Resource arrangement method and device and medium

Abstract

The application provides a resource arrangement method and device and medium, and relates to the technical field of networks. The method comprises the following steps: defining a resource allocation variable of a base station, a slice dormancy decision variable and a target rate variable; constructing an energy efficiency performance index optimal solution formula of the base station according to the resource allocation variable, the slice dormancy decision variable and the target rate variable; converting the optimal solution formula into an optimal solution formula of and an optimal solution formula of ; solving the optimal and according to the optimal solution formula of and the optimal solution formula of to obtain an optimal resource arrangement scheme; wherein, represents a rth time frame, represents a kth user, represents an mth transmission attempt, the transmission attempt comprises initial transmission and retransmission, and represents an nth slice. The application realizes high-energy-efficiency resource arrangement.

Description

Technical Field

[0001] This application relates at least to the field of network technology, and in particular to a resource orchestration method, apparatus and medium. Background Technology

[0002] Current hard slice resource orchestration schemes do not consider slice hybrid retransmission, nor do they consider the synergistic optimization of energy efficiency maximization and hybrid retransmission strategies. Under the background of limited resources, high business concurrency, and increasing demand for green energy saving, existing technologies cannot obtain high-energy-efficiency resource orchestration schemes. Summary of the Invention

[0003] To address the aforementioned shortcomings, this application provides a resource orchestration method, apparatus, and medium to solve the following technical problem: how to obtain a high-efficiency resource orchestration scheme.

[0004] Firstly, this application provides a resource orchestration method, the method comprising:

[0005] Define base station resource allocation variables Slicing dormant decision variables and target rate variable ;

[0006] according to , and Energy efficiency performance indicators for building base stations Optimal solution formula;

[0007] Will The optimal solution is transformed into , and Optimal solution formula;

[0008] according to , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration scheme;

[0009] in, This represents the r-th time frame. This represents the k-th user. This represents the m-th transmission attempt, which includes the initial transmission and retransmissions. This represents the nth slice.

[0010] Furthermore, define the resource allocation variables for the base station. Slicing dormant decision variables and target rate variable Specifically, it includes:

[0011] Define the set of resource blocks allocated to the base station: ,in, This represents the resource block allocated to the k-th user in the m-th transmission attempt within the r-th time frame. Let be the two-dimensional index of the resource block used by the k-th user in the m-th transmission attempt. This represents the subframe index of the resource block in the time dimension. This indicates the number of subframes contained in each time frame. This represents the sub-band index of the resource block in the frequency dimension. This indicates the number of subbands contained in each slice. This represents the total number of slices in the system;

[0012] definition Let the nth slice be in the sleep mode of the rth time frame, where, This means that the nth slice will sleep during the rth time frame. This means that the nth slice is active in the rth time frame;

[0013] definition Let be the target transmission rate for the k-th user in the r-th time frame.

[0014] Furthermore, according to , and Energy efficiency performance indicators for building base stations The optimal solution formula includes:

[0015] according to and Building base stations expression;

[0016] according to , and as well as Expression construction under the traditional HARQ protocol First optimal solution;

[0017] Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution.

[0018] Furthermore, according to and Building base stations Expressions, specifically including:

[0019] according to and Building base stations The expression is: ,in, Indicates the total number of time frames. Indicates the total number of users. According to The constructed set of equivalent channel capacities, This represents the distribution of service demand at the base station in the r-th time frame. According to The total energy consumption of the constructed base station in the r-th time frame.

[0020] Furthermore, according to and Building base stations Expressions, specifically including:

[0021] according to Constructing an equivalent channel capacity set ,in, Indicates the first Within the first time frame The first user accessed The signal-to-noise ratio corresponding to each slice. Indicates The normalized channel fading coefficients related to the signal-to-noise ratio for the indexed resource blocks;

[0022] Based on the business model, it can be set to Poisson arrival, Bernoulli arrival, or given by the actual business trajectory. The user in the first Packet arrival distribution within a time frame , Representing the The user in the first Business requirements for each time frame;

[0023] according to Construct the first Total energy consumption of base stations in each time frame ,in, Represents the static power consumption of a base station. Represents dynamic energy consumption. This represents the total number of slices in the system. This represents the total transmit power of the nth slice within the rth time frame. This represents the transmit power of the resource block indexed at (t,f). This represents the fixed RF link power consumption when the transmit link is enabled for the nth slice. This represents the set of resource blocks contained in the nth slice.

[0024] Furthermore, according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution formula specifically includes:

[0025] according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution is:

[0026]

[0027] Where k′ represents any other user besides the k-th user. For the first The user in the first The time frame in the ... Successful transfer indicator variable on each slice This represents the set of users accessing the nth slice. This represents the set of resource blocks allocated to the k-th user during the initial transmission of the r-th time frame. This represents the set of resource blocks contained in the nth slice. This represents the frequency domain index of the resource block used by the k-th user in the m-th transmission attempt. This represents the frequency domain index of the resource block used in the initial transmission of the k-th user. This represents the set of resource blocks corresponding to the initial transmission of user k in the previous time frame. According to and The first The user in the first The delay of one time frame, Indicates the maximum transmission delay. This indicates the total energy consumption limit.

[0028] Furthermore, according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution also includes:

[0029] according to and Construct the first The user in the first Throughput per time frame When an ACK is received =1, when a NACK is received or the transmission fails. =0;

[0030] according to and Construct the first The user in the first buffer load per time frame ,in, For the first The user in the first -1 time frame buffer load;

[0031] according to Construct the first The user in the first Delay of one time frame ,in, Indicates the total number of time frames. This indicates the maximum number of time frames that can be considered in system modeling or simulation.

[0032] Furthermore, The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution formula specifically includes:

[0033] Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution is:

[0034]

[0035] in, This represents the set of optimal resource block allocations obtained by the lower-level argmax algorithm. This represents the optimal solution. Indicates the first The user in the first The throughput per time frame This represents the total number of slices in the system. This represents the set of free resources for the nth slice.

[0036] Furthermore, The optimal solution is transformed into , and The optimal solution formula includes:

[0037] Using the Dinkelbach algorithm The second optimal solution is rewritten as follows: , and The optimal solution is:

[0038]

[0039] in, This represents the objective function after the Dinkelbach transformation.

[0040] Furthermore, according to , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration schemes, specifically including:

[0041] Get current Buffer load In this case, each user accesses only one slice;

[0042] Perform the following iterations:

[0043] calculate As the dynamic threshold for this round, if If it exceeds the dynamic threshold of this round, then ,otherwise To achieve optimization ;

[0044] Based on the dynamic threshold of this round and , Determine the retransmission ratio ,like Then adjust Otherwise keep To achieve optimization ;

[0045] The cost of matching user k and resource block b Where b=(t,f) represents a resource block uniquely determined by subframe index t and subband index f. Indicates The indexed resource block is associated with the normalized channel fading coefficients related to the signal-to-noise ratio, which are derived from the relationship between each user k at the base station and each resource block b. The cost matrix is ​​constructed, and the Hungarian algorithm is used to solve for the cost matrix in order to obtain the optimal solution that minimizes the matching cost. ;

[0046] According to optimization ,optimization ,optimization renew And continue iterating until If the value falls below the preset value or the maximum number of iterations is reached.

[0047] Secondly, this application provides a computer device, the computer device including a processor and a memory, the memory storing a computer program, and when the processor runs the computer program stored in the memory, the processor executes the resource orchestration method as described above.

[0048] Thirdly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the resource orchestration method described above.

