Social network-based node communication method, apparatus, medium, and device

Abstract

Embodiments of the present application provide a social network-based node communication method, device, medium and equipment. The method comprises: generating a first node feature corresponding to a stranger node at a current time according to a task requirement of the stranger node and interaction information between the stranger node and each network node in a target social network before the current time; generating initial environment information and updating to obtain target environment information according to association information between the network nodes in the target social network and second node features corresponding to each network node; inputting the target environment information and the first node feature into a reinforcement learning module to determine a reward corresponding to each possible action of the stranger node to obtain an attention value corresponding to each network node in the target social network, thereby determining a communication strategy of the stranger node. The technical solution of the embodiments of the present application can improve the rationality of communication resource allocation, thereby ensuring the efficiency and accuracy of information exchange between nodes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mobile network, in particular to a node communication method and device based on social network, medium and equipment. BACKGROUND

[0002] With the rapid development of modern mobile network technology, vehicles, mobile devices and other intelligent nodes have not only become simple tools or devices, but also become efficient nodes capable of collecting, processing and sharing data in real time. For example, in intelligent transportation systems (ITS), communication between vehicles and vehicles (V2V), vehicles and infrastructure (V2I), vehicles and pedestrians (V2P), and more widely, vehicles and anything (V2X), greatly improves road safety, increases traffic efficiency, and provides users with rich information services.

[0003] However, in a dynamic and constantly changing mobile network environment, traditional mobile networks often use static or predefined resource allocation rules, resulting in inefficient use of network resources, especially in high-density information flow and complex urban environments. In addition, the communication between nodes in existing mobile networks lacks effective personalization and context awareness, and the communication method often cannot meet the specific needs of different nodes at different times and places, affecting the efficiency and accuracy of information exchange. Therefore, how to improve the rationality of communication resource allocation and ensure the efficiency and accuracy of information exchange between nodes has become a technical problem to be solved. SUMMARY

[0004] Embodiments of the present application provide a node communication method and device based on social network, medium and equipment, which can at least improve the rationality of communication resource allocation and ensure the efficiency and accuracy of information exchange between nodes.

[0005] Other characteristics and advantages of the present application will become apparent from the following detailed description, or will be learned by practice of the present application.

[0006] According to an aspect of an embodiment of the present application, a node communication method based on social network is provided, comprising:

[0007] generating a first node feature corresponding to the stranger node at the current time according to the task demand of the stranger node and the interaction information between the stranger node and each network node in the target social network before the current time;

[0008] generating initial environment information according to the association information between the network nodes in the target social network and the second node features corresponding to each network node;

[0009] based on the initial environment information, updating the second node features corresponding to each of the network nodes by using a pre-constructed graph attention network to obtain target environment information;

[0010] inputting the target environment information and the first node features into a pre-constructed reinforcement learning module, so that the reinforcement learning module determines rewards corresponding to each possible action of the stranger node according to the first node features and the target environment information;

[0011] processing the rewards corresponding to each possible action of the stranger node to obtain attention values corresponding to each of the network nodes in the target social network;

[0012] determining a communication strategy between the stranger node and the target social network according to the attention values corresponding to each of the network nodes in the target social network.

[0013] According to an aspect of an embodiment of the present application, a node communication device based on a social network is provided, comprising:

[0014] a first generation module configured to generate first node features corresponding to the stranger node at a current time according to task requirements of the stranger node and interaction information between the stranger node and each network node in a target social network before the current time;

[0015] a second generation module configured to generate initial environment information according to association information between the network nodes in the target social network and second node features corresponding to each of the network nodes;

[0016] an environment updating module configured to update the second node features corresponding to each of the network nodes by using a pre-constructed graph attention network based on the initial environment information to obtain target environment information;

[0017] an input module configured to input the target environment information and the first node features into a pre-constructed reinforcement learning module, so that the reinforcement learning module determines rewards corresponding to each possible action of the stranger node according to the first node features and the target environment information;

[0018] a conversion module configured to process the rewards corresponding to each possible action of the stranger node to obtain attention values corresponding to each of the network nodes in the target social network;

[0019] a processing module configured to determine a communication strategy between the stranger node and the target social network according to the attention values corresponding to each of the network nodes in the target social network.

[0020] According to an aspect of some embodiments of the present application, a computer readable medium is provided, and the computer readable medium has a computer program stored thereon, and the computer program, when executed by a processor, implements the social network based node communication method as described in the above embodiments.

[0021] According to an aspect of some embodiments of the present application, an electronic device is provided, and the electronic device includes one or more processors, and a storage configured to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the social network based node communication method as described in the above embodiments.

