i,1 , P i,2 , P i,3 and P i,4 , multiple child nodes are divided into two categories, that is, child nodes 11 and 14 are one category, and child nodes 12 and 13 are another category.

[0089] In step S1003, the sub-node neural network model is trained for the multiple categories using the feature information to obtain an updated sub-node neural network model. Fig. 9 As shown, the master node 10 uses the feature information P of the first type of child nodes 11 and 14 i,1 and P i,4 , train the sub-node neural network models 111 and 141 of the first type of sub-nodes, and the master node 10 uses the feature information P of the second type of sub-nodes 12 and 13 i,2 and P i,3 , the sub-node neural network models 121 and 131 of the second type of sub-nodes are trained. That is, by training each sub-node 11, 12, 13 and 14 according to the classification, the available training data is expanded compared to the training process for a single sub-node, thereby improving the training efficiency and the accuracy of the sub-node neural network model obtained by training.

[0090] In step S1004, the sub-node neural network model of the plurality of sub-nodes is updated using the sub-node neural network model. Fig. 9 As shown, the master node 10 classifies and trains the sub-node neural network models of the multiple sub-nodes through signaling information P i+1,1, P i+1,2 , P i+1,3 and P i+1,4 Sent to child nodes 11, 12, 13 and 14, thereby configuring the child node neural network models 111, 121, 131 and 141 of the child nodes 11, 12, 13 and 14.

[0091] like Fig.11 As shown, an example of a communication system configuration method according to an embodiment of the present disclosure includes the following steps.

[0092] In step S1101, the feature information transmitted from each of the plurality of sub-nodes is received. Fig. 9 As shown, the master node 10 receives the characteristic information P transmitted from the child nodes 11, 12, 13 and 14. i,1 , P i,2 , P i,3 and P i,4 More specifically, in the scenario of training and updating each of the child nodes 11, 12, 13, and 14, the feature information P transmitted from each of the child nodes 11, 12, 13, and 14 is i,1 , P i,2 , P i,3 and P i,4 It is the online data generated for a specific task by each sub-node 11, 12, 13 and 14. For example, in the embodiment further described below, the online data is the historical optimal beam set predicted by the sub-node for the user equipment.

[0093] In step S1102, the plurality of child nodes are divided into a plurality of categories based on the characteristic information. Fig. 9 As shown, the master node 10 is based on the feature information P transmitted from each of the child nodes 11, 12, 13 and 14. i,1 , P i,2 , P i,3 and P i,4 , multiple child nodes are divided into two categories, that is, child nodes 11 and 14 are one category, and child nodes 12 and 13 are another category.

[0094] In step S1103, according to multiple categories, the feature information of the child nodes belonging to the same category among the multiple categories is notified to the child nodes of the same category. Fig.10 As shown, the training is performed by the master node 10. Fig.11 In the configuration method shown, the master node 10 is divided into multiple categories (such as Fig. 9The master node 10 notifies the subnodes of the same category of the feature information of the subnodes belonging to the same category in the multiple categories. For example, the master node 10 respectively transmits the feature information of the subnodes of the same category to the subnodes of the same category through the signaling P i+1,1 and P i+1,4 The first type of feature information (i.e., the first type of online data) P i,1 and P i,4 The first type of child nodes 11 and 14 are notified, and the master node 10 respectively sends a signal through the signaling P i+1,2 and P i+1,3 The second type of feature information (i.e., the second type of online data) P i,2 and P i,2 The second type of child nodes 12 and 13 are notified.

[0095] In step S1104, the sub-nodes of the same category perform training using the feature information of the sub-nodes of the same category to update the sub-node neural network model of the sub-nodes of the same category. Fig. 9 As shown, subnodes 11 and 14 use the first type of feature information P i,1 and P i,4 Execute training and update their own child node neural network models 111 and 141, and child nodes 12 and 13 use the second type of feature information P i,2 and P i,3 Execute training and update its own child node neural network models 121 and 131. That is, by using the feature information of the category to which each child node 11, 12, 13 and 14 is trained, the available training data is expanded compared to the training process in which a single child node only uses its own feature information, thereby improving the training efficiency and the accuracy of the trained child node neural network model.