[0049] This application provides a resource orchestration method, apparatus, and medium. Based on resource allocation variables, slice dormancy decision variables, and target rate variables, an optimal solution formula for energy efficiency performance indicators is constructed and transformed. By solving the resource allocation variables, slice dormancy decision variables, and target rate variables, the optimal energy efficiency performance indicator resource orchestration scheme is obtained, thereby achieving high-energy-efficiency resource orchestration and reducing algorithm complexity. Attached Figure Description

[0050] Figure 1 This is a flowchart of a resource orchestration method according to an embodiment of this application;

[0051] Figure 2 This is a flowchart illustrating a sleep strategy optimization based on dynamic threshold optimization according to an embodiment of this application;

[0052] Figure 3 This is a flowchart illustrating a target rate strategy optimization based on dynamic threshold optimization according to an embodiment of this application;

[0053] Figure 4 This is a comparison chart of IEE and throughput for different protocols according to an embodiment of this application;

[0054] Figure 5 This is a comparison chart of the running times of different algorithms according to an embodiment of this application;

[0055] Figure 6 This is a schematic diagram of the structure of a computer device according to an embodiment of this application;

[0056] Figure 7 This is a schematic diagram of the structure of a resource orchestration device according to an embodiment of this application;

[0057] Figure 8 This is a schematic diagram of the structure of a computer-readable storage medium according to an embodiment of this application. Detailed Implementation

[0058] To enable those skilled in the art to better understand the technical solution of this application, the embodiments of this application will be further described in detail below with reference to the accompanying drawings.

[0059] It is understood that the specific embodiments and accompanying drawings described herein are merely for explaining this application and are not intended to limit this application.

[0060] It is understood that, without conflict, the various embodiments and features in the embodiments of this application can be combined with each other.

[0061] It is understood that, for ease of description, only the parts relevant to this application are shown in the accompanying drawings, while parts unrelated to this application are not shown in the drawings.

[0062] It is understood that each module or unit involved in the embodiments of this application may correspond to only one entity structure, or may be composed of multiple entity structures, or multiple modules or units may be integrated into one entity structure.

[0063] It is understood that, without conflict, the functions and steps indicated in the flowcharts and block diagrams of this application may occur in a different order than that indicated in the accompanying drawings.

[0064] It is understood that the flowcharts and block diagrams of this application illustrate the possible architecture, functions, and operations of systems, apparatuses, devices, and methods according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, unit, program segment, or code, containing executable instructions for implementing the specified function. Furthermore, each block or combination of blocks in the block diagrams and flowcharts may be implemented using a hardware-based device to implement the specified function, or using a combination of hardware and computer instructions.

[0065] It is understood that the modules and units involved in the embodiments of this application can be implemented by software or by hardware. For example, the modules and units can be located in the processor.

[0066] Example 1:

[0067] like Figure 1 As shown, this application provides a resource orchestration method, the method comprising:

[0068] S1. Define the resource allocation variables for the base station. Slicing dormant decision variables and target rate variable ;

[0069] S2, according to , and Energy efficiency performance indicators for building base stations Optimal solution formula;

[0070] S3 will The optimal solution is transformed into , and Optimal solution formula;

[0071] S4, according to , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration scheme;

[0072] in, This represents the r-th time frame. This represents the k-th user. This represents the m-th transmission attempt, which includes the initial transmission and retransmissions. This represents the nth slice.

[0073] In this embodiment, the method constructs an optimal solution for energy efficiency performance indicators based on resource allocation variables, slice dormancy decision variables, and target rate variables, and transforms the solution to obtain the optimal energy efficiency performance indicator resource orchestration scheme by solving the resource allocation variables, slice dormancy decision variables, and target rate variables, thereby achieving high-energy-efficiency resource orchestration and reducing algorithm complexity.

[0074] Specifically, this embodiment provides a high-efficiency resource orchestration algorithm based on multi-level dynamic threshold optimization.

[0075] With the evolution of fifth-generation (5G) mobile communication systems, communication networks are transforming from a unified bearer to differentiated services to meet the diverse needs of vertical industries. Network slicing technology has attracted much attention because it can provide different Quality of Service (QoS) guarantees and service isolation on the same physical architecture.

[0076] In various typical scenarios such as the Industrial Internet, the Internet of Vehicles, Enhanced Mobile Broadband (eMBB), Ultra-Reliable and Low-Latency Communications (URLLC), and Massive Machine Type Communications (mMTC), different services have vastly different requirements in terms of throughput, latency, and reliability.

[0077] Therefore, it is necessary to study the resource scheduling and management mechanism for slices that cater to differentiated service needs.

[0078] As we move towards the Sixth Generation (6G) mobile communication era, "green and low-carbon" has been established as one of the core concepts from the very beginning of its design. This aims to significantly reduce carbon emissions from communication systems while meeting diverse business needs. Green sustainability has become a prerequisite for 6G network architecture, and energy efficiency optimization is considered a key driver of its development.

[0079] As communication services continue to evolve towards hyper-heterogeneity, the number of network slice instances and the types of services they support are constantly increasing, leading to a significant rise in energy consumption during the operation of base stations and wireless resources. This makes energy efficiency a core challenge for the sustainable operation of future networks. Although the energy efficiency of digital communication technologies has been continuously improving in recent years, some studies indicate that with the large-scale deployment of 5G networks, overall network energy consumption is still showing an upward trend.

[0080] Therefore, the academic community has proposed that future 6G networks need to achieve an average annual network capacity growth of 20% while doubling the energy efficiency of 5G to meet the development goals of green communications. Thus, the contradiction between the green and low-carbon vision and rising communication energy consumption urgently needs to be resolved at the system level. In multi-slice concurrent scenarios, conducting research on maximizing energy efficiency has become an important direction for promoting the green evolution of 6G networks.

[0081] To address unpredictable future business scenarios, deep reinforcement learning (DRL), as a data-driven self-learning method, has been widely applied to the slice resource orchestration problem. This method can automatically learn resource scheduling strategies through interaction with the environment, adapting to complex and dynamic network slice resource orchestration scenarios without explicitly modeling system transition rules, and is gradually becoming a research hotspot.

[0082] Some existing DRL methods use linear programming or integer linear programming models as policy evaluation benchmarks, apply heuristic results to pruning action spaces or initializing evolutionary populations, and combine deep neural networks to improve policy generalization ability and decision-making efficiency for 6G energy efficiency optimization and larger-scale scenarios. Reinforcement learning-based algorithms require a large amount of interactive data and computing power for training and convergence, and online deployment faces training-inference latency bottlenecks.

[0083] For resource orchestration problems in large-scale, multi-slice scenarios, where the objective function is non-convex, the environment is highly dynamic, and the optimization variables are discrete binary variables, reinforcement learning algorithms targeting discrete action spaces are often used, such as Deep Q-Network (DQN) and its variants. DQN is one of the representative algorithms of deep reinforcement learning.

[0084] Because reinforcement learning training samples originate from the agent's continuous exploration and experience accumulation in the environment, they are highly correlated and non-static, which can lead to instability and convergence difficulties during training. To address this issue, DQN introduces two key mechanisms: memory pool replay and a fixed target network. In practical training, DQN may encounter overestimation problems when solving resource orchestration problems. To solve this, Double Deep Q-Network (DDQN) proposes a mechanism to decouple action selection from action evaluation. Specifically, the estimation network selects the action corresponding to the maximum Q-value, and then the target network evaluates the value of that action. However, DDQN is still primarily applicable to discrete action space problems.

[0085] In large-scale, multi-slice scenarios where the objective function is non-convex, the environment is highly dynamic, and the optimization variables are continuous, algorithms targeting continuous action spaces, such as the DDPG algorithm, are more suitable. With improvements in model fusion and computational power, deep reinforcement learning will play an increasingly important role in cross-layer, multi-objective, and multi-slice resource collaborative optimization.