[0022] According to an aspect of some embodiments of the present application, a computer program product or a computer program is provided, and the computer program product or the computer program includes computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the social network based node communication method as described in the above embodiments.

[0023] In the technical solutions provided by some embodiments of the present application, the first node feature corresponding to the stranger node at the current moment is generated according to the task demand of the stranger node and the interaction information between the stranger node and each network node in the target social network before the current moment, then the initial environment information is generated according to the association relationship between the network nodes in the target social network and the second node features corresponding to the network nodes, the second node features corresponding to the network nodes are updated based on the initial environment information by using the pre-constructed graph attention network to obtain target environment information, and the first node feature and the target environment information are input into the pre-constructed reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the stranger node according to the first node feature and the target environment information, processes the reward corresponding to each possible action of the stranger node to obtain the attention value corresponding to each network node in the target social network, and determines the communication strategy between the stranger node and the target social network according to the attention value. Therefore, the rationality of the communication resource allocation can be improved, and the efficiency and accuracy of the information exchange between the nodes can be ensured.

[0024] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0025] The accompanying drawings, which are incorporated herein and constitute part of the specification, illustrate embodiments consistent with the application and, together with the description, further serve to explain the principles of the application. It is to be understood that the drawings are only schematic, and that they do not necessarily correspond to the actual relative sizes of the components. It is also to be understood that the embodiments shown in the drawings are merely meant as illustrative examples of the application, and that, for those skilled in the art, other drawings can be derived from the drawings without paying creative labor.

[0026] Figure 1 Fig. 1 shows a flow diagram of a method for social network based node communication according to an embodiment of the application;

[0027] Figure 2 Fig. 2 shows a processing flow diagram of a reinforcement learning module and a graph neural network module in a method for social network based node communication according to an embodiment of the application;

[0028] Figure 3 Fig. 3 shows a block diagram of a device for social network based node communication according to an embodiment of the application;

[0029] Figure 4 Fig. 4 shows a structure diagram of a computer system of an electronic device suitable for implementing embodiments of the application. DETAILED DESCRIPTION

[0030] Example implementations will now be described with reference to the drawings. However, example implementations can be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these implementations are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example implementations to those skilled in the art.

[0031] Moreover, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the application. One skilled in the relevant art will recognize, however, that the

[0032] The block diagrams in the drawings show only the functionality of the embodiments and do not necessarily imply a physical or an architectural arrangement of the same. That is, the functionality can be implemented in software, hardware, or a combination thereof. The embodiments can be implemented in a distributed manner, in a centralized manner, or in a combination of both.

[0033] The flowchart shown in the drawing is only an exemplary illustration, and is not necessarily required to include all contents and operations / steps, nor is it necessarily required to be executed in the order described. For example, some operations / steps can be further decomposed, and some operations / steps can be combined or partially combined, so that the actual execution order can be changed according to actual conditions.

[0034] Figure 1 A flowchart of a social network-based node communication method according to an embodiment of the present application is shown. The method can be applied in a terminal device or a server, where the terminal device can include one or more of a smartphone, a tablet computer, a portable computer, and a desktop computer, and the server can be a physical server or a cloud server. In an example, the method can be applied in a Road Side Unit (RSU) or a base station, i.e., an access point described below.

[0035] Referring to Figure 1 The social network-based node communication method includes at least steps S110 to S160, which are described in detail as follows (the method is described below by taking application in a server as an example):

[0036] In step S110, a first node feature corresponding to a current time of a stranger node is generated according to a task requirement of the stranger node and interaction information between the stranger node and each network node in a target social network before the current time.

[0037] The social network can be obtained by pre-dividing a plurality of existing nodes, i.e., the server can perform community division according to node information of a plurality of existing nodes according to a predetermined division rule to obtain at least one social network.

[0038] The target social network can be a social network corresponding to the stranger node, for example, the stranger node is within the communication range of a certain social network, and the stranger node communicates with the network nodes in the social network, so the social network is the target social network corresponding to the stranger node.

[0039] The stranger node can be a communication node that temporarily accesses the target social network or accesses the target social network for a short time and does not generate a sufficient number of interaction behaviors, and relatively, other nodes in the social network are acquaintance nodes.

[0040] In an embodiment, the server can obtain a task requirement of the stranger node, and the task requirement can be a parameter attribute of a current task of the stranger node. In an example, the task requirement can include three dimensions, i.e., communication, perception, and calculation, where the communication dimension includes transmit power Acceptance sensitivity Frequency bandwidth The perception dimension includes a vehicle perception range The sensor parameter The computation dimension includes a computation task size The computation complexity and so on.

