Channel resource allocation method, apparatus, device, and medium

Abstract

Embodiments of the present disclosure disclose a channel resource allocation method, device, equipment and medium. The method comprises: determining a target service type corresponding to a session data packet of a terminal device; determining a channel bandwidth and a channel occupation duration corresponding to the terminal device according to the target service type; obtaining target channel resource indication information corresponding to the target service type based on a first channel resource allocation model according to channel resources corresponding to an edge server, the channel bandwidth and the channel occupation duration; obtaining a channel utilization rate according to an occupation rate of the edge server by at least one terminal device in a first time slot; and updating the first channel resource allocation model according to the channel utilization rate. The channel bandwidth and the channel occupation duration required by the target service type are used to allocate channel resources to the device, and the first channel resource allocation model is updated based on the channel utilization rate, thereby improving the channel utilization rate under the condition of meeting the device requirements.

Description

Technical Field

[0001] This disclosure relates to the field of communications, specifically to methods, apparatus, equipment, and media for channel resource allocation. Background Technology

[0002] With the increasing prevalence of 5G technology, the types of communication services are becoming more and more diverse. Therefore, the allocation of channel resources is particularly important.

[0003] Currently, when allocating channel resources to edge servers or terminal devices, the industry typically uses pre-defined channel resource allocation strategies. However, these strategies are usually based on the experience of technical personnel and may not meet the actual needs of the edge servers or terminal devices. Allocating channel resources without meeting the actual needs of the devices can easily lead to low channel utilization.

[0004] Therefore, how to improve channel utilization while meeting equipment requirements has become an urgent technical problem to be solved. Summary of the Invention

[0005] To address the problems in related technologies, embodiments of this disclosure provide a channel resource allocation method, apparatus, device, and medium.

[0006] In a first aspect, this disclosure provides a channel resource allocation method.

[0007] Specifically, the method is applied to an edge server, and the method includes:

[0008] Determine the target service type corresponding to the session data packets of the terminal device;

[0009] Based on the target service type, determine the channel bandwidth and channel occupancy duration corresponding to the terminal device;

[0010] Based on the channel resources corresponding to the edge server, the channel bandwidth, and the channel occupancy duration, and based on a pre-acquired first channel resource allocation model, target channel resource indication information corresponding to the target service type is obtained. The target channel resource indication information is used to indicate at least a portion of the channel resources corresponding to the edge server.

[0011] The channel utilization rate of the edge server is obtained based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, wherein the first time slot is at least a portion of the time domain resources in the channel resources corresponding to the edge server.

[0012] The first channel resource allocation model is updated based on the channel utilization rate.

[0013] Secondly, this disclosure provides a channel resource allocation method.

[0014] Specifically, the method is applied to a cloud controller, and the method includes:

[0015] Obtain the communication traffic of the edge server in the first preset time period;

[0016] Based on the communication traffic, and using a pre-acquired second channel resource allocation model, first indication information is obtained, which is used to indicate the channel resources corresponding to the edge server.

[0017] The reward for the at least one edge server is obtained based on the channel utilization rate of the at least one edge server within a first preset time period;

[0018] The second channel resource allocation model is updated based on the reported returns.

[0019] Thirdly, this disclosure provides an information resource allocation device.

[0020] Specifically, the device is applied to an edge server and includes:

[0021] The first determining module is configured to determine the target service type corresponding to the session data packets of the terminal device.

[0022] The second determining module is configured to determine the channel bandwidth and channel occupancy duration corresponding to the terminal device based on the target service type.

[0023] The first acquisition module is configured to acquire target channel resource indication information corresponding to the target service type based on the channel resources corresponding to the edge server, the channel bandwidth, and the channel occupancy duration, according to a pre-acquired first channel resource allocation model. The target channel resource indication information is used to indicate at least a portion of the channel resources corresponding to the edge server.

[0024] The second acquisition module is configured to acquire the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in a first time slot, wherein the first time slot is at least a portion of the time domain resources in the channel resources corresponding to the edge server.

[0025] The first update module is configured to update the first channel resource allocation model based on the channel utilization.

[0026] Fourthly, embodiments of this disclosure provide a channel resource allocation system, including: at least one edge server and a cloud controller, wherein the edge server is used to execute the method in the first aspect and any possible implementation of the first aspect; and the cloud controller is used to execute the method in the second aspect and any possible implementation of the second aspect.

[0027] Fifthly, embodiments of this disclosure provide an electronic device, including a memory and at least one processor, wherein the memory is used to store one or more computer instructions, which are executed by the processor to implement the methods in the first to second aspects and any possible implementations of the first to second aspects.

[0028] In a sixth aspect, embodiments of this disclosure provide a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the methods of the first aspect to the second aspect and any possible implementation of the first aspect to the second aspect.

[0029] In a seventh aspect, embodiments of this disclosure provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the methods of the first aspect to the second aspect and any possible implementation of the first aspect to the second aspect.

[0030] The technical effects provided by the embodiments of this disclosure may include the following beneficial effects:

[0031] The above technical solution uses an edge server to determine the channel bandwidth and channel occupancy duration required for the target service type corresponding to the session data packets of a terminal device. Based on the requirements of the target service type and the channel resources available to the edge server, it allocates corresponding channel resources to the target service type of the terminal device using a first channel resource allocation model. After allocating corresponding channel resources for the target service type of at least one terminal device, the edge server can obtain the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server in the first time slot of at least one terminal device, and update the first channel resource allocation model based on this channel utilization rate. Since the edge server allocates channel resources based on the actual needs of the target service type of the terminal device, the probability of the terminal device using the allocated channel resources will increase if the allocated channel resources are more in line with the actual needs, thus ensuring channel utilization. Furthermore, the edge server updates the first channel resource allocation model in a timely manner based on the channel utilization rate, so that the updated first resource allocation model can allocate more suitable channel resources to the terminal device, further improving channel utilization.

[0032] The cloud controller acquires the communication traffic of edge servers within a first preset time period and allocates corresponding channel resources to the edge servers based on this traffic and a second channel resource allocation model. After allocating channel resources to at least one edge server, the cloud controller obtains a response from at least one edge server based on its channel utilization rate within the first preset time period and updates the second channel resource allocation model accordingly. Since the cloud controller allocates channel resources based on the edge server's communication traffic within the first preset time period, the probability of the edge server using the allocated channel resources increases if they better match actual needs, thus ensuring channel utilization. Furthermore, the cloud controller updates the second channel resource allocation model promptly based on the responses, ensuring that the updated model allocates more suitable channel resources to devices, further improving channel utilization.

[0033] Therefore, when edge servers and cloud controllers jointly allocate channel resources, channel utilization is further improved.

[0034] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0035] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments, taken in conjunction with the accompanying drawings. In the drawings:

[0036] Figure 1 A system architecture diagram of a channel resource allocation system according to an embodiment of the present disclosure is shown;

[0037] Figure 2 A flowchart illustrating a channel resource allocation method according to an embodiment of the present disclosure is shown;

[0038] Figure 3 Another flowchart of a channel resource allocation method according to an embodiment of the present disclosure is shown;

[0039] Figure 4 A structural block diagram of a channel resource allocation apparatus according to an embodiment of the present disclosure is shown;

[0040] Figure 5 Another structural block diagram of a channel resource allocation apparatus according to an embodiment of the present disclosure is shown;

[0041] Figure 6 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown;

[0042] Figure 7This is a schematic diagram of the structure of a computer system suitable for implementing a channel resource allocation method according to an embodiment of the present disclosure. Detailed Implementation

[0043] In the following, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings to enable those skilled in the art to readily implement them. Furthermore, for clarity, portions unrelated to the description of exemplary embodiments have been omitted from the drawings.

[0044] In this disclosure, it should be understood that terms such as “comprising” or “having” are intended to indicate the presence of features, figures, steps, behaviors, components, parts or combinations thereof disclosed in this specification, and are not intended to exclude the possibility of the presence or addition of one or more other features, figures, steps, behaviors, components, parts or combinations thereof.

[0045] It should also be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0046] As mentioned above, existing technologies allocate channel resources to devices based on pre-set channel resource allocation strategies. These strategies are based on the experience of technical personnel and may not meet the actual needs of edge servers or terminal devices. If the allocated channel resources cannot meet the actual needs of the devices, the devices may not use the allocated channel resources, leading to low channel utilization.

