Load balancing oriented cell handover method, device, equipment and storage medium

Abstract

The application provides a cell switching method and device for load balancing, equipment and storage medium, belonging to the field of communication technology, the method comprises the following steps: constructing a load distribution graph by using the neighbor node determined based on the candidate neighbor of the current cell and the user node determined based on the transfer user connected to the current cell; each edge of the load distribution graph is used to associate a neighbor node and a user node; predicting the edge weight of each edge in the load distribution graph by using an edge weight prediction model to obtain the load distribution graph after the edge weight is updated; and determining the target cell of the transfer user switching based on the load distribution graph after the edge weight is updated. The cell switching method and device for load balancing, equipment and storage medium provided by the application can fully utilize the topological structure information of the communication network by converting the load balancing problem into the edge prediction problem of the heterogeneous bipartite graph, accurately match the user to the most suitable target cell for switching, and thus realize the optimal load balancing between cells.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication technology, in particular to a cell switching method and device for load balancing, equipment and storage medium. BACKGROUND

[0002] Switching users in a high-load cell to a low-load neighboring cell of the high-load cell is an important way to achieve load balancing.

[0003] Currently, the main method for switching cells is threshold judgment, that is, according to the comparison result of the load interaction information, measurement report and other data of a certain candidate neighboring cell with the preset threshold value, it is determined whether the candidate neighboring cell can be used as the target cell for user cell switching.

[0004] However, the threshold judgment method cannot fully utilize the topology structure information of the communication network, cannot efficiently extract useful features from complex mass data, and is difficult to accurately match the most suitable target cell for the user among numerous candidate neighboring cells, and is difficult to achieve optimal load balancing between cells. SUMMARY

[0005] The present application provides a cell switching method and device for load balancing, equipment and storage medium, to solve the defect that the threshold judgment method in the prior art cannot accurately match the most suitable cell, and to achieve optimal load balancing between cells.

[0006] The present application provides a cell switching method for load balancing, comprising:

[0007] Constructing a load distribution graph using neighboring nodes and user nodes; the neighboring nodes are determined based on candidate neighboring cells of a current cell; the user nodes are determined based on transfer users connected to the current cell; each edge of the load distribution graph is used to associate a neighboring node and a user node;

[0008] Using an edge weight prediction model to predict the edge weight of each edge in the load distribution graph, to obtain an updated load distribution graph after updating the edge weight;

[0009] Based on the updated load distribution graph after updating the edge weight, determining the target cell for switching of the transfer user.

[0010] According to the cell switching method for load balancing provided by the present application, before constructing the load distribution graph using neighboring nodes and user nodes, the candidate neighboring cells are determined based on the neighboring cells of the current cell, comprising:

[0011] Based on the load index of each dimension of the neighboring cell, determining a comprehensive load index;

[0012] Based on the comprehensive load index, determining the candidate neighboring cell.

[0013] The cell handover method for load balancing provided by the present application, before the load distribution graph is constructed by using the neighbor node and the user node, the transfer user is determined based on the user connected to the current cell, comprising:

[0014] The transfer priority is determined based on the priority of each dimension of the user;

[0015] The transfer user is determined based on the transfer priority.

[0016] The cell handover method for load balancing provided by the present application, the load distribution graph is constructed by using the neighbor node and the user node, comprising:

[0017] The measurement report information is obtained; the measurement report information is measured by the transfer user after receiving the event information of the candidate neighbor node; the event information of the candidate neighbor node is sent to the transfer user by the current cell;

[0018] The initial edge weight of each edge for associating a neighbor node and a user node is determined based on the measurement report information;

[0019] The load distribution graph is constructed by using the node characteristics of the neighbor node, the node characteristics of the user node and the initial edge weight of each edge.

[0020] The cell handover method for load balancing provided by the present application, the measurement report information includes the reference signal receiving power and the signal to interference plus noise ratio; the edge weight of each edge for associating a neighbor node and a user node is determined based on the measurement report information, comprising:

[0021] The first quotient value is determined based on the reference signal receiving power of the candidate neighbor node associated with the neighbor node and the reference signal receiving power of the current cell;

[0022] The second quotient value is determined based on the signal to interference plus noise ratio of the candidate neighbor node associated with the neighbor node and the signal to interference plus noise ratio of the current cell;

[0023] The edge weight is determined based on the product of the first quotient value and the second quotient value.

[0024] The cell handover method for load balancing provided by the present application, the edge weight of each edge in the load distribution graph is predicted by using the edge weight prediction model, and the load distribution graph after the edge weight is updated, comprising:

[0025] determine the updated node feature of the neighbor node based on the fused node feature of the node feature of the neighbor node and the node feature of the user node adjacent to the neighbor node;

[0026] determine the updated node feature of the user node based on the fused node feature of the node feature of the user node and the node feature of the neighbor node adjacent to the user node;

[0027] determine the connection possibility score based on the updated node feature of the neighbor node and the updated node feature of the user node;

[0028] update the edge weight of each edge in the load distribution graph based on the connection possibility score to obtain the load distribution graph with updated edge weight.

[0029] The application further provides a cell switching device for load balancing, comprising:

[0030] a load distribution graph construction module configured to construct a load distribution graph by using neighbor nodes and user nodes, wherein the neighbor nodes are determined based on candidate neighbors of a current cell, the user nodes are determined based on transfer users connected to the current cell, and each edge of the load distribution graph is used to associate a neighbor node and a user node;

[0031] an edge weight updating module configured to predict the edge weight of each edge in the load distribution graph by using an edge weight prediction model to obtain the load distribution graph with updated edge weight;

[0032] a target cell switching module configured to determine a target cell for switching of the transfer user based on the load distribution graph with updated edge weight.

