An Associated Slice Resource Allocation Method Based on Task Timeliness

Abstract

The present invention discloses a method for allocating associated slice resources based on task timeliness, which relates to the technical field of network resource scheduling. The present invention provides a network slice framework considering slice correlation and a method for allocating inter-slice network resources (transmission, computing) based on task timeliness. The slice correlation is described by fusing the mean square error, and the analytical relationship between the inter-slice resource allocation strategy and the slice correlation is comprehensively characterized by task timeliness. By scaling and decoupling the original constrained optimization problem, an iterative particle swarm algorithm is proposed to minimize the estimation error of the entire system while ensuring the convergence of slice estimation, balance the resource occupancy between slices, and improve the overall performance.

Description

Technical Field

[0001] The present invention relates to the technical field of network resource scheduling, and particularly relates to a method for allocating associated slice resources based on task timeliness. Background Art

[0002] The technological development of industrial Internet of Things (IIoT) has laid a solid foundation for diversified industrial services such as intelligent manufacturing. A large number of sensing, transmission, and computing terminals are integrated in industrial sites, and a lot of computing-intensive tasks of IIoT services need to be processed. Processing such computing-intensive tasks by on-site devices requires adding expensive computing units; while using cloud computing for centralized processing will bring relatively large transmission delays. As a remedial measure for these limitations, a potential solution is to explore the edge computing (EC) paradigm, in which on-site devices can access computing resources and offload their computing tasks to edge devices for immediate processing by edge devices with stronger computing capabilities.

[0003] IIoT services have diverse quality of service (QoS) requirements. Taking the aircraft final assembly process as an example, the time-delay sensitive collaborative sensing service for measuring the angle of movable surfaces has strict requirements for time delay, such as 100 ms; while the resource-consuming wearable devices in human-machine interaction wiring harness installation require high throughput. To support these diverse IIoT services with different QoS requirements, a promising 5G network slicing method has emerged, aiming to build multiple logically isolated slices on the existing factory network infrastructure. The existing slice research in the context of IIoT networks usually designs slice request acceptance and transmission resource allocation strategies to maximize system benefits under limited physical resource constraints. By proposing an end-to-end solution to support the dynamic adaptation of slices to the environment and ensure the reliable operation of IIoT. Since the computing resources required for on-site tasks are different, it is necessary to appropriately allocate computing resources to achieve better service performance. Therefore, a comprehensive network slicing strategy should consider both transmission and computing resource allocation.

[0004] In addition, existing work focuses on network slicing under resource constraints and lacks consideration of the correlation between slices. For example, when monitoring the angle of the moving surface during aircraft final assembly, a combination of contact angle sensors and non-contact cameras is usually used to ensure the estimation performance. Contact measurement can ensure the timeliness of dense sampling edge estimation, but the accuracy of this observation method is low. Non-contact measurement requires a large amount of spectrum and computing resources, has high estimation accuracy, long processing time, and poor real-time performance. Network slicing can meet the heterogeneous requirements of delay-sensitive sensors and bandwidth-consuming cameras. The resource allocation decision between slices will affect the edge estimation performance of dozens of moving surface angles. In this case, formulating an optimal network slicing strategy faces many challenges. First, the impact of transmission and computing allocation decisions on estimation performance is coupled. Second, there is a high degree of heterogeneity in the QoS requirements of the two monitoring methods, and the slices need to meet multiple constraints such as delay, bandwidth, and estimation convergence.

[0005] To address these challenges, the concept of Age of Task (AoT) is introduced to capture the task freshness of edge estimation, which can simultaneously characterize resource consumption and delay metrics. AoT can quantify the information lag of the measurements received by edge devices, thus better representing the real-time estimation process. This paper proposes a new framework that considers the slice correlation model and network resource allocation between slices. The slice correlation model is characterized as the estimation error variance based on AoT and formulated as a constrained stochastic optimization problem to minimize the overall system estimation error while satisfying resource constraints.

