A resource scheduling method for differentiated quality of service guarantee

Abstract

The application relates to a resource scheduling method for differentiated quality of service guarantee, and belongs to the wireless communication field.The method comprises the following steps: S1, constructing a multi-user differentiated quality of service optimization scene in an edge network; S2, constructing a time delay and energy consumption weighted optimization problem, that is, maximizing the system utility weighted by the time delay and the energy consumption under the condition of meeting the differentiated service demand of the edge network user; S3, splitting the time delay and energy consumption weighted optimization problem into an uplink optimization subproblem and a computing resource allocation optimization subproblem of an edge server for solving; wherein the uplink optimization comprises two aspects of device transmission power and user pairing; and S4, alternately iterating the uplink optimization and the edge server computing resource allocation optimization through a time delay constraint relation to update an optimization optimal solution.The application obtains higher system utility with lower complexity under the condition of meeting the differentiated quality of service.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of wireless communication, and relates to the problem of providing deterministic latency guarantee for Internet of Things devices, in particular to a resource scheduling method for differentiated quality of service guarantee. BACKGROUND

[0002] The fifth generation mobile communication technology not only improves the communication quality, but also expands the effective connection between people and things, and things and things, and provides reliable communication for vehicle networking, industrial intelligent sensor network, etc. The network function and core elements of 5G have been transformed to serve and support various business scenarios, such as massive machine type communication (Massive Machine Type Communication, mMTC), ultra reliable and low latency communication (Ultra Reliable and Low Latency Communication, uRLLC) and enhanced mobile broadband communication (Enhance Mobile Broadband, eMBB). The International Telecommunication Union (International Telecommunication Union, ITU) names the scenario where mMTC and uRLLC coexist as machine type communication (Machine Type Communication, MTC). Among them, the former has the characteristics of massive connection and low cost, and is usually low bit rate and time delay tolerant. The latter is a low-power embedded device, sensitive to time delay and has very limited processing capability. With the gradual commercialization of 5G and the in-depth study of future 6G, the increasing number of mobile terminals has brought great challenges to the limited frequency spectrum resources. At present, the number of Internet of Things nodes at the access end is increasing, and it is unrealistic to transmit all the information collected by the expanding industrial field to the cloud for processing. Future production requirements and various personalized production customization are constantly being put forward, and enterprises not only face resource allocation problems, but also face the problems of too large time delay in wide area network data transmission and difficult guarantee of reliable data interaction.

[0003] The emergence of numerous computationally intensive and time-sensitive services poses challenges to the future of wireless cellular networks in providing large-scale connectivity and ensuring Quality of Service (QoS). In industrial IoT systems, IoT devices often only provide simple data upload functionality. In this environment, there is an urgent need for solutions that can alleviate pressure on the core network while addressing computational bottlenecks. Mobile Edge Computing (MEC) deploys cloud computing systems at the network edge, effectively relieving the pressure on cloud centers for data processing and providing users with low-latency computing services. This benefits many applications with low-latency, high-bandwidth requirements (such as VR / AR and V2X). From a key technology perspective, multiple users can be served within the same radio resource through Non-Orthogonal Multiple Access (NOMA). Similar radio resource allocation includes time slots in TDMA, frequency bands in FDMA (or subcarriers in OFDMA), spreading codes in CDMA, and space in SDMA. NOMA employs Successive Interference Cancellation (SIC) to recover valid data at the receiver, significantly improving the system's anti-interference capability. Therefore, integrating NOMA technology into edge networks helps improve system resource utilization. However, how to rationally and effectively integrate the communication and computing resources of NOMA-MEC systems is a challenging topic with significant research value.

[0004] Therefore, there is an urgent need for an optimized scheduling method to provide differentiated service quality assurance for equipment. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a resource scheduling method for differentiated service quality assurance, which comprehensively considers various business needs and system available resources to design an efficient resource scheduling strategy, thereby optimizing the latency and energy consumption of device terminals.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A resource scheduling method for differentiated quality of service assurance is proposed, which is a joint offloading scheduling and resource allocation optimization method. It optimizes the transmit power of devices, computing resources, and user pairing latency and energy consumption. First, due to the complexity of solving the optimization problem, a closed-form expression for the uplink power allocation problem is derived through formula derivation, proving that computing resource allocation is a convex programming problem. Second, to reduce the computational complexity of the matching problem, a stable matching method based on externalities is designed. Finally, through alternating optimization, the utility of the MEC system is maximized to balance latency and energy consumption, while simultaneously meeting the differentiated needs of devices.