[0047] To address the aforementioned shortcomings, this disclosure provides a channel resource allocation scheme. An edge server determines the channel bandwidth and channel occupancy duration required for the target service type corresponding to the session data packets of a terminal device. Based on the requirements of the target service type and the channel resources corresponding to the edge server, a first channel resource allocation model is used to allocate corresponding channel resources to the target service type of the terminal device. Furthermore, after allocating corresponding channel resources for the target service type of at least one terminal device, the edge server can also obtain the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in a first time slot, and update the first channel resource allocation model based on this channel utilization rate. A cloud controller obtains the communication traffic of the edge server in a first preset time period, and allocates corresponding channel resources to the edge server based on this communication traffic and a second channel resource allocation model. Furthermore, after allocating corresponding channel resources to at least one edge server, the cloud controller can also obtain the feedback from at least one edge server based on the channel utilization rate of at least one edge server in the first preset time period, and update the second channel resource allocation model based on the feedback.

[0048] It's understandable that the edge server allocates channel resources to the target service type of the terminal device based on the channel bandwidth and channel occupancy duration required by the terminal device's target type, while the cloud controller allocates corresponding channel resources based on the communication traffic of the edge server within a first preset time period. Therefore, if the channel resources allocated to the device better match the device's actual needs, the probability of the device using the allocated channel resources will increase, thus ensuring channel utilization. Furthermore, the edge server updates the first channel resource allocation model in a timely manner based on channel utilization, and the cloud controller updates the second channel resource allocation model in a timely manner based on feedback, so that the updated resource allocation model can allocate more suitable channel resources to the device, further improving channel utilization. In addition, the joint allocation of channel resources by the edge server and cloud controller further improves channel utilization.

[0049] Figure 1 A system architecture diagram of a channel resource allocation system according to an embodiment of the present disclosure is shown. Figure 1 As shown, the system 100 may include a cloud controller 110, edge servers 121 to 123, and terminal devices 131 to 140. Among them, edge servers 121 to 123 may be edge base stations.

[0050] Edge servers can provide communication coverage for specific geographical areas and can communicate with terminal devices within their coverage area. Edge servers can also receive signals from terminal devices within their coverage area. For example... Figure 1 As shown, terminal devices 131 to 134 are located in area #1 served by edge server 121, terminal devices 135 and 136 are located in area #2 served by edge server 122, and terminal devices 137 to 140 are located in area #3 served by edge server 123. Edge server 121 can communicate with any one or more of terminal devices 131 to 134 in area #1, edge server 122 can communicate with terminal devices 135 and / or 136 in area #2, and edge server 123 can communicate with any one or more of terminal devices 137 to 140 in area #3.

[0051] The edge server can also determine the service type corresponding to the session data packets sent by the terminal devices, and allocate corresponding channel resources for that service type of the terminal devices based on a pre-established channel resource allocation model. For example, in this embodiment, edge server 121 can determine the service type corresponding to each session data packet based on the session data packets sent by terminal devices 131 to 134 respectively. Edge server 122 can determine the service type corresponding to each session data packet based on the session data packets sent by terminal devices 135 and 136 respectively. Edge server 123 can also determine the service type corresponding to each session data packet based on the session data packets sent by terminal devices 137 to 140 respectively.

[0052] The cloud controller 110 can obtain dataset information from edge servers and allocate corresponding channel resources to each edge server based on a pre-established channel resource allocation model. For example... Figure 1 As shown, the cloud controller 110 can communicate with edge servers 121, 122, and 123, which have communication connections. For example, in this embodiment of the disclosure, the cloud controller 110 can obtain the size and signaling information of service packets reported by terminal devices 131 to 134 from edge server 121, the size and signaling information of service packets reported by terminal devices 135 and 136 from edge server 122, and the size and signaling information of service packets reported by terminal devices 137 to 140 from edge server 123. For example, the signaling information may include the number of evolved radio access bearer (E-RAB) connections, the number of radio resource control (RRC) connections, and the number of paging responses.

[0053] It should be understood that Figure 1 The cloud controller shown is merely an example and may be other devices with the same function; this application does not limit this.

[0054] It should be understood that Figure 1 The number of edge servers and terminal devices shown is for illustrative purposes only. Figure 1 The system shown may also include other numbers of edge servers and terminal devices, and each edge server may cover one or more small areas, which is not limited in this application.

[0055] Figure 2 A flowchart illustrating a channel resource allocation method according to an embodiment of the present disclosure is shown. This method can be applied to, for example... Figure 1The edge server shown can be executed by a physical device capable of providing edge server functions, or by a component (such as a chip) configured in the physical device, or by a module capable of implementing some or all edge server functions, etc. This application does not limit this.

[0056] For ease of understanding, this disclosure uses an edge server as an example to describe the method provided herein. The steps in method 200 are described in detail below.

[0057] In step 201, the target service type corresponding to the session data packet of the terminal device is determined;

[0058] In step 202, the channel bandwidth and channel occupancy duration corresponding to the terminal device are determined according to the target service type;

[0059] In step 203, based on the channel resources corresponding to the edge server, the channel bandwidth, and the channel occupancy duration, and based on the pre-acquired first channel resource allocation model, target channel resource indication information corresponding to the target service type is obtained. The target channel resource indication information is used to indicate at least a portion of the channel resources corresponding to the edge server.

[0060] In step 204, the channel utilization rate of the edge server is obtained based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, wherein the first time slot is at least a portion of the time domain resources in the channel resources corresponding to the edge server.

[0061] In step 205, the first channel resource allocation model is updated based on the channel utilization.

[0062] In one embodiment of this disclosure, a session can be established between a terminal device and an edge server, and multiple data packets can be transmitted within the session. The group of data packets corresponding to the session is referred to as a session data packet.

[0063] In one embodiment of this disclosure, each terminal device within the signal coverage area of ​​each edge server can establish a session with the edge server. The data packets transmitted within the session correspond to a service type; in other words, each session data packet corresponds to a service type.

[0064] In one embodiment of this disclosure, the service type may include at least one of the following: web page service, video service, or online game service. The target service type can be understood as determining whether a set of data packets sent by the terminal device to the edge server within a session corresponds to a specific service type: web page service, video service, or online game service.

[0065] In one embodiment of this disclosure, channel bandwidth can be understood as the difference between the maximum and minimum frequencies in a channel, which reflects the size of the frequency range covered by the channel.

[0066] In one embodiment of this disclosure, the channel occupancy duration can be understood as the length of time the channel is occupied.

[0067] In one embodiment of this disclosure, channel resources may include resources in two dimensions: channel time-domain resources and channel frequency-domain resources. Channel frequency-domain resources can be understood as channel bandwidth resources. Specifically, channel frequency-domain resources represent the size of the frequency range, and channel time-domain resources represent the start and end time range.

[0068] In one embodiment of this disclosure, the channel resources corresponding to the edge server are allocated by the cloud controller. In other words, the cloud controller can allocate corresponding channel time-domain resources and channel frequency-domain resources to each edge server.

[0069] It should be noted that the specific implementation method for the cloud controller to allocate channel resources to the edge server can be referred to the description in subsequent method 300, and will not be introduced here.

[0070] In one embodiment of this disclosure, the target channel resource indication information includes target channel frequency domain resource indication information and target channel time domain resource indication information. The target channel frequency domain resource indication information indicates the frequency range that the terminal device can occupy when transmitting service data corresponding to the target service type. The target channel time domain resource indication information indicates the time range that the terminal device can occupy when transmitting service data corresponding to the target service type.

[0071] It can be understood that when transmitting service data corresponding to the target service type, the target channel frequency domain resources that the terminal device can occupy are at least a part of the channel frequency domain resources corresponding to the edge server, and the target channel time domain resources that the terminal device can occupy are at least a part of the channel time domain resources corresponding to the edge server.

[0072] In one embodiment of this disclosure, channel utilization can be understood as the percentage of the channel that is occupied.