[0033] The application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the cell switching method for load balancing when executing the computer program.

[0034] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable on a processor to implement the cell switching method for load balancing.

[0035] The application further provides a computer program product comprising a computer program, wherein the computer program is executable on a processor to implement the cell switching method for load balancing.

[0036] The application provides a cell handover method, device and equipment for load balancing and a storage medium. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0038] Figure 1 FIG. 1 is a flowchart of the cell handover method for load balancing provided by the application.

[0039] Figure 2 FIG. 2 is an example diagram of the load distribution graph provided by the application.

[0040] Figure 3 FIG. 3 is a structural diagram of the cell handover device for load balancing provided by the application.

[0041] Figure 4 FIG. 4 is a structural diagram of the electronic device provided by the application. DETAILED DESCRIPTION

[0042] In order to make the objects, technical solutions and advantages of the application clearer, the technical solutions in the application will be described clearly and completely in combination with the drawings in the application. Obviously, the described embodiments are some embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the application.

[0043] It is to be understood that the terminology "includes", "has", "holds", "contains" or any other variant thereof is intended to cover a non-exclusive inclusion, such that processes, methods, articles, or apparatuses that comprise a list of elements are not limited to those elements but can include other elements not expressly listed or inherent to such processes, methods, articles, or apparatuses.

[0044] The terms "first", "second", and the like in the description do not denote any ordinal, chronological or spatial precedence or sequence, but are used for the purpose of differentiating a similar object. It should be understood that the data thus used can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than that illustrated or described herein, and the objects differentiated by "first", "second", etc. are generally a category and do not limit the number of objects, for example, the first object can be one or more.

[0045] The following describes the load balancing oriented cell handover method, device, equipment and storage medium provided by the present application. Figures 1-4 The present application provides a load balancing oriented cell handover method, device, equipment and storage medium.

[0046] In the 5G mobile communication system, the time-frequency resources of each cell are limited, so the number of users and the amount of traffic that can be accommodated are also limited.

[0047] On the one hand, in the spatial dimension, terminal users move randomly in the entire network, often showing uneven distribution. For example, in the scenario of large-scale sports events, concerts, etc., the crowd is densely gathered in the same area, which will concentrate the use of cell resources in that area.

[0048] On the other hand, in the time dimension, even for users facing relatively stable positions, the business demand will be different at different time periods in a day. For example, near a construction site, the data traffic volume during the lunch and evening break periods is much higher than at other times.

[0049] Therefore, the cell in the hotspot area and the hotspot time period is prone to high load problems, and resource congestion will lead to cell performance degradation and user experience rate decline. At the same time, other relatively idle low-load cells with the same coverage of different frequencies have the problem of resource waste. This phenomenon of load imbalance among multiple cells will lead to a decrease in the throughput of the entire network, seriously affecting user perception. In order to efficiently and fully utilize the resources of each cell, load balancing technology is needed to achieve traffic sharing among each cell, and users in high-load cells are transferred to low-load cells, i.e. switching the cell where the user is located, to achieve load balancing.

[0050] It should be noted that the execution subject of the load balancing oriented cell handover method provided by the application can be a server, a computer device, such as a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted electronic device, a wearable device, an Ultra-Mobile Personal Computer (UMPC), a netbook, a Personal Digital Assistant (PDA) or the like.

[0051] Figure 1 is a flowchart of the load balancing oriented cell handover method provided by the application, as shown in Figure 1 The load balancing oriented cell handover method includes but is not limited to steps 101 to 103.

[0052] Step 101: constructing a load distribution graph by using neighbor nodes and user nodes.

[0053] The neighbor nodes are determined based on candidate neighbor cells of a current cell; the user nodes are determined based on transfer users connected to the current cell; and each edge of the load distribution graph is used to associate a neighbor node and a user node.

[0054] The current cell, also referred to as a current primary serving cell, is a cell whose load exceeds a set threshold and is determined by a base station to be in a high load state. The cells in the same base station can directly exchange load information, and the cells in different base stations can send resource state request messages to other base stations through an interface between the base stations, and then accept the resource state information returned by the other base stations.

[0055] The cell configured with a neighbor relationship, in an active state, and different in frequency from the current cell and allowed to exchange load balancing information with the current cell is a neighbor cell (abbreviated as a neighbor cell). The neighbor cell and the current cell can be in the same base station or in different base stations.

[0056] The candidate neighbor cell is part or all of the neighbor cells that can be used for load redistribution and cell handover for users.

[0057] The transfer user is part or all of the users determined from all the users connected to the current cell and needing to hand over the cell.

[0058] Specifically, candidate neighbor cells are determined from all neighbor cells of the current cell, and transfer users are determined from all users connected to the current cell. Each candidate neighbor cell is taken as each neighbor cell node in the load distribution graph, and each transfer user is taken as each user node in the load distribution graph. All nodes in the load distribution graph are determined. Further, an edge is used to connect any neighbor cell node and any user node, so that each edge in the load distribution graph is used to associate a neighbor cell node and a user node.