[0008] The method specifically includes the following steps:

[0009] S1: Building a multi-user differentiated service quality optimization scenario in edge networks;

[0010] S2: Construct a latency and energy consumption weighted optimization problem through channel modeling, resource allocation and task scheduling, that is, to maximize the system utility measured by latency and energy consumption weighting while meeting the differentiated service needs of edge network users;

[0011] S3: The latency and energy consumption weighted optimization problem is broken down into an uplink optimization subproblem and an edge server computing resource allocation optimization subproblem for solution; among them, the uplink optimization includes two aspects: device transmit power and user pairing.

[0012] S4: Uplink Joint Optimization Resource Scheduling: Using an alternating iterative method, uplink optimization and edge server computing resource allocation optimization are performed alternately and iteratively through latency constraints to update the optimal solution.

[0013] Furthermore, step S1 specifically includes the following steps:

[0014] S11: In the multi-user edge network covered by the base station cell, there are terminal device sets (such as a large number of sensors, smartphones, etc.), mainly including two types: uRLLC (ultra-reliable low-latency communication) devices and mMTC (massive machine communication) devices. Considering the indivisible computing tasks and the limited computing and storage capabilities of terminal devices, a complete offloading method is adopted to upload multiple computing tasks to the edge server for processing.

[0015] S12: Deploy MEC servers in the network that can provide limited computing and storage services to multiple terminal devices simultaneously. The base station adopts hybrid NOMA technology as the access scheme. uRLLC devices and mMTC devices interact with the base station through sub-channels in a dedicated or shared manner.

[0016] S13: The device generates a task request, calculates the local offloading latency of the task and the computing latency of the edge server, and defines a computing resource allocation strategy.

[0017] S14: Calculate total latency and total energy consumption: Since the calculated result data is much smaller than the offloaded data, the uplink data transmission rate is much lower than the downlink, and the MEC (Mobile Edge Computing) server device is directly connected to the power grid, for simplicity, the return latency and energy consumption of the result are ignored. Due to differentiated requirements, the latency threshold of uRLLC devices is more stringent (smaller) than that of mMTC devices, and the energy consumption threshold of mMTC devices is more stringent (smaller) than that of uRLLC devices.

[0018] Furthermore, step S13 specifically includes: device i generating a task request. Among them, l i This refers to the size of the data that needs to be unloaded to the MEC server for processing. i It refers to the workload, which is the number of CPU cycles required to complete a task; This indicates that there is a set of terminal devices in the multi-user edge network covered by the base station cell, including two types: U uRLLC devices and M mMTC devices, denoted as follows: and

[0019] Local uninstallation latency is: in Let i be the task offloading rate of user i on subchannel k;

[0020] The computation latency of the edge server is: Where c i It is workload, f i The computing frequency assigned to device i by the server;

[0021] The computing resource allocation strategy is defined as follows:

[0022] Furthermore, in step S14, the total delay t i The calculation formula is: in, For the time delay threshold, The latency threshold for uRLLC devices. The latency threshold for mMTC devices;

[0023] Total energy consumption The calculation formula is: Where, p i Indicates the transmission power, l i This refers to the size of the data that needs to be unloaded to the MEC server for processing. Energy consumption threshold Energy consumption threshold of uRLLC devices This represents the energy consumption threshold of mMTC devices.

[0024] Furthermore, in step S2, a latency and energy consumption weighted optimization problem is constructed, which specifically includes the following steps:

[0025] S21: Define the task scheduling strategy. The scheduling strategy is related to the access method of the device. Use binary variables to represent the matching relationship between the device and the sub-channel, and analyze the conditions that user pairing needs to meet.

[0026] S22: Define the channel gain and transmit power allocation strategy, and calculate the task offload rate;

[0027] S23: Considering the order-of-magnitude differences in latency and energy consumption thresholds for different types of devices, normalize and weight the latency and energy consumption.