[0073] In the above implementation, the edge server determines the channel bandwidth and channel occupancy duration required for the target service type corresponding to the session data packets of the terminal device. Based on the channel resources available to the edge server and the channel bandwidth and channel occupancy duration requirements of the target service type of the terminal device, and using a first channel resource allocation model, it obtains target channel resource indication information to indicate the target service type, thereby allocating the corresponding target channel resources for the target service type. After allocating the corresponding target channel resources for the target service type of at least one terminal device in this manner, the edge server can also obtain the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, and update the first channel resource allocation model based on the channel utilization rate.

[0074] It is understandable that because the edge server allocates channel resources to the target service type of the terminal device based on the channel bandwidth and channel occupancy duration required by the target type of the terminal device, the allocated channel resources are more in line with the actual needs of the terminal device, and the probability of the terminal device using the allocated channel resources will increase, thereby ensuring channel utilization. At the same time, the edge server will also update the first channel resource allocation model based on the channel utilization, so that the updated first channel resource allocation model can allocate more suitable channel resources for the target service type of the terminal device, further improving channel utilization. Therefore, channel resource utilization can be improved while meeting the needs of the terminal device.

[0075] In one embodiment of this disclosure, step 201, namely determining the target service type corresponding to the session data packet of the terminal device, can be implemented through the following execution processes (i) to (iiiii).

[0076] (i) Receive a session data packet sent by the terminal device. The session data packet includes R sub-data packets, each of which includes Z bytes, where R > 1 and Z > 1.

[0077] (ii) Based on the preset number of sub-data packets Y, determine Y target data packets from R sub-data packets, where Y < R;

[0078] (iii) Based on the preset number of bytes X, determine X target bytes in each of the Y target data packets, where X < Z;

[0079] (iiii) Determine the number of rows of the service matrix as X based on X target bytes, and determine the number of columns of the service matrix as Y based on Y target data packets, and generate a service matrix of X×Y dimensions. Each element in the service matrix corresponds to the byte data under different target data packets and different combinations of target bytes.

[0080] (iiiii) Input the business matrix into the preset business identification model to obtain the target business type corresponding to the session data packet.

[0081] In one embodiment of this disclosure, after the terminal device establishes a session with the edge server, the set of data packets sent within the session may specifically include R sub-data packets.

[0082] In one embodiment of this disclosure, each sub-data packet includes Z bytes, and each byte can be represented by two hexadecimal symbols. The data of each byte can be converted from hexadecimal to decimal, and the value range can be [0, 255].

[0083] Those skilled in the art can set the number of target data packets and target bytes according to actual needs, as long as the number of target data packets Y ≤ the number of sub-data packets R and the number of target bytes Z are not limited in this disclosure and are all within the scope of protection.

[0084] In one embodiment of this disclosure, the Y target data packets are the Y consecutive sub-data packets arranged at the top of the R sub-data packets.

[0085] In one embodiment of this disclosure, the X target bytes are the Z consecutive bytes arranged in the first column of the Z bytes.

[0086] In one embodiment of this disclosure, the service identification model can be deployed on an edge server, and each edge server can deploy the service identification model.

[0087] In one embodiment of this disclosure, the business identification model can be a two-dimensional convolutional neural network, which may include a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a flattening layer, a first fully connected layer, a dropout layer, and a second fully connected layer. The first and second convolutional layers have 64 convolutional kernels, with kernel sizes of 5×5 and 3×3, respectively. The first and second pooling layers can be the same pooling layer with a size of 3×3. The flattening layer can be used to convert multidimensional data into one-dimensional data. The first fully connected layer can use 128 neurons. The number of neurons in the second fully connected layer is the same as the type of business. For example, if existing business types include web page business, video business, and online game business, then the second fully connected layer would have 3 neurons. In one implementation, Softmax is used as the activation function for the second fully connected layer.

[0088] Furthermore, the determined service matrix is ​​used as a node in the input layer of the service identification model, with each byte of data in the service matrix corresponding to one node. The target service type is used as a node in the output layer of the service identification model, meaning the output layer outputs the target service type corresponding to the determined session data packet.

[0089] In the above implementation, when the terminal device establishes a session with the edge server and sends session data packets within that session, the edge server can continuously select the first Y target data packets from R sub-data packets, and then select the first X target bytes from each of the Y target data packets. The hexadecimal data corresponding to each target byte is converted to decimal data, and a service matrix is ​​generated. The rows of this service matrix physically represent target bytes, with the number of rows equal to the number of target bytes. The columns physically represent target data packets, with the number of columns equal to the number of target data packets. Each element in the service matrix represents decimal byte data under different combinations of target data packets and target bytes. Once the specific service matrix corresponding to the session data packet is determined, it can be input into the input layer of the service identification model. The service identification model, through algorithmic calculation, can output the target service type corresponding to the service matrix through the output layer. It can be understood that in a scenario where the terminal device and the edge server have established a session, the edge server can determine the target service type corresponding to each session data packet of each terminal within its communication coverage area using this method.

[0090] In one embodiment of this disclosure, the business identification model in step (iiiii) above can be trained through the following execution process (1) to (3).

[0091] (1) Obtain the sample dataset.

[0092] Specifically, the edge server can determine the X×Y business matrix corresponding to a single terminal device within its communication coverage area, using millisecond-level time granularity and a single session as the unit. This determination is performed locally in the same manner as steps (i) to (iiii) above. This business matrix is ​​then used as a sample business matrix, and a business type label is added to it. By repeating this process, multiple sample business matrices corresponding to each business type can be obtained. These multiple business matrices are used as a sample dataset, with each matrix labeled with its corresponding business type. For example, 500 sample business matrices can be obtained for web page services, 500 for video services, and 500 for online game services. These 1500 sample business matrices can then be used as a sample dataset.

[0093] It should be understood that multiple sample service matrices in the sample dataset can be service matrices corresponding to session data packets of different service types from different terminal devices.

[0094] (2) Determine the training dataset and validation dataset based on the preset sample training ratio and the sample dataset.

[0095] Specifically, the sample business matrices corresponding to each business type in the sample dataset are divided in an 8:2 ratio to obtain the training dataset and the validation dataset. For example, the 500 sample business matrices corresponding to the web page business are divided in an 8:2 ratio, with 400 of them used as the training business matrices and the remaining 100 used as the validation business matrices. Similarly, 400 of the 500 sample business matrices corresponding to the video business and the online game business are used as the training business matrices, and the remaining 100 used as the validation business matrices. Therefore, the 400 training business matrices corresponding to each of the web page business, video business, and online game business can be used together to form the training dataset, and the 100 validation business matrices corresponding to each of the web page business, video business, and online game business can be used together to form the validation dataset.

[0096] (3) Train the business identification model using the training dataset, and verify the accuracy of the business identification model using the verification dataset.

[0097] Specifically, multiple training business matrices from the training dataset are sequentially input into a pre-defined business recognition model. Supervised learning is used to enable the model to learn the correspondence between the business matrices and business types, thus training the model. After training, multiple verification business matrices from the verification dataset are sequentially input into the trained model. The model outputs the business type corresponding to each verification matrix. By comparing the output business type with the pre-added business type labels to the matrix, the model's error is determined. If the error is too large, the Adam algorithm (Adaptive Momentum) can be used to optimize the model and reduce its error.

[0098] In one embodiment of this disclosure, step 202 involves determining the channel bandwidth and channel occupancy duration corresponding to the terminal device based on the target service type.

[0099] Specifically, based on the service type corresponding to the session data packets determined in step 201, the edge server can further determine the frequency range of channel coverage and the duration of channel occupation required by the terminal device under that service type. It is known that different service types require different channel bandwidths and channel occupation durations. For example, for video services, the terminal device requires a larger channel bandwidth and a longer channel occupation duration. Clearly, for different target service types of different terminal devices, the edge server can determine the required channel bandwidth and channel occupation duration for each target service type.

[0100] In one embodiment of this disclosure, step 203, namely, obtaining target channel resource indication information corresponding to the target service type based on the channel resources corresponding to the edge server, the channel bandwidth and the channel occupancy duration, and the pre-acquired first channel resource allocation model, can be achieved by obtaining the target channel resource indication information in the following manner.

[0101] The channel resources allocated by the cloud controller to the edge server include channel time-domain resources and channel frequency-domain resources. The channel time-domain resources can be divided into M time slots, each time slot corresponding to a time period. Furthermore, the channel frequency-domain resources can be divided into N unit bandwidths, each unit bandwidth corresponding to a frequency range, where M > 0 and N > 0. In a preferred implementation, each unit bandwidth is one byte of bandwidth.