[0041] Meanwhile, the server can obtain, according to the interaction behaviors between the stranger node and the network nodes in the target social network, interaction information between the stranger node and the network nodes in the target social network before the current moment, which can include but is not limited to a time stamp T0 of the stranger node entering the target social network (i.e., a moment when the stranger node first interacts with the center node in the target social network), coordinates (X0, Y0) of the stranger node at the current moment (the coordinates can take the coordinates of the center node of the target social network as the coordinate origin), an interaction frequency of the stranger node with the access points and the acquaintance nodes (such as vehicles, mobile devices carried by pedestrians, etc.) in the target social network before the current moment and so on.

[0042] After obtaining the above information, the server can generate a first node feature corresponding to the stranger node at the current moment according to the task demand and the interaction information. Specifically, the server can encode and splice the above information to generate the corresponding first node feature. For example, the first node feature can be represented as

[0043] In step S120, initial environment information is generated according to the association information between the network nodes in the target social network and the second node features corresponding to the network nodes.

[0044] In this embodiment, the server can obtain the association information between the network nodes in the target social network and the second node features corresponding to the network nodes. The association information can be the connection relationship between the network nodes in the social network, i.e., if there is a connected edge between two network nodes, there is an association relationship between the two network nodes. The second node feature can be obtained by encoding the node information corresponding to each network node. In an example, the node information can include but is not limited to one or more of position, speed, capacity, service type, coverage range. The server can integrate the two to obtain the corresponding initial environment information.

[0045] In step S130, based on the initial environment information, the second node features corresponding to the network nodes are updated by using a pre-constructed graph attention network to obtain target environment information.

[0046] In this embodiment, the graph attention network can be pre-constructed and stored by those skilled in the art. The graph attention network can use an attention mechanism to emphasize the features of important nodes, thereby assisting the subsequent reinforcement learning module to more accurately evaluate the impact of each network node on the decision.

[0047] Specifically, the server can extract the graph relationships (i.e., the associations between network nodes) and the second node features of each network node from the initial environment information. Then, the obtained second node features are merged to obtain a feature matrix. This feature matrix and graph relationships are input into a graph attention network to update the second node features of each network node, and the updated second node features are used as the target environment information.

[0048] In one example, the graph attention network can update the feature matrix according to the following formula:

[0049]

[0050] Where, α ij It is the attention coefficient, h j σ represents the features of neighboring nodes, W is the weight matrix, and σ is the activation function.

[0051] It's worth noting that during feature matrix updates, graph attention networks can also utilize temporal and spatial information to optimize the calculation of attention coefficients, better reflecting the dynamic relationships between nodes. Specifically, regarding temporal information, each node's feature information includes the timestamp of its entry into or update within the social network. When calculating attention coefficients, weights can be adjusted based on the proximity of these timestamps. Interactions between nodes with more recent timestamps are more important than those with more distant timestamps, thus edges with more recent timestamps can be given higher attention weights. As for spatial information, each node's feature information includes its current coordinates. Graph networks inherently possess spatial information; therefore, attention coefficients with different weights can be assigned by considering the physical distance between nodes.

[0052] Please continue to refer to this. Figure 1 In step S140, the target environment information and the first node features are input into a pre-built reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the unfamiliar node based on the first node features and the target environment information.

[0053] In this embodiment, after updating the second node features corresponding to each network node to obtain the target environment information, the target environment information and the first node features corresponding to the stranger node can be input into the pre-constructed reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the stranger node according to the first node features and the target environment information.

[0054] It should be noted that when constructing the reinforcement learning module, the actions of the agent need to be defined, which includes how many resources the agent needs to allocate when interacting with the network nodes in the environment, and the state transition function of the agent and the reward function of the agent after interacting with the environment also need to be defined. The reward function can include the feedback of each network node in the environment to the agent.

[0055] Specifically, the reward function can be dynamically adjusted based on the interaction results between the agent and the network nodes. For example, if the interaction between the agent and a certain network node brings positive results (such as successful data exchange, good communication quality, or short response time, etc.), a positive reward is given; if it brings negative results (such as connection failure, data loss, or response time exceeding the delay tolerance, etc.), a negative reward is given.

[0056] In an example, the reward function can be as shown in the following formula:

[0057]

[0058] wherein R t represents the total reward of time step t; r t represents the immediate reward of time t; γ is a discount factor, which measures the importance of future rewards, 0≤γ≤1; Tmax represents the maximum value of time steps.

