A network slice dynamic scheduling system and method based on power service differences

Abstract

The present invention discloses a network slice dynamic scheduling system and method based on power service differences, belonging to the field of power wireless communication technology. The scheduling system includes a slice layer, a service layer, and a network layer. The slice layer calculates the network slice resource reservation ratio according to the power service type, service delay, and service arrival rate of the service layer, and calculates and sorts the weights of the network slices; the service layer calculates and sorts the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request delay, and allocates the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed; the network layer configures the network slice links with the physical link load balancing as a constraint and the minimum virtual link delay of the network slice as a target, and outputs the optimal link configuration result of the network slice. The present invention comprehensively considers link delay and load balancing, and ensures the optimal allocation of resources and the service requirements of services.

Description

Technical Field

[0001] The present invention relates to the technical field of power wireless communication, and in particular to a network slice dynamic scheduling system and method based on power service differences. Background Art

[0002] Distributed power services can be divided into three categories: control services with fixed periods, control services with random bursts, and collection services. Different types of services have different requirements for latency. Control services are mainly involved in real-time production regulation and control, and have extremely high requirements for latency accuracy and stability. Especially in variable environmental conditions, the impact of latency jitter is more significant. Therefore, control services usually have the most stringent requirements for communication latency, and it is necessary to ensure the timeliness and reliability of data transmission.

[0003] In the real-time regulation and control of distributed power clusters, the communication latency usually needs to reach the millisecond level. However, traditional communication technologies often have difficulty meeting the dual requirements of low latency and high reliability at the same time. Therefore, the 5G network slicing technology has become an effective means to solve this problem. The 5G network slicing technology can provide dedicated virtual network resource pools for different types of services, and through flexible slicing and customized configuration of network resources, provide different latency, bandwidth, and reliability guarantees for each type of service.

[0004] In a slice-based network architecture, how to make full use of limited spectrum resources, flexibly allocate network resources through resource reservation and slicing, etc., and ensure the real-time and reliable bearer of various power services in the 5G network with large differences in communication requirements and different focuses, is an urgent problem to be solved for realizing the secure and flexible bearer of power 5G multi-services. Summary of the Invention

[0005] The purpose of the present invention is to provide a network slice dynamic scheduling system and method based on power service differences, which solve the problems of low resource utilization rate, insufficient latency guarantee for key services, and weak network dynamic response ability in traditional solutions by constructing a three-layer model of slice-service-network and a resource scheduling system.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, the present invention provides a network slice dynamic scheduling system based on power service differences, including:

[0008] A slice layer, configured to calculate the network slice resource reservation ratio according to the power service type, service latency, and service arrival rate in the service layer, and calculate and sort the weights of the network slices;

[0009] The service layer is configured to calculate and sort the priorities of the power services to be processed according to the resource requirements and service request delays of different power service requests, and allocate the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed;

[0010] The network layer is configured to perform network slice link configuration with the physical link load balancing as the constraint and the minimum virtual link delay of the network slice as the goal, output the optimal link configuration result of the network slice, and complete the resource allocation for the power services to be processed.

[0011] Preferably, the types of power services include control services and collection services;

[0012] The control services include fixed-period control services and random burst control services.

[0013] Preferably, the slice layer is specifically used for

[0014] Dividing the reserved resources of the slice layer into static reserved resources and dynamic reserved resources;

[0015] The static reserved resources are used by the fixed-period control services, and the dynamic reserved resources are used by the random burst control services. If there are no random burst control services, the dynamic reserved resources are used by the collection services, and the remaining resources except the static reserved resources and the dynamic reserved resources are used by the collection services.

[0016] Preferably, the slice layer calculates the network slice resource reservation ratio according to the power service type, service delay and service arrival rate of the service layer, expressed as:

[0017] ;

[0018] ;

[0019] Among them, is the reservation ratio of the static reserved resources, is the reservation ratio of the dynamic reserved resources, is the average rate of the data arriving for the fixed-period control services, represents the maximum delay that the packets of the fixed-period control services can tolerate, is the service rate of the scheduling system, is the maximum error rate that the packets of the fixed-period control services can tolerate, is the maximum error rate that the packets of the random burst control services can tolerate, represents the slope of the random arrival curve of the random burst control services;

[0020] Expressed as:

[0021] ;

[0022] is expressed as:

[0023] ;

[0024] is expressed as:

[0025] ;

[0026] wherein, and are both normal constants related to the randomness of the service process, represents a variable in the boundary function of the random arrival curve, is a free parameter greater than 0 related to the randomness of the arrival of random burst control type services, is a parameter related to the random service curve of random burst control type services, represents time.