[0102] It should be understood that in this disclosure, the channel resources allocated by the cloud controller to the edge server are channel frequency domain resources within a first preset time period. The first preset time period is the time period after the current time node, that is, the M time slots are the time slots after the current time node.

[0103] The first channel resource allocation model can be deployed on edge servers.

[0104] The first channel resource allocation model adopts the Dueling Deep Q Network (DQN) algorithm from deep reinforcement learning algorithms. This algorithm contains two basic elements: state and action.

[0105] The state is defined as whether each unit of bandwidth corresponding to the edge server in the i-th time slot has been pre-booked, and can be represented as si = {s1, s2, ..., sN}. Here, si represents the set of cases where each unit of bandwidth is occupied in the i-th time slot. sN represents the case where the N-th unit of bandwidth is occupied in the i-th time slot. When sN = 0, it means the N-th unit of bandwidth is not occupied; when sN = 1, it means the N-th unit of bandwidth is occupied. It should be understood that 1 ≤ i ≤ M.

[0106] For example, suppose the channel frequency domain resources corresponding to the edge server are divided into 5 units of bandwidth, namely the first unit bandwidth to the fifth unit bandwidth, and suppose the channel time domain resources corresponding to the edge server are divided into 6 time slots, each time slot having a duration of 1 second. Specifically, the time slot 1 corresponds to 00:00–00:01, the second time slot to 00:01–00:02, the third time slot to 00:02–00:03, the fourth time slot to 00:03–00:04, the fifth time slot to 00:04–00:05, and the sixth time slot to 00:05–00:06.

[0107] Assuming s1 = {1, 1, 0, 1, 1}, then in the first time slot, the first, second, fourth, and fifth units of bandwidth are already occupied. Assuming s2 = {1, 1, 1, 1, 1}, then in the second time slot, all five units of bandwidth are occupied. Assuming s3 = {0, 1, 0, 1, 1}, then in the third time slot, the second, fourth, and fifth units of bandwidth are already occupied. Assuming s4 = {0, 0, 0, 0, 1}, then in the fourth time slot, the fifth unit of bandwidth is already occupied. Similarly, assuming s5 = {0, 0, 0, 0, 0} and s6 = {0, 0, 0, 0, 0}, then in the fifth and sixth time slots, none of the five units of bandwidth are occupied.

[0108] The action is defined as the target unit bandwidth allocated to the target service type of the terminal device in the i-th time slot, which can be represented as ai = {a1, a2, ..., aN}. Here, ai represents the set of cases where each unit bandwidth is allocated in the i-th time slot. aN represents whether the N-th unit bandwidth is allocated in the i-th time slot. When aN = 0, it means the N-th unit bandwidth is not allocated; when aN = 1, it means the N-th unit bandwidth is definitely allocated.

[0109] For example, assuming a4, a5, a6 = {0, 1, 1, 1, 0}, it can be seen that in time slots 4 to 6, the second unit bandwidth, third unit bandwidth, and fourth unit bandwidth are allocated to the target service type of the terminal device. That is, the second unit bandwidth to the fourth unit bandwidth in time slots 4 to 6 is the target unit bandwidth. In other words, the target channel frequency domain resources corresponding to the target service type of the terminal device are the second unit bandwidth to the fourth unit bandwidth, and the corresponding target channel time domain resources are time slots 4 to 6. Therefore, the terminal device can transmit service data related to the target service type in time slots 4 to 6 and on the second unit bandwidth to the fourth unit bandwidth.

[0110] It should be noted that, in one embodiment of this disclosure, the actions involved in the above algorithm are determined comprehensively based on whether each unit bandwidth in each time slot is pre-occupied, and the channel bandwidth and channel occupancy duration required for the target service type of the terminal device determined by step 202.

[0111] Taking the example above, by obtaining the status of whether each unit bandwidth in each time slot is pre-occupied, it is determined that the first unit bandwidth is not pre-occupied in time slots 3 to 6, the second unit bandwidth is not pre-occupied in time slots 4 to 6, the third unit bandwidth is not pre-occupied in time slots 1 and 3 to 6, the fourth unit bandwidth is not pre-occupied in time slots 3 to 6, and the fifth unit bandwidth is not pre-occupied in time slots 5 and 6. Assume that step 202 determines that the required channel bandwidth for the target service type of the terminal device is 3, and the channel occupancy time is 3 seconds. Therefore, considering both aspects, the above actions can be allocated, that is, the target channel frequency domain resources for the target service type of the terminal device are allocated as the second to fourth unit bandwidths, and the corresponding target channel time domain resources are allocated as time slots 4 to 6.

[0112] In the above implementation, the edge server obtains the channel resources allocated to it by the cloud controller. After determining the channel bandwidth and channel occupancy duration required for the target service type, the edge server can use the Dueling DQN algorithm in the first channel resource allocation model to determine the pre-occupancy of each unit bandwidth in each time slot of the channel resources corresponding to the edge server. Based on the channel bandwidth and channel occupancy duration required for the target service type, the edge server determines the action corresponding to the target service type, that is, the target channel resources allocated to the terminal device for that service type, thereby generating target channel resource indication information. This allows the terminal device to determine the target channel resources corresponding to the target service type based on the target channel resource indication information. It can be understood that, through this method, each edge server can allocate corresponding channel resources to each target service type of each terminal device within the signal coverage area. Since the channel resources allocated to each target service type of each terminal device are based on the channel bandwidth and channel occupancy duration required by each terminal for each target service type, the allocation of channel resources can meet the needs of the terminal devices, realizing resource allocation for differentiated user groups.

[0113] In one embodiment of this disclosure, step 204, namely obtaining the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, can be implemented through the following execution process (j) to (jj).

[0114] (j) Obtain the number of units of bandwidth occupied by at least one terminal device in the first time slot;

[0115] (jj) Obtain the channel utilization rate based on the ratio of the number of units of bandwidth to the number of units of bandwidth N corresponding to the edge server.

[0116] In one embodiment of this disclosure, the first time slot can be any one of the M time slots.

[0117] In this embodiment, after the edge server allocates corresponding channel resources for each target service type of multiple terminal devices within the signal coverage area in step 203, each terminal device may or may not use the allocated channel resources. To monitor the utilization of the allocated channels, the edge server can monitor the channel bandwidth usage of a specific time slot out of the M time slots. When monitoring the channel bandwidth usage of the first time slot, the actual number of units of bandwidth occupied within the first time slot can be obtained. Then, the ratio of the actual number of units of bandwidth occupied to the total number of units of bandwidth N corresponding to the edge server can be calculated, thus determining the channel utilization rate for the first time slot. For example, if the first to third units of bandwidth are actually occupied in the first time slot, then the number of sub-channels occupied in the first time slot (3) divided by the number of units of bandwidth N corresponding to the edge server (5) determines the channel utilization rate to be 60%. It can be understood that, through this method, each edge server can determine the channel utilization rate for each of its corresponding M time slots.

[0118] In one embodiment of this disclosure, step 205, namely updating the first channel resource allocation model according to the channel utilization rate, can be achieved by updating the first channel resource allocation model in the following manner.

[0119] As mentioned earlier, the Dueling DQN algorithm used in the first channel resource allocation model includes two basic elements: state and action. To update this algorithm, the Dueling DQN algorithm can also include a reward. The reward is defined as the channel utilization rate pi in the i-th time slot, representing the ratio of the number of units of bandwidth occupied in the i-th time slot to the total number of units of bandwidth corresponding to the edge servers.

[0120] In this implementation, after the edge server obtains the channel utilization rate in the first time slot through step 204, it also learns the feedback from the Dueling DQN algorithm. This feedback can then be used as a parameter to solve the Bellman equation, thereby updating the first channel resource model. In other words, the channel utilization rate obtained in step 204 indicates whether the channel resources allocated by the Dueling DQN algorithm for the target service type of the terminal device can achieve a high channel occupancy rate within the same time slot. If the channel occupancy rate is insufficient, the logical algorithm for allocating channel resources for the target service type of the terminal device can be adjusted based on the feedback feedback, i.e., the channel utilization rate. This updates the Dueling DQN algorithm in the first channel resource allocation model, ensuring that the allocated channel resources are utilized more effectively in the next allocation of channel resources for the target service type of the terminal device, thus improving channel utilization.