[0096] Above, refer to Figures 9 to 11 It is described that when training and updating the neural network model of each sub-node in a communication system, instead of training a single sub-node using only its own training data, the sub-nodes are classified according to their own characteristics, so that all training data of sub-nodes of the same type are used to perform training and updating of the neural network model for the sub-nodes of this type, thereby expanding the available training data and improving the training efficiency and the accuracy of the neural network model of the trained sub-nodes.

[0097] Below, we will further refer to Figure 12 to Figure 14 A specific example of training a neural network model of a subnode for the purpose of providing an optimal beam candidate set for a user equipment is described.

[0098] Fig.12It is a schematic diagram illustrating a communication system performing an optimal beam scanning task according to an embodiment of the present disclosure.

[0099] like Fig.12 As shown, the subnode 11 is, for example, a base station using NR massive MIMO. When the user equipment 20 is in a mobile state, the optimal beam of the user equipment 20 at different times T1 and T2 will change significantly.

[0100] In the communication system according to the embodiment of the present disclosure, the neural network model 111 configured in the subnode 11 can perform the prediction task of the future optimal beam candidate set on the user equipment 20. It should be understood that in order to perform the prediction task of the future optimal beam candidate set, it is necessary to train the neural network model 111 configured in the subnode 11. Figures 3 to 11 In the communication system configuration method according to the embodiment of the present disclosure, the master node 10 or the sub-node 11 performs training.

[0101] Fig.13 It is a schematic diagram illustrating the training and prediction process of a neural network model configured in a communication system according to an embodiment of the present disclosure.

[0102] Fig.13 The training phase 130 and the prediction phase 140 of the neural network model are shown respectively. In the training phase 130, the historical optimal beam set is used as the training data set 1301. More specifically, in the embodiment of the present disclosure, the relative index of the historical optimal beam is used as the training data.

[0103] For example, in one embodiment, multiple optimal beams Idx at multiple consecutive time points are used. t1 ,Idx t2 ,…Idx tn-1 The optimal beam Idx at the nearest time point tn The difference sequence {{Idxt1-Idxtn},{Idxt2-Idxtn},...,{Idxtn-1-Idxtn},{0}} is taken as the historical optimal beam set.

[0104] In another embodiment, a difference sequence {{0}, {Idxt2-Idxt1}, {Idxt3-Idxt2}, {Idxt4-Idxt3}, ..., {Idxtn-Idxtn-1}} between the optimal beams of two adjacent time points in a plurality of consecutive time points is used as the historical optimal beam set.

[0105] By configuring the representation of the training data in this way, the same change trend of the optimal beam will be recognized by the neural network model as the same training data, thereby reducing the redundancy of the training data.

[0106] In addition, in an embodiment of the present disclosure, a weighted binary cross entropy is used to construct a loss function required for training. In one example, the loss function required for training is expressed as:

[0107] L n =-w n [y n ·log(x n )+(1-y n )·log(1-x n )]

[0108] Among them, x n is the prediction result of the neural network model during training, y n is the prediction target of the neural network model, w n is the corresponding weight of the corresponding beam. In the initial training, each beam is assigned the same initial weight. As the training progresses, each time a beam becomes the optimal beam, its corresponding weight is incremented, while also keeping all beam weights normalized. In this way, a more accurate training result can be achieved by using a loss function constructed by considering the frequency of different beams becoming the optimal beam.

[0109] In the prediction stage 140 , using the historical optimal beam set 1401 as input, the trained sub-node neural network model 111 of the sub-node 11 will output the corresponding candidate beam set 1501 .

[0110] Fig.14 is a schematic diagram illustrating a neural network model configured in a communication system according to an embodiment of the present disclosure.

[0111] like Fig.14 As shown, in one embodiment of the present disclosure, the subnode neural network model 111 of the subnode 11 uses a cascaded gated recurrent unit (GRU) module to extract the long-term variation trend of the beam. Fig.14 As shown, an attention layer 400 is introduced into the neural network model to more effectively extract valuable information from the input sequence, thereby effectively improving the accuracy of the prediction of the optimal beam candidate set, especially improving the accuracy in the long-term prediction case.

[0112] Fig.15 1004, an input device 1005, an output device 1006, a bus 1007, and the like.

[0113] In addition, in the following description, the word "device" may also be replaced by circuit, equipment, unit, etc. The hardware structure of the subnodes 11, 12, 13, 14 and the user equipment 20 may include one or more devices shown in the figure, or may not include some devices.