[0006] The closest implementation solution is the Chinese patent application number: 202010348612.5, titled: Joint Allocation Method of Wireless and Computing Resources for Network Slicing. The specific approach is as follows: Considering network fluctuations, slice work tasks, and resource requirements, wireless and computing resources are allocated at the slice level to improve the overall efficiency of the system. Patent application number: 202110886452.4, titled: An Industrial Software-Defined Network Slicing Method Based on Edge Collaboration. The specific approach is as follows: According to heterogeneous control performance requirements, slice creation and network planning are carried out, and a problem of minimizing system power consumption under the constraint of information timeliness overrun is constructed within the slice, and joint allocation of communication, computing, and caching resources is performed. Although both patents consider the current situation of multi-dimensional resource coupling, they ignore the impact of the correlation between slices themselves on system performance, that is, when multiple slices serve a task simultaneously, the relationship between the inter-slice resource allocation decision and the overall performance.

[0007] Current research on network slicing only focuses on the impact of physical resource constraints on slices, ignoring the correlation between slices themselves. Under this mode, resource allocation decisions may not meet the overall performance requirements of the system. There is a high degree of heterogeneity in slice requirements, and it is difficult to explicitly express the impact of multi-dimensional network resources on slice requirements and performance. Existing research mostly considers slice design under single-constraint influence, making it difficult to ensure the effective operation of the system. Existing network slicing research mostly focuses on transmission resource allocation, ignoring the impact of computing resources on network slice performance. Moreover, the impact of transmission and computing resource decisions on service performance is coupled, and separate design will cause resource waste and reduce resource utilization.

[0008] Therefore, those skilled in the art are committed to developing a correlated slice resource allocation method based on task timeliness to meet the overall performance requirements of the system, ensure the effective operation of the system, avoid resource waste, and improve resource utilization. Summary of the Invention

[0009] In view of the above-mentioned defects of the prior art, the technical problems to be solved by the present invention are how to comprehensively consider slice correlation and resource constraints, design a suitable network slice framework to meet heterogeneous slice requirements and coupled performance indicators, and at the same time have the ability to adapt to network dynamic changes; how to reveal the coupled impact of inter-slice transmission and computing resource decisions on system performance, establish a slice correlation model, and analyze the mathematical expressions of network slices and comprehensive performance evaluation indicators; how to reasonably allocate transmission and computing resources in the case of mutual coupling of multiple slices and multiple resources, and improve system performance on the premise of meeting slice heterogeneous measurement requirements and physical resource limitations.

[0010] To achieve the above object, the present invention provides a correlated slice resource allocation method based on task timeliness, including a network slice framework considering slice correlation and an inter-slice network resource allocation method based on task timeliness; the network slice framework includes network resource allocation at a large time scale and scheduling and estimation at a small time scale.

[0011] Further, the network slice framework considers scenarios of fusion estimation using multiple measurement methods, and uses network slices to achieve logical isolation to meet heterogeneous measurement requirements.

[0012] Further, for the network resource allocation at a large time scale, taking the fusion estimation error as the slice correlation model, by allocating transmission and computing resources between slices, the timeliness and accuracy of edge estimation are ensured.

[0013] Further, for the scheduling and estimation at a small time scale, each slice uploads and processes measurement values based on the allocated transmission and computing resources, and performs fusion estimation on edge devices to provide a basis for subsequent control.

[0014] Furthermore, the network resource allocation at the large time scale remains unchanged before the slice request is updated, and the scheduling and estimation at the small time scale cyclically repeat within the slice window until the window ends.

[0015] Furthermore, the inter-slice network resource allocation method includes the following steps:

[0016] Step 1, Initialize different measurement devices, as well as the state change models of each slice and configure relevant parameters;

[0017] Step 2, After the SDN module in the edge device receives a slice creation request, collect network information, perform pooling virtualization of physical resources, and establish a slice association model and inter-slice resource constraint conditions;

[0018] Step 3, Construct a performance evaluation index based on slice correlation, and establish a resource optimization scheduling problem under the constraints of resource finiteness and system convergence;

[0019] Step 4, Use scaling and decoupling methods to transform the optimization problem into a single-resource allocation sub-problem for a single time slot, and use an iterative solution method to obtain transmission and computing resource allocation decisions to complete inter-slice resource allocation;

[0020] Step 5, The SDN module executes the slice decision, maps the virtualized slice resources to the corresponding physical network node set, and completes the construction of the network slice;

[0021] Step 6, When the slice request is updated, the SDN module re-performs pooling virtualization of physical resources and constructs the association model, returns to execute Step 1, and re-performs network slicing and resource allocation.