[0059] It can be understood that each neighbor cell node in the load distribution graph corresponds to a candidate neighbor cell one by one, and each user node corresponds to a transfer user one by one.

[0060] Figure 2 is an example of the load distribution graph provided by the present application, as Figure 2 As shown, the left dashed box is a node set composed of neighbor cell nodes corresponding to three candidate neighbor cells, and the right dashed box is a node set composed of user nodes corresponding to five transfer users. A black undirected edge is used to associate each candidate neighbor cell and each transfer user. Finally, all neighbor cell nodes, all user nodes, and all edges jointly constitute the load distribution graph.

[0061] The load distribution graph belongs to a heterogeneous bipartite graph. On the one hand, a heterogeneous graph refers to at least one of the nodes and edges in the graph having multiple types. In the load distribution graph, there are two types of nodes, neighbor cell nodes and user nodes, so the load distribution graph belongs to a heterogeneous graph. On the other hand, the set of neighbor cell nodes and the set of user nodes are disjoint, and each edge in the load distribution graph is associated with two nodes belonging to the set of cell nodes and the set of user nodes, respectively, so the load distribution graph is a bipartite graph.

[0062] Step 102: predicting the edge weight of each edge in the load distribution graph using an edge weight prediction model to obtain an updated load distribution graph.

[0063] The edge weight prediction model is trained using a training sample set, and the training sample set includes multiple training samples. Each training sample includes a load distribution graph sample and its corresponding edge weight label.

[0064] Specifically, after constructing the load distribution graph using the neighbor cell nodes and the user nodes, the load distribution graph is input into a pre-trained edge weight prediction model. The pre-trained edge weight prediction model is used to predict the edge weight of each edge in the load distribution graph, and an updated load distribution graph is obtained.

[0065] Optionally, the edge weight prediction model is constructed based on a graph neural network model.

[0066] It should be noted that the edge weight prediction model can be constructed based on any one of a graph neural network (GNN), a relation-aware graph convolutional network (RAGCN), or the like, and the loss function of the edge weight prediction model can be defined as a commonly used loss function such as cross entropy, without limitation.

[0067] In an embodiment, each training sample is obtained based on the following manner:

[0068] In the same manner as the construction of the load distribution graph, the candidate neighboring area data and the transfer user data at the first historical moment are used to construct a load distribution graph sample at the first historical moment;

[0069] The candidate neighboring area data and the transfer user data at the second historical moment are used to construct a load distribution graph at the second historical moment, and the edge weight of each edge in the load distribution graph at the second historical moment is used as the edge weight label corresponding to the load distribution graph sample.

[0070] The first historical moment is prior to the second historical moment.

[0071] The batch training of the edge weight prediction model is performed using the training sample set, and after the model accuracy reaches the preset requirement, the edge prediction for cell switching can be performed.

[0072] Step 103: determining the target cell for switching of the transfer user based on the load distribution graph with updated edge weights.

[0073] Specifically, after obtaining the load distribution graph with updated edge weights, for each transfer user, the candidate neighboring area corresponding to the neighboring area node associated with the edge with the largest edge weight among all edges of the user node of the transfer user is determined as the target cell for switching of the transfer user, so as to switch the transfer user from the current cell to the target cell. In this way, after determining the target cells for switching of all transfer users, all transfer users are switched from the current cell with high load to the neighboring area with low load, and finally the load balancing is achieved.

[0074] For example, in the un-updated load distribution graph G, if the user node corresponding to the transfer user 1 is associated with the neighboring area nodes a, b, and c corresponding to the candidate neighboring areas A, B, and C through the edges a, β, and γ, respectively. The edge weights of the edges a, β, and γ are 0.2, 0.4, and 0.4, respectively.

[0075] After the load distribution graph G is input into the edge weight prediction model, the edge weight prediction model updates the edge weight of each edge in the load distribution graph, and in the load distribution graph G' after the edge weight is updated, the edge weights of the edge α, the edge β and the edge γ are 0.8, 0.1 and 0.1 respectively.

[0076] Therefore, the candidate neighboring cell A corresponding to the neighboring cell node a is taken as the target cell for the transfer user 1 to switch from the current cell to the candidate neighboring cell A. In this way, the target cells for all the transfer users to switch from the current cell to the target cells are determined, so that the users are switched from the current cell with high load to the neighboring cells with low load, and finally the load balancing is achieved.

[0077] The cell switching method for load balancing provided by the application converts the load balancing problem into the edge prediction problem of the heterogeneous bipartite graph by using the candidate neighboring cells and the transfer users as the nodes of the load distribution graph, and each edge of the load distribution graph is associated with a neighboring cell node and a user node, and further predicts the edge weight of the load distribution graph by using the neural network model, so as to determine the candidate neighboring cell corresponding to the neighboring cell node associated with the edge with the highest edge weight as the target cell of the transfer user, thereby fully utilizing the topological structure information of the communication network, efficiently mining and extracting useful features from the data, accurately matching the users to the most suitable target cells for switching, and achieving the optimal load balancing between cells.

[0078] Based on the above embodiment, as an optional embodiment, before the load distribution graph is constructed by using the neighboring cell nodes and the user nodes, the candidate neighboring cells are determined based on the neighboring cells of the current cell, including:

[0079] Based on the dimension load indexes of the neighboring cells, a comprehensive load index is determined.

[0080] Based on the comprehensive load index, the candidate neighboring cells are determined.