[0086] Current hard slice resource orchestration schemes are generally designed separately from the physical layer HARQ (Hybrid Automatic Repeat Request) protocol, lacking consideration of scheduling coupling with core transmission mechanisms such as HARQ. Especially in the context of resource constraints, high business concurrency, and increasing demand for green energy conservation, there is still a lack of systematic research on how to achieve collaborative optimization between the HARQ protocol and the resource layer. In addition, existing traditional algorithms cannot fully meet the problems of maximizing energy efficiency and retransmission policy conflicts in high-energy-efficiency scenarios; existing reinforcement learning algorithms have the problem of excessive complexity. Therefore, it is necessary to design a slice resource orchestration algorithm for the new HARQ protocol.

[0087] This embodiment addresses the green communication requirements of 6G networks by proposing a multi-level dynamic threshold resource orchestration algorithm that supports cross-slice collaboration. It solves core problems in traditional resource scheduling, such as insufficient energy efficiency optimization, conflict between retransmission and sleep strategies, and high computational complexity of reinforcement learning algorithms. By constructing a two-layer optimization framework and a cross-slice HARQ protocol, it achieves coordinated optimization of energy consumption and throughput, significantly improving system integrated energy efficiency (IEE).

[0088] The technical means of this embodiment include: constructing a cross-slice HARQ-RAN slice collaborative model for maximizing IEE (Infrastructure Energy Efficiency), unifying the modeling of slice dormancy state, user target rate, HARQ initial / retransmission resource allocation and energy consumption constraints, and allowing retransmission resources to be collaboratively orchestrated among the idle resource sets of each slice, constructing an end-to-end collaborative optimization algorithm system for the energy efficiency maximization scenario, and designing corresponding slice resource orchestration algorithms.

[0089] In one implementation, S1, define the resource allocation variables for the base station. Slicing dormant decision variables and target rate variable Specifically, it includes:

[0090] Define the set of resource blocks allocated to the base station: ,in, This represents the resource block allocated to the k-th user in the m-th transmission attempt within the r-th time frame. Let be the two-dimensional index of the resource block used by the k-th user in the m-th transmission attempt. This represents the subframe index of the resource block in the time dimension. This indicates the number of subframes contained in each time frame. This represents the sub-band index of the resource block in the frequency dimension. This indicates the number of subbands contained in each slice. This represents the total number of slices in the system;

[0091] definition Let the nth slice be in the sleep mode of the rth time frame, where, This means that the nth slice will sleep during the rth time frame. This means that the nth slice is active in the rth time frame;

[0092] definition Let be the target transmission rate for the k-th user in the r-th time frame.

[0093] In this embodiment, the method includes:

[0094] Step 1: Construct a system model for the energy efficiency maximization scenario and propose problem modeling:

[0095] Step 1.1: Constructing a resource allocation and transmission model: In the scenario of maximizing energy efficiency, the resource allocation and transmission model is defined, mainly including the definition of resource allocation variables, slice dormancy decision variables, and target rate variables.

[0096] Definition 1.1.1 (Resource Block Allocation Set) )definition Indicates the first In the time frame, it is the _ A set of resource blocks allocated to each user.

[0097] in, Indicates the number of transmission attempts, when This indicates the first transmission, and Indicates the first The second pass, here Indicates the first Within the [number] time frame, the [number]th [time frame] is [time frame]. The user in the first The set of resource blocks allocated for each transmission attempt, including the initial transmission or retransmission.

[0098] in, This represents the subframe index of the resource block in the time dimension, and its value range is limited to the [number]th [frame]. Each time frame contains Within each subframe; This represents the subband index of the resource block in the frequency dimension, and its value range covers the frequency domain resources of all slices, i.e. arrive ; This indicates the number of subframes contained in each time frame. This indicates the number of subbands contained in each RAN (Radio Access Network) slice. This represents the total number of RAN slices in the system.

[0099] Therefore, a resource block can be generated by A unique identifier, that is, one that determines the subframe and subband position of the element. It can be understood as the first The user in the first A two-dimensional index of a particular RB (Resource Block) used in the next transmission attempt.

[0100] in, This indicates the subframe position where the RB is located. Indicates the subband location of the RB; , , Limiting the optional range of the RB index: Within each time frame, there are a total of Subframes, so Must fall arrive Between; the system has There are slices, each slice has There are 1 sub-band, therefore the total frequency domain resources of the entire system are 100 sub-bands. Height belt, so The range is arrive .

[0101] Definition 1.1.2 (Slice Hibernation Mode) )definition For slices exist The sleep mode at any time, in which, Representative slice Carrier-level shutdown, sleep mode. This indicates that the slice resource is active and can transmit and retransmit data packets normally.

[0102] Definition 1.1.3 (User Target Transmission Rate) )definition For users In the User target transmission rate within a time frame.

[0103] In one embodiment, S2, according to , and Energy efficiency performance indicators for building base stations The optimal solution formula includes:

[0104] according to and Building base stations expression;

[0105] according to , and as well as Expression construction under the traditional HARQ protocol First optimal solution;

[0106] Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution.

[0107] In this embodiment, in the system model oriented towards maximizing energy efficiency, an energy efficiency performance index is constructed, and the problem is represented as the energy efficiency maximization problem under the traditional HARQ protocol by optimizing the sleep strategy, target rate and resource allocation. A cross-slice HARQ protocol oriented towards maximizing energy efficiency is designed. Combined with the analysis of the systemic problems of the traditional HARQ protocol under the energy efficiency maximization problem, it is transformed into the IEE maximization problem under the cross-slice HARQ protocol.

[0108] In one embodiment, according to and Building base stations Expressions, specifically including:

[0109] according to and Building base stations The expression is: ,in, Indicates the total number of time frames. Indicates the total number of users. According to The constructed set of equivalent channel capacities, This represents the distribution of service demand at the base station in the r-th time frame. According to The total energy consumption of the constructed base station in the r-th time frame.

[0110] In one embodiment, according to and Building base stations Expressions, specifically including:

[0111] according to Constructing an equivalent channel capacity set ,in, Indicates the first Within the first time frame The first user accessed The signal-to-noise ratio corresponding to each slice. Indicates The normalized channel fading coefficients related to the signal-to-noise ratio for the indexed resource blocks;

[0112] Based on the business model, it can be set to Poisson arrival, Bernoulli arrival, or given by the actual business trajectory. The user in the first Packet arrival distribution within a time frame , Representing the The user in the first Business requirements for each time frame;

[0113] according to Construct the first Total energy consumption of base stations in each time frame ,in, Represents the static power consumption of a base station. Represents dynamic energy consumption. This represents the total number of slices in the system. This represents the total transmit power of the nth slice within the rth time frame. This represents the transmit power of the resource block indexed at (t,f). This represents the fixed RF link power consumption when the transmit link is enabled for the nth slice. This represents the set of resource blocks contained in the nth slice.

[0114] In this embodiment, the method includes:

[0115] Step 1.2: Constructing the energy consumption model and performance indicators:

[0116] Based on the above resource block allocation set The definition, in the first User at time frame The equivalent channel capacity can be expressed as:

[0117]

[0118] in, SNR (Signal to Noise Ratio) represents the signal-to-noise ratio, specifically indicating the... Within the first time frame The first user accessed The SNR corresponding to each slice, Indicates The normalized channel fading coefficients associated with the SNR of the indexed resource block RB are used to characterize the channel gain variation on that time-frequency resource block.

[0119] Therefore, in the capacity formula, Together they reflect the equivalent received channel quality on the RB, where Give the average SNR at the user-slice level. The channel fading impact on specific RBs is given.

[0120] Since, under normal circumstances, a user only accesses one RAN slice, therefore This can be omitted; the first step of user access... The total number of RAN slices in the system is [number of RAN slices]. express.

