A MAC downlink scheduling method, device and equipment

Abstract

The present invention provides a media access control MAC downlink scheduling method, device and equipment, the MAC downlink scheduling method includes: using digital twin technology to simulate downlink scheduling online, and outputting predicted scheduling results of N scheduling periods, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling; the predicted scheduling results of the current scheduling period are corrected according to the input data to obtain the actual scheduling results of the current scheduling period. The embodiment of the present application can improve the real-time performance of the downlink scheduling process.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a medium access control (MAC) downlink scheduling method, device and equipment. Background Art

[0002] Driven by new theories and technologies such as mobile Internet, big data, supercomputing, sensor networks, and brain science, as well as strong demands for economic and social development, artificial intelligence (AI) technology is accelerating its development. In 5G networks, operators have introduced AI technology into network planning, construction, and maintenance, enhancing the network's capabilities in intelligent networking, flexible operation, and efficient business support, while reducing network construction, maintenance, and management costs. For example, AI technology has been introduced into the radio resource management (RRM) function of the radio resource control (RRC) layer at Layer 3 (Layer 3).

[0003] However, since the current and future mobile communication networks are networks of massive Internet of Things and intelligent Internet of Everything, current AI technology cannot adapt to challenges of this order of magnitude. When applied to MAC downlink scheduling, there is still the problem of poor real-time performance of downlink scheduling due to long calculation time. Summary of the invention

[0004] The embodiments of the present invention provide a MAC downlink scheduling method, apparatus and device to solve the problem of poor real-time performance of downlink scheduling when AI technology is applied to MAC downlink scheduling in the related art.

[0005] To solve the above technical problems, the present invention is achieved as follows:

[0006] In a first aspect, an embodiment of the present invention provides a MAC downlink scheduling method, including:

[0007] Use digital twin technology to simulate downlink scheduling online and output predicted scheduling results of N scheduling cycles, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling;

[0008] The predicted scheduling result of the current scheduling period is corrected according to the input data to obtain the actual scheduling result of the current scheduling period.

[0009] In a second aspect, an embodiment of the present invention provides a MAC downlink scheduling device, including:

[0010] An online simulation module, used to simulate downlink scheduling online using digital twin technology, and output predicted scheduling results of N scheduling cycles, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling;

[0011] The scheduling correction module is used to correct the predicted scheduling result of the current scheduling period according to the input data to obtain the actual scheduling result of the current scheduling period.

[0012] In a third aspect, an embodiment of the present invention provides a network side device, including: a processor and a transceiver;

[0013] The processor is used to simulate downlink scheduling online using digital twin technology, and output predicted scheduling results of N scheduling cycles, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling;

[0014] The processor is further used to correct the predicted scheduling result of the current scheduling period according to the input data to obtain the actual scheduling result of the current scheduling period.

[0015] In a fourth aspect, an embodiment of the present invention provides a network side device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps in the MAC downlink scheduling method based on digital twin technology as described in the first aspect are implemented.

[0016] In a fifth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the MAC downlink scheduling method based on digital twin technology as described in the first aspect are implemented.

[0017] In an embodiment of the present invention, digital twin technology is used to simulate downlink scheduling online, and the predicted scheduling results of N scheduling cycles are output, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling; the predicted scheduling results of the current scheduling cycle are corrected according to the input data to obtain the actual scheduling results of the current scheduling cycle. In this way, the scheduling results are predicted in advance according to the historical scheduling data. During the scheduling process, it is only necessary to correct the predicted scheduling results according to the input data, without executing the entire calculation process to determine the actual scheduling results, which greatly reduces the amount of calculation to determine the actual scheduling results. The online simulation method can reduce the transmission of data such as the prediction results, thereby reducing the time-consuming process of determining the actual scheduling results, and achieving the effect of improving the real-time performance of downlink scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0019] Figure 1 It is a schematic diagram of data interaction during downlink scheduling;

[0020] Figure 2 This is one of the flow charts of a MAC downlink scheduling method based on digital twin technology provided in an embodiment of the present invention;

[0021] Figure 3a This is a network architecture diagram to which the MAC downlink scheduling method based on the digital twin technology provided by an embodiment of the present invention can be applied;

[0022] Figure 3b It is a structural diagram of a functional module in a MAC downlink scheduler based on digital twin technology provided by an embodiment of the present invention;

[0023] Figure 4 It is a structural diagram of a MAC downlink scheduling device based on digital twin technology provided by an embodiment of the present invention;

[0024] Figure 5 It is a structural diagram of a network side device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] With the increase in wireless network application scenarios, 5G networks need to support three major scenarios: enhanced mobile broadband (eMBB), massive machine type of communication (mMTC) and ultra-reliable and low latency communications (URLLC). The future mobile communication network will be a network of massive Internet of Things and intelligent connection of all things.

[0027] With the increase in wireless network application scenarios, the increase in devices involved in communication and higher requirements, the amount of data and calculation in the downlink scheduling process will also increase greatly, thereby increasing the latency of downlink scheduling and failing to meet the high real-time performance of wireless networks in terminal scheduling and resource allocation at the air interface.