[0081] The dimension load indexes include at least two of the user number load index, the physical resource block (PRB) utilization rate load index and the control channel element (CCE) utilization rate load index, but are not limited to the above.

[0082] The PRB utilization rate load index is an index for comparing and measuring the service channel resource conditions of the neighboring cells and the current cell.

[0083] The CCE utilization rate load index is an index for comparing and measuring the control channel resource conditions of the neighboring cells and the current cell.

[0084] Specifically, before constructing the load distribution map by using the neighbor nodes and the user nodes, it is necessary to determine the candidate neighbor cells from the neighbor cells of the current cell and determine the transfer users from the users connected to the current cell.

[0085] For determining the candidate neighbor cells from the neighbor cells of the current cell, first, it is ensured that the neighbor cells are not in a high load state or in an interference state. Then, according to the dimension load indexes of each neighbor cell, a comprehensive load index of each neighbor cell is calculated. Further, according to the comprehensive load indexes in descending order, a first number of neighbor cells are determined as the candidate neighbor cells. Correspondingly, in the load distribution map, there are also a first number of neighbor nodes corresponding to the candidate neighbor cells one by one.

[0086] It can be understood that for judging whether the neighbor cells are in a high load state or in an interference state, the data required for calculating the user number load index, the PRB utilization rate load index and the CCE utilization rate load index can be obtained from the load interaction information between the current cell and the neighbor cells.

[0087] Optionally, the first number is determined according to a pre-set number threshold.

[0088] Optionally, the first number is determined according to a pre-set proportion value and the total number of neighbor cells.

[0089] In an embodiment, the user number load index is determined based on a downlink Radio Resource Control (RRC) connected state user number of the neighbor cell, a downlink RRC connected state user number of the current cell, a downlink RRC connected state user proportion of the neighbor cell and a downlink RRC connected state user proportion of the current cell.

[0090] Optionally, the expression for determining the user number load index of each neighbor cell based on the downlink RRC connected state user number of the neighbor cell, the downlink RRC connected state user number of the current cell, the downlink RRC connected state user proportion of the neighbor cell and the downlink RRC connected state user proportion of the current cell is as follows:

[0091] .

[0092] In an embodiment, the PRB utilization rate load index is determined based on a PRB number of a downlink data channel actually occupied by the neighbor cell, a PRB number of a downlink data channel available to the neighbor cell, a PRB number of a downlink data channel actually occupied by the current cell and a PRB number of a downlink data channel available to the current cell.

[0093] Optionally, the expression of the PRB utilization load index of each neighbor cell is determined based on the number of PRBs of the downlink data channel actually occupied by the neighbor cell, the number of PRBs of the downlink data channel available to the neighbor cell, the number of PRBs of the downlink data channel actually occupied by the current cell, and the number of PRBs of the downlink data channel available to the current cell, and is shown as follows:

[0094] .

[0095] In an embodiment, the CCE utilization load index is determined based on the number of CCEs of the equivalent-to-reference power used by the neighbor cell in downlink, the number of CCEs of the equivalent-to-reference power available to the neighbor cell in downlink, the number of CCEs of the equivalent-to-reference power used by the current cell in downlink, and the number of CCEs of the equivalent-to-reference power available to the current cell in downlink.

[0096] Optionally, the expression of the CCE utilization load index is determined based on the number of CCEs of the equivalent-to-reference power used by the neighbor cell in downlink, the number of CCEs of the equivalent-to-reference power available to the neighbor cell in downlink, the number of CCEs of the equivalent-to-reference power used by the current cell in downlink, and the number of CCEs of the equivalent-to-reference power available to the current cell in downlink, and is shown as follows:

[0097] .

[0098] In an embodiment, based on the dimension load indexes of the neighbor cell, a comprehensive load index is determined, including: determining the comprehensive load index based on the product of the user number load index, the PRB utilization load index, and the CCE utilization load index.

[0099] It can be seen that the user number load index, the PRB utilization load index, and the CCE utilization load index are all positively correlated with the comprehensive load index, and the lower the comprehensive load index, the greater the difference between the user number, the PRB utilization, and the CCE utilization of the neighbor cell and the current cell, and the more suitable the neighbor cell is as a candidate neighbor cell for the migrated user. Therefore, the neighbor cells of the current cell can be arranged in ascending order according to the calculated comprehensive load indexes, and the first M neighbor cells are selected as the candidate neighbor cells.

[0100] The cell handover method for load balancing provided by the application calculates a comprehensive load index from multi-dimensional load indexes such as a user number load index, a PRB utilization rate load index and a CCE utilization rate load index according to load interaction information of a current cell and neighboring cells, considers the network use of each neighboring cell by the user number load index, and considers the mutual influence between the current cell and the neighboring cells by the PRB utilization rate load index and the CCE utilization rate load index. The candidate neighboring cells are determined according to the size of the comprehensive load index of all neighboring cells, and the selection of the candidate neighboring cells is not dependent on the decision threshold of a single index set by experience, and thus has good generalization performance in different scenarios.

[0101] Based on the above embodiment, as an optional embodiment, before the load distribution map is constructed by using the neighboring cell nodes and the user nodes, the transfer user is determined based on the users connected to the current cell, and the method comprises the following steps of:

[0102] The transfer priority is determined based on the multi-dimensional priority of the user.

[0103] The transfer user is determined based on the transfer priority.