[0028] S24: The interaction between device-side wireless access requirements and resource allocation is mainly reflected in task offloading scheduling and resource allocation. By jointly offloading scheduling and resource allocation, the optimization objective function is described as an optimization problem that maximizes system utility. The constraints include: offloading power budget constraints, otherwise the task cannot be completely offloaded to the edge server and will be discarded; latency and energy consumption limits, the device cannot exceed the threshold, otherwise it is considered a transmission failure; the computing resources allocated by the server to the device cannot exceed its computing capacity; scheduling policy constraints. The optimization problem is expressed as an integer nonlinear programming problem.

[0029] Furthermore, in step S24, the optimization problem for maximizing system utility is:

[0030]

[0031]

[0032]

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039] Where C1 is the device's offload power budget constraint, otherwise the task cannot be completely offloaded to the edge server and will be discarded; C2 and C3 are latency and energy consumption limits, the device cannot exceed the threshold, otherwise it is considered a transmission failure; C4 and C5 indicate that the computing resources allocated by the server to the device cannot exceed its computing capacity; C6-C8 represent scheduling policy constraints; it is worth noting that this problem is an integer nonlinear programming problem. Represents system utility, where J i Represents the offloading utility function for device i; The task scheduling strategy, where It is a binary variable. This indicates that uRLLC device u and device u′ share subchannel k, and when u=u′, it means that uRLLC device u occupies subchannel k alone; It is the same definition; This indicates that uRLLC device u and mMTC device m share subchannel k; the system bandwidth resources are evenly divided into K subchannels, and the channel set is... It can accommodate a maximum of n at the same time max One device; For transmit power allocation strategy, For computing resource allocation strategies; F represents total energy consumption. s This indicates the threshold size for calculating resources.

[0040] Furthermore, in step S3, the computational resource sub-problem of the edge server is decoupled from the optimization problem, and is expressed as:

[0041]

[0042]

[0043] C4,C5

[0044] Where C3-a is the computational delay constraint, and the parameters are... This indicates the stringency of device i's time delay constraints.

[0045] Furthermore, in step S3, the uplink transmit power sub-problem is decoupled from the optimization problem, and is expressed as:

[0046]

[0047] stC1,C2

[0048]

[0049] Among them, parameters B is the sub-channel bandwidth. This indicates the stringency of energy consumption constraints imposed on device i.

[0050] Solving this subproblem requires calculating the minimum point and discussing the positional relationship between the minimum point and the domain.

[0051] Furthermore, in step S3, the uplink user pairing sub-problem is decoupled from the optimization problem, and is expressed as:

[0052]

[0053] stC 6-C8

[0054] in, Let N be the channel gain, where N0 is the power spectral density of additive white Gaussian noise. It is the exponential channel gain of device i on subchannel k, where a represents the path loss exponent. This represents the Euclidean distance from device i to the base station;

[0055] If a swap pair is allowed to be swapped, it is defined as a swap-blocking pair of the match. When there is no swap-blocking pair, it has bilateral stability. By iteratively swapping the match, a bilaterally stable match is eventually obtained.

[0056] The beneficial effects of this invention are as follows: Addressing the issue of differentiated quality of service (QoS) assurance for different types of smart terminals, this invention considers using non-orthogonal multiple access (NOMA) technology as the access technology for edge networks. To ensure the differentiated service requirements of terminal devices in terms of latency and energy consumption, a system utility maximization model weighted by latency and energy consumption is constructed. The optimization objective is expressed as a mixed-integer nonlinear programming problem. Due to the complexity of the joint optimization problem, it is difficult to solve directly. Therefore, this invention proposes an optimization method for joint offloading scheduling and resource allocation, decoupling the optimization objective into an uplink offloading scheduling problem, a power allocation problem, and a computational resource allocation problem. This invention designs a stable matching method based on externalities to solve the offloading scheduling problem, and obtains a closed-form expression for the uplink power allocation problem through formula derivation. Then, through alternating iterations, joint optimization is performed, effectively reducing computational complexity while significantly improving system utility, demonstrating broad application prospects.

[0057] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0058] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0059] Fig. 1 This invention provides a communication system architecture diagram for a differentiated service quality optimization scenario.