[0028] In the downlink scheduling process of the embodiment of the present application, the downlink scheduler needs to perform scheduling calculations based on relevant data of the terminal and the network side to obtain the carrying capacity of the network side device, and predict the terminal data reception and transmission behavior of a network side device within a certain period of time according to the actual situation of data reception and transmission of each terminal and the actual scheduling information of each scheduling, and then correct the predicted scheduling result according to the actual measurement conditions (for example: in the process of real-time scheduling of the downlink scheduler at each transmission time interval (Transmission Time Interval, TTI), the predicted scheduling result is corrected based on the digital twin technology and the predicted scheduling result at each TTI, combined with real-time information (such as CQI, the latest amount of data to be sent, etc.), so as to obtain the actual scheduling result of this TTI), and schedule according to the corrected actual scheduling result.

[0029] Among them, digital twin technology makes full use of physical models, sensor updates, operation history and other data, integrates multi-disciplinary, multi-physical quantity, multi-scale, and multi-probability simulation processes, completes mapping in virtual space, and thus reflects the entire life cycle of the corresponding physical equipment.

[0030] In application, the predicted scheduling result of each TTI obtained based on the digital twin technology can also be called: online simulation of downlink scheduling using a scheduling prediction algorithm, which can be specifically: using digital twin technology to obtain data of the downlink scheduling system during system operation, based on the intermediate results of the downlink scheduling system and at least one of the historical data of the downlink scheduling system, to replicate the downlink scheduling process of the system, thereby forming an online simulation system for downlink scheduling, and using the online simulation system to predict the downlink scheduling result (i.e., predicted scheduling result) in the next time period. For example: Figure 1 As shown, the input data in the downlink scheduling process may include the following data:

[0031] UE capabilities, QOS parameters, channel state information (CSI), downlink buffer status, downlink transmit power, acknowledgment (ACK) or negative acknowledgment (NACK) feedback information, downlink MIMO scheme, and edge and center frequency bands, etc.

[0032] When the downlink scheduler obtains the above data, it can perform downlink scheduling, for example:

[0033] The maximum number of bits and layers that can be transmitted in each TTI is determined based on the UE capability. Specifically, the UE capability may include capability parameters such as whether the terminal supports multiple input multiple output (MIMO) capability, and the QOS parameter may include at least one of the following parameters: channel quality indicator (CQI) parameter, guaranteed bit rate (GBR) parameter, and aggregate maximum bit rate (AMBR) parameter.

[0034] For example, in actual use, the downlink transmission power is shared by all users in the same service cell, so the downlink available power should also be considered in the scheduling process;

[0035] For another example, the downlink scheduler can also decide to retransmit data or newly transmit data based on ACK or NACK feedback information;

[0036] For example, the downlink scheduler can also perform resource allocation and priority calculation based on QOS parameters;

[0037] For another example, the downlink scheduler may also select a modulation and coding scheme (MCS) according to the edge frequency and center frequency of the serving cell.

[0038] Specifically, CSI may include at least one of a rank indicator (RI), a precoding matrix indicator (PMI) and a CQI. The downlink scheduler will consider information such as channel quality during the process of scheduling the terminal.

[0039] It can be seen from the above that in the downlink scheduling process, massive amounts of data need to be calculated, which prolongs the calculation time of the downlink scheduling. In view of the substantial increase in the number of terminals and the higher demand for service quality from the terminals, the embodiments of the present application can calculate massive amounts of data in advance. In the downlink scheduling process, it is only necessary to correct the calculation of the predicted scheduling results according to the actual scheduling environment, thereby reducing the amount of calculation in the downlink scheduling process and shortening the calculation delay of the downlink scheduling, thereby achieving the effect of improving the real-time performance of the downlink scheduling.

[0040] In an embodiment of the present application, a downlink scheduling method based on the MAC layer is provided, which performs online prediction of the scheduling result in advance through a scheduling prediction algorithm. In this way, in the actual scheduling process, it is only necessary to correct the predicted scheduling result according to the actual input data, which can greatly reduce the amount of calculation in the scheduling process and reduce the delay of the scheduling algorithm.

[0041] See also Figure 2 , Figure 2 is a flowchart of a MAC downlink scheduling method based on digital twin technology provided by an embodiment of the present invention, such as Figure 2 As shown, the MAC downlink scheduling method based on digital twin technology may include the following steps:

[0042] Step 201: Use digital twin technology to simulate downlink scheduling online and output predicted scheduling results for N scheduling cycles, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results for downlink scheduling.

[0043] Step 202: Correct the predicted scheduling result of the current scheduling period according to the input data to obtain the actual scheduling result of the current scheduling period.

[0044] The above-mentioned prediction scheduling result may be a prediction result of one scheduling period (i.e., N is equal to 1), or a prediction result of multiple scheduling periods (i.e., N>1), which is not specifically limited here. In an application, the above-mentioned prediction scheduling result may include a prediction scheduling result within a future time period T, for example, if T is 10ms and the scheduling period TTI is 1ms, then N is equal to 10.

[0045] In a specific implementation, the above-mentioned scheduling prediction algorithm can be a prediction algorithm model trained based on historical scheduling data.