[0104] The multi-dimensional priority includes at least two of the following priorities: an Allocation and Retention Priority (ARP) priority, a Quality of Service (QoS) Class Identifier (QCI) priority, a PRB utilization rate priority and a far point priority.

[0105] The ARP priority is a priority based on an ARP threshold user selection strategy, and the user with a higher ARP priority should be more easily selected as the transfer user. The QCI priority is a priority based on a QCI threshold user selection strategy, and the user with a higher QCI priority should be more easily selected as the transfer user. The PRB utilization rate priority is a priority based on a PRB utilization rate threshold user selection strategy, and the user with a higher PRB utilization rate priority should be more easily selected as the transfer user. The far point priority is a priority based on a far point threshold user selection strategy, and the user with a higher far point priority should be more easily selected as the transfer user.

[0106] Specifically, for determining the transfer users from the users connected to the current cell, a transfer priority of each user is calculated according to a dimension priority of each user. The dimension priority includes at least two of the following priorities: ARP priority, QCI priority, PRB utilization priority, and far point priority. Then, the second number of users is determined as the transfer users in descending order of the transfer priority, such as selecting the first N users as the transfer users. Correspondingly, in the load distribution graph, there are also the second number of user nodes corresponding to the transfer users one by one.

[0107] Optionally, the second number is determined according to a pre-set number threshold.

[0108] Optionally, the second number is determined according to a product of a pre-set ratio value and the total number of users.

[0109] In an embodiment, the transfer priority is determined based on the dimension priority of the user, including: determining the transfer priority based on a weighted sum of the ARP priority, the QCI priority, the PRB utilization priority, and the far point priority of the user.

[0110] Optionally, in an embodiment, the expression for determining the transfer priority based on the dimension priority of the user is as follows:

[0111] ;

[0112] wherein, is a weight coefficient of the ARP priority; is a weight coefficient of the QCI priority; is a weight coefficient of the PRB utilization priority; is a weight coefficient of the far point priority.

[0113] Optionally, .

[0114] The cell switching method for load balancing provided by the application calculates the transfer priority from the ARP priority, the QCI priority, the PRB utilization priority, and the far point priority of the user in multiple dimensions, and makes the selection of the transfer user independent of the decision threshold of a single index relying on experience, and thus has good generalization performance in different scenarios.

[0115] In another embodiment, before the load distribution graph is constructed by using the neighbor node and the user node, the candidate neighbor is determined based on the neighbor of the current cell, and then the transfer user is determined based on the user connected to the current cell.

[0116] By optimizing the determination method of the candidate neighboring cell and the transfer user, a more suitable candidate neighboring cell can be matched as a target cell for the transfer user, which helps to effectively reduce the load of the current cell, avoids causing high load of the neighboring cell, and reduces the impact on the transfer user as much as possible.

[0117] Based on the above embodiments, as an optional embodiment, the load distribution graph is constructed by using the neighboring cell nodes and the user nodes, including:

[0118] Obtaining measurement report information; the measurement report information is measured by the transfer user after receiving the event information of the candidate neighboring cell; the event information of the candidate neighboring cell is issued by the current cell to the transfer user;

[0119] Based on the measurement report information, determining the initial edge weight of each edge for associating a neighboring cell node and a user node;

[0120] Using the node characteristics of the neighboring cell nodes, the node characteristics of the user nodes, and the initial edge weight of each edge, the load distribution graph is constructed.

[0121] The node characteristics are information related to a specific node.

[0122] Specifically, the current cell issues a measurement command to the determined N transfer users, and issues the event information of M candidate neighboring cells to the transfer users. After receiving the measurement command and the event information, the transfer users perform measurement and upload the measurement report information to the current cell if the event decision condition is met, so that the current cell obtains the measurement report information. Among them, the measurement report information uploaded by the transfer user includes but is not limited to the reference signal receiving power (Reference Signal Receiving Power, RSRP) and signal to interference plus noise ratio (Signal to Interference plus Noise Ratio, SINR) of the candidate neighboring cell and the current cell.

[0123] According to the measurement report information uploaded by the transfer user, the initial edge weight of each edge for associating the user node corresponding to the transfer user and each neighboring cell node can be determined by a certain calculation method, so as to determine the initial edge weight of each edge for associating a neighboring cell node and a user node in the load distribution graph. Further, the node characteristics of the neighboring cell nodes, the user characteristics of the user nodes, and the initial edge weight of each edge are used to construct the load distribution graph.

[0124] Optionally, the event information includes but is not limited to at least one of the related information of the frequency point A4 or A5 event of the candidate neighboring cell and other event information.

[0125] Optionally, the node feature of each neighbor node and the node feature of each user node are determined based on the following manner:

[0126] The node set of all candidate neighbors is determined as A The node set of all transfer users is determined as B .

[0127] The node feature of each neighbor node is defined as a d A dimensional vector, which is composed of the values of the fields of the number of cell RRC connected users, the proportion of cell RRC connected users, the cell PRB utilization rate, the cell interference, the cell uplink bandwidth, the cell downlink bandwidth, etc. of the candidate neighbor corresponding to the neighbor node, and is obtained through encoding conversion.

[0128] The node feature of each user node is defined as a d B dimensional vector, which is composed of the values of the fields of the ARP priority, the QCI priority, the PRB utilization rate priority and the far point priority, etc. of the transfer user corresponding to the user node, and is obtained through encoding conversion.