[0060] Fig. 2 This is a network block diagram related to the present invention;

[0061] Fig. 3 This invention relates to the user pairing process;

[0062] Fig. 4 This invention relates to the SIC decoding process. Detailed Implementation

[0063] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0064] Please see Figs. 1-4 This invention provides a resource scheduling method for ensuring differentiated service quality. First, it proposes a differentiated service quality optimization scenario, employing non-orthogonal multiple access (NOMA) technology as the access technology for the edge network. Second, it studies the problem of ensuring differentiated service quality for different types of smart terminals. Finally, it constructs a system utility maximization model weighted by latency and energy consumption. By considering the impact of power, offloading scheduling, and computing resources, an optimization scheme for joint offloading scheduling and resource allocation is proposed. Computational resource allocation is a convex programming problem, solvable using convex optimization methods. Power allocation is a single-peak function, yielding a closed-form solution. To reduce the complexity of the matching problem, a stable matching algorithm based on externalities is used to determine the bidirectional selection of users and sub-channels. Then, through alternating optimization, the utility of the MEC system is maximized to balance latency and energy consumption, while simultaneously meeting the differentiated needs of the two types of devices. The method specifically includes the following steps:

[0065] Step 1: Differentiated Service Quality Optimization Scenario Design: For uRLLC devices with low latency requirements and mMTC devices with low power consumption requirements, this study investigates the differentiated service problem in edge networks based on non-orthogonal multiple access technology. In multi-task scenarios, communication is mainly uplink-based, with terminal devices frequently sending short data packets on the uplink. Specifically, the following steps are included:

[0066] Step 1.1: A set of terminal devices exists in the multi-user edge network covered by the base station cell. (For example, a large number of sensors, smartphones, etc.) mainly include two types: U uRLLC devices and M mMTC devices, respectively represented as and Considering the indivisible computing tasks and the limited computing and storage capabilities of terminal devices, a complete offloading approach is adopted to upload multiple computing tasks to edge servers for processing.

[0067] Step 1.2: An MEC server capable of providing limited computing and storage services to multiple terminal devices simultaneously was deployed in the network, with a bandwidth and computing resource threshold size of B. s and F s The base station uses hybrid NOMA technology as the access scheme. uRLLC and mMTC devices interact with the base station via sub-channels in a dedicated or shared manner. System bandwidth resources are evenly allocated to K sub-channels, and the channel set is as follows: The sub-channel bandwidth is B = B s / K, can hold a maximum of n at a time. max One device.

[0068] Step 1.3: Device i generates a task request. l i This refers to the size of the data that needs to be unloaded to the MEC server for processing. i This refers to the workload, specifically the number of CPU cycles required to complete a task. The local offload latency is... Let be the offloading rate of user i on subchannel k, and the computation latency of the edge server be... f i The computing resource allocation strategy is defined as the computing frequency allocated to device i by the server.

[0069] Step 1.4: Since the calculated result data is much smaller than the unloaded data, the uplink data transmission rate is much lower than the downlink rate, and the MEC server device is directly connected to the power grid. For simplicity, the return latency and energy consumption are ignored. Therefore, the total latency is... Total energy consumption is Among them, parameters and These are the latency thresholds, and the latency thresholds and energy consumption thresholds have implicit conditions, respectively.

[0070] Step 2: Constructing a Latency and Energy Weighted Optimization Problem: In edge networks, the QoS requirements of devices are mainly reflected in the latency and energy consumption of task completion. To meet the differentiated service needs of two types of devices and maximize the system utility measured by latency and energy consumption, a latency and energy weighted model is designed. This includes the following steps:

[0071] Step 2.1: Define the task scheduling strategy as follows in It is a binary variable. This indicates that uRLLC device u and device u′ share subchannel k, and when u=u′, it means that uRLLC device u occupies subchannel k alone; It is the same definition; This indicates that uRLLC device u and mMTC device m share subchannel k. The offloading scheduling policy χ is related to the device's access method and must meet the following conditions:

[0072]

[0073]

[0074] Step 2.2: Channel gain mainly considers the effects of path loss and antenna gain; therefore, channel gain is defined as... in It is the exponential channel gain of device i on subchannel k, where a represents the path loss exponent, and the transmit power allocation strategy is defined as The random normalized gain of device i on subchannel k is N0 is the power spectral density of additive white Gaussian noise, and the task unloading rate can reach:

[0075]

[0076] If the device occupies the uplink subchannel alone to transmit data, the interference comes only from noise. If the device shares the uplink subchannel with other devices, and if the devices in the same channel are of the same type, the stronger user will be subject to interference from the weaker user. Therefore, the signal-to-noise ratio can be expressed as:

[0077]

[0078] Step 2.3: Considering the order-of-magnitude differences in latency and energy consumption thresholds among different types of devices, and that the relative improvement in task completion latency and energy consumption is expressed as (T max -t) / Tmax (E) max -e) / E max Therefore, the offloading utility function for device i is defined as:

[0079]

[0080] Among them, parameters and This indicates the stringency of device i's requirements regarding latency and energy consumption. A higher value indicates stricter requirements for latency and energy consumption. The parameters must be satisfied... For uRLLC, it can be achieved by adding This involves reducing overall task execution latency at the expense of increased energy consumption. Similarly, mMTC saves energy by sacrificing latency, i.e., increasing...

[0081] Step 2.4: The interaction between device-side wireless access requirements and resource allocation is mainly reflected in task offloading scheduling and resource allocation. The system utility is defined as follows: By jointly unloading scheduling and resource allocation, the optimization objective function is described as an optimization problem that maximizes system utility, as shown in the following structure:

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] Where C1 is the device's offload power budget constraint, otherwise the task cannot be completely offloaded to the edge server and will be discarded; C2 and C3 are latency and energy consumption limits, the device cannot exceed the threshold, otherwise it is considered a transmission failure; C4 and C5 indicate that the computing resources allocated by the server to the device cannot exceed its computing capacity; C6-C8 represent scheduling policy constraints; it is worth noting that this problem is an integer nonlinear programming problem.

[0092] Step 3: Computational Resource Allocation: Directly solving the optimization problem is very complex. It can be broken down into an uplink optimization sub-problem and an edge server computational resource allocation sub-problem. Under the condition of meeting the requirements of the terminal devices, it is derived that computational resource allocation is a convex programming problem. Specifically, it includes the following steps:

[0093] Step 3.1: The computational resource subproblem can be decoupled from the optimization problem, represented as:

[0094]

[0095]

[0096] C4,C5

[0097] Where C3-a is the computational delay constraint, and the parameters are...

[0098] Step 3.2: The objective function and constraint C3-a in Step 3.1 are both f i The nonlinear convex function, with constraints C4 and C5 having convex properties, means that the computational resource subproblem is formulated as a convex programming problem for solution.

[0099] Step 4: Transmit Power Allocation: Uplink optimization includes both device transmit power and user pairing, while the signal-to-noise ratio of device i's uplink is... Not only with p i This also depends on the transmit power of other users on the same sub-channel. An optimal transmit power solution was found through formula derivation. Specifically, the following steps are included:

[0100] Step 4.1: The uplink transmit power subproblem can be decoupled from the optimization problem, expressed as:

[0101]

[0102] stC1,C2

[0103]

[0104] Where parameters

[0105] Step 4.2: The second derivative of the objective function in Step 4.1 is not always greater than zero, making it a strictly quasi-convex function. A local optimum exists at the zero of the first derivative. For a strictly quasi-convex function, the local optimum is equivalent to the global optimum, and the minimum point... The conditions are met. Where parameters From step 2.2, we know that parameter v3 satisfies the following relationship:

[0106]

[0107] Step 4.3: Step 4.2 analyzed the impact of the optimization variables on the objective function and solved P. * Further discussion is needed regarding the positional relationship between the extreme points and the domain. The derivation shows that the domain has the following expression:

[0108]

[0109]

[0110] Among them, parameters It is the predetermined maximum transmission power, parameters satisfy:

[0111]

[0112] Therefore, the optimal power solution for device i satisfies the following condition:

[0113]

[0114] Step 5: User Pairing: Device Collection and sub-channel set These are two completely non-overlapping sets, with devices and sub-channels influencing each other's scheduling. Therefore, this scheduling problem can be described as a many-to-one matching problem with externalities. To reduce computational complexity, a low-complexity, externality-based stable matching method is designed using matching theory. Specifically, it includes the following steps:

[0115] Step 5.1: The uplink user pairing sub-problem can be decoupled from the optimization problem, represented as:

[0116]

[0117] stC 6-C8

[0118] The above problem is an integer programming problem, which can be solved by the branch and bound method and the cutting plane method, but it is very cumbersome, especially when the dimension of the optimization variables increases, resulting in huge computational complexity.