[0114] For example, only one processor 1001 is shown, but there may be multiple processors. In addition, the processing may be performed by one processor, or by more than one processor simultaneously, sequentially, or by other methods. In addition, the processor 1001 may be installed by more than one chip.

[0115] The functions in the sub-nodes 11, 12, 13, 14 and the user device 20 are implemented, for example, in the following manner: by reading the specified software (program) into the hardware such as the processor 1001 and the memory 1002, so that the processor 1001 performs calculations, controls the communication performed by the communication device 1004, and controls the reading and / or writing of data in the memory 1002 and the storage 1003.

[0116] The processor 1001 controls the entire computer by, for example, operating an operating system. The processor 1001 may be composed of a central processing unit (CPU) including an interface with peripheral devices, a control device, a computing device, a register, and the like. In addition, the processor 1001 reads programs (program codes), software modules, data, and the like from the storage 1003 and / or the communication device 1004 to the memory 1002, and performs various processes based on them. As a program, a program that enables a computer to perform at least a part of the actions described in the above embodiments may be used. For example, the polarization encoder 300 may be implemented by a control program stored in the memory 1002 and operated by the processor 1001, and other functional blocks may be implemented in the same manner. The memory 1002 is a computer-readable recording medium, and may be composed of, for example, at least one of a read-only memory (ROM), an erasable programmable ROM (EPROM), an electrically programmable read-only memory (EEPROM), a random access memory (RAM), and other appropriate storage media. The memory 1002 may also be called a register, a cache, a main memory (main storage device), etc. The memory 1002 may store an executable program (program code), a software module, etc. for implementing the wireless communication method according to an embodiment of the present invention.

[0117] The memory 1003 is a computer-readable recording medium, and may be composed of at least one of a flexible disk, a floppy disk, a magneto-optical disk (e.g., a read-only disk (CD-ROM (CompactDiscROM)), a digital versatile disk, a Blu-ray (registered trademark) optical disk), a removable disk, a hard disk drive, a smart card, a flash memory device (e.g., a card, a stick, a key driver), a magnetic stripe, a database, a server, and other appropriate storage media. The memory 1003 may also be referred to as an auxiliary storage device.

[0118] The communication device 1004 is a hardware (transmitting and receiving device) for communicating between computers via a wired and / or wireless network, and is also referred to as a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc., in order to implement, for example, frequency division duplex (FDD) and / or time division duplex (TDD). For example, the transmitter 202 described above may be implemented by the communication device 1004.

[0119] The input device 1005 is an input device that receives input from the outside (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.). The output device 1006 is an output device that implements output to the outside (e.g., a display, a speaker, a light emitting diode (LED, Light Emitting Diode) lamp, etc.). In addition, the input device 1005 and the output device 1006 may also be an integrated structure (e.g., a touch panel).

[0120] In addition, the processor 1001, the memory 1002 and other devices are connected via a bus 1007 for communicating information. The bus 1007 may be composed of a single bus or different buses between devices.

[0121] In addition, the subnodes 11, 12, 13, 14 and the user equipment 20 may include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), etc., and part or all of each functional block may be implemented by the hardware. For example, the processor 1001 may be installed by at least one of these hardware.

[0122] Above, refer to Figures 1 to 15 A communication system based on a neural network model and a configuration method thereof according to the present disclosure are described, which realizes the dynamic configuration of the neural network model of a new sub-node in the communication system, and fully utilizes online data during operation, and the master node realizes centralized update at the master node or distributed update at each sub-node. In the process of dynamic configuration and update, the full sharing and utilization of training data and neural network models between the same or similar sub-nodes are considered, which improves the efficiency of training and the accuracy of the obtained model. In addition, in the configuration process of the neural network model, various characteristics of the sub-nodes are fully considered, such as the height of the sub-node, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, and historical configuration information, and the neural network model is represented in different ways such as neural network model index, model weight of the neural network model, model weight change of the neural network model and semantic representation of the neural network model, which further improves the efficiency of training and the accuracy of the obtained model. Furthermore, when a neural network model is used for a specific task such as configuring an optimal beam candidate set for a user device, a lightweight recurrent neural network (RNN) is adopted, and a gated recurrent unit (GRU) module is used to capture the long-term dependency information of the input sequence, an appropriate training data representation method is selected, and the construction of the loss function is improved in a targeted manner. At the same time, an attention mechanism is introduced into the neural network model to effectively extract valuable information from the input sequence, thereby effectively improving the accuracy of the prediction of the optimal beam candidate set, especially improving the accuracy in the long-term prediction case.