[0129] The cell switching method for load balancing provided by the application, after determining the candidate neighbors and the transfer users, the measurement report information of each transfer user relative to each candidate neighbor is obtained by issuing the measurement command and the event information of the candidate neighbor to the transfer user, and the initial edge weight of each edge for associating a neighbor node and a user node in the load distribution graph is determined according to the measurement report information, thereby improving the accuracy of determining the edge weight of each edge in the load distribution graph, and helping to improve the accuracy of matching the most suitable target cell for the transfer user.

[0130] Based on the above embodiment, as an optional embodiment, the measurement report information includes RSRP and SINR; the determination of the edge weight of each edge for associating a neighbor node and a user node based on the measurement report information includes:

[0131] determining a first quotient value based on the RSRP of the candidate neighbor associated with the neighbor node and the RSRP of the current cell;

[0132] determining a second quotient value based on the SINR of the candidate neighbor associated with the neighbor node and the SINR of the current cell;

[0133] determining the edge weight based on the product of the first quotient value and the second quotient value.

[0134] Specifically, the RSRP of a plurality of candidate neighboring cells covering the transfer user and the RSRP of the current cell of the transfer user are included in the measurement report information uploaded by the transfer user, when determining the edge weight of each edge, a first quotient value is determined according to the RSRP of the candidate neighboring cell associated with the neighboring cell node and the RSRP of the current cell, a second quotient value is determined according to the SINR of the candidate neighboring cell associated with the neighboring cell node and the SINR of the current cell, and finally the product of the first quotient value and the second quotient value is determined as the edge weight.

[0135] Optionally, the expression for determining the edge weight of each edge for associating a neighboring cell node and a user node based on the measurement report information is as follows:

[0136] .

[0137] For example, the edge weight between the neighboring cell node of a candidate neighboring cell and the user node of a transfer user is equal to the product of a first quotient value and a second quotient value, the first quotient value is equal to the quotient of the RSRP of the candidate neighboring cell measured by the transfer user and the RSRP of the current cell measured by the transfer user, and the second quotient value is equal to the SINR of the candidate neighboring cell measured by the transfer user and the SINR of the current cell measured by the transfer user. A i B i A i A i

[0138] The cell handover method for load balancing provided by the present application can set a larger edge weight between the user node corresponding to the transfer user and the neighboring cell node corresponding to the candidate neighboring cell of the current cell when the RSRP and the SINR measured by the transfer user are closer to the candidate neighboring cell of the current cell, so that when determining the target cell for the transfer user, the candidate neighboring cell similar to the current cell can be determined as the target cell, thereby reducing the impact on the user in the connected state when switching cells.

[0139] Based on the above embodiment, as an optional embodiment, the edge weight of each edge in the load distribution graph is predicted by using the edge weight prediction model to obtain the load distribution graph after the edge weight is updated, which includes:

[0140] Based on the node feature of the neighboring cell node and the node feature of the user node adjacent to the neighboring cell node after fusion, the updated node feature of the neighboring cell node is determined;

[0141] Based on the node feature of the user node and the node feature of the neighboring cell node adjacent to the user node after fusion, the updated node feature of the user node is determined.​​​​

[0142] Based on the updated node characteristics of the neighboring nodes and the updated node characteristics of the user nodes, a connection probability score is determined;

[0143] Based on the connection probability score, the edge weight of each edge in the load distribution graph is updated to obtain the load distribution graph with updated edge weights.

[0144] Specifically, after constructing a load distribution graph using the node features of all neighboring nodes, the node features of all user nodes, and the initial edge weights of each edge, the load distribution graph is input into a trained edge weight prediction model. Each layer of the edge weight prediction model updates the node features of neighboring nodes and user nodes, and calculates the connection probability score based on the updated node features of neighboring nodes and user nodes. The calculated connection probability score is used as the new edge weight of the edge connecting neighboring nodes and user nodes, resulting in a load distribution graph with updated edge weights.

[0145] In one embodiment, the expression for determining the updated node features of the neighboring node based on the fused node features of the neighboring node and the node features of the user nodes adjacent to the neighboring node is as follows:

[0146] ;

[0147] in, l It is the first edge weight prediction model l layer; It is a neighboring node A i Updated node features; It is a neighboring node A i Its own d A Dimensional node features; N A,i It is a neighboring node A i The set of all adjacent user nodes; It is a set N A,i The absolute value of the set represents the absolute value of the set. N A,i The number of elements in the data, i.e., neighboring nodes. A i The number of all adjacent user nodes; It is a set N A,i The Middle j User nodes B j Node characteristics; is is the corresponding weight matrix, which is a parameter obtained by training the edge weight prediction model using a training sample set; is is the corresponding weight matrix, which is a parameter obtained by training the edge weight prediction model using a training sample set; is an activation function.

[0148] In an embodiment, the expression for determining the updated node feature of the user node based on the fused node features of the node features of the user node and the node features of the adjacent area nodes adjacent to the user node is as follows:

[0149] ;

[0150] wherein, l is the i-th layer of the edge weight prediction model; l ; user node B i updated node feature; is the node feature of the user node B i itself; d B dimension node feature; N B,i is the node feature of the user node B i all adjacent adjacent area nodes; is the set of adjacent area nodes N B,i , which represents the number of elements in the set N B,i , i.e., the number of adjacent area nodes B i adjacent to the user node; is the i-th adjacent area node N B,i in the set j ; A j node feature; is the corresponding weight matrix, which is a parameter obtained by training the edge weight prediction model using a training sample set; is the corresponding weight matrix, which is a parameter obtained by training the edge weight prediction model using a training sample set; is an activation function.