[0121] consider In the first The total energy consumption of a base station at any given time can be broken down into two parts: static energy consumption and dynamic energy consumption, as shown below:

[0122]

[0123] in, Static power consumption mainly includes resident overhead such as baseband processing, cooling, and signal control, which is unrelated to base station load. Dynamic power consumption fluctuates with wireless resource utilization and transmit power, mainly due to power amplifier and RF link losses.

[0124] in, Slice In the The total transmit power within a time frame is modeled as follows:

[0125]

[0126] in, This represents the transmit power of RB indexed at (t,f). This represents the fixed RF link power consumption of slice n when the transmit link is enabled; This represents the set of radio resource blocks contained in the nth RAN slice, that is, all (t,f) resource blocks covered by this slice.

[0127] Decoupling slice energy consumption from base station load helps to independently control slice activation decisions in subsequent optimization.

[0128] In a system model aimed at maximizing energy efficiency, based on the energy consumption model described above, the following performance indicators should be considered:

[0129]

[0130] in, In a physical sense, it represents the number of new arriving data packets or the arrival process of service requests for user k in the r-th time frame. It represents the user's needs in the r-th time frame. Based on the above model, it is the packet arrival distribution of user k, which can be set to Poisson arrival, Bernoulli arrival, or given by the actual service trajectory according to the service model.

[0131] In one embodiment, according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution formula specifically includes:

[0132] according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution is:

[0133]

[0134] Where k′ represents any other user besides the k-th user. For the first The user in the first The time frame in the ... Successful transfer indicator variable on each slice This represents the set of users accessing the nth slice. This represents the set of resource blocks allocated to the k-th user during the initial transmission of the r-th time frame. This represents the set of resource blocks contained in the nth slice. This represents the frequency domain index of the resource block used by the k-th user in the m-th transmission attempt. This represents the frequency domain index of the resource block used in the initial transmission of the k-th user. This represents the set of resource blocks corresponding to the initial transmission of user k in the previous time frame. According to and The first The user in the first The delay of one time frame, Indicates the maximum transmission delay. This indicates the total energy consumption limit.

[0135] In this embodiment, the method includes:

[0136] Step 1.3: Construct and transform the problem towards maximizing energy efficiency:

[0137] Problem 1.3.1 (IEE Maximization Problem of Traditional HARQ Protocol) This problem refers to maximizing energy efficiency under the traditional HARQ protocol by optimizing the sleep strategy, target rate, and resource allocation:

[0138]

[0139] in, This represents the set of users accessing the nth slice. This represents the set of RBs allocated to user k during the initial transmission of the r-th time frame. This represents the set of RBs contained in the nth slice. This represents the frequency domain index of the RB used by user k in its initial transmission. This represents the set of resource blocks corresponding to the initial transmission of user k in the previous time frame.

[0140] The constraints are as follows: Equation (1.1) indicates that the resource block allocation of user k and any other user k′ in the same transmission round does not overlap, which is used to avoid the same RB being occupied by multiple users at the same time; Equation (1.2) indicates that when there are still unsuccessful transmissions or retransmission constraints for users, the slice state remains the previous frame; Equation (1.3) indicates that the initial transmission resources belong to the same active slice. Traditional HARQ requires the initial transmission and retransmission to be on the same non-dormant slice because soft merging, ACK / NACK status and hard slice isolation are usually bound to the same HARQ process; Equation (1.4) indicates that retransmission can only use the resources remaining after the initial transmission within the same active slice; Equation (1.5) indicates that retransmission under traditional HARQ stop-and-wait uses the frequency domain resources of the initial transmission; Equation (1.6) limits the success indicator variable and the slice state variable to be binary variables; Equation (1.7) indicates the user delay limit. The maximum transmission delay is given by equation (1.8), which represents the energy consumption constraint. This represents the total energy consumption limit. Equations (1.2) to (1.5) represent the traditional HARQ protocol limits.

[0141] In one embodiment, according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution also includes:

[0142] according to and Construct the first The user in the first Throughput per time frame When an ACK is received =1, when a NACK is received or the transmission fails. =0;

[0143] according to and Construct the first The user in the first buffer load per time frame ,in, For the first The user in the first -1 time frame buffer load;

[0144] according to Construct the first The user in the first Delay of one time frame ,in, Indicates the total number of time frames. This indicates the maximum number of time frames that can be considered in system modeling or simulation.

[0145] In this embodiment, the method includes:

[0146] user In the The throughput of the nth time frame can be expressed as the throughput of the nth time frame. The data packets successfully transmitted on the active slice within a time frame can be represented as:

[0147]

[0148] in, For user k, a successful HARQ transmission indicator variable is provided in the r-th time frame and slice n; when an ACK is received... =1, when a NACK is received or the transmission fails. =0, therefore the throughput only counts the amount of data successfully transmitted on the active slice.

[0149] Based on the throughput, the buffer state of user k in the r-th time frame can be obtained. as follows:

[0150]

[0151] in, This represents the number of new arriving data packets or the arrival process of service requests for user k in the r-th time frame. It can be set to Poisson arrival, Bernoulli arrival, or given by the actual service trajectory, depending on the service model.

[0152] Then the delay of user k in the r-th time frame is:

[0153]

[0154] Where T∈{1,..., } represents the total number of time frames transmitted, and is the observation time window used to calculate the average delay. This indicates the maximum number of time frames that can be considered in system modeling or simulation. This indicates the number of data packets to be transmitted for user k.

[0155] In one embodiment, The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution formula specifically includes:

[0156] Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution is:

[0157]

[0158] in, This represents the set of optimal resource block allocations obtained by the lower-level argmax algorithm. This represents the optimal solution. Indicates the first The user in the first The throughput per time frame This represents the total number of slices in the system. This represents the set of free resources for the nth slice.

[0159] It is important to note that the traditional HARQ protocol has the following drawbacks in scenarios requiring maximum energy efficiency:

[0160] Defect 1: Inability to dynamically adapt to load fluctuations: Traditional retransmission parameters (such as maximum number of retransmissions and power level) are statically configured and are difficult to adjust in time when faced with sudden changes in slice traffic, resulting in excessive retransmission or excessive sleep.

[0161] Defect 2: The hibernation strategy is limited to single-slice services: Slice hibernation is only based on the load of the slice itself, without the awareness of retransmission services. It may enter hibernation during peak retransmission periods, causing latency defaults. Each slice runs HARQ retransmission independently, and it is impossible to borrow idle retransmission resources across slices, which easily leads to resource fragmentation and low energy efficiency.

[0162] Defect 3: Conflict in time scale between HARQ retransmission and slice sleep decisions: Traditional HARQ retransmission immediately occupies retransmission resources within millisecond-level time slots, but slice sleep strategies typically make decisions on a longer time scale (seconds or more). The two are difficult to coordinate, leading to low resource utilization and energy waste. Furthermore, most solutions only optimize the number of retransmissions and power for throughput or reliability, without considering overall energy minimization, making it difficult to meet green RAN requirements.

[0163] In this embodiment, the method includes:

[0164] Step 2: Design and problem transformation of cross-slice protocol for hibernation strategy, target rate and resource allocation.

[0165] Step 2.1: Design a cross-slice HARQ protocol for maximizing energy efficiency. Combine the analysis of the systematic problems of the traditional HARQ protocol under the problem of maximizing energy efficiency, and extend the unified retransmission framework to the field of multi-dimensional parameter optimization.

[0166] Addressing the shortcomings of traditional protocols, this section presents an enhanced HARQ-IR (Incremental Redundancy) protocol for maximizing energy efficiency in multi-slice scenarios. This protocol proposes a two-layer problem structure, decoupling resource orchestration from sleep decisions and introducing a cross-slice collaborative mechanism, as detailed below:

[0167] 1. Two-layer optimization framework

[0168] Upper layer – Sleep strategy decision and target rate decision: Based on the load of each slice, the length of the retransmission queue and historical energy consumption indicators, the sleep mode selection problem is solved with a period of second-level time slots.