[0027] Preferably, the slice layer calculates and sorts the weights of network slices. Specifically,

[0028] initialize an empty network slice queue;

[0029] calculate the weights of each network slice, sort them in descending order, and add them to the network slice queue;

[0030] when allocating power services to be processed each time, preferentially select the network slice with the largest weight;

[0031] after the power service to be processed is processed, update the weights of each network slice and re - sort them in descending order;

[0032] calculate the weights of network slices in the following manner:

[0033] ;

[0034] wherein: represents the network slice weight of the transmission time interval of, is the weighting coefficient of the network slice, and the weighting coefficients of each network slice are different, represents the utilization rate of bandwidth resource blocks from the start to the transmission time interval of, is the utilization rate of bandwidth resource blocks from the start to the transmission time interval of.

[0035] Preferably, the service layer calculates and sorts the priorities of the power services to be processed according to the resource requirements and service request delays of different power service requests. Specifically,

[0036] Calculate the priorities of the power services to be processed according to the resource requirements and service request delays of different power service requests;

[0037] Initialize an empty queue of power services to be processed;

[0038] In each iteration, select the power service to be processed with a priority of 0 and add it to the queue of power services to be processed for processing. If there is no power service to be processed with a priority of 0, then select the power service to be processed with the highest priority, and so on until the power service to be processed with the lowest priority, and remove it from the queue of power services to be processed after processing;

[0039] After each iteration, recalculate the priorities of the power services to be processed, and then add them to the queue of power services to be processed according to the priorities;

[0040] When all the power services to be processed are completed, the iteration ends;

[0041] Calculate the priorities of the power services to be processed in the following way:

[0042] ;

[0043] Among them, is the priority of the power service to be processed, is the service resource requirement of the virtual node, is the duration of the power service request to be processed, is the ratio of the waiting time of the power service request to be processed to the longest waiting time, is a binary value;

[0044] It is expressed as:

[0045] ;

[0046] Among them, is the virtual node The cpu computing power required for the power service request to be processed, is the th virtual node, is the virtual link Remaining bandwidth resources, Represents the set of virtual links, Represents the set of virtual nodes ; The virtual nodes are intelligent terminals, base stations and control centers in the distribution network;

[0047] Expressed as:

[0048] ;

[0049] Wherein, is the waiting time for the power service request to be processed, is the maximum waiting time for the power service request to be processed.

[0050] Preferably, the network layer configures the network slice link with the physical link load balancing as the constraint and the minimum virtual link delay of the network slice as the goal. Specifically,

[0051] Determine the optimal network slice link configuration problem, expressed as:

[0052] ;

[0053] ;

[0054] ;

[0055] And the following constraints need to be satisfied:

[0056] ;

[0057] ;

[0058] Wherein, Min represents minimization, is the reliability penalty function, is the weight value, is the virtual link delay, is the variance of the physical link load ratio, is the physical link in the network slice, is the set of physical links, is the transmission delay, is the propagation delay, is the queuing delay, is the processing delay, is a binary value, representing the mapping index variable, When = 1, it means that the physical link is successfully mapped to the virtual link, When = 0, it means that the physical link is not successfully mapped, is the physical link load ratio, represents the average value of the load ratios of all physical links, is the maximum allowable delay of the service request, is the virtual node The bandwidth requirement of the service request, is a binary value, indicating that the virtual node of the service request is successfully mapped to the physical link, for the remaining bandwidth resources of the physical link;

[0059] The genetic algorithm is used to solve the optimal network slice link configuration problem, and the optimal network slice link configuration result is obtained.

[0060] In a second aspect, the present invention provides a network slice dynamic scheduling method based on power service differences, which is implemented based on the above-mentioned network slice dynamic scheduling system based on power service differences. The method includes:

[0061] The slice layer calculates the network slice resource reservation ratio according to the power service type, service delay, and service arrival rate of the service layer, and calculates and sorts the weights of the network slices;

[0062] The service layer calculates and sorts the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request delay;

[0063] The power services to be processed are assigned to the network slices according to the network slice weights and the priorities of the power services to be processed;

[0064] The network layer configures the network slice links with the physical link load balancing as the constraint and the minimum virtual link delay of the network slice as the goal, outputs the optimal network slice link configuration result, and completes the resource allocation for the power services to be processed.