[0046] The prediction algorithm model may include a collection of multiple algorithms, such as sorting algorithms, wireless resource allocation algorithms, power control algorithms, CQI correction algorithms, service QoS identification algorithms, physical resource control algorithms, power control and allocation algorithms, control channel allocation algorithms, and other algorithms. In actual operation, when the scheduling prediction algorithm of the downlink scheduler is running, it can use different sub-class algorithms from the downlink scheduler when it is actually running (each of the above algorithms can have multiple specific sub-class algorithms, such as power control algorithms include open-loop power control algorithms and closed-loop function algorithms, etc.), and it can also use different algorithms respectively.

[0047] Driven by new theories and technologies such as mobile Internet, big data, supercomputing, sensor networks, and brain science, as well as strong demands for economic and social development, AI technology is also accelerating its development. In 5G networks, operators have introduced AI technology into network planning, construction, and maintenance, enhancing the network's capabilities in intelligent networking, flexible operation, and efficient business support, and reducing network construction, maintenance, and management costs. For example, AI technology has been introduced into the RRM function of the L3 RRC layer.

[0048] Therefore, in a specific embodiment of the present invention, the above-mentioned prediction algorithm model may be an AI algorithm, but may also be other algorithms that can predict scheduling results based on historical scheduling data. For ease of explanation, the following only uses the AI ​​algorithm as an example for illustration, which does not constitute a specific limitation.

[0049] In a specific embodiment of the present invention, "online simulation of downlink scheduling using digital twin technology" can be selected to be executed when computing resources are idle, so as to better utilize computing resources and avoid conflicts with other calculations.

[0050] At the same time, in a specific embodiment of the present invention, the above-mentioned scheduling prediction algorithm uses historical scheduling data for online simulation, which can be understood as: the scheduling prediction algorithm can obtain the historical scheduling data of the downlink scheduler online to simulate the scheduling of the existing downlink scheduler based on the historical scheduling data.

[0051] The historical scheduling data may include input data for downlink scheduling and actual scheduling results of downlink scheduling. In addition, the historical scheduling data may also include intermediate data generated during the downlink scheduling process.

[0052] For example, historical scheduling data includes: the wireless resource allocation result (Transport Format Resource Control (TFRC)) of each terminal scheduling, the terminal sorting result at each scheduling, the priority and multiplexing combination of the user data bearer scheduled each time, each wireless physical resource block (PRB), the control channel content and format, the timing relationship of the air interface, the amount of valid data on each bearer, the MAC PDU sent each time, and the size and number of the MAC CE and MAC SDU included, and other scheduling data. It can also include: the time interval between the reception of two adjacent data packets, the size of each data packet, the length of time the data packet is cached, the ratio of the data cache being scheduled empty, and other intermediate data.

[0053] In the application, the prediction results of the scheduling prediction algorithm can be verified according to the actual scheduling results of the downlink scheduling, so that the scheduling prediction algorithm can be used for prediction only when the accuracy of the scheduling prediction algorithm is high enough, otherwise, the original downlink scheduling algorithm can continue to be used for downlink scheduling.

[0054] That is, in an optional implementation, before the online simulation of downlink scheduling using the digital twin technology, the method further includes:

[0055] Training the scheduling prediction algorithm using the historical scheduling data, the predicted scheduling result and the actual scheduling result until the deviation between the predicted scheduling result and the actual scheduling result is less than a predetermined threshold;

[0056] Before the training of the scheduling prediction algorithm is completed, the downlink scheduling algorithm is used to perform scheduling to obtain the actual scheduling result.

[0057] The scheduling prediction algorithm is trained using the historical scheduling data, the predicted scheduling result and the actual scheduling result. Specifically, the historical scheduling data can be input online into the initial scheduling prediction algorithm model to obtain the predicted scheduling result, and the predicted scheduling result is verified using the actual scheduling result corresponding to the same historical scheduling data to obtain the deviation of the scheduling prediction algorithm, and the scheduling prediction algorithm is adaptively adjusted according to the deviation to reduce the deviation. In this way, when the deviation between the predicted scheduling result and the actual scheduling result is less than a predetermined threshold, it can be indicated that the scheduling prediction algorithm has met the accuracy requirement, so that the scheduling prediction algorithm can be used to predict the scheduling result of the downlink scheduling.

[0058] In addition, the training process of the scheduling prediction algorithm needs to last for a certain period of time. At this time, the original downlink scheduling algorithm can be used to obtain the actual scheduling result.

[0059] In actual applications, the specific process of downlink scheduling is as follows:

[0060] The downlink scheduler first uses the existing scheduling algorithm to perform scheduling and outputs the actual scheduling results of the first stage;

[0061] The input data, output data (i.e., actual scheduling results) and intermediate data in the downlink scheduling process are stored locally in the downlink scheduler;

[0062] The scheduling prediction algorithm is trained using the data stored locally in the downlink scheduler until the deviation between the predicted scheduling result output by the scheduling prediction algorithm and the actual scheduling result of the first stage is within a reasonable range, and then the second stage is entered;

[0063] In the second stage, the scheduling prediction algorithm directly uses the locally stored data to output the predicted scheduling results;

[0064] The downlink scheduler directly modifies the predicted scheduling result according to the changes in the actual situation and outputs it as the actual scheduling result.

[0065] The existing downlink scheduling algorithm may be the same as the downlink scheduling algorithm in the prior art, and will not be described in detail here.