[0022] Furthermore, in Step 1, initializing the measurement device includes the sampling period, data packet size, computing intensity, available frequency band resources and computing resources, inter-slice transmission resource allocation rate, and computing resource allocation rate.

[0023] Furthermore, Step 2 includes the following steps:

[0024] Step 2.1, Define the sampling period, transmission delay, and computing delay of different slices respectively, and calculate the task timeliness of different slices To measure the information lag of the measured values received by different slices, where Δ i (t) is the task timeliness of slice i at time slot t, is the generation time of the latest data packet received by slice i;

[0025] Step 2.2, Construct an estimation error model for different slices based on task timeliness

[0026]

[0027] Among them, K i (t) is a function of the estimated error variance and the measurement error variance, A is the system state matrix, ω i and v i are the system noise and the observation noise of slice i, respectively;

[0028] Step 2.3: Calculate the fusion estimation errors of all slices, and use the mean square error to evaluate the estimation performance, which is used as the mathematical expression of the slice association model to reveal the coupling effect of the inter-slice resource allocation decision on the system performance;

[0029] Step 2.4: Considering the estimation convergence of different slices, give the delay constraint conditions for the delay-sensitive slices and the bandwidth constraint conditions for the bandwidth-consuming slices respectively.

[0030] Furthermore, the said step 4 includes the following steps:

[0031] Step 4.1: Transform the original time-accumulative objective function into a single-slot objective function through term cancellation and logarithmic operations;

[0032] Step 4.2: Considering the zero-waiting strategy, the average AoT of a slice can be expressed as a linear function of the transmission delay and the computing delay, that is, the single-slot objective function increases as the transmission and computing delays increase;

[0033] Step 4.3: Decouple the optimization problem into a transmission resource allocation sub-problem P1 and a computing resource allocation sub-problem P2, and use the particle swarm optimization algorithm to solve the objective variables respectively;

[0034] Step 4.4: Take the objective functions of the two sub-problems as the fitness values of the particle swarm optimization algorithm, and use the iterative particle swarm optimization algorithm to continuously iterate and optimize to seek the optimal solution, and obtain the optimal inter-slice transmission and computing resource allocation decision of the optimization objective.

[0035] Furthermore, the said step 4.4 includes the following steps:

[0036] Step 4.4.1: Given the initial value of the inter-slice computing resource allocation rate and the loop termination condition;

[0037] Step 4.4.2: For the transmission resource allocation sub-problem P1, based on the given inter-slice computing resource allocation rate, use the particle swarm optimization algorithm to optimize and obtain the current optimal transmission resource allocation rate;

[0038] Step 4.4.3: For the computing resource allocation sub-problem P2, based on the current inter-slice transmission resource allocation rate, use the particle swarm optimization algorithm to optimize and obtain the current optimal computing resource allocation rate;

[0039] Step 4.4.4: Update the computing resource allocation rate in P1, re-optimize, and regularly follow this rule until the difference between the optimized objective function value and the function value of the previous iteration is less than the termination condition or the number of loops is greater than the maximum number of iterations, to obtain the optimal inter-slice spectrum and computing resource allocation decision.

[0040] In a preferred embodiment of the present invention, the object of the present invention is to provide a network slice framework considering slice correlation and an inter-slice network resource (transmission, computing) allocation method based on task timeliness. The slice correlation is described by estimation error, and the analytical relationship between the inter-slice resource allocation strategy and slice correlation is comprehensively characterized by task timeliness. By scaling and decoupling the original constrained optimization problem, an iterative particle swarm algorithm is proposed to minimize the estimation error of the entire system on the premise of ensuring the convergence of slice estimation, balance the resource occupancy between slices, and improve the overall performance.

[0041] The present invention provides the following technical solutions: A network slice framework considering slice correlation includes network resource allocation on a large time scale and scheduling and estimation on a small time scale. Considering the scenario of fusion estimation with multiple measurement methods, in order to meet heterogeneous measurement requirements, network slicing is used to achieve logical isolation. The network resource allocation on the large time scale uses the fusion estimation performance as the slice correlation model, and by reasonably allocating the transmission and computing resources between slices, the timeliness and accuracy of edge estimation are ensured. The scheduling and estimation on the small time scale, each slice uploads and processes measurement values based on the allocated transmission and computing resources, and performs fusion estimation on the edge device to provide a basis for subsequent control. The network resource allocation on the large time scale does not change before the slice request is updated, and the scheduling and estimation on the small time scale periodically loop within the slice window until the window ends.