[0121] In one embodiment of this disclosure, step 205, that is, after updating the first channel resource allocation model according to the channel utilization rate, may further include: determining whether the Dueling DQN algorithm of the first channel resource allocation model has converged; if it has not converged, then return to step 201; if it has converged, then end.

[0122] In this implementation, by determining whether the Dueling DQN algorithm has converged, the Dueling DQN algorithm is iteratively updated until it converges. This indicates that the Dueling DQN algorithm in the first channel resource allocation model at this time can maximize the channel utilization by allocating channel resources for the target service type of the terminal device.

[0123] In one embodiment of this disclosure, before step 203, which is to obtain the target channel resource indication information corresponding to the target service type based on the channel resources, channel bandwidth and channel occupancy duration corresponding to the edge server and the pre-acquired first channel resource allocation model, method 200 further includes the following steps (k1) to (k2).

[0124] k1. Receive the first indication information sent by the cloud controller, which is used to indicate the channel resources corresponding to the edge server;

[0125] k2. Determine the channel resources corresponding to the edge server based on the first instruction information.

[0126] As previously described, during step 203, the channel resources allocated by the first channel resource allocation model to the target service type of the terminal device are a portion of the channel resources corresponding to the edge server, which are allocated by the cloud controller. Therefore, the edge server will also receive a first indication message from the cloud controller to determine the corresponding channel resources based on this message. It should be understood that the channel resources allocated by the cloud controller to the edge server can be referred to the relevant descriptions in steps 301 and 302 of method 300, and will not be repeated here.

[0127] In one embodiment of this disclosure, after step 204, i.e., obtaining the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, method 200 may further include the following step(s).

[0128] s. Send a second instruction to the cloud controller. The second instruction is used to indicate the different channel utilization rates of the edge server in different first time slots within a first preset time period. The first time slot is part of the first preset time period, and the first preset time period corresponds to the channel time domain of the edge server.

[0129] The first preset time period is the time range of the channel time domain resources corresponding to the edge server.

[0130] In this embodiment, the edge server can obtain the channel utilization rate in each of the M time slots corresponding to the edge server through step 204, and can inform the cloud controller of the channel utilization rate in each time slot so that the cloud controller can use it in step 303 of method 300.

[0131] It should be understood that the specific method by which the cloud controller uses the channel utilization sent by the edge server can be found in the relevant description of step 303 in method 300, and will not be repeated here.

[0132] In one embodiment of this disclosure, method 200 may further include the following steps (t1) to (t2):

[0133] t1. Determine the data packet size and signaling information of the service packets sent by at least one terminal device to the edge server within the second preset time period. The signaling information includes: the number of Evolved Radio Access Bearer (E-RAB) connections, the number of Radio Resource Control (RRC) connections, and the number of paging responses.

[0134] t2. Based on the data packet volume and signaling information, generate dataset information and send the dataset information to the cloud controller;

[0135] The second preset time period is the time period preceding the first preset time period corresponding to the edge server, and the duration of the second time period is an integer multiple of the duration of the first preset time period.

[0136] In this embodiment, in order to ensure the cloud controller's reasonable allocation of channel resources to the edge server, the edge server needs to obtain the data volume and related signaling information of the service packets of all terminal devices within the communication coverage area over a long period of time, and send them to the cloud controller for use by the cloud controller when executing method 300.

[0137] It should be understood that the specific usage of the dataset information sent by the cloud controller to the edge server can be found in the relevant description in method 300, and will not be repeated here.

[0138] Figure 3 This diagram illustrates another flowchart of a channel resource allocation method according to an embodiment of the present disclosure, which can be applied to, for example... Figure 1 The cloud controller shown may be executed by a physical device capable of providing cloud controller functions, or by a component (such as a chip) configured in the physical device, or by a module capable of implementing some or all of the cloud controller functions, etc. This application does not limit this.

[0139] For ease of understanding, this disclosure uses a cloud controller as an example to describe the method provided herein. The steps in method 300 are described in detail below.

[0140] In step 301, the communication traffic of the edge server in the first preset time period is obtained;

[0141] In step 302, based on the communication traffic and a pre-acquired second channel resource allocation model, first indication information is obtained, which is used to indicate the channel resources corresponding to the edge server.

[0142] In step 303, the reward for at least one edge server is obtained based on the channel bandwidth occupancy rate of at least one edge server during a first preset time period.

[0143] In step 304, the second channel resource allocation model is updated based on the feedback.

[0144] In one embodiment of this disclosure, the first indication information specifically indicates the channel frequency domain resources of the edge server within a first preset time period. It should be understood that channel frequency resources are also channel bandwidth resources.

[0145] In the above implementation, the cloud controller acquires the communication traffic of the edge server within a first preset time period. Based on this communication traffic and a second channel resource allocation model, it obtains first indication information to indicate the corresponding channel resources for the edge server, thus allocating the corresponding channel resources to the edge server. After allocating corresponding channel resources to at least one edge server in this way, the cloud controller can also obtain the feedback from the at least one edge server based on its channel utilization within the first preset time period, and update the second channel resource allocation model based on the feedback. Since the cloud controller allocates corresponding channel resources to the edge server based on its communication traffic within the first preset time period, the allocated channel resources are more in line with the actual needs of the edge server, and the probability of the edge server using the allocated channel resources will increase, thereby ensuring channel utilization. At the same time, since the cloud controller monitors the feedback on the allocated channel resources and updates the first channel resource allocation model based on the feedback, the updated second channel resource allocation model can allocate more suitable channel resources to the edge server, further improving channel utilization. Therefore, channel resource utilization can be improved while meeting the needs of the edge server.

[0146] In one embodiment of this disclosure, before step 301, i.e., obtaining the communication traffic of the edge server in the first preset time period, method 300 may further include the following step (x).

[0147] x. Receive dataset information sent by the edge server. The dataset information includes the data volume and signaling information of service packets sent by at least one terminal device to the edge server within a second preset time period. The signaling information includes: E-RAB connection count, RRC connection count, and paging response count.

[0148] In one embodiment of this disclosure, the second preset time period is the time period preceding the first preset time period, and the duration of the second preset time period is Q times the duration of the first preset time period, where Q > 1.

[0149] In the above implementation, before step 301, the edge server can count the total size of service packet data reported by each terminal device within the communication coverage area, as well as the number of E-RAB connections, RRC connections, and paging responses, on an hourly or daily basis, thereby generating dataset information and sending this dataset information to the cloud controller. For example, the edge server can continuously count the total size of service packet data, the number of E-RAB connections, the number of RRC connections, and the number of paging responses for 24 hours on an hourly basis and send it to the cloud controller. It can be understood that each edge server can use this method to send dataset information to the cloud controller, that is, the cloud controller can receive dataset information from each edge server.

[0150] In one embodiment of this disclosure, step 301, namely obtaining the communication traffic of the edge server within a first preset time period, includes the following step (y).

[0151] y. Based on the dataset information and the pre-acquired traffic prediction model, obtain the communication traffic of the edge server in a first preset time period. The first preset time period is the time period after the second preset time period. The duration of the second preset time period is Q times the duration of the first preset time period, where Q > 1.

[0152] It should be understood that the first preset time period is a future time period after the second preset time period, that is, a time period after the current time node.

[0153] In one embodiment of this disclosure, the traffic prediction model can be deployed in a cloud controller.

[0154] In one embodiment of this disclosure, the traffic prediction model may include K one-dimensional convolutional neural networks and one long short-term memory neural network. The number of one-dimensional convolutional neural networks is determined by the data type in the dataset information. For example, if the dataset information includes four data types—total data volume of data packets in a service packet, E-RAB connection count, RRC connection count, and paging response count—then the number of one-dimensional convolutional neural networks K = 4.