[0119] Step 5.2: User pairing externalities are based on system utility and are reflected in the device's pairing decisions. They depend not only on the benefits of pairing with a sub-channel but also on the benefits gained by other devices in the same channel. Therefore, a parameter ξ is defined. i (k) represents the preference value of device i in subchannel k, and its magnitude is expressed as... The preference value of subchannel k for device i is defined as ξ. k (Π(k))=∑ i∈Π(k) ξi (k).

[0120] Step 5.3: >> is defined as a preference relation. For possible matches Π(i) = k1 and Π(i) = k2 of the devices, if ξ i (k1)>ξ i If (k2), then (i,k1) >> (i,k2), meaning device i can gain greater benefits by matching to channel k1; similarly, for possible matching of sub-channels... and if but This means that the sub-channel matching device set The benefits obtained are better.

[0121] Step 5.4: Since externalities also affect the preference relationship between device and sub-channel matching, thus affecting the stability of this problem, the stability of bilateral switching is analyzed to solve the externality problem. To achieve switching stability, a switching matching pair is defined:

[0122]

[0123] At this point, Π is a matching pair vector, where the matches before the swap are Π(m) = k1 and Π(m) = k2. This means that while keeping the channel matching of other devices unchanged, devices m and n exchange their sub-channels.

[0124] Step 5.5: If If swapping is allowed, then define For a given pair of blocking devices (pairs Π), if the preference values ​​of all participating devices and sub-channels do not decrease after the swap, and at least one of them increases, then the blocking pair is allowed to be swapped. Assuming that a blocking pair consisting of devices m and n is allowed to be swapped, the following condition must be met:

[0125]

[0126] It exhibits bilateral stability when there are no swap blocking pairs, and eventually a bilaterally stable match is obtained through iterative swap matching.

[0127] Step 6: Uplink Joint Optimization Resource Scheduling Method: An alternating iterative method is used to alternately execute the uplink optimization in Steps 4 and 5 and the edge server computing resource optimization in Step 3 through latency constraints, so as to update and optimize the optimal solution.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A resource scheduling method for differentiated quality of service assurance, characterized in that, The method specifically includes the following steps: S1: Building a multi-user differentiated service quality optimization scenario in edge networks; S2: Construct a latency and energy consumption weighted optimization problem through channel modeling, resource allocation and task scheduling, that is, to maximize the system utility measured by latency and energy consumption weighting while meeting the differentiated service needs of edge network users; In step S2, a latency and energy consumption weighted optimization problem is constructed, which specifically includes the following steps: S21: Define the task scheduling strategy. The scheduling strategy is related to the access method of the device. Use binary variables to represent the matching relationship between the device and the sub-channel, and analyze the conditions that user pairing needs to meet. S22: Define the channel gain and transmit power allocation strategy, and calculate the task offload rate; S23: Normalize and then weight the latency and energy consumption; S24: By jointly offloading scheduling and resource allocation, the objective function is described as an optimization problem maximizing system utility. Constraints include: offloading power budget constraints (otherwise, tasks cannot be fully offloaded to edge servers and will be discarded); latency and energy consumption limits (devices cannot exceed thresholds, otherwise transmission is considered a failure); the computing resources allocated to devices by servers cannot exceed their computing capabilities; and scheduling policy constraints. The optimization problem is formulated as an integer nonlinear programming problem. The optimization problem maximizing system utility is as follows: C1 represents the device's offload power budget constraint; otherwise, the task cannot be completely offloaded to the edge server and will be discarded. C2 and C3 are latency and energy consumption limits; the device cannot exceed the threshold, otherwise it is considered a transmission failure. C4 and C5 indicate that the computing resources allocated by the server to the device cannot exceed its computing capacity. C6-C8 represent scheduling policy constraints. Represents system utility, where Indicates equipment The unloading utility function is defined as follows: ,in and Indicates equipment The stringency of the constraints on latency and energy consumption, satisfying ; This indicates that there is a set of terminal devices in the multi-user edge network covered by the base station cell, which includes two types: one uRLLC device and Each mMTC device is represented as follows: and ; The task scheduling strategy, where , , It is a binary variable. Indicates uRLLC class device With equipment Shared subchannel ,when Time indicates uRLLC device Occupying a sub-channel alone ; It is the same definition; Indicates uRLLC device With mMTC devices Shared subchannel System bandwidth resources are evenly distributed to There are 1 sub-channels, and the channel set is 1 It can accommodate at most [number] people at the same time. One device; For transmit power allocation strategy, For computing resource allocation strategies, , Assigning devices to the server The calculation frequency; Indicates the transmission power; Indicates total energy consumption. Energy consumption threshold; For the total delay, This is the time delay threshold; This indicates the threshold size for calculating resources; S3: The latency and energy consumption weighted optimization problem is broken down into an uplink optimization subproblem and an edge server computing resource allocation optimization subproblem for solution; among them, the uplink optimization includes two aspects: device transmit power and user pairing. S4: Uplink Joint Optimization Resource Scheduling: Using an alternating iterative method, uplink optimization and edge server computing resource allocation optimization are performed alternately and iteratively through latency constraints to update the optimal solution.