[0065] Preferably, the step of assigning the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed includes:

[0066] When assigning the power services to be processed each time, the power services to be processed with a priority of 0 are preferentially selected to join the queue of power services to be processed. If there are no power services to be processed with a priority of 0, the power services to be processed with the maximum priority are selected, and so on until the power services to be processed with the minimum priority;

[0067] The selected power services to be processed are assigned to the network slice with the largest weight.

[0068] Preferably, after the power services to be processed are assigned and processed, the weights of each network slice are updated, sorted again; and the priorities of the power services to be processed are recalculated, and then added to the queue of power services to be processed according to the priorities.

[0069] The beneficial effects achieved by the present invention are: ​

[0070] The present invention provides a network slice dynamic scheduling system based on power service differences. First, the slicing layer reduces resource waste through dynamic reservation. By modeling with stochastic network calculus theory, the reserved resources for control services are changed from a fixed peak value to dynamically adjusted reservation ratios, which not only guarantees the delay and reliability requirements of bursty control services but also improves the utilization rate of system resources. Secondly, the link configuration model in the network layer comprehensively considers link delay and load balancing, and finds the optimal link with smaller link delay on the basis of guaranteeing reliability, ensuring the optimal allocation of resources and the service requirements of services. The service layer first sets the highest priority for control services, and then sorts the services according to the delay requirements and bandwidth requirements of the services, effectively guaranteeing the performance requirements of critical services.

[0071] The present invention proposes a three-layer model, and solves the problems of low system resource utilization rate, insufficient delay guarantee for critical services, and weak network dynamic response ability in traditional solutions through a three-level cooperation mechanism of the slicing layer, service layer, and network layer. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a flowchart of a network slice dynamic resource scheduling method based on multi-service differentiation provided by an embodiment of the present invention;

[0073] Figure 2 is a schematic diagram of a network slice dynamic resource scheduling system based on power multi-service differentiation provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in combination with the embodiments and the drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but do not limit the present invention.

[0075] Here, it also needs to be noted that in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution of the present invention are shown in the drawings, while other details less related to the present invention are omitted.

[0076] It should be emphasized that the term "including / comprising" when used herein refers to the presence of features, elements, steps, or components, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0077] It should be emphasized here that the step labels mentioned hereinafter do not limit the sequence of each step. Instead, it should be understood that the steps can be executed in the sequence mentioned in the embodiments, can be different from the sequence in the embodiments, or several steps can be executed simultaneously.

[0078] In the power grid system based on 5G, different application scenarios have different service requirements, including control services and acquisition services. Different services have different requirements for factors such as latency and bandwidth. As shown in Table 1:

[0079] Table 1 Analysis of Distributed Energy Regulation Services

[0080]

[0081] In actual application scenarios, control services may arrive suddenly. The arrival time of randomly burst control services is very short, so ultra-low latency and high reliability are also required.

[0082] To optimize the resource allocation of the entire system, the embodiments of the present invention provide a dynamic resource scheduling system for network slices based on power multi-service differentiation. Refer to Figure 2 , which is a three-layer structure of slice layer - service layer - network layer.

[0083] The slice layer first considers the situation of randomly burst arrival of control services and proposes a resource reservation scheme to reserve some idle network resources for control services. However, if the reservation is made according to the peak rate of randomly burst control services and there is no randomly burst control service in a certain time interval, it will cause waste of system resources. Therefore, the embodiments of the present invention divide the reserved resources into two parts: static reserved resources and dynamic reserved resources. The static reserved resources are used by fixed-cycle control services, and the dynamic reserved resources are used by randomly burst control services. If there is no randomly burst control service, the dynamic reserved resources are used by acquisition services, and the remaining resources except the static reserved resources and dynamic reserved resources are used by acquisition services. After calculating the resource reservation ratio, the slice layer considers the scheduling problem of slices. The core is how to reasonably arrange the resources of different network slices to meet the needs of various services to the greatest extent.