[0155] Each convolutional neural network (CNN) consists of an input layer, a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, and an output layer. The first and second convolutional layers have 128 and 64 convolutional kernels, respectively, both with a 3×3 kernel size. The first and second pooling layers have a 2×2 kernel size. Each CNN's input layer corresponds to a data type in the dataset. The second pooling layer flattens the data from multidimensional to one-dimensional vectors, which serve as the output of the CNN. Each CNN's output layer outputs a one-dimensional vector. All one-dimensional vectors from the CNNs are concatenated to form a single input for a Long Short-Term Memory (LSTM) neural network. The LSM contains 128 neurons, each corresponding to a node in the LSM's output layer, meaning the LSM can output 128 possible predicted communication flows. A Dropout layer follows the LSM, which limits the probability of each node being dropped, for example, by setting it to 0.1. After the Dropout layer, an output layer is connected, which can output the predicted communication traffic of the edge server in the first preset time period.

[0156] In other words, the four data types—total data volume of data packets, number of E-RAB connections, number of RRC connections, and number of paging responses—are used as four nodes in the input layer of the traffic prediction model, with one node corresponding to each data type. The communication traffic of the edge server in the first preset time period is used as the node in the output layer of this traffic prediction model; that is, the output layer outputs the predicted communication traffic of the edge server in the future.

[0157] In the above implementation, after the cloud controller obtains the dataset information sent by the edge server through step (x), it can input the total data volume of the service packets, the number of E-RAB connections, the number of RRC connections, and the number of paging responses from the dataset information into the four input layers of the traffic prediction model. The traffic prediction model, through algorithm execution, can output the predicted communication traffic of the edge server within a first preset time period in the future through the output layer. For example, if the input layer of the traffic prediction model contains the total data volume of service packets, the number of E-RAB connections, the number of RRC connections, and the number of paging responses for 24 consecutive hours, and the output layer of the traffic prediction model outputs the predicted communication traffic of the edge server for the next hour, then the cloud controller can obtain the predicted value of the communication traffic of the edge server for the next hour. It can be understood that the cloud controller can obtain the predicted value of the communication traffic of each edge server within a future period through this method.

[0158] In one embodiment of this disclosure, the traffic prediction model in step (y) above can also be trained by the following execution process (1) to (3).

[0159] (1) Obtain the sample dataset.

[0160] Specifically, the edge server statistically analyzes the data packet volume and signaling information of all terminal devices within its communication coverage area, using hours or days as the time granularity. It generates a sample dataset and sends it to the cloud controller, which then retrieves the sample dataset. In other words, the cloud controller obtains the sample dataset in the same way as in step (x) described above, using the acquired data as sample data. For example, the cloud controller might obtain a sample dataset from an edge server containing the total size of the service packet data, E-RAB connection counts, RRC connection counts, and paging response counts for all terminal devices within the edge server's communication coverage area over a continuous 30-hour period.

[0161] (2) Determine the training dataset and validation dataset based on the preset sample training ratio and the sample dataset.

[0162] Specifically, using time as the scale, the data in the sample dataset is divided into a training dataset and a validation dataset in an 8:2 ratio. For example, the total data volume of service packets, the number of E-RAB connections, the number of RRC connections, and the number of paging responses from the first 24 hours of the sample dataset are divided into training data, forming the training dataset. Similarly, the total data volume of service packets, the number of E-RAB connections, the number of RRC connections, and the number of paging responses from the last 6 hours of the sample dataset are divided into validation data, forming the validation dataset.

[0163] (3) Train the traffic prediction model using the training dataset and verify the accuracy of the traffic model using the validation dataset.

[0164] Specifically, training data from the training dataset is input into a pre-defined traffic prediction model. Supervised learning is used to enable the model to learn the correspondence between four types of data—total data volume of service packets, E-RAB connection count, RRC connection count, and paging response count—and communication traffic, thus training the traffic prediction model. After training, multiple validation datasets from the validation dataset are input into the trained model, allowing it to output the predicted communication traffic for a future time period. The prediction accuracy is determined by the root mean squared error (RMSE) between the predicted and actual traffic. When the accuracy is low (i.e., the RMSE is high), the Adam algorithm is used to optimize the model and improve its prediction accuracy.

[0165] In one embodiment of this disclosure, step 302 involves obtaining first indication information based on the communication traffic and a pre-acquired second channel resource allocation model. This first indication information is used to indicate the channel resources corresponding to the edge server and can be obtained in the following manner.

[0166] The second channel resource allocation model can be deployed in the cloud controller.

[0167] The second channel resource allocation model adopts the Deep Deterministic Policy Gradient (DDGP) algorithm, which includes two basic elements: state and action.

[0168] Here, the state is defined as the channel bandwidth of each edge server at the current time, which can be expressed as Si = {S1, S2, ..., Sn}. Here, Si represents the channel bandwidth of the i-th edge server at the current time.

[0169] The action is defined as the channel bandwidth allocated to each edge server during the first preset time period after the current moment, which can be represented as Ai = {A1, A2, ..., An}. Here, Ai represents the channel bandwidth allocated to the i-th edge server during the first preset time period after the current moment.

[0170] It should be understood that the channel bandwidth involved in this action is represented as a normalized value, ranging from [0 to 1]. When Ai = 0, it means that no channel bandwidth is allocated to the i-th edge server during the first preset time period after the current time. When Ai = 1, it means that all channel bandwidth is allocated to the i-th edge server during the first preset time period after the current time.

[0171] In one embodiment of this disclosure, the DDGP algorithm can be implemented based on an actor-critic network. Preferably, the actor network has 100 neurons in its hidden layer, and the critic network has 150 neurons in its hidden layer.

[0172] In the above implementation, after the cloud controller obtains the communication traffic of the edge server predicted by the traffic prediction model in step 201 for the first preset time period, it can first convert the communication traffic into the corresponding bandwidth and then normalize the bandwidth to obtain a bandwidth reference value. Based on this bandwidth reference value and a preset deviation value, the bandwidth range corresponding to the edge server is determined. For example, if the bandwidth reference value obtained after the cloud controller converts the communication traffic is Bp, and the preset deviation value is 0.1, then the bandwidth range corresponding to the edge server can be determined to be [Bp-0.1, Bp+0.1]. The cloud controller can obtain the channel bandwidth of each edge server at the current moment, and the channel bandwidth allocated to each edge server in the first preset time period after the current moment, thus obtaining the two basic elements of state and action in the DDGP algorithm of the second channel resource allocation model. Furthermore, by combining the acquired state and action data, and determining the bandwidth range [Bp-0.1, Bp+0.1] for the edge server, the DDGP algorithm can determine the channel bandwidth allocated to the edge server for the first preset time period after the current moment. This means that the corresponding channel resources are allocated to the edge server through the second channel resource allocation model. In this way, the cloud controller can allocate corresponding channel resources to each edge server, that is, allocate channel bandwidth within the first preset time period. Since the channel resources allocated to the edge server are based on the data packet volume and signaling information of the edge server's service packets in the past second preset time period, it better matches the actual traffic situation of the edge server. Therefore, it can achieve channel resource allocation that meets the needs of the edge server and realize resource allocation for differentiated user groups.

[0173] In one embodiment of this disclosure, step 303, namely obtaining the reward of at least one edge server based on the channel utilization rate of at least one edge server within a first preset time period, includes the following step (z).

[0174] z. Based on the channel utilization of each edge server in the first preset time period, the transmission delay of each data packet of each terminal device, and the preset delay threshold, determine the reward of at least one edge server according to the preset reward calculation formula.

[0175] In one embodiment of this disclosure, the return calculation formula is as follows:

[0176]

[0177] Among them, R i This represents the reward for the edge server; α represents the weight, with a value ranging from [0 to 1]; p i This represents the channel utilization rate of the edge server within the first time slot; All represents the sum of channel utilization of the edge server within the first preset time period; All represents the total number of data packets sent by all terminals within the communication coverage area of ​​the edge server within the first preset time period; Success represents the number of data packets whose transmission delay is less than the delay threshold among all data packets sent by all terminals within the communication coverage area of ​​the edge server within the first preset time period.

[0178] It should be understood that the length of the first preset time period is T, which is 1 / Q of the length of the second preset time period, i.e., T = 1 / Q.

[0179] It should be understood that different service types have different data packet latency tolerance thresholds.

[0180] In the above implementation, as described in method 200, each edge server can send its channel utilization rate for each time slot within a first preset time period to the cloud controller. After receiving the channel utilization rate of each edge server for each time slot within the first preset time period, the cloud controller can determine the reward of each edge server based on the reward calculation formula, calculate the sum of the rewards of all edge servers, and thus obtain the total reward sum.