[0066] In this embodiment, the scheduling prediction algorithm is trained online using the historical scheduling data, the predicted scheduling results and the actual scheduling results, so that the trained scheduling prediction algorithm is more closely matched with the actual operating scenario of the downlink scheduler, thereby improving the accuracy of the predicted scheduling results.

[0067] In one implementation, the scheduling prediction algorithm can be trained by reporting data by the downlink scheduler. The data reporting method can adopt the reporting method used in the AI-based Minimization Drive Test (MDT) / Self-Organized Networks (SON) project research of the 5G 3rd Generation Partnership Project Radio Access Network (3GPP RAN3).

[0068] In another embodiment, the above-mentioned training of the scheduling prediction algorithm can be an online training process, that is, in an endogenous manner, the MAC downlink scheduler uses its own storage unit to store the input data, output data and intermediate data (such as wireless resource information, scheduling results, control channel information, terminal context information, HARQ allocation information, terminal sorting information, data information of each wireless bearer of the terminal, and information of each data packet sent) during the operation of the MAC downlink scheduler, and the historical data stored locally in the downlink scheduler is obtained online during training to perform online training on the scheduling prediction algorithm based on the historical data.

[0069] When this approach is used, there is no need to report a large amount of data. It is only necessary to perform digital twin cloning of the downlink scheduler during the downlink scheduling process, which can reduce the overhead of the wireless communication system.

[0070] Of course, trained or untrained AI models can also be configured by the operation and maintenance system or RRC signaling.

[0071] That is to say, in this implementation, before the online simulation of downlink scheduling using the digital twin technology, the MAC downlink scheduling method based on the digital twin technology further includes:

[0072] The scheduling prediction algorithm configured by an operation and maintenance system or RRC signaling is received.

[0073] Among them, in the implementation method where the AI ​​model is configured by the operation and maintenance system, the configuration can be performed when the MAC entity is started, or during the operation of the MAC entity.

[0074] In addition, in an implementation where the AI ​​model is configured by RRC signaling, the AI ​​model may be configured by RRC signaling when establishing or reconfiguring the MAC system.

[0075] In this implementation, the above-mentioned AI model can be obtained by operating the maintenance system or RRC signaling configuration, which simplifies the complexity of obtaining the AI ​​model.

[0076] In a specific embodiment of the present invention, the predicted scheduling result of the current scheduling period needs to be corrected according to the input data to obtain the actual scheduling result of the current scheduling period. In an optional implementation, the predicted scheduling result includes at least one of the following data: QoS requirement data, CQI data, bearer information data, scheduling strategy data, transmission resource data, and channel information data.

[0077] For example, in the process of predicting the scheduling result, if the CQI is A1, the predicted scheduling result includes the allocated resource quantity B1; at this time, if the CQI in the actual downlink scheduling process is A1+delta, the allocated resource quantity in the actual scheduling result should be appropriately increased.

[0078] For another example: in the process of predicting the scheduling result, if the CQI is A2, the predicted scheduling result includes an MCS level of B2; at this time, if the CQI in the actual downlink scheduling process is A1+delta, the MCS level in the actual scheduling result can be improved accordingly.

[0079] For another example: in the process of predicting the scheduling results, if the number of terminals to be scheduled is A3, the predicted scheduling result includes the number of allocated resources being B3; at this time, if the number of terminals to be scheduled in the actual downlink scheduling process is less than A3, the number of resources in the actual scheduling result can be appropriately less than B3.

[0080] For another example, when the QoS requirement in the actual scheduling process is different from the QoS requirement in the predicted scheduling result, the bearer should be adjusted to meet the QoS requirement.

[0081] For example: during predictive scheduling, based on the UE's service characteristics, M wireless resources (PRBs) in each of N subframes are continuously allocated to the UE. During actual scheduling, the scheduler makes real-time corrections based on the M PRBs according to the system load in each scheduling cycle to obtain the actual number of PRBs used by the UE in this subframe, which is MX or M+X.

[0082] In this implementation, the downlink scheduler needs to have functions corresponding to the various data in the above-mentioned predicted scheduling results, for example: the MAC downlink scheduler has QoS control function (QoS Controlling), CQI modification function (CQIIimprovement), bearer mapping (Bearer Mapping), business model monitoring and orchestration (Traffic ModelMonitoring and Orchestration), physical channel and resource quality monitoring (The Quality Monitoringof Physical Channel and Physical Resource Element), resource allocation (Scheduling andTFRC per UE) function.

[0083] QoS control function, used to make each scheduled MAC layer meet the QoS requirements of the upper layer bearer and determine the lower layer bearer that meets the QoS requirements;

[0084] A CQI correction function is used to correct the channel quality of the terminal according to the feedback of the terminal;

[0085] A bearer mapping function, used to select a lower layer bearer according to a QoS parameter and determine a mapping relationship between the lower layer bearer and the upper layer bearer;

[0086] The service model monitoring and orchestration function is used to model the services that need to be scheduled and guaranteed by the MAC downlink scheduler according to the regularity of the received data packets on each upper layer bearer and the configured QoS characteristic value of the bearer, and provide appropriate scheduling strategy information for the MAC downlink scheduler;

[0087] Physical channel and resource quality monitoring function, used to monitor the quality of air interface physical channels and terminal transmission resources;

[0088] The resource allocation function is used to allocate transmission resources, scheduled bearers, data volume and air interface channel parameters to terminals.