[0059] In an embodiment of the present application, as Figure 2 shown, after the reinforcement learning module is constructed, the reinforcement learning module can generate corresponding initial environment information as the environment when the agent acts according to the pre-constructed social network, that is, the step S120 can be performed by the reinforcement learning module. When needed, the reinforcement learning module can input the generated environment (i.e., the aforementioned initial environment information) into the graph neural network module to extract the graph relationship and construct the feature matrix as described above, and input them into the graph attention network to update the node features to obtain the target environment information. The graph attention network can feed back the target environment information, i.e., the updated second node features, to the reinforcement learning module for subsequent processing by the reinforcement learning module according to the target environment information. In this way, the complex node relationship processed by the graph attention network can optimize the decision quality of the reinforcement learning strategy.

[0060] In an embodiment of the present application, after the target environment information and the first feature vector are input into the pre-constructed reinforcement learning module to make the reinforcement learning module determine the reward corresponding to each possible action of the stranger node according to the first node feature and the target environment information, the method further comprises:

[0061] The graph attention network is optimized according to the reward and state generated by the reinforcement learning module.

[0062] In this embodiment, it should be understood that the parameters in the graph attention network are randomly initialized, and for a social network, there is no real label for each node and edge for training, so, as shown in Figure 2 The action result of the reinforcement learning module, i.e., the generated reward and state, can feedback the adjustment of the edge weight and the update of the node feature of the graph attention network, thereby forming a closed-loop control to ensure the accuracy of the output result of the graph attention network.

[0063] Please continue to refer to Figure 1 In step S150, the reward corresponding to each possible action of the stranger node is processed to obtain the attention value corresponding to each network node in the target social network.

[0064] In this embodiment, the server can process the reward corresponding to each possible action of the stranger node output by the reinforcement learning module to generate the attention value corresponding to each network node in the target social network. It should be noted that the higher the attention value, the higher the priority of the stranger node to communicate with the network node.

[0065] In an embodiment of the present application, it should be understood that each action a corresponds to an interaction with one or more nodes, such as selecting a specific access point or network node within the target social network to communicate with. For a given state s, the reinforcement learning module can select the action a that maximizes Q(s, a) (i.e., the reward) * The server can normalize Q(s, a * ) to an attention value according to the following formula:

[0066]

[0067] where A is the set of all possible actions, is the optimal action involving network node i.

[0068] In step S160, the communication strategy between the stranger node and the target social network is determined according to the attention value corresponding to each network node in the target social network.

[0069] In this embodiment, the communication strategy can include a resource allocation strategy and a communication priority, the server can guide the stranger node to communicate with which network node in the target social network according to the attention value corresponding to each network node in the target social network, and how many resources the network node should allocate for corresponding processing.

[0070] Thus, in Figure 1 In the embodiment shown, by generating the first node feature corresponding to the stranger node at the current moment according to the task demand of the stranger node and the interaction information between the stranger node and each network node in the target social network before the current moment, then generating the initial environment information according to the association relationship between the network nodes in the target social network and the second node features corresponding to each network node, using the pre-constructed graph attention network to update the second node features corresponding to each network node based on the initial environment information to obtain the target environment information, and inputting the target environment information and the first feature vector into the pre-constructed reinforcement learning module to make the reinforcement learning module determine the reward corresponding to each possible action of the stranger node according to the first node feature and the target environment information, processing the reward corresponding to each possible action of the stranger node to obtain the attention value corresponding to each network node in the target social network, and determining the communication strategy between the stranger node and the target social network according to the attention value. Thus, the rationality of the communication resource allocation can be improved, and the efficiency and accuracy of the information exchange between nodes can be ensured.

[0071] In an embodiment of the present application, before generating the first node feature corresponding to the stranger node at the current moment according to the task demand of the stranger node and the interaction information between the stranger node and each network node in the target social network before the current moment, the method further comprises:

[0072] According to the received first node information of other nodes except the network nodes included in the target social network, determining the node intimacy between the other nodes and the target social network;

[0073] When the node intimacy is less than a preset threshold, the other nodes are determined as stranger nodes.

[0074] In this embodiment, when the center node in the target social network receives the first node information of other nodes except the network nodes (i.e. familiar nodes) included in the target social network, the target social network can determine the node intimacy between the other nodes and each network node in the target social network according to the first node information. It should be understood that the higher the node intimacy, the more the interaction times between the other nodes and each network node in the target social network.

[0075] Therefore, whether the other node is a stranger node can be determined according to the determined node intimacy, and if the node intimacy is less than a preset threshold, it indicates that the number of interactions is small, and therefore the other node can be determined as a stranger node. Thus, the stranger node can be accurately identified, and a corresponding communication strategy can be adopted.