[0123] In addition, the terms described in this specification and / or the terms required for understanding this specification can be interchanged with terms having the same or similar meanings. For example, a channel and / or a symbol can also be a signal (signaling). In addition, a signal can also be a message. A reference signal can also be referred to as RS (Reference Signal), and can also be referred to as a pilot (Pilot), a pilot signal, etc. depending on the applicable standard. In addition, a component carrier (CC) can also be referred to as a cell, a frequency carrier, a carrier frequency, etc.

[0124] In addition, the information, parameters, etc. described in this specification may be expressed as absolute values, as relative values ​​to a specified value, or as other corresponding information. For example, a wireless resource may be indicated by a specified index. Furthermore, the formulas, etc. using these parameters may also be different from those explicitly disclosed in this specification.

[0125] The names used for parameters and the like in this specification are not limiting in any respect. For example, various channels (Physical Uplink Control Channel (PUCCH), Physical Downlink Control Channel (PDCCH), etc.) and information units may be identified by any appropriate names, and therefore the various names assigned to these various channels and information units are not limiting in any respect.

[0126] The information, signals, etc. described in this specification may be represented by any of a variety of different technologies. For example, data, commands, instructions, information, signals, bits, symbols, chips, etc. that may be mentioned in the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or photons, or any combination thereof.

[0127] In addition, information, signals, etc. can be output from an upper layer to a lower layer, and / or from a lower layer to an upper layer. Information, signals, etc. can be input or output via multiple network nodes.

[0128] Input or output information, signals, etc. can be stored in a specific place (such as memory), or managed through a management table. Input or output information, signals, etc. can be overwritten, updated, or supplemented. Output information, signals, etc. can be deleted. Input information, signals, etc. can be sent to other devices.

[0129] The notification of information is not limited to the methods / implementations described in this specification, and may also be performed by other methods. For example, the notification of information may be implemented by physical layer signaling (e.g., downlink control information (DCI, DownlinkControlInformation), uplink control information (UCI, UplinkControlInformation)), upper layer signaling (e.g., radio resource control (RRC, RadioResourceControl) signaling, broadcast information (master information block (MIB, MasterInformationBlock), system information block (SIB, SystemInformationBlock), etc.), media access control (MAC, MediumAccessControl) signaling), other signals or a combination thereof.

[0130] In addition, physical layer signaling may also be referred to as L1 / L2 (Layer 1 / Layer 2) control information (L1 / L2 control signal), L1 control information (L1 control signal), etc. In addition, RRC signaling may also be referred to as RRC message, such as RRC Connection Setup message, RRC Connection Reconfiguration message, etc. In addition, MAC signaling may be notified, for example, by a MAC control unit (MAC CE (Control Element)).

[0131] Furthermore, notification of specified information (eg, notification of “ACK” or “NACK”) is not limited to being performed explicitly, but may be performed implicitly (eg, by not notifying the specified information or by notifying other information).

[0132] The judgment can be made by a value represented by 1 bit (0 or 1), by a true or false value (Boolean value) represented by true (true) or false (false), or by comparing numerical values ​​(for example, comparing with a specified value).

[0133] Software, whether referred to as software, firmware, middleware, microcode, hardware description language, or by other names, shall be interpreted broadly to refer to commands, sets of commands, codes, code segments, program code, programs, subroutines, software modules, applications, software applications, software packages, routines, subroutines, objects, executables, execution threads, steps, functions, etc.

[0134] In addition, software, commands, information, etc. can be sent or received via a transmission medium. For example, when software is sent from a website, server, or other remote resource using wired technology (coaxial cable, optical cable, twisted pair, digital subscriber line (DSL, Digital Subscriber Line), etc.) and / or wireless technology (infrared, microwave, etc.), these wired technology and / or wireless technology are included in the definition of transmission medium.

[0135] The terms "system" and "network" used in this specification are used interchangeably.

[0136] In this specification, the terms "base station (BS)," "wireless base station", "eNB", "gNB", "cell", "sector", "cell group", "carrier" and "component carrier" can be used interchangeably. Base stations are sometimes also referred to as fixed stations, NodeB, eNodeB (eNB), access points, transmission points, reception points, micro-micro cells, small cells, etc.