2. The resource scheduling method according to claim 1, characterized in that, Step S1 specifically includes the following steps: S11: In the multi-user edge network covered by the base station cell, there are two types of terminal devices: uRLLC devices and mMTC devices. Multiple computing tasks are uploaded to the edge server for processing using a complete offload method. S12: Deploy MEC servers in the network that can simultaneously provide limited computing and storage services to multiple terminal devices. The base station adopts hybrid NOMA technology as the access scheme. uRLLC devices and mMTC devices interact with the base station through sub-channels in a dedicated or shared manner. S13: The device generates a task request, calculates the local offloading latency of the task and the computing latency of the edge server, and defines a computing resource allocation strategy. S14: Calculate total delay and total energy consumption.

3. The resource scheduling method according to claim 2, characterized in that, Step S13 specifically includes: equipment Generate a task request ,in, This refers to the size of the data that needs to be unloaded to the MEC server for processing. It refers to the workload, which is the number of CPU cycles required to complete a task. This indicates that there is a set of terminal devices in the multi-user edge network covered by the base station cell, which includes two types: one uRLLC device and Each mMTC device is represented as follows: and ; Local uninstallation latency is: ,in For users In sub-channel Task unloading rate; The computation latency of the edge server is: ,in It is workload. Assigning devices to the server The calculation frequency; The computing resource allocation strategy is defined as follows: .

4. The resource scheduling method according to claim 3, characterized in that, In step S14, the total delay The calculation formula is: ,in, For the time delay threshold, , The latency threshold for uRLLC devices. The latency threshold for mMTC devices; Total energy consumption The calculation formula is: ,in, Indicates the transmission power. This refers to the size of the data that needs to be unloaded to the MEC server for processing. Energy consumption threshold , Energy consumption threshold of uRLLC devices This represents the energy consumption threshold of mMTC devices.

5. The resource scheduling method according to claim 1, characterized in that, In step S3, the computational resource sub-problem of the edge server is decoupled from the optimization problem, and is expressed as: Where C3-a is the computational delay constraint, and the parameters are... , Indicates equipment The stringency of the time delay constraint.

6. The resource scheduling method according to claim 5, characterized in that, In step S3, the uplink transmit power sub-problem is decoupled from the optimization problem, and is expressed as: Among them, parameters , , B Sub-channel bandwidth, Indicates equipment The stringency of energy consumption constraints; Solving this subproblem requires calculating the minimum point and discussing the positional relationship between the minimum point and the domain.

7. The resource scheduling method according to claim 6, characterized in that, In step S3, the uplink user pairing sub-problem is decoupled from the optimization problem, and is represented as: in, , , For channel gain, where It is the power spectral density of additive white Gaussian noise. It is a sub-channel Up equipment The exponential channel gain, This represents the path loss index. Indicates equipment Euclidean distance to the base station Indicates equipment The optimal power solution; Define swapped matching pairs: at this time It is a matching pair vector, where the matching before the swap is... and , This means that while keeping the channel matching of other devices unchanged, the device and equipment Exchange their sub-channels; and Sub-channel; if If swapping is allowed, then define For matching A swap blocking pair; for a given match If the preference values ​​of all participating devices and sub-channels do not decrease after the exchange, and at least one increases, then the exchange blocking pair is allowed to exchange; assuming the devices and For a blocking pair to be allowed to swap, the following conditions must be met: in, The preference value of the sub-channel to the device; it has bilateral stability when there is no exchange blocking pair, and finally obtains a bilaterally stable match through iterative exchange matching.

Images