[0076] In an example, the first node information can include the position, speed, task requirement and node intimacy of the other node. The position can be obtained by using a GPS positioning technology to obtain the real-time latitude and longitude coordinates of the node, denoted as (X0, Y0); the speed can be obtained by using a speed sensor built in the node to obtain the current speed of the node, denoted as R0; and the task requirement is the parameter attribute of the current task of the node.

[0077] In an embodiment of the present application, the node intimacy between the other node and the target social network can be determined according to the following formula:

[0078]

[0079] wherein, denotes the node intimacy between node i and access point j in the target social network, denotes the number of communications between node i and access point j in period T, denotes the total communication duration between node i and access point j in period T, is a weight coefficient determined according to the communication content and type between node i and access point j.

[0080] In an embodiment of the present application, the node communication method based on a social network further comprises:

[0081] obtaining second node information corresponding to each access point in a target area, and third node information corresponding to each mobile node;

[0082] performing community division according to the second node information corresponding to each access point and the third node information corresponding to each mobile node, to generate at least one social network, each social network taking an access point as a center node and containing at least one mobile node.

[0083] In this embodiment, the server can pre-obtain second node information of access points and third node information of mobile nodes existing in a target area, wherein the access points can be roadside devices or base stations, and the mobile nodes can be vehicles, mobile terminals, etc.

[0084] In an example, it is assumed that there are a set of access points R = {r1, r2, r3,..., r j} in a target area, and the second node information corresponding to each access point can include one or more of the following:

[0085] Location: This refers to the installation location of the access point, which can be represented using latitude and longitude coordinates, denoted as .

[0086] Coverage range: Based on the technical specifications of the access point, its effective communication range is determined and denoted as...

[0087] Capacity: Determined based on the processor and storage capacity of the access point, denoted as M.

[0088] Service type: Based on the access point's functional settings, such as traffic information updates, emergency message broadcasts, etc.

[0089] Let the set of mobile nodes existing in the target area be V = {v1, v2, v3, ..., v...} i The third-party information corresponding to each mobile node may include one or more of the following:

[0090] Location: The real-time latitude and longitude coordinates of the node are obtained using GPS technology, denoted as...

[0091] Speed: The current speed is obtained through the node's built-in speed sensor and denoted as...

[0092] Direction: The direction of movement of the node is determined by the direction sensor built into the node, denoted as .

[0093] Type: Classified according to the physical characteristics and uses of nodes, denoted as...

[0094] Node affinity: This is represented by the number of communications and the frequency of communication between the mobile node and each access point. For the specific calculation method, please refer to the previous text.

[0095] Therefore, after obtaining the second node information of the access point and the third node information of the mobile node, the server can divide the community based on the two information, thereby generating at least one social network. It should be understood that since the location of the access point is usually fixed and always active within the target area, the central node of the social network is selected only from the access points.

[0096] In one embodiment of this application, community segmentation is performed based on the second node information corresponding to each access point and the third node information corresponding to each mobile node to generate at least one social network, including:

[0097] According to the second node information corresponding to each access point and the third node information corresponding to each mobile node, a weight degree centrality corresponding to each access point is determined, the weight degree centrality being positively correlated with the node intimacy between the access point and its neighbor nodes;

[0098] According to the weight degree centrality in descending order, one access point is selected as a center node of a social network, and a target mobile node belonging to the social network is determined according to the increase or decrease of community fitness of each mobile node before and after joining the social network, the community fitness being determined based on a first total edge weight corresponding to the inside of the social network and a second total edge weight corresponding to the outside of the social network connecting the inside of the social network.

[0099] In this embodiment, the server can determine the weight degree centrality corresponding to each access point according to the acquired second node information corresponding to each access point and the third node information corresponding to each mobile node, the weight degree centrality being positively correlated with the node intimacy between the access point and its neighbor nodes. It should be understood that the neighbor nodes are the mobile nodes having interaction behaviors with the access point.

[0100] In an example, the weight degree centrality of each access point can be calculated according to the following formula:

[0101]

[0102] wherein N(r) is a neighbor node set of the access point r, and w(r, v) is an edge weight of the access point r and the neighbor node v, the value being equal to the node intimacy.

[0103] Therefore, after determining the weight degree centrality corresponding to each access point, the server can sort the access points according to the weight degree centrality in descending order, and preferentially select the access point with the highest weight degree centrality as the center node of the social network to be divided.