[0042] The present invention also provides an inter-slice network resource allocation method based on task timeliness, including the following steps:

[0043] Step 1: Initialize the sampling period T of different measurement devices i , the data packet size λ i , the computing intensity X i , the available frequency band resource M T and the computing resource M C , the inter-slice transmission resource allocation rate β and the computing resource allocation rate γ, as well as the state change models of each slice and configure relevant parameters;

[0044] Step 2: After the SDN module in the edge device receives the slice creation request, collect network information, perform pooling virtualization of physical resources, and establish a slice correlation model and inter-slice resource constraint conditions;

[0045] Step 3: Construct a performance evaluation index based on slice correlation, and establish a resource optimization scheduling problem under resource finiteness and system convergence constraints;

[0046] Step 4: Since the problem cannot be explicitly expressed using decision variables, the original optimization problem is transformed into a single-resource allocation sub-problem for a single time slot by means of scaling and decoupling, and an iterative solution method is used to obtain the transmission and computing resource allocation decisions, completing the inter-slice resource allocation;

[0047] Step 5: The SDN module is responsible for executing the slice decision, mapping the virtualized slice resources to the corresponding physical network node set, and completing the network slice construction;

[0048] Step 6: When the slice request is updated, the SDN module re-performs the pooling virtualization and association model construction of physical resources, returns to execute the first step, and re-performs network slicing and resource allocation.

[0049] In Step 2, the slice association model and the inter-slice resource constraint conditions are established, and the specific steps are as follows:

[0050] Step 2.1 Define the sampling period, transmission delay, and computing delay of different slices as T i , and respectively, and calculate the task timeliness of different slices where This is used to measure the information lag of the measured values received by different slices, where is the generation time of the latest received data packet of slice i;

[0051] Step 2.2 Construct an estimation error model for different slices based on task timeliness, where K i (t) is a function of the estimation error variance and the measurement error variance, A is the system state matrix, ω i and v i are the system noise and observation noise of slice i respectively;

[0052] Step 2.3 Calculate the fusion estimation error of all slices, and use the mean square error to evaluate the estimation performance, where ∑μ i = 1 is the weight factor of each slice, and this is used as the mathematical expression of the slice association model to reveal the coupled impact of the inter-slice resource allocation decision on the system performance;

[0053] Step 2.4 Considering the estimation convergence of different slices, respectively give the delay constraint condition for the delay-sensitive slice and the bandwidth constraint condition for the bandwidth-consuming slice;

[0054] In step 4, the scaling, decoupling and iterative solution methods are as follows:

[0055] In step 4.1, through term elimination and logarithmic operations, the original time-accumulative objective function is transformed into a single-slot objective function.

[0056] In step 4.2, considering the zero-waiting strategy, that is the average AoT of the slice can be expressed as a linear function of the transmission delay and the computing delay that is, the single-slot objective function increases as the transmission and computing delays increase.

[0057] In step 4.3, the original optimization problem can be decoupled into a transmission resource allocation sub-problem P1 and a computing resource allocation sub-problem P2. Since both problems are still complex and difficult to directly obtain a closed-form solution, the particle swarm optimization algorithm can be used to solve the objective variables respectively.

[0058] In step 4.4, to improve the overall performance of the system, the objective functions of the two sub-problems are used as the fitness values of the particle swarm optimization algorithm. The iterative particle swarm optimization algorithm is adopted to continuously iterate and optimize to seek the optimal solution, so as to obtain the optimal inter-slice transmission and computing resource allocation decision for this optimization objective.

[0059] The iterative particle swarm optimization algorithm in step 4.4 is as follows:

[0060] In step 4.4.1, an initial value of the inter-slice computing resource allocation rate and a loop termination condition are given.

[0061] In step 4.4.2, for the transmission resource allocation problem P1, based on the given inter-slice computing resource allocation rate γ, the particle swarm optimization algorithm is used for optimization to obtain the current optimal transmission resource allocation rate.