[0137] A base station can accommodate one or more (for example, three) cells (also called sectors). When a base station accommodates multiple cells, the entire coverage area of ​​the base station can be divided into multiple smaller areas, and each smaller area can also provide communication services through a base station subsystem (for example, a small base station for indoor use (Radio Remote Head (RRH))). Terms such as "cell" or "sector" refer to a part or the entire coverage area of ​​a base station and / or a base station subsystem that provides communication services in the coverage.

[0138] In this specification, the terms "mobile station (MS), "user terminal", "user equipment (UE), and "terminal" are used interchangeably. Base stations are sometimes referred to as fixed stations, NodeBs, eNodeBs (eNBs), access points, transmission points, reception points, femtocells, and small cells.

[0139] Mobile stations are sometimes also referred to by those skilled in the art as user stations, mobile units, user units, wireless units, remote units, mobile devices, wireless devices, wireless communication devices, remote devices, mobile user stations, access terminals, mobile terminals, wireless terminals, remote terminals, handsets, user agents, mobile clients, clients or several other appropriate terms.

[0140] In addition, the wireless base station in this specification can also be replaced by a user terminal. For example, for a structure in which the communication between a wireless base station and a user terminal is replaced by the communication between multiple user terminals (D2D, Device-to-Device), the various methods / implementations of the present invention can also be applied. In this case, the functions of the above-mentioned sub-nodes 11, 12, 13, and 14 can be regarded as the functions of the user terminal 20. In addition, words such as "uplink" and "downlink" can also be replaced by "side". For example, the uplink channel can also be replaced by the side channel.

[0141] Likewise, the user terminal in this specification can also be replaced by a wireless base station. In this case, the functions of the user terminal 20 described above can be regarded as the functions of the sub-nodes 11, 12, 13, and 14.

[0142] In this specification, a specific action performed by a base station is sometimes performed by its upper node according to the situation. Obviously, in a network composed of one or more network nodes having a base station, various actions performed for communication with a terminal can be performed by a base station, one or more network nodes other than the base station (for example, a mobile management entity (MME), a serving gateway (S-GW), etc., but not limited to this), or a combination thereof.

[0143] The various methods / implementations described in this specification may be used alone or in combination, and may be switched during execution. In addition, the processing steps, sequences, flow charts, etc. of the various methods / implementations described in this specification may be changed in order as long as there is no contradiction. For example, with respect to the methods described in this specification, various step units are given in an exemplary order, and are not limited to the specific order given.

[0144] The various modes and embodiments described in this specification can be applied to long term evolution (LTE), long term evolution-advanced (LTE-A), long term evolution-beyond (LTE-B), super 3G, IMT-Advanced, 4G, 5G, future radio access (FRA), new radio access technology (New-RAT), new radio (NR), new radio access (NX), future generation radio access (FX), global system for mobile communications (GSM (registered trademark), code division multiple access 2000 (CDMA2000), ultra mobile broadband (UMB), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE802.20, Ultra-WideBand (UWB), Bluetooth (registered trademark), systems of other appropriate wireless communication methods, and / or next-generation systems expanded based on them.

[0145] The phrase "based on" used in this specification does not mean "only based on" unless otherwise specified in other paragraphs. In other words, the phrase "based on" means both "only based on" and "at least based on".

[0146] Any reference to a unit using the names "first", "second", etc. used in this specification does not fully limit the number or order of these units. These names may be used in this specification as a convenient method to distinguish between two or more units. Therefore, a reference to a first unit and a second unit does not mean that only two units can be used or that the first unit must precede the second unit in some form.

[0147] The term "determining" used in this specification sometimes includes a variety of actions. For example, with respect to "determining", calculating, computing, processing, deriving, investigating, looking up (e.g., searching in a table, database, or other data structure), ascertaining, etc. can be regarded as performing "determining". In addition, with respect to "determining", receiving (e.g., receiving information), transmitting (e.g., transmitting information), input, output, accessing (e.g., accessing data in a memory), etc. can also be regarded as performing "determining". In addition, with respect to "determining", resolving, selecting, choosing, establishing, comparing, etc. can also be regarded as performing "determining". That is, with respect to "determining", several actions can be regarded as performing "determining".