[0104] Then, the server can expand the community from the neighbor nodes of the selected center node, evaluate the mobile nodes by using the community fitness index, and gradually add the mobile nodes into the social network until no more mobile nodes can be added.

[0105] In an example, the formula for calculating the community fitness is as follows:

[0106]

[0107] wherein is the first total edge weight in the inside of the social network, is the second total edge weight in the outside of the social network connecting the inside of the social network, and α is an adjustment factor.

[0108] According to the determined community fitness, if a new mobile node is to be added to the social network, f G is increased, the mobile node can be included in the social network; if f G is decreased, the mobile node is not included in the social network. After all the scalable mobile nodes are included in the social network, a new central node can be selected from the remaining access points to establish a new social network until all the nodes are assigned.

[0109] In an embodiment of the present application, if there is a mobile node belonging to multiple social networks simultaneously according to the above division, the weight centrality of the mobile node in different social networks can be determined, which can refer to the aforementioned calculation method. The mobile node is planned into the social network with the highest weight centrality, so as to ensure that each mobile node belongs to only one social network.

[0110] In an example, after the division of the social networks is completed, the network nodes in each social network run independently, and the server can periodically or according to the changes in the node intimacy to re-plan and adjust the social networks to respond to the changes in the internal and external environment of the intelligent transportation system.

[0111] In an embodiment of the present application, after the division of the mobile nodes included in each social network is completed, the corresponding edges between the network nodes and the corresponding edge weights can be determined. In the present application, the types of edges can be divided into two categories, one of which is the edge between the network nodes in the social network, and the other of which is the edge of the social network merging the stranger nodes.

[0112] For the edge between the network nodes in the social network, the corresponding edge weight value can be the node intimacy of one of the two network nodes connected by the edge or the average of the node intimacies of the two nodes.

[0113] For the edge of the social network merging the stranger nodes, the corresponding edge weight value needs to consider the three factors of communication, perception and calculation. Specifically, for the communication quality, the free space propagation model or the parameters such as the actually measured signal strength, communication delay and packet loss rate can be used for calculation, for example, the signal strength can be estimated according to the distance between the two nodes and the specifications of the sending / receiving devices. For the perception coincidence rate, it can be determined based on the historical interaction frequency of the nodes, the interaction history of the mobile path with other nodes, and the common mobile mode. For the calculation demand, it can be evaluated based on the size of the current calculation task of the stranger node, the calculation complexity, the calculation delay tolerance, etc.

[0114] When the values corresponding to the above three types of factors are determined, the server can integrate the values of the three types of factors, for example, weighted sum, etc., to obtain the edge weight value corresponding to the edge.

[0115] The device embodiments of the present application are introduced below, which can be used to execute the social network-based node communication method in the above-mentioned embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the above-mentioned embodiments of the social network-based node communication method.

[0116] Figure 3 A block diagram of a social network-based node communication device according to an embodiment of the present application is shown.

[0117] Referring to Figure 3 According to an embodiment of the present application, the social network-based node communication device comprises:

[0118] The first generation module is configured to generate a first node feature corresponding to the stranger node at the current moment according to the task demand of the stranger node and the interaction information between the stranger node and each network node in the target social network before the current moment.

[0119] The second generation module is configured to generate initial environment information according to the association information between the network nodes in the target social network and the second node features corresponding to each network node.

[0120] The environment updating module is configured to update the second node features corresponding to each network node based on the initial environment information by using a pre-constructed graph attention network to obtain target environment information.

[0121] The input module is configured to input the target environment information and the first node feature into a pre-constructed reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the stranger node according to the first node feature and the target environment information.

[0122] The conversion module is configured to process the reward corresponding to each possible action of the stranger node to obtain an attention value corresponding to each network node in the target social network.

[0123] The processing module is configured to determine a communication strategy between the stranger node and the target social network according to the attention value corresponding to each network node in the target social network.

[0124] In an embodiment of the present application, before generating a first node feature corresponding to the stranger node at the current moment according to the task demand of the stranger node and the interaction information between the stranger node and each network node in the target social network before the current moment, the method further comprises:

[0125] determine a node intimacy between the other node and the target social network according to the received first node information of the other node except the network nodes contained in the target social network;

[0126] determine the other node as a stranger node when the node intimacy is less than a preset threshold.

[0127] In an embodiment of the present application, the node intimacy between the other node and the target social network is determined according to the following formula:

[0128]

[0129] wherein, denotes the node intimacy between node i and access point j in the target social network, denotes the communication frequency between node i and access point j in period T, denotes the total communication duration between node i and access point j in period T, is a weight coefficient determined according to the communication content and type between node i and access point j.