[0062] In step 4.4.3, for the computing resource allocation problem P2, based on the current inter-slice transmission resource allocation rate β, the particle swarm optimization algorithm is used for optimization to obtain the current optimal computing resource allocation rate.

[0063] Update the computing resource allocation rate in P1, re-optimize, and regularly follow this rule until the difference between the objective function value of this optimization and the function value of the previous iteration is less than the termination condition or the number of loops is greater than the maximum number of iterations, that is, the optimal inter-slice transmission and computing resource allocation decision is obtained.

[0064] Compared with the prior art, the present invention has the following obvious substantial features and remarkable advantages:

[0065] 1. A network slice framework considering slice correlation is proposed, which simultaneously considers network resource allocation at a large time scale and scheduling and estimation at a small time scale to support multiple IIoT services and improve the overall performance of the system.

[0066] 2. Considering the timeliness of tasks, the information lag caused by transmission and computing delays can be comprehensively characterized. An estimation error model based on task timeliness is constructed, and the system performance is evaluated by fusing the mean square errors of all slices, which is used as the mathematical expression of the slice association model, revealing the coupling effects of different slices on the system performance.

[0067] 3. Aiming at the current situation of slice correlation and multi-dimensional resource coupling, an iterative particle swarm optimization (PSO) algorithm is proposed to optimize the multi-dimensional resource allocation, improve the overall system performance, and reduce the computational complexity at the same time.

[0068] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, features and effects of the present invention. Description of the Drawings

[0069] Figure 1 is a schematic diagram of a network slice framework based on slice correlation in a preferred embodiment of the present invention;

[0070] Figure 2 is a schematic diagram of a scheduling and estimation model on a small time scale in a preferred embodiment of the present invention;

[0071] Figure 3 is a flow chart of an iterative particle swarm algorithm for joint optimization of multiple resources in a preferred embodiment of the present invention;

[0072] Figure 4 is a flow chart of a particle swarm algorithm for single resource allocation optimization in a preferred embodiment of the present invention. Detailed Embodiments

[0073] The following introduces multiple preferred embodiments of the present invention with reference to the accompanying drawings of the specification to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0074] In the drawings, components with the same structure are denoted by the same numeral labels, and components with similar structures or functions are denoted by similar numeral labels. The size and thickness of each component shown in the drawings are arbitrarily shown, and the present invention does not limit the size and thickness of each component. In order to make the illustration clearer, the thickness of some parts in the drawings is appropriately exaggerated.

[0075] Figure 1It is a schematic diagram of a network slicing framework based on slice correlation. Taking the angle measurement of the aircraft final assembly activity surface as an example, in order to improve the overall performance of the system, a scenario of using delay-sensitive sensors and bandwidth-consuming cameras for simultaneous observation for state estimation is considered. The considered edge-assisted IIoT system includes S delay-sensitive sensors, a bandwidth-consuming camera, and an edge device. To meet the requirements of the two measurement methods, network slicing is used to achieve resource isolation. Slice 1 represents the transmission and computing resources allocated to delay-sensitive sensors, and slice 2 represents the resources allocated to the bandwidth-consuming camera. By dynamically allocating inter-slice transmission and computing resources, the measurement values of slice 1 and slice 2 are combined for fusion estimation to ensure the timeliness and accuracy of edge estimation. To better describe this process, a dual time scale is designed, including the allocation of transmission and computing resources at the large time scale (slice window), and the scheduling and estimation at the small time scale (time slot). At the beginning of each slice window, the edge device virtualizes the computing resources and spectrum resources, and allocates resources according to slice correlation and resource constraints to establish heterogeneous network slices. Without loss of generality, it is assumed that the resources of the network slice do not change within the slice window. Within the slice window, different measurement devices upload the measurement values according to the allocated resources, and perform fusion estimation on the edge device to provide a basis for subsequent control.