[0169] Lower layer – retransmission resource orchestration: Within millisecond-level time slots, HARQ retransmission resource allocation is performed for active slices, transforming traditional single-slice scheduling into cross-slice optimization to maximize throughput, thereby maximizing system energy efficiency at the upper layer.

[0170] 2. Cross-slice resource orchestration and flexible hibernation

[0171] Cross-slice resource borrowing: Allows active slices to call on idle retransmission resources of other slices in the lower-level orchestration, eliminating resource silos and realizing the reuse of fragmented resources.

[0172] Load / queue driven sleep switching: Sleep decisions are no longer limited by the traditional retransmission service cycle. Slices can flexibly enter or exit sleep based on real-time load and queue information, ensuring that retransmission is not affected by sleep.

[0173] Energy efficiency gain ranking: Calculate the marginal energy efficiency gain for each slice, and prioritize the idle or inefficient slices to go into hibernation. Their resources are then allocated to efficient slices to improve overall energy efficiency.

[0174] Closed-loop energy consumption feedback: Users transmit actual energy consumption back in ACK / NACK, and the base station dynamically updates the sleep and retransmission model parameters to ensure the coordinated convergence of the two-layer optimization.

[0175] Based on the above cross-slice HARQ protocol design, the IEE maximization problem under the traditional protocol can be transformed into the following form:

[0176] Problem 2.1.1 (IEE Maximization Problem under Cross-Slice HARQ Protocol) The energy efficiency maximization problem under the enhanced HARQ protocol can be formulated as follows:

[0177]

[0178] in, This represents the optimal RB allocation set obtained by the lower-level argmax, with the asterisk indicating the optimal solution; This represents the total set of system users, with a total number of [number]. ; The set of free resources in the nth slice can be defined as follows: .

[0179] Equation (2.3) indicates that the cross-slice HARQ retransmission resources belong to the union of the idle resource sets of each slice, and the user index still comes from the user set of the corresponding slice. .

[0180] In one embodiment, S3, will The optimal solution is transformed into , and The optimal solution formula includes:

[0181] Using the Dinkelbach algorithm The second optimal solution is rewritten as follows: , and The optimal solution is:

[0182]

[0183] in, This represents the objective function after the Dinkelbach transformation.

[0184] Step 2.2 Analysis and Design: This problem is difficult to solve using traditional methods, mainly due to the following two difficulties:

[0185] 1. Mixed variable space: Variable types include continuous variables. set variables and binary variables This leads to a complex solution space, making it difficult to directly apply traditional optimization methods.

[0186] 2. Objective Function: The objective function is a fractional programming problem, which adds nonlinearity and makes the optimization process more difficult.

[0187] Therefore, to address the above challenges, a low-complexity resource orchestration algorithm based on dynamic threshold optimization-Hungarian algorithm is proposed.

[0188] In this embodiment, the method includes:

[0189] Step 3: Design a low-complexity resource orchestration algorithm based on dynamic threshold optimization and the Hungarian algorithm.

[0190] Step 3.1: The original IEE maximization problem can be rewritten into a linear form using the Dinkelbach algorithm (which transforms the nonlinear problem into subproblems through parameterization). The original IEE maximization problem can be decomposed into the following two processes to obtain the optimal solution. That is: in the first process, for a given The variables can be optimized by addressing the following issues. In the second process, updates are made to the given variables. The problem can be further transformed into the following form:

[0191] Problem 3.1.1 (Transformed Given IEE Maximization Problem) can be transformed using the Dinkelbach algorithm, resulting in the following final problem:

[0192]

[0193] in, This represents the objective function after the Dinkelbach transformation. The statement indicates that the actual serviceable volume cannot exceed the smaller of the channel capacity and the service demand; the overall objective function remains maximizing the value after deducting the energy consumption penalty. For example, C=8, D=10, P=3. When, the objective function takes the value .

[0194] In one embodiment, S4, according to , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration schemes, specifically including:

[0195] Get current Buffer load In this case, each user accesses only one slice;

[0196] Perform the following iterations:

[0197] calculate As the dynamic threshold for this round, if If it exceeds the dynamic threshold of this round, then ,otherwise To achieve optimization ;

[0198] Based on the dynamic threshold of this round and , Determine the retransmission ratio ,like Then adjust Otherwise keep To achieve optimization ;

[0199] The cost of matching user k and resource block b Where b=(t,f) represents a resource block uniquely determined by subframe index t and subband index f. Indicates The indexed resource block is associated with the normalized channel fading coefficients related to the signal-to-noise ratio, which are derived from the relationship between each user k at the base station and each resource block b. The cost matrix is ​​constructed, and the Hungarian algorithm is used to solve for the cost matrix in order to obtain the optimal solution that minimizes the matching cost. ;

[0200] According to optimization ,optimization ,optimization renew And continue iterating until If the value falls below the preset value or the maximum number of iterations is reached.

[0201] The complexity and challenges of solving Problem 3.1.1 are mainly reflected in the following two aspects:

[0202] Hybrid variable space construction: The set of optimization variables consists of three types of heterogeneous optimization variables, including dormant mode variables. Given a binary discrete space, the target rate variable The variables constituting the continuous solution space and resource allocation set under non-convex constraints This forms a combinatorial space of discrete finite sets.

[0203] Energy efficiency maximization scenarios: In energy efficiency maximization scenarios, traditional optimization methods may not meet real-time requirements. Furthermore, due to data scarcity and real-time requirements, machine learning algorithms may not be suitable for such problems.

[0204] In network resource scheduling scenarios aimed at maximizing energy efficiency, the design of low-complexity algorithms is of great significance.

[0205] On the one hand, energy efficiency optimization goals are not limited to wireless transmission energy consumption, but also need to cover the computational energy consumption of the algorithm execution itself. Therefore, highly complex algorithms may introduce additional power consumption, which may reduce overall energy efficiency.

[0206] On the other hand, such scenarios are usually accompanied by large-scale terminal access and concurrent scheduling of data slices, requiring optimization algorithms to have good real-time performance and scalability. If the algorithm complexity is too high, it may lead to the inability to complete the calculation within the scheduling cycle, thereby causing resource misallocation and energy efficiency degradation.

[0207] In addition, low-complexity algorithms can significantly reduce system deployment costs and improve their applicability in edge computing platforms and energy-constrained terminals, thus becoming one of the core requirements for energy efficiency optimization system design.

[0208] Therefore, this problem belongs to the category of mixed integer nonconvex optimization problems, and a low-complexity hierarchical optimization strategy is proposed.

[0209] In this embodiment, the method includes:

[0210] Step 3.2: Construct a multi-level dynamic threshold optimization upper-level algorithm.

[0211] Step 3.2.1: Optimization of Sleep Strategy Based on Dynamic Threshold: In scenarios aiming to maximize throughput, the system typically performs sleep strategy optimization on each time frame. This approach uses a switching decision-making method to achieve dynamic resource allocation. However, it ignores the time cost of switching hibernation strategies, which may lead to a decrease in system performance.

[0212] To address this issue, a state transition-based optimization method is proposed. This method models system state changes as transitions from sleep mode to active mode or from active mode to sleep mode, rather than making frequent switching decisions in each time frame. This approach effectively reduces the number of switching operations, lowers system switching overhead, and thus improves overall system performance.

[0213] Specifically, such as Figure 2 As shown, the system's state space is defined as S = {sleep mode, active mode}, and the triggering condition for state switching is controlled by a dynamic threshold mechanism. When the network load exceeds a certain preset threshold, the system switches from sleep mode to active mode; conversely, when the network load falls below another preset threshold, the system switches from active mode to sleep mode. This threshold-based switching strategy can adjust the system's operating mode in real time according to changes in network status, thereby improving resource utilization efficiency and overall system performance. Furthermore, considering the system's latency sensitivity, a dynamic threshold mechanism is introduced to dynamically adjust the switching threshold of the sleep strategy based on network status, adapting to different network environments and further improving system performance.