[0181] In one embodiment of this disclosure, step 304, namely updating the second channel resource allocation model according to the feedback, can be achieved by updating the second channel resource allocation model in the following manner.

[0182] As mentioned earlier, the DDGP algorithm used in the second channel resource allocation model includes two basic elements: state and action. To update this algorithm, the DDGP algorithm can also include a reward. The reward is defined as the sum of the total rewards from all edge base stations within a first preset time period.

[0183] In this implementation, the cloud controller obtains the total reward of all edge base stations within the first preset time period through step 303, thus knowing the reward of the DDGP algorithm. The reward can then be used as a parameter to solve the Bellman equation, thereby updating the second channel resource model.

[0184] In other words, by using the total reward obtained in step 303, it can be determined whether the channel bandwidth allocated to the edge server by the DDGP algorithm can achieve a high channel occupancy rate within the first preset time period. If the channel occupancy rate is not high enough, the DDGP algorithm in the second channel resource allocation model can be adjusted based on the total reward feedback, so that when allocating channel bandwidth to each edge server for the first preset time period in the next round, the allocated channel resources can be utilized to a greater extent, thereby improving channel utilization.

[0185] In one embodiment of this disclosure, after step 304, i.e., updating the second channel resource allocation model according to the reward, method 300 may further include: determining whether the DDPG algorithm of the second channel resource allocation model has converged; if it has not converged, then return to step 301; if it has converged, then end.

[0186] In this implementation, by determining whether the DDPG algorithm has converged, the DDPG algorithm is iteratively updated until it converges. This indicates that the DDPG algorithm in the second channel resource allocation model at this time can maximize the channel utilization by allocating the channel bandwidth to each edge server within the first preset time period.

[0187] In one embodiment of this disclosure, after step 302, i.e., obtaining the first indication information based on the communication traffic and a pre-acquired second channel resource allocation model, method 300 may further include: sending the first indication information to the edge server.

[0188] In this embodiment, after the cloud controller allocates the channel bandwidth corresponding to the first preset time period to each edge server in step 302, it can notify each edge server so that each edge server can allocate the channel resources corresponding to each target service type of each terminal device in step 203 of method 200 based on the allocated channel resources.

[0189] In one embodiment of this disclosure, method 300 may further include: receiving second indication information from an edge server, the second indication information being used to indicate the channel utilization rate of the edge server in different first time slots within a first preset time period, the first time slot being a part of the first preset time period; and determining the channel utilization rate of the edge server within the first preset time period based on the second indication information.

[0190] In this implementation, as described in method 200, each edge server can send its channel utilization rate for each time slot within a first preset time period to the cloud controller. After receiving the channel utilization rates for each time slot from the edge servers within the first preset time period, the cloud controller can determine the sum of the channel utilization rates. In this way, the cloud controller can determine the channel utilization rate of each edge server within a first preset time period.

[0191] It should be noted that in this disclosure, method 200 and method 300 can be implemented separately or in combination. When method 200 and method 300 are implemented in combination, the channel utilization can be further improved.

[0192] The following are embodiments of the apparatus disclosed herein, which can be used to execute embodiments of the method disclosed herein.

[0193] Figure 4 This diagram illustrates a structural block diagram of a channel resource allocation apparatus according to an embodiment of the present disclosure. This apparatus can be implemented as part or all of an electronic device through software, hardware, or a combination of both. Figure 4 As shown, the channel resource allocation device 400 includes:

[0194] The first determining module 410 is configured to determine the target service type corresponding to the session data packet of the terminal device;

[0195] The second determining module 420 is configured to determine the channel bandwidth and channel occupancy duration corresponding to the terminal device based on the target service type.

[0196] The first acquisition module 430 is configured to acquire target channel resource indication information corresponding to the target service type based on the channel resources corresponding to the edge server, the channel bandwidth and the channel occupancy duration, and a pre-acquired first channel resource allocation model. The target channel resource indication information is used to indicate at least a portion of the channel resources corresponding to the edge server.

[0197] The second acquisition module 440 is configured to acquire the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in a first time slot, wherein the first time slot is at least a portion of the time domain resources in the channel resources corresponding to the edge server.

[0198] The first update module 450 is configured to update the first channel resource allocation model based on the channel utilization.

[0199] In one embodiment of this disclosure, the first determining module 410 includes:

[0200] The first receiving module is configured to receive the session data packet sent by the terminal device. The session data packet includes R sub-data packets, each sub-data packet includes Z bytes, where R > 1 and Z > 1.

[0201] The first determining submodule is configured to determine Y target data packets from the R subdata packets based on a preset number Y of subdata packets, where Y < R;

[0202] The second determining submodule is configured to determine X target bytes in each of the Y target data packets according to a preset number of bytes X, where X < Z;

[0203] The first generation module is configured to determine the number of rows of the service matrix as X based on the X target bytes and the number of columns of the service matrix as Y based on the Y target data packets, and generate the service matrix with an X×Y dimension, wherein each element in the service matrix corresponds to byte data under different combinations of target data packets and different target bytes;

[0204] The first input module is configured to input the service matrix into a preset service identification model to obtain the target service type corresponding to the session data packet.

[0205] In one embodiment of this disclosure, the second acquisition module 420 is specifically configured to acquire the number of unit bandwidths occupied by at least one terminal device in a first time slot; and to acquire the channel utilization rate based on the ratio of the number of unit bandwidths to the number of unit bandwidths N corresponding to the edge server.

[0206] In one embodiment of this disclosure, the device 400 further includes:

[0207] The second receiving module is configured to receive first indication information sent by the cloud controller, the first indication information being used to indicate the channel resources corresponding to the edge server;

[0208] The third determining module is configured to determine the channel resources corresponding to the edge server based on the first indication information.

[0209] In one embodiment of this disclosure, the device 400 further includes:

[0210] The first sending module is configured to send second indication information to the cloud controller. The second indication information is used to indicate the channel utilization rate of the edge server in different first time slots within a first preset time period. The first time slot is a part of the first preset time period, and the first preset time period corresponds to the channel time domain of the edge server.

[0211] In one embodiment of this disclosure, the device 400 further includes:

[0212] The fourth determining module is configured to determine the data packet size and signaling information of the service packets sent by at least one terminal device to the edge server within a second preset time period. The signaling information includes: the number of Evolved Radio Access Bearer (E-RAB) connections, the number of Radio Resource Control (RRC) connections, and the number of paging responses.

[0213] The second sending module is configured to generate dataset information based on the data volume of the data packet and the signaling information, and send the dataset information to the cloud controller;

[0214] Wherein, the second preset time period is the time period preceding the first preset time period corresponding to the edge server, and the duration of the second time period is an integer multiple of the duration of the first preset time period.

[0215] Figure 5 This diagram illustrates another structural block diagram of a channel resource allocation apparatus according to an embodiment of the present disclosure. This apparatus can be implemented as part or all of an electronic device through software, hardware, or a combination of both. Figure 5 As shown, the channel resource allocation device 500 includes:

[0216] The third acquisition module 510 is configured to acquire the communication traffic of the edge server in a first preset time period;

[0217] The fourth acquisition module 520 is configured to acquire first indication information based on the communication traffic and a pre-acquired second channel resource allocation model, wherein the first indication information is used to indicate the channel resources corresponding to the edge server.

[0218] The fifth acquisition module 530 is configured to acquire the reward of the at least one edge server based on the channel utilization rate of the at least one edge server within a first preset time period.

[0219] The second update module 540 is configured to update the second channel resource allocation model based on the reported feedback.

[0220] In one embodiment of this disclosure, the device 500 further includes:

[0221] The third receiving module is configured to receive dataset information sent by the edge server. The dataset information includes the data volume and signaling information of service packets sent by at least one terminal device to the edge server within a second preset time period. The signaling information includes: Evolved Radio Access Bearer (E-RAB) connection count, Radio Resource Control (RRC) connection count, and paging response count.

[0222] In one embodiment of this disclosure, the third acquisition module 510 is specifically configured to acquire the communication traffic of the edge server in a first preset time period based on the dataset information and a pre-acquired traffic prediction model. The first preset time period is the time period after the second preset time period, and the duration of the second preset time period is Q times the duration of the first preset time period, where Q > 1.