[0148] The terms "connected", "coupled" or any variation thereof used in this specification refer to any direct or indirect connection or combination between two or more units, and may include the following situations: there are one or more intermediate units between two units that are "connected" or "coupled" to each other. The combination or connection between units can be physical, logical, or a combination of the two. For example, "connect" can also be replaced by "access". When used in this specification, two units can be considered to be "connected" or "coupled" to each other by using one or more wires, cables, and / or printed electrical connections, and as several non-limiting and non-exhaustive examples, by using electromagnetic energy with wavelengths in the radio frequency region, microwave region, and / or light (both visible and invisible light) region, etc.

[0149] When the terms "including", "comprising", and their variations are used in this specification or claims, these terms are open ended like the term "having". Furthermore, the term "or" used in this specification or claims is not an exclusive or.

[0150] The present invention has been described in detail above, but it is obvious to those skilled in the art that the present invention is not limited to the embodiments described in this specification. The present invention can be implemented as a modification and variation without departing from the purpose and scope of the present invention as determined by the description of the claims. Therefore, the description of this specification is for the purpose of illustration and does not have any restrictive meaning for the present invention.

Claims

1. A method for configuring a communication system based on a neural network model, wherein the communication system comprises at least one master node and a plurality of sub-nodes communicatively connected to the master node, and each of the plurality of sub-nodes is configured with a sub-node neural network model, and the method for configuring the communication system comprises: Acquire characteristic information of the multiple child nodes; as well as Based on the acquired feature information, dynamically configuring the sub-node neural network model includes: Based on the feature information, the plurality of sub-nodes are divided into a plurality of categories; The sub-node neural network model is dynamically configured according to multiple categories.

2. The communication system configuration method according to claim 1, wherein: The acquiring the characteristic information of the plurality of child nodes comprises: The characteristic information transmitted from one of the plurality of child nodes is received.

3. The communication system configuration method according to claim 1, wherein: The acquiring the characteristic information of the plurality of child nodes comprises: receiving initial information transmitted from one of the plurality of child nodes; and Based on the initial information, the feature information of the one child node is predicted.

4. The communication system configuration method according to claim 2, wherein: The dynamically configuring the sub-node neural network model based on the acquired feature information includes: Based on the feature information, selecting a neural network model from a plurality of predetermined neural network models; and The sub-node neural network model of the one sub-node is configured using the selected one neural network model.

5. The communication system configuration method according to claim 2, wherein: The dynamically configuring the sub-node neural network model based on the acquired feature information includes: Based on the feature information, selecting a matching child node that matches the one child node from the multiple child nodes; receiving a sub-node neural network model of the matching sub-node from the matching sub-node; and The sub-node neural network model of the one sub-node is configured using the sub-node neural network model of the matching sub-node.

6. The communication system configuration method according to claim 1, wherein: The one child node is a child node newly added to the communication system.

7. The communication system configuration method according to claim 1, wherein: The acquiring the characteristic information of the plurality of child nodes comprises: The characteristic information transmitted from each of the plurality of child nodes is received.

8. The communication system configuration method according to claim 7, wherein: The dynamically configuring the sub-node neural network model based on the acquired feature information includes: Using the feature information, training the sub-node neural network model for the multiple categories to obtain an updated sub-node neural network model; and The sub-node neural network model is used to update the sub-node neural network models of the multiple sub-nodes.

9. The communication system configuration method according to claim 7, wherein: The dynamically configuring the sub-node neural network model based on the acquired feature information includes: According to a plurality of categories, notifying the child nodes of the same category of the characteristic information of the child nodes belonging to the same category among the plurality of categories; and The sub-nodes of the same category perform training using the feature information of the sub-nodes of the same category to update the sub-node neural network model of the sub-nodes of the same category.

10. The communication system configuration method according to claim 1, wherein: The characteristic information includes the height of the subnode, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, and historical configuration information.

11. The communication system configuration method according to claim 1, wherein: The configuring of the sub-node neural network model includes one of the following: Establishing an index of a plurality of neural network models, and using the index to indicate that the sub-node neural network model is one of the plurality of neural network models; Indicating the child node neural network model using a model weight of the neural network model; Indicating the child node neural network model using the model weight change of the neural network model; and The sub-node neural network model is indicated using the semantic representation of the neural network model.

12. The communication system configuration method according to claim 7, wherein: The characteristic information is a historical optimal beam set of the child node corresponding to the user equipment, and The historical optimal beam set includes: a sequence of differences between a plurality of optimal beams at a plurality of consecutive time points and an optimal beam at a most recent time point; or The difference sequence between the optimal beams at two adjacent time points in a series of multiple time points.