[0076] Figure 2 It is a schematic diagram of the scheduling and estimation model at the small time scale. The transmission and computing processes are regarded as separate queues. According to the inter-slice resource allocation decision at the start time of the slice window, the transmission and computing resources occupied by each slice remain unchanged within the slice window. Both slice 1 and slice 2 adopt the zero-waiting strategy, with T D and T B for periodic sampling. When new measurement values are generated, they are transmitted to the computing unit of the edge device through 5G for effective information extraction and estimation; if the computing unit does not receive new measurement values, estimation is performed based on the existing measurement values. The estimation unit fuses the estimation values of all slices according to certain criteria to improve the fault tolerance and self-adaptability of edge estimation. The considered slice 1 is a homogeneous network, that is, the packet sizes and computing intensities generated by each sensor are the same, and the transmission and computing resources of slice 1 are evenly divided, so the transmission and computing delays of all sensors within slice 1 are the same, and the measurement value of slice 1 is the mean value of the measurement values of all sensors.

[0077] Figure 3 It is a flowchart of the iterative particle swarm algorithm for joint optimization of multiple resources. The specific steps are as follows:

[0078] Step 1 Set the tolerance threshold ε = 10 -4 , and the maximum number of iterations σ = 30;

[0079] Step 2 Initialize the computing resource allocation rate The number of loops num = 0;

[0080] Step 3: For the transmission resource allocation problem P1, based on the given inter-chip computing resource allocation rate, use the particle swarm optimization algorithm for optimization to obtain the current optimal transmission resource allocation decision;

[0081] Step 4: Update the inter-chip transmission resource allocation rate;

[0082] Step 5: For the computing resource allocation problem P2, based on the given inter-chip computing resource allocation rate, use the particle swarm optimization algorithm for optimization to obtain the current optimal transmission resource allocation decision;

[0083] Step 6: Update the inter-chip computing resource allocation rate;

[0084] Step 7: Substitute the transmission and computing allocation rates, calculate the value of the optimization objective function, and the number of loops num = num + 1;

[0085] Step 8: Determine whether the difference between the two function values is less than the tolerance threshold ε or the number of loops is greater than the maximum iteration number σ. If the termination condition is satisfied, go to Step 9; otherwise, jump to Step 3 and repeat the iteration;

[0086] Step 9: Output the optimal inter-chip transmission and computing resource allocation rates and end the entire algorithm.

[0087] Figure 4 It is the flowchart of the particle swarm optimization algorithm for single resource allocation optimization, and the specific steps are as follows:

[0088] Step 1: Initialize the population size N, the velocity v i v i and the position x i ;

[0089] Step 2: Calculate the fitness value F it [i] of each particle according to P1 / P2;

[0090] Step 3: Calculate the individual optimal value P best (i) of each particle. Compare the fitness value F it [i] of each particle with the individual optimal value P best (i). If F it [i] < P best (i), then replace P it (i) with F best (i);

[0091] Step 4: Calculate the global optimal value g best of the entire population. Compare the fitness value F it [i] of each particle with the global optimal value g best and if Fit [i]<g best Then replace g with F it [i]; best ;

[0092] Step 5 Update the particle velocity v i = w * v i + c1r1(P best (i) - x i ) + c2r2(g best - x i ) and the position x i = x i + v i ;

[0093] Step 6 Perform boundary condition processing and update the velocities and positions of the particles that do not meet the conditions;

[0094] Step 7 Determine whether the end conditions (error threshold and number of iterations) are met. If the end conditions are met, go to Step 8; otherwise, return to Step 2 and repeat the optimization;

[0095] Step 8 Output the optimal transmission / calculation resource allocation rate, and the algorithm ends.

[0096] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field according to the concept of the present invention through logical analysis, reasoning, or limited experiments on the basis of the prior art should be within the protection scope determined by the claims.