[0130] In an embodiment of the present application, after inputting the target environment information and the first feature vector into the pre-constructed reinforcement learning module to make the reinforcement learning module determine the reward corresponding to each possible action of the stranger node according to the first node feature and the target environment information, the input module is further configured to optimize the graph attention network according to the reward and state generated by the reinforcement learning module.

[0131] In an embodiment of the present application, the processing module is further configured to: obtain second node information corresponding to each access point in the target area and third node information corresponding to each mobile node; perform community division according to the second node information corresponding to each access point and the third node information corresponding to each mobile node, and generate at least one social network, each of the social networks taking one of the access points as a center node and containing at least one of the mobile nodes.

[0132] In an embodiment of the present application, the community division is performed according to the second node information corresponding to each access point and the third node information corresponding to each mobile node, and at least one social network is generated, including: determining a weight centrality corresponding to each access point according to the second node information corresponding to each access point and the third node information corresponding to each mobile node, the weight centrality being positively correlated with the node intimacy between the access point and its neighbor nodes; selecting a said access point as a center node of a social network in order according to the weight centrality from large to small, and determining a target mobile node belonging to the social network according to the rise and fall of the community fitness before and after each mobile node joins the social network, the community fitness being determined based on the first total edge weight corresponding to the internal of the social network and the second total edge weight corresponding to the external connection of the internal of the social network.

[0133] In an embodiment of the present application, if there is a mobile node belonging to multiple social networks, the weight centrality of the mobile node in different social networks is determined, and the mobile node is divided into the social network with the largest weight centrality.

[0134] Figure 4 A structural schematic diagram of a computer system of an electronic device suitable for implementing the embodiments of the present application is shown.

[0135] It should be noted that, Figure 4 The computer system of the electronic device shown is only an example, and should not bring any limitation to the functions and use range of the embodiments of the present application.

[0136] As Figure 4 shown, the computer system includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 402 or programs loaded from a storage portion 408 to a random access memory (RAM) 403, such as performing the methods described in the above embodiments. In the RAM 403, various programs and data required for system operation are also stored. The CPU 401, the ROM 402, and the RAM 403 are connected to each other through a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0137] The following components are connected to the I / O interface 405: an input part 406 including a keyboard, a mouse, etc.; an output part 407 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc., and a speaker, etc.; a storage part 408 including a hard disk, etc.; and a communication part 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication part 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as necessary. A removable medium 411 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive 410 as necessary, so that a computer program read out therefrom is installed in the storage part 408 as necessary.

[0138] In particular, according to embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing a computer program for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication part 409, and / or installed from the removable medium 411. When the computer program is executed by the central processing unit (CPU) 401, various functions defined in the system of the present application are executed.

[0139] It should be noted that the computer-readable medium in the embodiments of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination thereof. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination thereof. More specific examples of the computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (Compact Disc Read-Only Memory, CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus or device. In this application, the computer-readable signal medium can include a data signal carrying computer-readable computer programs in a baseband or as a part of a carrier wave. Such a propagated data signal can take on various forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium that can transmit, propagate or transport programs for use by or in connection with an instruction execution system, apparatus or device. The computer programs contained in the computer-readable medium can be transmitted by any suitable medium, including but not limited to wireless, wired, or the like, or any suitable combination thereof.

[0140] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In the flowcharts or block diagrams, each block can represent a module, a program segment or a part of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different orders than that shown in the drawings. For example, two blocks that are shown in succession can actually be executed substantially in parallel, and sometimes in reverse order, depending on the involved functions. It should also be noted that each block in the block diagrams or flowcharts, and the combination of blocks in the block diagrams or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0141] The units described in the embodiments of the present application can be implemented by software, or by hardware, or by a combination of software and hardware. The units described can also be located in a single processor. In some cases, the names of the units do not limit the units themselves.

[0142] As another aspect, the present application provides a computer readable medium, which can be included in the electronic device described in the above embodiments, or can exist separately without being assembled into the electronic device. The computer readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to implement the method described in the above embodiments.

[0143] It should be noted that although several modules or units for performing actions are mentioned in the above detailed description, the division into the modules or units is not mandatory. In fact, according to the embodiments of the present application, features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, features and functions of one module or unit described above can be further divided into a plurality of modules or units.

[0144] From the above description of the embodiments, those skilled in the art will readily appreciate that the example embodiments described herein can be implemented by software and / or by hardware coupled with software. Accordingly, the technical solutions of the embodiments of the present application can be embodied in the form of a software product. The software product can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, or the like) or on a network, and includes a number of instructions for causing a computing device (which can be a personal computer, a server, a terminal, or a network device, etc.) to perform the methods according to the embodiments of the present application.