[0151] In an embodiment, the expression for determining the connection possibility score based on the updated node features of the adjacent area nodes and the updated node features of the user node is as follows:

[0152] ;

[0153] wherein, l is the i-th layer of the edge weight prediction model; l ; is the connection probability score of the neighboring node A i and the user node B i ; is the updated node feature of the neighboring node A i ; is the updated node feature of the user node B i ; is a weight matrix with a dimension of d A ; d B is a parameter obtained by training the edge weight prediction model using the training sample set.

[0154] It can be understood that, based on the heterogeneous bipartite graph property of the load distribution graph, any neighboring node of a neighboring node is a user node, and any neighboring node of a user node is a neighboring node.

[0155] The cell handover method for load balancing provided by the application updates itself by using the node features of all neighboring user nodes of the neighboring node, updates itself by using the node features of all neighboring neighboring nodes of the user node, and updates the edge weight of each edge by calculating the connection probability score using the updated node features, so as to determine the candidate neighboring area corresponding to the neighboring node associated with the edge with the highest updated edge weight as the target cell of the transferred user, thereby fully utilizing the topological structure information of the communication network, efficiently mining and extracting useful features from data, accurately matching the user to the most suitable target cell for handover, and achieving optimal load balancing between cells.

[0156] Overall, the cell handover method for load balancing provided by the application is aimed at the load balancing problem existing in the actual 5G network. The load distribution graph is designed by abstractly modeling the candidate neighboring cells, the transferred users and the relationship between the two. A load distribution graph dataset is constructed using historical data such as network load state and load balancing strategy. The edge weight prediction model is trained through the definition of input and output, and the technical problem of load balancing is converted into the edge prediction problem of the heterogeneous bipartite graph. The resource gap between the current cell and the neighboring cell is compared based on multiple comprehensive indicators such as the number of users, PRB utilization and CCE utilization, and the candidate neighboring cell is determined according to the index ranking, avoiding setting the candidate neighboring cell threshold which depends on experience and improving the generalization ability for different scenarios. Active feature learning is performed based on the graph neural network model, which can fully integrate the topological structure, load state and measurement results of the 5G network and other multi-dimensional information, efficiently extract useful features from complex network data, and provide a basis for accurate matching of candidate neighboring cells and transferred users from the perspective of data structure design. The prediction ability of the trained model is used to realize one-to-many matching between the candidate neighboring cells and the transferred users, and the most suitable neighboring cell is selected for the transferred user to switch according to the predicted edge weight, which can improve the decision efficiency of load balancing and provide an efficient auxiliary tool for intelligent and automated future decision-making of load balancing.

[0157] Figure 3 is a structural schematic diagram of the cell handover device for load balancing provided by the application, as Figure 3 indicated, the cell handover device for load balancing includes but is not limited to a load distribution graph construction module 301, an edge weight updating module 302 and a target cell handover module 303.

[0158] The load distribution graph construction module 301 is used to construct a load distribution graph by using neighboring cell nodes and user nodes. The neighboring cell nodes are determined based on the candidate neighboring cells of the current cell. The user nodes are determined based on the transferred users connected to the current cell. Each edge of the load distribution graph is used to associate a neighboring cell node and a user node.

[0159] The edge weight updating module 302 is used to predict the edge weight of each edge in the load distribution graph by using an edge weight prediction model, and obtain an updated load distribution graph.

[0160] The target cell handover module 303 is used to determine the target cell for the transferred user to switch based on the updated load distribution graph.

[0161] It should be noted that the cell handover device for load balancing provided by the application can execute the cell handover method for load balancing described in any of the above embodiments when it is actually running, and this embodiment will not be repeated here.

[0162] The application provides a cell switching device for load balancing, which is constructed as a heterogeneous bipartite graph by using candidate neighboring cells and transfer users as nodes of a load distribution graph, and each edge of the load distribution graph is associated with a neighboring cell node and a user node, so as to convert the load balancing problem into an edge prediction problem of the heterogeneous bipartite graph, and further predict the edge weight of the load distribution graph by using a neural network model, so as to determine the candidate neighboring cell corresponding to the neighboring cell node associated with the edge with the highest edge weight as the target cell of the transfer user, thereby fully utilizing the topological structure information of the communication network, efficiently mining and extracting useful features from data, accurately matching the user to the target cell most suitable for switching, and achieving optimal load balancing between cells.

[0163] Figure 4 is a structural schematic diagram of an electronic device provided by the application, as shown in Figure 4 The electronic device can include a processor 410, a communications interface 420, a memory 430 and a communications bus 440, wherein the processor 410, the communications interface 420 and the memory 430 complete mutual communication through the communications bus 440. The processor 410 can invoke a logical instruction in the memory 430 to execute the load balancing-oriented cell switching method provided by any of the above-mentioned embodiments, which includes but is not limited to the following steps: constructing a load distribution graph by using neighboring cell nodes and user nodes; the neighboring cell nodes are determined based on candidate neighboring cells of a current cell; the user nodes are determined based on transfer users connected to the current cell; each edge of the load distribution graph is used to associate a neighboring cell node and a user node; predicting the edge weight of each edge in the load distribution graph by using an edge weight prediction model to obtain a load distribution graph with updated edge weight; and determining a target cell for switching of the transfer user based on the load distribution graph with updated edge weight.