[0089] In application, the above-mentioned scheduling prediction algorithm may include prediction algorithms corresponding to the above-mentioned functions, so as to simulate each function separately and obtain prediction results. In this way, each functional module only needs to correct its corresponding prediction results to obtain the actual scheduling results.

[0090] For example: Figure 3b As shown, the downlink scheduling processing module includes: a QoS control unit for performing a QoS control function, a CQI correction unit for performing a bearer mapping function, a bearer mapping unit for performing a bearer mapping function, a business model monitoring and orchestration unit for performing a business model monitoring and orchestration function, a physical channel and resource quality monitoring unit for performing a physical channel and resource quality monitoring function, and a resource allocation unit for performing a resource allocation function.

[0091] The specific correction process of each unit is as follows: Figure 2 In the implementation of the method shown, the predicted scheduling result of the current scheduling period is corrected according to the input data, and the process of obtaining the actual scheduling result of the current scheduling period is similar and will not be repeated here.

[0092] In this embodiment, the algorithm in the above-mentioned digital twin module can respectively perform online simulation on each unit included in the downlink scheduling processing module, so that each unit only needs to correct the corresponding simulation prediction results to obtain QoS control, CQI correction, determination of bearer mapping relationship, determination of scheduling strategy and resource allocation and other functions, which can simplify the calculation amount and calculation time of each unit, thereby improving the real-time performance of each unit included in the downlink scheduling processing module.

[0093] In this embodiment, the results of each functional module in the downlink scheduler can be predicted through the predictive scheduling algorithm, so that each functional module only needs to correct its corresponding prediction result without executing the complete calculation process, thereby enhancing the effect of each functional module in the downlink scheduler.

[0094] In this embodiment, the above-mentioned predicted scheduling results are corrected, and the terminal's transmission resources, scheduling strategies, etc. can be adaptively adjusted according to the time-varying characteristics of mobile communications, so as to improve the utilization rate of transmission resources while providing communication services that meet the communication quality.

[0095] In this implementation, historical scheduling data can be stored in the MAC downlink scheduler body, and digital twin technology can be used to obtain real-time local data of the MAC downlink scheduler, and the scheduling prediction algorithm can be used to simulate the function of the MAC downlink scheduler to obtain the predicted results of user scheduling and wireless resource allocation by the MAC downlink scheduler, that is, the predicted scheduling results. In this way, the predicted scheduling results of the MAC downlink scheduler are obtained by means of endogenous digital twins and endogenous prediction scheduling algorithms, so that the calculation of the AI ​​tool does not need to be based on measurement and reporting of wireless networks and terminals, which can reduce the waste of resources caused by measurement and reporting of wireless networks and terminals.

[0096] The following is Figure 3a Taking the wireless network architecture shown in the figure as an example, the MAC downlink scheduling method based on digital twin technology provided in the embodiment of the present application is specifically described:

[0097] like Figure 3a In the network architecture shown, the MAC layer is provided with: a downlink scheduler (DL schedule), a digital twin (Digital Twin, DT) functional entity, an AI warehouse (AI Chest) functional entity, a database of MAC subsystem (Database of MAC), a MAC layer data processing (Data Process for MAC) module and a hybrid automatic repeat request (HARQ) module.

[0098] Among them, the AI ​​warehouse functional entity is used to store and train AI models.

[0099] The database of the MAC subsystem is used to store wireless resource information, scheduling results, control channel information, terminal context information, HARQ allocation information, terminal sorting information, data information of each wireless bearer of the terminal, and information of each data packet sent during the operation of the MAC downlink scheduler.

[0100] In the specific implementation, the MAC subsystem database does not need to store each MAC protocol data unit (Protocol Data Unit, PDU), but only needs to record the key information of each MAC PDU (the digital twin of each MAC PDU, for example: including the total length of MAC PDU, the number of MAC service data units (Service Data Unit, SDU), the size of each MAC SDU, the size and type of MAC channel element (Channel Element, CE), the number of HARQ transmissions and whether they are successful or not, etc.).

[0101] The database of the MAC subsystem is the basis for the operation of the downlink scheduler and the twin of the downlink scheduler. Through the data stored in the database, the simulation operation of the downlink scheduler twin is realized in the form of "0" measurement and "0" reporting.

[0102] The DT functional entity is a full-scale simulation system of the downlink scheduler. The system supports the same scheduling algorithms as the scheduler, including sorting algorithm, wireless resource allocation algorithm, power control algorithm, CQI correction algorithm, service QoS identification algorithm, physical resource control algorithm, power control and allocation algorithm, control channel allocation algorithm, etc. The DT functional entity can also call various AI models in the AI ​​warehouse as needed to realize the training and enhancement of the above algorithms.

[0103] In the application, the DT function obtains the real-time scheduling data of the downlink scheduler through the database of the MAC subsystem. The DT function obtains the intermediate scheduling data of the MAC data processing system through the data of the MAC subsystem, so as to obtain the data reception characteristic value of each data bearer in this way.