[0214] Figure 2 In this context, the input variables for the sleep strategy include channel capacity. Service requirements for frame r Total energy consumption and current buffer / load The base station first... Calculate the slice state switching threshold Then compare the current buffer / load |B| with the threshold: when |B| is below the threshold, the slice either enters or remains in sleep mode; when |B| is below the threshold, the slice enters or remains in sleep mode. When the value exceeds a threshold, the slice remains active or is switched to active. The output shows the slice state for the next time frame. For example, if C=8, D_r=5, , The threshold value is 3; when the current load is below 3, it enters sleep mode, and when it is above 3, it remains active.

[0215] State transition-based optimization methods can effectively solve the problem of frequent switching in traditional methods, reduce system switching overhead, and improve the overall system performance, especially in latency-sensitive scenarios where they have significant application value.

[0216] Step 3.2.2: Target rate strategy optimization based on dynamic threshold optimization: when the system capacity Less than the target rate In such cases, the system ensures reliable data transmission through retransmission. The retransmission rate is related to a dynamic threshold. Closely related.

[0217] Specifically, the system calculates the proportion of data that needs to be retransmitted based on the current network status, and adjusts the target rate accordingly. The process of setting the dynamic threshold is as follows:

[0218] 1. Retransmission Ratio Setting: When the capacity is less than the target rate, a dynamic threshold determines the proportion of retransmissions required. The dynamic threshold is adjusted based on the current network capacity and transmission conditions to ensure effective transmission within a reasonable latency. The specific formula is as follows:

[0219]

[0220] in, It is the target rate. It is the capacity of the current time slot and the user. It is a dynamically adjusted retransmission ratio.

[0221] 2. Adjusted target rate: When the capacity is less than the target rate, the system adjusts the target rate according to the retransmission ratio, specifically as follows:

[0222]

[0223] The meaning of this formula is that if the current quantity cannot meet the target rate, the system will determine the proportion of retransmissions based on the dynamic threshold and adjust the target rate to avoid system overload or excessive latency as much as possible.

[0224] To maximize system throughput and minimize latency, the dynamic threshold is dynamically adjusted based on the real-time channel status. This adjustment is made according to the current network conditions and transmission demands. It is obtained by solving the optimization problem.

[0225] The specific algorithm is as follows: Figure 3 As shown, Figure 3 The input variables for the target rate strategy include the current target rate. Channel capacity Dynamic retransmission ratio θ and energy consumption penalty term The system first determines the target rate threshold based on capacity, demand, and energy consumption penalty, and then adjusts the retransmission ratio that needs to be compensated by θ; when At that time, the adjusted target rate is C. .For example , , At this time, the retransmission compensation is 1.5, and the adjusted target rate is 8.5. The output result is the updated target rate, which is used for subsequent resource matching.

[0226] Step 3.3: Lower-level algorithm based on Hungarian matching: Lower-level resource allocation is one of the key tasks in wireless communication systems, especially when considering the allocation of multiple users and multiple resources. To achieve efficient resource matching and allocation, a lower-level algorithm based on the Hungarian algorithm is proposed, which can efficiently match multiple resource blocks and multiple users.

[0227] The basic idea of ​​the Hungarian algorithm is to achieve optimal allocation by minimizing the matching cost. In this algorithm, a cost matrix is ​​first constructed, where each element represents the matching cost between a user and a resource block. Then, the Hungarian algorithm is used to minimize both rows and columns, ultimately obtaining the optimal matching scheme.

[0228] The key steps in this process include:

[0229] Construction cost matrix: for each user and resource blocks ,in, Indicates indexing by subframe and sub-index A uniquely identified resource block. Its matching cost can typically be calculated in the following way:

[0230]

[0231] Applying the Hungarian algorithm: The cost matrix is ​​processed using the Hungarian algorithm to find a matching scheme that minimizes the cost.

[0232] Resource allocation: Based on the results of the Hungarian algorithm, resource blocks are allocated to each user to ensure efficient resource utilization and maximize system performance.

[0233] The advantage of the Hungarian algorithm lies in its ability to find the optimal solution in polynomial time, making it suitable for large-scale resource allocation problems. Using this method, the system can efficiently allocate appropriate resources to each user, maximizing system throughput and minimizing latency. The complete algorithm is shown in pseudocode.

[0234] Algorithm: Lower-level resource allocation algorithm based on Hungarian matching

[0235] enter:

[0236] Set of users to be scheduled ;

[0237] The set of free resource blocks for activating a slice ;

[0238] resource blocks ;

[0239] User-resource block matching cost ;

[0240] Number of transmission attempts .

[0241] Output:

[0242] Resource block allocation set .

[0243] 1: Initialization ;

[0244] 2: ;

[0245] 3: FOR each user DO

[0246] 4: FOR each resource block DO

[0247] 5: IF The constraints of power, delay, and non-overlapping resources are satisfied.

[0248] 6: ;

[0249] 7: ELSE

[0250] 8: ;

[0251] 9: END IF

[0252] 10: END FOR

[0253] 11: END FOR

[0254] 12: ;

[0255] 13: IF no Square Array THEN

[0256] 14: Add virtual rows or virtual columns to... Expand to phalanx;

[0257] 15: END IF

[0258] 16: FOR DO

[0259] 17: ;

[0260] 18: FOR DO

[0261] 19: ;

[0262] 20: END FOR

[0263] 21: END FOR

[0264] 22: FOR DO

[0265] twenty three: ;

[0266] 24: FOR DO

[0267] 25: ;

[0268] 26: END FOR

[0269] 27: END FOR

[0270] 28: WHILE TRUE DO

[0271] 29: Cover all zero elements in C using the fewest number of straight lines;

[0272] 30: The number of lines required to cover all zero elements;

[0273] 31: IF THEN

[0274] 32: BREAK;

[0275] 33: END IF

[0276] 34: The minimum value among all uncovered elements;

[0277] 35: FOR Each element in DO

[0278] 36: IF Not covered by any straight line THEN

[0279] 37: ;

[0280] 38: ELSE IF The point at the intersection of the two straight lines THEN

[0281] 39: ;

[0282] 40: END IF

[0283] 41: END FOR

[0284] 42: End When

[0285] 43: from Select an independent set of zero elements that are distinct from each other in both rows and columns;

[0286] 44: FOR each independent zero element DO

[0287] 45: IF k and b are not virtual elements AND THEN

[0288] 46: ;

[0289] 47: ;

[0290] 48: END IF

[0291] 49: END FOR

[0292] 50: RETURN .

[0293] In this embodiment, to verify the effectiveness of the method, the method further includes:

[0294] Step 4: Construct the actual algorithm and obtain numerical results:

[0295] Step 4.1: Numerical results of protocol validity: Figure 4 Three protocol strategies were compared: the traditional retransmission protocol, the enhanced protocol aimed at throughput, and the cross-slice HARQ protocol proposed in this embodiment aimed at maximizing energy efficiency. Evaluations were made on both average IEE and throughput dimensions.

[0296] Experimental results show that the proposed energy efficiency protocol significantly outperforms the cross-slice protocol focused on throughput optimization in terms of IEE, improving performance. Compared to traditional protocols, it improves In terms of throughput, the proposed protocol achieves 6.0 Mbps, only slightly lower than the 6.3 Mbps of the pure throughput optimization strategy, and significantly higher than the 3.5 Mbps of the traditional protocol. This demonstrates that the proposed method achieves a good balance between significantly improving energy efficiency and maintaining superior throughput performance.