[0164] Further, the logic instructions in the memory 430 described above can be implemented in the form of software functional units and sold or used as standalone products, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or partially contribute to the prior art, or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0165] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to execute the load balancing oriented cell handover method provided by any of the above embodiments. The load balancing oriented cell handover method comprises but is not limited to the following steps: constructing a load distribution graph by using neighbor nodes and user nodes; the neighbor nodes are determined based on candidate neighbor cells of a current cell; the user nodes are determined based on transfer users connected to the current cell; each edge of the load distribution graph is used to associate a neighbor node and a user node; predicting edge weights of each edge in the load distribution graph by using an edge weight prediction model to obtain an updated load distribution graph of the edge weights; and determining a target cell for the transfer user to hand over based on the updated load distribution graph of the edge weights.

[0166] In another aspect, the present application also provides a non-transitory computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the load balancing oriented cell handover method provided by any of the above embodiments. The load balancing oriented cell handover method comprises but is not limited to the following steps: constructing a load distribution graph by using neighbor nodes and user nodes; the neighbor nodes are determined based on candidate neighbor cells of a current cell; the user nodes are determined based on transfer users connected to the current cell; each edge of the load distribution graph is used to associate a neighbor node and a user node; predicting edge weights of each edge in the load distribution graph by using an edge weight prediction model to obtain an updated load distribution graph of the edge weights; and determining a target cell for the transfer user to hand over based on the updated load distribution graph of the edge weights.

[0167] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0168] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0169] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A load balancing oriented cell handover method, characterized in that, The method comprises the following steps: constructing a load distribution graph by using neighbor nodes and user nodes; the neighbor nodes are determined based on candidate neighbors of a current cell; the user nodes are determined based on transfer users connected to the current cell; each edge of the load distribution graph is used to associate a neighbor node and a user node; predicting an edge weight of each edge in the load distribution graph by using an edge weight prediction model to obtain an updated load distribution graph after the edge weight; 2. The load balancing oriented cell handover method according to claim 1, characterized in that, determining a target cell for handover of the transfer user based on the updated load distribution graph after the edge weight. Before the step of constructing the load distribution graph by using the neighbor nodes and the user nodes, the candidate neighbors are determined based on neighbors of the current cell, and the method comprises the following steps: determining a comprehensive load index based on each dimension load index of the neighbors; 3. The load balancing oriented cell handover method according to claim 1, c h a r a c t e r i z e d b y, determining the candidate neighbors based on the comprehensive load index. Before the step of constructing the load distribution graph by using the neighbor nodes and the user nodes, the transfer users are determined based on users connected to the current cell, and the method comprises the following steps: determining a transfer priority based on each dimension priority of the users; 4. The load balancing oriented cell handover method according to claim 1, characterized in that, determining the transfer users based on the transfer priority. The step of constructing the load distribution graph by using the neighbor nodes and the user nodes comprises the following steps: obtaining measurement report information; the measurement report information is measured by the transfer users after receiving event information of the candidate neighbors; the event information of the candidate neighbors is sent to the transfer users by the current cell; determining an initial edge weight of each edge used to associate a neighbor node and a user node based on the measurement report information; 5. The load balancing oriented cell handover method according to claim 4, characterized in that, constructing the load distribution graph by using node features of the neighbor nodes, node features of the user nodes and the initial edge weight of each edge. The measurement report information comprises a reference signal receiving power and a signal to interference plus noise ratio; The step of determining the edge weight of each edge used to associate a neighbor node and a user node based on the measurement report information comprises the following steps: determining a first quotient value based on a reference signal receiving power of a candidate neighbor associated with the neighbor node and a reference signal receiving power of the current cell; determining a second quotient value based on a signal to interference plus noise ratio of the candidate neighbor associated with the neighbor node and a signal to interference plus noise ratio of the current cell; 6. The load balancing oriented cell handover method according to claim 1, characterized in that, determining the edge weight based on a product of the first quotient value and the second quotient value. The step of predicting the edge weight of each edge in the load distribution graph by using the edge weight prediction model to obtain the updated load distribution graph after the edge weight comprises the following steps: determining updated node features of the neighbor nodes based on node features of the neighbor nodes and node features of user nodes adjacent to the neighbor nodes after fusion; determining updated node features of the user nodes based on node features of the user nodes and node features of neighbor nodes adjacent to the user nodes after fusion; determining a connection possibility score based on the updated node features of the neighbor nodes and the updated node features of the user nodes; Based on the connection possibility score, an edge weight of each edge in the load distribution graph is updated to obtain an updated load distribution graph.

7. A load balancing oriented cell handover apparatus, characterized by comprising: Comprise: A load distribution graph construction module, configured to construct a load distribution graph by using neighbor cell nodes and user nodes; The neighbor cell nodes are determined based on candidate neighbor cells of a current cell; The user nodes are determined based on transfer users connected to the current cell; each edge of the load distribution graph is used to associate a neighbor cell node and a user node; An edge weight updating module, configured to predict an edge weight of each edge in the load distribution graph by using an edge weight prediction model to obtain an updated load distribution graph; A target cell switching module, configured to determine a target cell for switching of the transfer users based on the updated load distribution graph.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the load balancing-oriented cell switching method of any one of claims 1 to 6. 9.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the load balancing-oriented cell switching method of any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the load balancing-oriented cell switching method of any one of claims 1 to 6.

Images