Claims

1. A method for allocating associated slice resources based on task timeliness, characterized in that, It includes a network slice framework considering slice correlation and an inter-slice network resource allocation method based on task timeliness; the network slice framework includes network resource allocation at a large time scale and scheduling and estimation at a small time scale; the inter-slice network resource allocation method includes the following steps: Step 1, Initialize different measurement devices, as well as the state change models of each slice and configure relevant parameters; Step 2, After the SDN module in the edge device receives a slice creation request, collect network information, perform pooling virtualization of physical resources, and establish a slice correlation model and inter-slice resource constraint conditions; Step 3, Construct a performance evaluation index based on slice correlation, and establish a resource optimization scheduling problem under the constraints of resource finiteness and system convergence; Step 4, Use the scaling and decoupling method to transform the optimization problem into a single-resource allocation sub-problem for a single time slot, and use the iterative solution method to obtain the transmission and computing resource allocation decisions to complete the inter-slice resource allocation; Step 5, The SDN module executes the slice decision, maps the virtualized slice resources to the corresponding physical network node set, and completes the network slice construction; Step 6, When the slice request is updated, the SDN module re-performs the pooling virtualization of physical resources and the construction of the correlation model, returns to execute Step 1, and re-performs network slicing and resource allocation; The said Step 2 includes the following steps: Step 2.1: Define the sampling period, transmission delay, and computing delay of different slices respectively, and calculate the task timeliness of different slices To measure the information lag of the measured values received by different slices, where Δ u (t) is the task timeliness of slice i at time slot t, is the generation time of the latest received data packet of slice i; Step 2.2, Construct an estimation error model for different slices based on task timeliness Among them, K i (t) is a function of the estimated error variance and the measurement error variance, A is the system state matrix, ω i and v i are the system noise and the observation noise of slice i, respectively; Step 2.3, Calculate the fusion estimation error of all slices, and use the mean square error to evaluate the estimation performance, which is used as the mathematical expression of the slice correlation model to reveal the coupling effect of inter-slice resource allocation decisions on system performance; Step 2.4, Considering the estimation convergence of different slices, respectively give the delay constraint conditions for delay-sensitive slices and the bandwidth constraint conditions for bandwidth-consuming slices; The said Step 4 includes the following steps: Step 4.1, Through term elimination and logarithmic operations, transform the original time-accumulative objective function into a single-time-slot objective function; Step 4.2, Considering the zero-waiting strategy, the average AoT of a slice can be expressed as a linear function of transmission delay and computing delay, that is, the single-time-slot objective function increases as the transmission and computing delays increase; Step 4.3, Decouple the optimization problem into a transmission resource allocation sub-problem P1 and a computing resource allocation sub-problem P2, and use the particle swarm algorithm to solve the objective variables respectively; Step 4.4, Take the objective functions of the two sub-problems as the fitness values of the particle swarm algorithm, use the iterative particle swarm algorithm, continuously iterate and optimize to seek the optimal solution, and obtain the optimal inter-slice transmission and computing resource allocation decisions for the optimization objective.

2. The method for allocating associated slice resources based on task timeliness according to claim 1, wherein The said network slice framework uses network slicing to achieve logical isolation and meet heterogeneous measurement requirements.

3. The method for allocating associated slice resources based on task timeliness according to claim 1, wherein For the network resource allocation at a large time scale, the fusion estimation error is used as the slice correlation model, and by allocating the transmission and computing resources between slices, the timeliness and accuracy of edge estimation are guaranteed.

4. The method for allocating associated slice resources based on task timeliness according to claim 1, wherein For the scheduling and estimation at a small time scale, each slice uploads and processes measurement values based on the allocated transmission and computing resources, and performs fusion estimation at the edge device to provide a basis for subsequent control.

5. The method for associative slice resource allocation based on task timeliness according to claim 1, wherein The network resource allocation at the large time scale does not change before the slice request is updated, and the scheduling and estimation at the small time scale cycle periodically within the slice window until the window ends.

6. The method for allocating associated slice resources based on task timeliness according to claim 1, wherein, In step 1, the measurement device is initialized to include the sampling period, packet size, computing intensity, available frequency band resources and computing resources, the inter-slice transmission resource allocation rate, and the computing resource allocation rate.

7. The method for allocating associated slice resources based on task timeliness according to claim 1, characterized in that, Step 4.4 includes the following steps: Step 4.4.1: Given the initial value of the inter-slice computing resource allocation rate and the loop termination condition; Step 4.4.2: For the transmission resource allocation sub-problem P1, based on the given inter-slice computing resource allocation rate, use the particle swarm optimization algorithm to optimize and obtain the current optimal transmission resource allocation rate; Step 4.4.3: For the computing resource allocation sub-problem P2, based on the current inter-slice transmission resource allocation rate, use the particle swarm optimization algorithm to optimize and obtain the current optimal computing resource allocation rate; Step 4.4.4: Update the computing resource allocation rate in P1, re-optimize, and regularly follow this rule until the difference between the optimized objective function value and the function value of the previous iteration is less than the termination condition or the number of loops is greater than the maximum number of iterations, to obtain the optimal inter-slice spectrum and computing resource allocation decision.

Images