[0223] In one embodiment of this disclosure, the fifth acquisition module 530 is specifically configured to acquire the reward of the at least one edge server based on the channel utilization of each edge server in a first preset time period, the transmission delay of each data packet of each terminal device, and a preset delay threshold, and a preset reward calculation formula.

[0224] In one embodiment of this disclosure, the device 500 further includes:

[0225] The third sending module is configured to send the first indication information to the edge server.

[0226] In one embodiment of this disclosure, the device 500 further includes:

[0227] The fourth receiving module is configured to receive second indication information from the edge server, the second indication information being used to indicate the channel utilization rate of the edge server in different first time slots within the first preset time period, the first time slot being a part of the first preset time period;

[0228] The fifth determining module is configured to determine the channel utilization rate of the edge server within a first preset time period based on the second indication information. This disclosure also discloses an electronic device. Figure 6 This diagram illustrates a structural block diagram of an electronic device according to an embodiment of the present disclosure, such as... Figure 6 As shown, the electronic device 600 includes a memory 601 and a processor 602; wherein,

[0229] The memory 601 is used to store one or more computer instructions, which are executed by the processor 602 to implement... Figure 2 and / or Figure 3 The methods and steps involved.

[0230] Figure 7 This is a schematic diagram of the structure of a computer system suitable for implementing a channel resource allocation method according to an embodiment of the present disclosure.

[0231] like Figure 7 As shown, the computer system 700 includes a processing unit 701, which can execute various processes described above based on a program stored in a read-only memory (ROM) 702 or a program loaded from a storage section 708 into a random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the computer system 700. The processing unit 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0232] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed into the storage section 708 as needed. The processing unit 701 can be implemented as a CPU, GPU, TPU, FPGA, NPU, etc.

[0233] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion 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 indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0234] The units or modules described in the embodiments of this disclosure can be implemented in software or hardware. The described units or modules can also be located in a processor, and the names of these units or modules do not necessarily constitute a limitation on the unit or module itself.

[0235] In another aspect, this disclosure also provides a channel resource allocation system, comprising: at least one edge server and a cloud controller, wherein the edge server is used to perform, for example... Figure 2 The method shown; the cloud controller is used to perform, as... Figure 3 The method shown.

[0236] In another aspect, this disclosure also provides a computer-readable storage medium, which may be a computer-readable storage medium included in the apparatus described in the above embodiments; or it may be a standalone computer-readable storage medium not assembled into the apparatus. The computer-readable storage medium stores one or more programs that are used by one or more processors to execute the channel resource allocation method described in this disclosure.

[0237] In another aspect, this disclosure also provides a computer program product, including a computer program / instruction that, when permitted, implements a channel resource allocation method.

[0238] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A channel resource allocation method, characterized in that, The method is applied to an edge server, and the method includes: Determine the target service type corresponding to the session data packets of the terminal device; Based on the target service type, determine the channel bandwidth and channel occupancy duration corresponding to the terminal device; Based on the channel resources corresponding to the edge server, the channel bandwidth, and the channel occupancy duration, and based on a pre-acquired first channel resource allocation model, target channel resource indication information corresponding to the target service type is obtained. The target channel resource indication information is used to indicate at least a portion of the channel resources corresponding to the edge server, so that the terminal device can determine the target channel resources corresponding to the target service type based on the target channel resource indication information. The channel utilization rate of the edge server is obtained based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, wherein the first time slot is at least a portion of the time domain resources in the channel resources corresponding to the edge server. The first channel resource allocation model is updated based on the channel utilization rate.

2. The method according to claim 1, characterized in that, The determination of the target service type corresponding to the session data packet of the terminal device includes: The terminal device receives the session data packet, which includes R sub-data packets, each sub-data packet including Z bytes, where R > 1 and Z > 1. Based on the preset number Y of sub-data packets, Y target data packets are determined from the R sub-data packets, where Y < R; Based on a preset number of bytes X, X target bytes are determined in each of the Y target data packets, where X < Z; The number of rows of the service matrix is ​​determined as X based on the X target bytes, and the number of columns of the service matrix is ​​determined as Y based on the Y target data packets. The service matrix with dimensions X×Y is generated, and each element in the service matrix corresponds to the byte data under different combinations of target data packets and target bytes. The service matrix is ​​input into a preset service identification model to obtain the target service type corresponding to the session data packet.

3. The method according to claim 1 or 2, characterized in that, The step of obtaining the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot includes: Obtain the number of units of bandwidth occupied by at least one terminal device in the first time slot; The channel utilization rate is obtained based on the ratio of the number of units of bandwidth to the number of units of bandwidth N corresponding to the edge server.

4. The method according to claim 1, characterized in that, Before obtaining the target channel resource indication information corresponding to the target service type based on the channel resources corresponding to the edge server, the channel bandwidth, and the channel occupancy duration, according to a pre-acquired first channel resource allocation model, the method further includes: Receive first indication information sent by the cloud controller, the first indication information being used to indicate the channel resources corresponding to the edge server; The channel resources corresponding to the edge server are determined based on the first indication information.

5. The method according to claim 1, characterized in that, After obtaining the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in the first time slot, the method further includes: Send a second indication message to the cloud controller. The second indication message is used to indicate the channel utilization rate of the edge server in different first time slots within a first preset time period. The first time slot is a part of the first preset time period, and the first preset time period corresponds to the channel time domain of the edge server.

6. The method according to claim 5, characterized in that, The method further includes: The data packet size and signaling information of the service packets sent by at least one terminal device to the edge server within a second preset time period are determined. The signaling information includes: the number of Evolved Radio Access Bearer (E-RAB) connections, the number of Radio Resource Control (RRC) connections, and the number of paging responses. Based on the data volume of the data packet and the signaling information, a dataset information is generated and sent to the cloud controller; Wherein, the second preset time period is the time period preceding the first preset time period corresponding to the edge server, and the duration of the second preset time period is an integer multiple of the duration of the first preset time period.

7. A channel resource allocation method, characterized in that, The method is applied to a cloud controller, and the method includes: Obtain the communication traffic of the edge server in the first preset time period; Based on the communication traffic, and using a pre-acquired second channel resource allocation model, first indication information is obtained, which is used to indicate the channel resources corresponding to the edge server. The reward for the at least one edge server is obtained based on the channel utilization rate of the at least one edge server within a first preset time period; Update the second channel resource allocation model based on the reported feedback; The formula for calculating returns is as follows: Among them, R i This represents the reward for the edge server; α represents the weight, with a value ranging from [0 to 1]; p i This represents the channel utilization rate of the edge server within the first time slot; All represents the sum of channel utilization of the edge server within the first preset time period; All represents the total number of data packets sent by all terminals within the communication coverage area of ​​the edge server within the first preset time period; Success represents the number of data packets whose transmission delay is less than the delay threshold among all data packets sent by all terminals within the communication coverage area of ​​the edge server within the first preset time period.

8. A channel resource allocation device, characterized in that, The device is used in an edge server, and the device includes: The first determining module is configured to determine the target service type corresponding to the session data packets of the terminal device. The second determining module is configured to determine the channel bandwidth and channel occupancy duration corresponding to the terminal device based on the target service type. The first acquisition module is configured to acquire target channel resource indication information corresponding to the target service type based on the channel resources corresponding to the edge server, the channel bandwidth, and the channel occupancy duration, according to a pre-acquired first channel resource allocation model. The target channel resource indication information is used to indicate at least a portion of the channel resources corresponding to the edge server, so that the terminal device can determine the target channel resources corresponding to the target service type based on the target channel resource indication information. The second acquisition module is configured to acquire the channel utilization rate of the edge server based on the occupancy rate of the channel bandwidth corresponding to the edge server by at least one terminal device in a first time slot, wherein the first time slot is at least a portion of the time domain resources in the channel resources corresponding to the edge server. The first update module is configured to update the first channel resource allocation model based on the channel utilization.

9. An electronic device, characterized in that, include: A memory and at least one processor; wherein the memory is used to store one or more computer instructions, which are executed by the processor to implement the method steps as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed by the processor, they implement the method steps as described in any one of claims 1 to 7.

Images