[0145] Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the embodiments disclosed herein. The present application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such

[0146] It should be understood that the present application is not limited to the precise construction that has been described above and illustrated in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the present application. The scope of the present application is limited only by the appended claims.

Claims

1. A node communication method based on a social network, characterized in that, include: Based on the task requirements of the unfamiliar node and the interaction information between the unfamiliar node and each network node in the target social network before the current moment, the first node feature corresponding to the unfamiliar node at the current moment is generated. Initial environment information is generated based on the association information between network nodes in the target social network and the second node characteristics corresponding to each network node. Based on the initial environment information, a pre-constructed graph attention network is used to update the second node features corresponding to each network node to obtain the target environment information; The target environment information and the first node features are input into a pre-built reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the unfamiliar node based on the first node features and the target environment information. The reward corresponding to each possible action of the unfamiliar node is processed to obtain the attention value corresponding to each network node in the target social network. Based on the attention value corresponding to each network node in the target social network, a communication strategy between the unfamiliar node and the target social network is determined.

2. The method according to claim 1, characterized in that, Before generating the first node feature corresponding to the unfamiliar node at the current moment based on the task requirements of the unfamiliar node and the interaction information between the unfamiliar node and various network nodes in the target social network before the current moment, the method further includes: Based on the first node information received from other nodes besides those included in the target social network, the node affinity between the other nodes and the target social network is determined. When the node affinity is less than a preset threshold, the other nodes are identified as unfamiliar nodes.

3. The method according to claim 2, characterized in that, The node affinity between the other nodes and the target social network is determined using the following formula: in, This represents the node affinity between node i and access point j in the target social network. This represents the number of communications between node i and access point j within period T. This represents the total communication duration between node i and access point j within period T. The weighting coefficient is determined based on the communication content and type between node i and access point j.

4. The method according to claim 1, characterized in that, After inputting the target environment information and the first node features into a pre-built reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the unfamiliar node based on the first node features and the target environment information, the method further includes: The graph attention network is optimized based on the rewards and states generated by the reinforcement learning module.

5. The method according to any one of claims 1-4, characterized in that, The method further includes: Obtain the second node information corresponding to each access point within the target area, and the third node information corresponding to each mobile node; The community is divided according to the second node information corresponding to each access point and the third node information corresponding to each mobile node, and at least one social network is generated. Each social network takes one of the access points as the central node and includes at least one of the mobile nodes.

6. The method according to claim 5, characterized in that, Based on the second node information corresponding to each access point and the third node information corresponding to each mobile node, communities are divided to generate at least one social network, including: Based on the second node information corresponding to each access point and the third node information corresponding to each mobile node, the weight degree centrality of each access point is determined. The weight degree centrality is positively correlated with the node affinity between the access point and its neighboring nodes. According to the order of the centrality of the weights from largest to smallest, an access point is selected as the central node of the social network. Based on the change in community fitness of each mobile node before and after joining the social network, the target mobile node belonging to the social network is determined. The community fitness is determined based on the first total edge weight corresponding to the inside of the social network and the second total edge weight corresponding to the connection between the outside of the social network and the inside of the social network.

7. The method according to claim 6, characterized in that, If a mobile node belongs to multiple social networks simultaneously, determine the weight centrality of the mobile node in different social networks, and assign the mobile node to the social network with the largest corresponding weight centrality.

8. A node communication device based on a social network, characterized in that, include: The first generation module is used to generate the first node feature corresponding to the unfamiliar node at the current moment based on the task requirements of the unfamiliar node and the interaction information between the unfamiliar node and each network node in the target social network before the current moment. The second generation module is used to generate initial environment information based on the association information between the network nodes in the target social network and the second node features corresponding to each network node. The environment update module is used to update the second node features corresponding to each network node based on the initial environment information using a pre-constructed graph attention network, so as to obtain the target environment information. An input module is used to input the target environment information and the first node features into a pre-built reinforcement learning module, so that the reinforcement learning module determines the reward corresponding to each possible action of the unfamiliar node based on the first node features and the target environment information. The conversion module is used to process the reward corresponding to each possible action of the unfamiliar node to obtain the attention value corresponding to each network node in the target social network. The processing module is used to determine the communication strategy between the unfamiliar node and the target social network based on the attention value corresponding to each network node in the target social network.

9. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the node communication method based on a social network as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the node communication method based on a social network as described in any one of claims 1 to 7.

Images