[0297] Step 4.2: Numerical results of time complexity: as follows Figure 5 As shown, the algorithm in this embodiment can complete a resource scheduling decision in just 10 ms, which is significantly better than the 24 ms of the greedy strategy and the 420 ms of the reinforcement learning strategy. Although the reinforcement learning strategy has the best performance, its training and inference complexity is extremely high, making it unsuitable for latency-sensitive applications; while the proposed method balances performance and complexity, making it suitable for deployment in scenarios that maximize energy efficiency.

[0298] In this embodiment of dynamic threshold-Hungarian resource orchestration, when the size of the Hungarian matching matrix is ​​n, the core matching complexity is... If combined with an L-layer neural network inference module, the overall complexity can be expressed as: Where L is the number of layers, , This represents the number of nodes in adjacent layers. Compared to the training and inference complexity of reinforcement learning strategies, this embodiment uses dynamic thresholding and Hungarian matching as its core methods to reduce the complexity of online resource orchestration.

[0299] This embodiment constructs a cross-slice HARQ-RAN slice coordination model oriented towards maximizing IEE. It unifies the modeling of slice sleep state, user target rate, HARQ initial / retransmission resource allocation, and energy consumption constraints, and allows retransmission resources to be coordinated and orchestrated among the idle resource sets of each slice. Through ACK / NACK success indication, cross-slice idle resource sets, dynamic threshold sleep control, and Hungarian resource matching, sleep decisions no longer conflict with HARQ retransmission scheduling, and fragmented idle resources are preferentially allocated to users who need retransmission or high energy efficiency gains. Without destroying hard slice isolation, it improves the utilization rate of idle RBs, reduces the risk of retransmission blocking and sleep mishandling, and improves IEE and latency performance.

[0300] The dynamic threshold-Hungarian low-complexity algorithm in this embodiment decouples the upper-layer sleep / target rate control from the lower-layer RB matching: the upper layer according to... The system generates sleep thresholds and target rate thresholds. The lower layer uses Hungarian matching to find the optimal allocation among multiple users and multiple RBs. This structure reduces the online solution pressure of global mixed integer non-convex problems, making the solution more suitable for energy-constrained and latency-sensitive RAN slice deployments.

[0301] Example 2:

[0302] like Figure 6 As shown, Embodiment 2 of this application provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the resource orchestration method as described in Embodiment 1.

[0303] The memory is connected to the processor. The memory can be flash memory, read-only memory or other types of memory. The processor can be a central processing unit or a microcontroller.

[0304] More specifically, the computer device is as follows Figure 7 The diagram shows a resource orchestration apparatus, the apparatus comprising:

[0305] Define module 1, which is used to define the resource allocation variables of the base station. Slicing dormant decision variables and target rate variable ;

[0306] Module 2 is constructed and connected to Module 1 for use in accordance with... , and Energy efficiency performance indicators for building base stations Optimal solution formula;

[0307] Transformation module 3, connected to construction module 2, is used to convert... The optimal solution is transformed into , and Optimal solution formula;

[0308] Solver module 4, connected to transformation module 3, is used to solve according to... , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration scheme;

[0309] in, This represents the r-th time frame. This represents the k-th user. This represents the m-th transmission attempt, which includes the initial transmission and retransmissions. This represents the nth slice.

[0310] In one embodiment, module 1 is defined, specifically including:

[0311] The first defining unit is used to define the set of resource blocks allocated by the base station: ,in, This represents the resource block allocated to the k-th user in the m-th transmission attempt within the r-th time frame. Let be the two-dimensional index of the resource block used by the k-th user in the m-th transmission attempt. This represents the subframe index of the resource block in the time dimension. This indicates the number of subframes contained in each time frame. This represents the sub-band index of the resource block in the frequency dimension. This indicates the number of subbands contained in each slice. This represents the total number of slices in the system;

[0312] The second defining unit is used to define... Let the nth slice be in the sleep mode of the rth time frame, where, This means that the nth slice will sleep during the rth time frame. This means that the nth slice is active in the rth time frame;

[0313] The third defining unit is used to define Let be the target transmission rate for the k-th user in the r-th time frame.

[0314] In one embodiment, the construction module 2 specifically includes:

[0315] The first building unit is used to determine... and Building base stations expression;

[0316] The second building unit, connected to the first building unit, is used to... , and as well as Expression construction under the traditional HARQ protocol First optimal solution;

[0317] The first transformation unit, connected to the second construction unit, is used to convert... The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution.

[0318] In one embodiment, the first building unit is specifically used for:

[0319] according to and Building base stations The expression is: ,in, Indicates the total number of time frames. Indicates the total number of users. According to The constructed set of equivalent channel capacities, This represents the distribution of service demand at the base station in the r-th time frame. According to The total energy consumption of the constructed base station in the r-th time frame.

[0320] In one embodiment, the first building unit is further configured to:

[0321] according to Constructing an equivalent channel capacity set ,in, Indicates the first Within the first time frame The first user accessed The signal-to-noise ratio corresponding to each slice. Indicates The normalized channel fading coefficients related to the signal-to-noise ratio for the indexed resource blocks;

[0322] Based on the business model, it can be set to Poisson arrival, Bernoulli arrival, or given by the actual business trajectory. The user in the first Packet arrival distribution within a time frame , Representing the The user in the first Business requirements for each time frame;

[0323] according to Construct the first Total energy consumption of base stations in each time frame ,in, Represents the static power consumption of a base station. Represents dynamic energy consumption. This represents the total number of slices in the system. This represents the total transmit power of the nth slice within the rth time frame. This represents the transmit power of the resource block indexed at (t,f). This represents the fixed RF link power consumption when the transmit link is enabled for the nth slice. This represents the set of resource blocks contained in the nth slice.

[0324] In one embodiment, the second building unit is specifically used for:

[0325] according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution is:

[0326]

[0327] Where k′ represents any other user besides the k-th user. For the first The user in the first The time frame in the ... Successful transfer indicator variable on each slice This represents the set of users accessing the nth slice. This represents the set of resource blocks allocated to the k-th user during the initial transmission of the r-th time frame. This represents the set of resource blocks contained in the nth slice. This represents the frequency domain index of the resource block used by the k-th user in the m-th transmission attempt. This represents the frequency domain index of the resource block used in the initial transmission of the k-th user. This represents the set of resource blocks corresponding to the initial transmission of user k in the previous time frame. According to and The first The user in the first The delay of one time frame, Indicates the maximum transmission delay. This indicates the total energy consumption limit.

[0328] In one embodiment, the second building unit is further configured to:

[0329] according to and Construct the first The user in the first Throughput per time frame When an ACK is received =1, when a NACK is received or the transmission fails. =0;

[0330] according to and Construct the first The user in the first buffer load per time frame ,in, For the first The user in the first -1 time frame buffer load;

[0331] according to Construct the first The user in the first Delay of one time frame ,in, Indicates the total number of time frames. This indicates the maximum number of time frames that can be considered in system modeling or simulation.

[0332] In one embodiment, the first conversion unit is specifically used for:

[0333] Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution is:

[0334]

[0335] in, This represents the set of optimal resource block allocations obtained by the lower-level argmax algorithm. This represents the optimal solution. Indicates the first The user in the first The throughput per time frame This represents the total number of slices in the system. This represents the set of free resources for the nth slice.

[0336] In one embodiment, the conversion module 3 is specifically used for:

[0337] Using the Dinkelbach algorithm The second optimal solution is rewritten as follows: , and The optimal solution is:

[0338]

[0339] in, This represents the objective function after the Dinkelbach transformation.