[0104] After the DT function module obtains the scheduling data from the MAC subsystem, it runs according to the TTI of the downlink scheduler and compares the generated results with the actual results of the downlink scheduler. When the difference between the two is less than a certain threshold range, it means that the DT function is already available for the system. Then the DT system independently runs the function of the simulated downlink scheduler based on the information in the database of the MAC subsystem, and predicts the next scheduling results, or predicts various sub-functions such as user sorting, resource allocation, power control, etc. in the scheduling, and inputs the results of the advance operation into the downlink scheduler. In this way, the downlink scheduler only needs to correct the predicted results obtained by the DT function according to the input data in the actual operation process to obtain the actual scheduling results, thereby reducing the computational complexity and operation time of the downlink scheduler during operation.

[0105] The DT function generates the QoS characteristic values ​​of the bearers at the lower layer of the MAC layer, as well as the mapping relationship between the upper layer wireless bearers and the lower layer bearers, based on the QoS characteristic values ​​of the wireless bearers at the upper layer of the MAC layer and the characteristics of data reception and transmission.

[0106] During the use of the AI ​​model, the AI ​​model is trained according to the usage results fed back by the DT functional module to improve the effectiveness of the AI ​​model. For example: in the model that predicts the user data packet sending rules, the AI ​​model is trained according to the degree of match between the actual received data packet rules and the rules provided by the model; in the model that predicts the number of wireless resources allocated to the user, the AI ​​model is trained according to the degree of match between the actual wireless resources allocated each time and the results provided by the model; in the model for selecting scheduled users, the AI ​​model is trained according to the degree of match between the actual scheduled users each time and the potential scheduled user list provided by the model, as well as the number or ratio of valid users in the potential scheduled user list, etc., which helps to reduce the running time of the downlink scheduler and improve the scheduling efficiency of various AI models.

[0107] In actual applications, the DT function and the downlink scheduler are parallel functional entities that can run on an independent processor or a processor's processing core, or can run as a thread or process when the processor is idle. In this way, the DT function can provide operation support for the MAC downlink scheduler without competing for computing resources with the actual downlink scheduler.

[0108] In this implementation, the above-mentioned downlink scheduling method is applied to the MAC downlink scheduler to simplify the calculation process of the MAC downlink scheduler. It should be noted that, in actual applications, the above-mentioned downlink scheduling method can also be applied to other network devices that are communicatively connected to the MAC downlink scheduler, which is not specifically limited here.

[0109] The operation of the above scheduler is illustrated as follows.

[0110] During each scheduling operation, the downlink scheduler stores intermediate calculation data, scheduling result information (data) finally sent to the air interface, and information related to the UE and the cell, for example:

[0111] The above data of N scheduling cycles, for example, N = 10000;

[0112] The characteristic values ​​of the data packets received and sent by each UE on each data bearer within a certain period of time, including the number of data packets, the average size of data packets, the maximum and minimum sizes of data packets, the average time interval between adjacent data packets, etc.; and

[0113] UE channel quality related parameters, such as initial BLER, residual BLER, the ratio of a whole data packet being split into small parts for transmission, HARQ information, etc.

[0114] The number of users accessing the cell within a certain period of time T_Cell, the number of users waiting for scheduling, the average time the UE waits for scheduling, PRB utilization, the maximum throughput of the cell, etc.

[0115] Based on the above stored information, DT calculates the carrying capacity of the entire cell, and based on the actual data transmission and reception of each UE and the actual scheduling information of each scheduling, it predicts the user data transmission and reception behavior in a cell within a certain period of time (including pre-scheduling, UE data transmission and reception status in a certain period of time in the future, resource requirements, channel quality changes, candidate UEs to be scheduled, possible allocated resources, data blocks that can be sent, etc.), and outputs large-scale, large-granularity pre-scheduling results.

[0116] The scheduler performs comparison and calculation based on the pre-scheduling result in combination with the real-time channel information of the UE and the real-time conditions (such as the amount of data to be sent, the number of NACKs, etc.), and modifies the pre-scheduling result according to the degree of difference.

[0117] In an embodiment of the present invention, digital twin technology is used to simulate downlink scheduling online, and the predicted scheduling results of N scheduling cycles are output, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling; the predicted scheduling results of the current scheduling cycle are corrected according to the input data to obtain the actual scheduling results of the current scheduling cycle. In this way, the scheduling results are predicted in advance according to the historical scheduling data. During the scheduling process, it is only necessary to correct the predicted scheduling results according to the input data, without executing the entire calculation process to determine the actual scheduling results, which greatly reduces the amount of calculation to determine the actual scheduling results. The online simulation method can reduce the transmission of data such as the prediction results, thereby reducing the time-consuming process of determining the actual scheduling results, and achieving the effect of improving the real-time performance of downlink scheduling.

[0118] See also Figure 4 , is a structural diagram of a MAC downlink scheduling device based on digital twin technology provided by an embodiment of the present invention, such as Figure 4 As shown, the MAC downlink scheduling device 400 based on digital twin technology includes:

[0119] The online simulation module 401 is used to simulate downlink scheduling online using the digital twin technology, and output the predicted scheduling results of N scheduling cycles, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling;

[0120] The scheduling correction module 402 is used to correct the predicted scheduling result of the current scheduling period according to the input data to obtain the actual scheduling result of the current scheduling period.