13. The communication system configuration method according to claim 12, wherein: Updating the sub-node neural network model using the feature information includes: Determine the weight of each historical optimal beam by using the number of occurrences of each historical optimal beam in the historical optimal beam set; and According to the weight of each historical optimal beam and the historical optimal beam set, a weighted loss function is constructed to perform training to update the sub-node neural network model.

14. The communication system configuration method according to claim 12 or 13, wherein: Updating the sub-node neural network model using the feature information includes: An attention layer is configured in the sub-node neural network model, and training is performed using the sub-node neural network model including the attention layer to update the sub-node neural network model.

15. A communication system based on a neural network model, comprising: At least one master node; A plurality of sub-nodes are connected to the master node for communication, and each of the plurality of sub-nodes is configured with a sub-node neural network model, The at least one master node obtains characteristic information of the plurality of child nodes; and Based on the acquired feature information, dynamically configure the sub-node neural network model; The at least one master node divides the plurality of child nodes into a plurality of categories based on the characteristic information; The sub-node neural network model is dynamically configured according to multiple categories.

16. The communication system according to claim 15, wherein: The at least one master node receives the feature information transmitted from one of the plurality of child nodes.

17. The communication system according to claim 15, wherein: The at least one master node receives initial information transmitted from one of the plurality of child nodes; as well as Based on the initial information, the feature information of the one child node is predicted.

18. The communication system according to claim 16, wherein: The at least one master node selects a neural network model from a plurality of predetermined neural network models based on the feature information; and The sub-node neural network model of the one sub-node is configured using the selected one neural network model.

19. The communication system according to claim 16, wherein: The at least one main node selects a matching child node that matches the one child node from the plurality of child nodes based on the feature information; receiving a sub-node neural network model of the matching sub-node from the matching sub-node; as well as The sub-node neural network model of the one sub-node is configured using the sub-node neural network model of the matching sub-node.

20. The communication system of claim 15, wherein: The one child node is a child node newly added to the communication system.

21. The communication system of claim 15, wherein: The at least one master node receives the characteristic information transmitted from each of the plurality of child nodes.

22. The communication system of claim 21, wherein: Using the feature information, training the sub-node neural network model for the multiple categories to obtain an updated sub-node neural network model; and The sub-node neural network model is used to update the sub-node neural network models of the multiple sub-nodes.

23. The communication system of claim 21, wherein: According to a plurality of categories, notifying the child nodes of the same category of the characteristic information of the child nodes belonging to the same category of the plurality of categories; as well as The sub-nodes of the same category perform training using the feature information of the sub-nodes of the same category to update the sub-node neural network model of the sub-nodes of the same category.

24. The communication system of claim 15, wherein: The characteristic information includes the height of the subnode, antenna configuration, coverage area size, service type, service volume, user distribution, environmental information, and historical configuration information.

25. The communication system of claim 15, wherein: The configuring of the sub-node neural network model includes one of the following: Establishing an index of a plurality of neural network models, and using the index to indicate that the sub-node neural network model is one of the plurality of neural network models; Indicating the child node neural network model using a model weight of the neural network model; Indicating the child node neural network model using the model weight change of the neural network model; and The sub-node neural network model is indicated using the semantic representation of the neural network model.

26. The communication system of claim 21, wherein: The characteristic information is a historical optimal beam set of the child node corresponding to the user equipment, and The historical optimal beam set includes: a sequence of differences between a plurality of optimal beams at a plurality of consecutive time points and an optimal beam at a most recent time point; or The difference sequence between the optimal beams at two adjacent time points in a series of multiple time points.

27. The communication system of claim 26, wherein: The at least one master node or the subnode determines the weight of each historical optimal beam by using the number of occurrences of each historical optimal beam in the historical optimal beam set; as well as According to the weight of each historical optimal beam and the historical optimal beam set, a weighted loss function is constructed to perform training to update the sub-node neural network model.

28. The communication system according to claim 26 or 27, wherein: The at least one main node or the sub-node configures an attention layer in the sub-node neural network model, and performs training using the sub-node neural network model including the attention layer to update the sub-node neural network model.

Citation Information

Patent Citations

  • An image recognition method and system

    CN109086789A

  • Load adjustment method, device and system and storage medium

    CN110676845A

  • Method and apparatus for machine learning based wide beam optimization in cellular network

    US20190372644A1