[0340] In one embodiment, the solution module 4 specifically includes:

[0341] Initial unit, used to obtain the current Buffer load In this case, each user accesses only one slice;

[0342] An iteration unit, connected to the initial unit, is used to perform the following iterations:

[0343] calculate As the dynamic threshold for this round, if If it exceeds the dynamic threshold of this round, then ,otherwise To achieve optimization ;

[0344] Based on the dynamic threshold of this round and , Determine the retransmission ratio ,like Then adjust Otherwise keep To achieve optimization ;

[0345] The cost of matching user k and resource block b Where b=(t,f) represents a resource block uniquely determined by subframe index t and subband index f. Indicates The indexed resource block is associated with the normalized channel fading coefficients related to the signal-to-noise ratio, which are derived from the relationship between each user k at the base station and each resource block b. The cost matrix is ​​constructed, and the Hungarian algorithm is used to solve for the cost matrix in order to obtain the optimal solution that minimizes the matching cost. ;

[0346] According to optimization ,optimization ,optimization renew And continue iterating until If the value falls below the preset value or the maximum number of iterations is reached.

[0347] Example 3:

[0348] like Figure 8 As shown, Embodiment 3 of this application provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it implements the resource orchestration method as described in Embodiment 1.

[0349] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program units, or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory or other memory technologies, CD-ROM (Compact Disc Read-Only Memory), DVD or other optical disc storage, cartridges, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer.

[0350] Embodiments 1-3 of this application provide a resource orchestration method, apparatus, and medium. Based on resource allocation variables, slice dormancy decision variables, and target rate variables, an optimal solution formula for energy efficiency performance indicators is constructed and transformed. By solving the resource allocation variables, slice dormancy decision variables, and target rate variables, the optimal energy efficiency performance indicator resource orchestration scheme is obtained, thereby achieving high-energy-efficiency resource orchestration and reducing algorithm complexity.

[0351] It is understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of this application, and this application is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of this application, and these modifications and improvements are also considered to be within the scope of protection of this application.

Claims

1. A resource orchestration method, characterized in that, The method includes: Defining base station resource allocation variables , slice dormancy decision variables and target rate variables ; According to , and constructing the energy efficiency performance index of base station optimal solution formula; Will The optimal solution is transformed into , and Optimal solution formula; according to , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration scheme; in, This represents the r-th time frame. This represents the k-th user. This represents the m-th transmission attempt, which includes the initial transmission and retransmissions. This represents the nth slice.

2. The method according to claim 1, characterized in that, Define base station resource allocation variables Slicing dormant decision variables and target rate variable Specifically, it includes: Define the set of resource blocks allocated to the base station: ,in, This represents the resource block allocated to the k-th user in the m-th transmission attempt within the r-th time frame. Let be the two-dimensional index of the resource block used by the k-th user in the m-th transmission attempt. This represents the subframe index of the resource block in the time dimension. This indicates the number of subframes contained in each time frame. This represents the sub-band index of the resource block in the frequency dimension. This indicates the number of subbands contained in each slice. This represents the total number of slices in the system; definition Let the nth slice be in the sleep mode of the rth time frame, where, This means that the nth slice will sleep during the rth time frame. This means that the nth slice is active in the rth time frame; definition Let be the target transmission rate for the k-th user in the r-th time frame.

3. The method according to claim 1 or 2, characterized in that, according to , and Energy efficiency performance indicators for building base stations The optimal solution formula includes: according to and Building base stations expression; according to , and as well as Expression construction under the traditional HARQ protocol First optimal solution; Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution.

4. The method according to claim 3, characterized in that, according to and Building base stations Expressions, specifically including: according to and Building base stations The expression is: ,in, Indicates the total number of time frames. Indicates the total number of users. According to The constructed set of equivalent channel capacities, This represents the distribution of service demand at the base station in the r-th time frame. According to The total energy consumption of the constructed base station in the r-th time frame.

5. The method according to claim 4, characterized in that, according to and Building base stations Expressions, specifically including: according to Constructing an equivalent channel capacity set ,in, Indicates the first Within the first time frame The first user accessed The signal-to-noise ratio corresponding to each slice. Indicates The normalized channel fading coefficients related to the signal-to-noise ratio for the indexed resource blocks; Based on the business model, it can be set to Poisson arrival, Bernoulli arrival, or given by the actual business trajectory. The user in the first Packet arrival distribution within a time frame , Representing the The user in the first Business requirements for each time frame; according to Construct the first Total energy consumption of base stations in each time frame ,in, Represents the static power consumption of a base station. Represents dynamic energy consumption. This represents the total number of slices in the system. This represents the total transmit power of the nth slice within the rth time frame. This represents the transmit power of the resource block indexed at (t,f). This represents the fixed RF link power consumption when the transmit link is enabled for the nth slice. This represents the set of resource blocks contained in the nth slice.

6. The method according to claim 4, characterized in that, according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution formula specifically includes: according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution is: ， Where k′ represents any other user besides the k-th user. For the first The user in the first The time frame in the ... Successful transfer indicator variable on each slice This represents the set of users accessing the nth slice. This represents the set of resource blocks allocated to the k-th user during the initial transmission of the r-th time frame. This represents the set of resource blocks contained in the nth slice. This represents the frequency domain index of the resource block used by the k-th user in the m-th transmission attempt. This represents the frequency domain index of the resource block used in the initial transmission of the k-th user. This represents the set of resource blocks corresponding to the initial transmission of user k in the previous time frame. According to and The first The user in the first The delay of one time frame, Indicates the maximum transmission delay. This indicates the total energy consumption limit.

7. The method according to claim 6, characterized in that, according to , and as well as Expression construction under the traditional HARQ protocol The first optimal solution also includes: according to and Construct the first The user in the first Throughput per time frame When an ACK is received =1, when a NACK is received or the transmission fails. =0; according to and Construct the first The user in the first buffer load per time frame ,in, For the first The user in the first -1 time frame buffer load; according to Construct the first The user in the first Delay of one time frame ,in, Indicates the total number of time frames. This indicates the maximum number of time frames that can be considered in system modeling or simulation.

8. The method according to claim 6, characterized in that, Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution formula specifically includes: Will The first optimal solution is transformed into the cross-slice HARQ protocol. The second optimal solution is: ， in, This represents the set of optimal resource block allocations obtained by the lower-level argmax algorithm. This represents the optimal solution. Indicates the first The user in the first The throughput per time frame This represents the total number of slices in the system. This represents the set of free resources for the nth slice.

9. The method according to claim 8, characterized in that, Will The optimal solution is transformed into , and The optimal solution formula includes: Using the Dinkelbach algorithm The second optimal solution is rewritten as follows: , and The optimal solution is: ， in, This represents the objective function after the Dinkelbach transformation.

10. The method according to claim 9, characterized in that, according to , and Optimal solution formula for optimal solution , and To obtain the optimal Resource orchestration schemes, specifically including: Get current Buffer load In this case, each user accesses only one slice; Perform the following iterations: calculate As the dynamic threshold for this round, if If it exceeds the dynamic threshold of this round, then ,otherwise To achieve optimization ; Based on the dynamic threshold of this round and Determine the retransmission ratio ,like Then adjust Otherwise keep To achieve optimization ; The cost of matching user k and resource block b Where b=(t,f) represents a resource block uniquely determined by subframe index t and subband index f. Indicates The indexed resource block is associated with the normalized channel fading coefficients related to the signal-to-noise ratio, which are derived from the relationship between each user k at the base station and each resource block b. The cost matrix is ​​constructed, and the Hungarian algorithm is used to solve for the cost matrix in order to obtain the optimal solution that minimizes the matching cost. ; According to optimization ,optimization ,optimization renew And continue iterating until If the value falls below the preset value or the maximum number of iterations is reached.

11. A computer device, characterized in that, The computer device includes a processor and a memory, the memory storing a computer program, and when the processor runs the computer program stored in the memory, the processor executes the resource orchestration method as described in any one of claims 1-10.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the resource orchestration method as described in any one of claims 1-10.

Images