[0121] Optionally, the scheduling prediction algorithm is an artificial intelligence AI algorithm.

[0122] Optionally, the historical scheduling data also includes intermediate data generated during the downlink scheduling process.

[0123] Optionally, the historical scheduling data is stored locally in the MAC downlink scheduling device.

[0124] Optionally, the MAC downlink scheduling device based on digital twin technology also includes:

[0125] The receiving module is used to receive the scheduling prediction algorithm configured by the operation and maintenance system or RRC signaling.

[0126] Optionally, the MAC downlink scheduling device based on digital twin technology also includes:

[0127] A training module, used to train the scheduling prediction algorithm using the historical scheduling data, the predicted scheduling result and the actual scheduling result until the deviation between the predicted scheduling result and the actual scheduling result is less than a predetermined threshold;

[0128] The temporary scheduling module is used to perform scheduling using a downlink scheduling algorithm before the training of the scheduling prediction algorithm is completed to obtain the actual scheduling result.

[0129] In a specific implementation, the temporary scheduling module may include two working modes. In the first working mode, the temporary scheduling module adopts a scheduling prediction algorithm for scheduling. In the second working mode, the temporary scheduling module adopts a downlink scheduling algorithm for scheduling. Among them, the first working mode means: the prediction scheduling algorithm used to predict the predicted scheduling result is trained until the deviation between the predicted scheduling result obtained by the trained prediction scheduling algorithm and the actual scheduling result is less than a predetermined threshold, and the downlink scheduling is simulated and calculated by the prediction scheduling algorithm to obtain the predicted scheduling result. The second working mode means: in the process of training the prediction scheduling algorithm, when the deviation between the predicted scheduling result obtained by the prediction scheduling algorithm and the actual scheduling result is greater than or equal to the predetermined threshold, that is, when the training of the prediction scheduling algorithm is not completed, the existing downlink scheduling algorithm is used for scheduling.

[0130] The existing downlink scheduling algorithm may be a downlink scheduling algorithm in the prior art, which will not be described in detail here.

[0131] Optionally, the predicted scheduling result includes at least one of the following data: quality of service QoS requirement data, channel quality indication CQI data, bearer information data, scheduling strategy data, transmission resource data and channel information data.

[0132] The MAC downlink scheduling device based on digital twin technology provided in the embodiment of the present application can perform the following Figure 2 The various processes in the method embodiment shown can achieve the same beneficial effects as the method embodiment, and will not be described again here to avoid repetition.

[0133] See also Figure 5 The embodiment of the present invention further provides a network side device, which includes a bus 501, a transceiver 502, an antenna 503, a bus interface 504, a processor 505 and a memory 506.

[0134] The processor 505 is used for:

[0135] Use digital twin technology to simulate downlink scheduling online and output predicted scheduling results of N scheduling cycles, wherein the scheduling prediction algorithm uses historical scheduling data for online simulation, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling;

[0136] The predicted scheduling result of the current scheduling period is corrected according to the input data to obtain the actual scheduling result of the current scheduling period.

[0137] Furthermore, the scheduling prediction algorithm is an artificial intelligence AI algorithm.

[0138] Furthermore, the historical scheduling data also includes intermediate data generated during the downlink scheduling process.

[0139] Furthermore, the MAC downlink scheduling method based on digital twin technology is applied to a MAC downlink scheduler, and the historical scheduling data is stored locally in the MAC downlink scheduler.

[0140] Furthermore, before the processor 505 executes the online simulation of downlink scheduling using the digital twin technology, the transceiver 502 is used to receive the scheduling prediction algorithm configured by the operation and maintenance system or RRC signaling.

[0141] Further, before executing the online simulation of downlink scheduling using the digital twin technology, the processor 505 is also used to:

[0142] Training the scheduling prediction algorithm using the historical scheduling data, the predicted scheduling result and the actual scheduling result until the deviation between the predicted scheduling result and the actual scheduling result is less than a predetermined threshold;

[0143] Before the training of the scheduling prediction algorithm is completed, the downlink scheduling algorithm is used to perform scheduling to obtain the actual scheduling result.

[0144] Furthermore, the predicted scheduling result includes at least one of the following data: quality of service QoS requirement data, channel quality indication CQI data, bearer information data, scheduling strategy data, transmission resource data and channel information data.

[0145] The network side device provided by the embodiment of the present invention can realize Figure 2 To avoid repetition, each process in the illustrated method embodiment will not be described again here.

[0146] exist Figure 5 In the embodiment, the bus architecture (represented by bus 501) is shown, and bus 501 may include any number of interconnected buses and bridges, and bus 501 links various circuits including one or more processors represented by processor 505 and memory represented by memory 506. Bus 501 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. Bus interface 504 provides an interface between bus 501 and transceiver 502. Transceiver 502 may be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. Data processed by processor 505 is transmitted on a wireless medium via antenna 503, and further, antenna 503 also receives data and transmits the data to processor 505.

[0147] The processor 505 is responsible for managing the bus 501 and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management and other control functions. The memory 506 can be used to store data used by the processor 505 when performing operations.

[0148] Optionally, the processor 505 may be a CPU, an ASIC, an FPGA or a CPLD.

[0149] Preferably, the embodiment of the present invention further provides a network side device, including a processor 505, a memory 506, and a computer program stored in the memory 506 and executable on the processor 505, wherein the computer program is executed by the processor 505 to implement the above Figure 2 The various processes of the MAC downlink scheduling method embodiment based on the digital twin technology are shown, and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0150] The embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, which implements the above-mentioned Figure 2 The various processes of the embodiment of the MAC downlink scheduling method based on digital twin technology shown in the figure can achieve the same technical effect. To avoid repetition, they will not be repeated here.

[0151] The computer-readable storage medium is, for example, a ROM, RAM, a magnetic disk or an optical disk.

[0152] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0153] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0154] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A media access control MAC downlink scheduling method, characterized in that: include: Using digital twin technology to simulate downlink scheduling online, and output predicted scheduling results of N scheduling cycles, wherein the online simulation of downlink scheduling using digital twin technology includes using a scheduling prediction algorithm for online simulation, the scheduling prediction algorithm is trained based on historical scheduling data, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling; The predicted scheduling result of the current scheduling period is corrected according to the input data of the current scheduling period to obtain the actual scheduling result of the current scheduling period.

2. The MAC downlink scheduling method according to claim 1, characterized in that: The scheduling prediction algorithm is an artificial intelligence AI algorithm.

3. The MAC downlink scheduling method according to claim 1, characterized in that: The historical scheduling data also includes intermediate data generated during the downlink scheduling process.

4. The MAC downlink scheduling method according to claim 1, characterized in that: The MAC downlink scheduling method is applied to a MAC downlink scheduler, and the historical scheduling data is stored locally in the MAC downlink scheduler.

5. The MAC downlink scheduling method according to claim 1, characterized in that: Before the online simulation of downlink scheduling using the digital twin technology, the method further includes: The scheduling prediction algorithm configured by an operation and maintenance system or a radio resource control RRC signaling is received.

6. The MAC downlink scheduling method according to claim 1, characterized in that: Before the online simulation of downlink scheduling using the digital twin technology, the method further includes: Training the scheduling prediction algorithm using the historical scheduling data, the predicted scheduling result and the actual scheduling result until the deviation between the predicted scheduling result and the actual scheduling result is less than a predetermined threshold; Before the training of the scheduling prediction algorithm is completed, the downlink scheduling algorithm is used to perform scheduling to obtain the actual scheduling result.

7. The MAC downlink scheduling method according to claim 1, characterized in that: The predicted scheduling result includes at least one of the following data: quality of service QoS requirement data, channel quality indication CQI data, bearer information data, scheduling strategy data, transmission resource data and channel information data.

8. A media access control MAC downlink scheduling device, characterized in that: include: An online simulation module, used to simulate downlink scheduling online using digital twin technology, and output predicted scheduling results of N scheduling cycles, wherein the online simulation of downlink scheduling using digital twin technology includes online simulation using a scheduling prediction algorithm, the scheduling prediction algorithm is trained based on historical scheduling data, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling; The scheduling correction module is used to correct the predicted scheduling result of the current scheduling period according to the input data of the current scheduling period to obtain the actual scheduling result of the current scheduling period.

9. The MAC downlink scheduling device according to claim 8, characterized in that: The scheduling prediction algorithm is an artificial intelligence AI algorithm.

10. The MAC downlink scheduling device according to claim 8, characterized in that: The historical scheduling data also includes intermediate data generated during the downlink scheduling process.

11. The MAC downlink scheduling device according to claim 8 or 10, characterized in that: The historical scheduling data is stored locally in the MAC downlink scheduling device.

12. The MAC downlink scheduling device according to claim 8, characterized in that: Also includes: The receiving module is used to receive the scheduling prediction algorithm configured by the operation and maintenance system or the radio resource control RRC signaling.

13. The MAC downlink scheduling device according to claim 8, characterized in that: Also includes: A training module, used to train the scheduling prediction algorithm using the historical scheduling data, the predicted scheduling result and the actual scheduling result until the deviation between the predicted scheduling result and the actual scheduling result is less than a predetermined threshold; The temporary scheduling module is used to perform scheduling using a downlink scheduling algorithm before the training of the scheduling prediction algorithm is completed to obtain the actual scheduling result.

14. The MAC downlink scheduling device according to claim 8, characterized in that: The predicted scheduling result includes at least one of the following data: quality of service QoS requirement data, channel quality indication CQI data, bearer information data, scheduling strategy data, transmission resource data and channel information data.

15. A network side device, characterized in that: include: processor and transceiver; The processor is used to simulate downlink scheduling online using digital twin technology, and output predicted scheduling results of N scheduling cycles, wherein the online simulation of downlink scheduling using digital twin technology includes online simulation using a scheduling prediction algorithm, the scheduling prediction algorithm is obtained by training based on historical scheduling data, and N is an integer greater than or equal to 1; the historical scheduling data includes input data for downlink scheduling and actual scheduling results of downlink scheduling; The processor is further used to correct the predicted scheduling result of the current scheduling period according to the input data of the current scheduling period to obtain the actual scheduling result of the current scheduling period.

16. A network side device, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps in the media access control MAC downlink scheduling method as claimed in any one of claims 1 to 7.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps in the media access control MAC downlink scheduling method according to any one of claims 1 to 7 are implemented.

Images