[0084] The service layer mainly solves the resource allocation problem by evaluating the importance and real-time nature of different services. The service layer can preferentially allocate more resources to high-priority services to ensure timely and reliable services in the network. At the same time, the service layer also considers the needs of low-priority services to reasonably arrange the remaining resources, so as to optimize the overall resource utilization efficiency while meeting key services.

[0085] The network layer mainly considers two factors: link latency and load balancing. The embodiments of the present invention design a link configuration method, construct a penalty function for service reliability, and output an optimal network slice link configuration result.

[0086] The specific description of the scheduling system of this embodiment is as follows:

[0087] In this embodiment, the slice layer is mainly used for calculating the resource reservation ratio and the slice weight.

[0088] Specifically, the slice layer determines the amount of resources to be allocated to the network slice according to the instructions issued by the slice control layer. At the same time, in the virtual network layer, the network functions of the control layer can allocate the IDs of the network slice type and the service. Based on this, this embodiment proposes a dynamic resource reservation method based on stochastic network calculus.

[0089] Stochastic network calculus can well analyze the performance of power networks with strong randomness and burstiness. Through the min-plus algebra theory, various network elements in the distribution network (such as intelligent terminals, base stations, control centers, etc.) are abstracted as virtual nodes to facilitate the analysis and optimization of performance boundaries. These node models are described using stochastic network calculus based on the arrival process and service process of the service.

[0090] In the embodiment of the present invention, control services and collection services coexist in a resource reservation manner, where the proportion of resources reserved for fixed-cycle control services is and the proportion of resources reserved for random burst control services is , then the proportion of resources left for collection services is .

[0091] The delay referred to in the present invention is defined as the time difference from the generation of the data packet to the end of the data packet transmission. When the delay experienced by the data packet exceeds the maximum tolerable delay, the packet will be discarded. The discarded data packet is regarded as an incorrect reception at the receiving end. Although the incorrect reception at the receiving end in the actual system includes the error caused by packet loss and the reception failure caused by reasons such as channel conditions, the error rate mentioned in the embodiment of the present invention only considers the reception failure caused by packet loss. On this basis, the probabilistic delay constraint is defined to realize the joint consideration of delay and error rate.

[0092] Therefore, to meet the performance requirements of control services, reliability and delay can be expressed by formula (1):

[0093] (1)

[0094] Where, represents the actual delay of the data packet at time; represents the maximum tolerable delay of the data packet; represents the maximum tolerable error rate of the data packet. Formula (1) means that the system network has successfully transmitted the data of control services under the condition of meeting the reliability and delay requirements.

[0095] In network traffic modeling, the Poisson distribution is often used to describe the situation. In control traffic, data usually arrives at fixed intervals, and the size of each data packet is a fixed unit size. Assume that the data arriving in the fixed-interval control traffic approximately follows a Poisson distribution with an average rate So the arrival curve , and the boundary function is:

[0096] (2)

[0097] In the formula, represents a variable in the boundary function of the stochastic arrival curve, which is used to describe the randomness of the data stream arrival process. In stochastic network calculus, the boundary function is used to represent the probability that the data stream arrival process exceeds a certain value.

[0098] Introduce the definition of the stochastic arrival curve:

[0099] If for all there is:

[0100] (3)

[0101] holds, then the stochastic arrival curve of the data stream is , denoted as . In the formula: describes the data accumulated at the server from time to time ; is used to express the boundary function, assuming that there is a probability that the arrival process exceeds the stochastic arrival curve, but is restricted by the boundary curve function. represents the supremum. In mathematics, the supremum is the least upper bound of a set. In formula (3), represents the maximum value of the difference between the data stream and the average arrival curve within the time interval . represents the average arrival curve of the data stream, which is a deterministic function that describes the average arrival rate of the data stream within time .

[0102] Let the arrival process of the fixed-interval control traffic following the Poisson distribution be: , indicating that the arrival process is related to the arrival curve and the boundary function, is the arrival curve , is the boundary function .

[0103] The arrival of random burst control type services is randomly uncertain. Therefore, in the embodiments of the present invention, the ON / OFF model is used to perform network modeling on random burst control type services.

[0104] Let the data flow of random burst control type services be a Markov modulated ON / OFF model. The data flow switches between two states: the ON active state and the OFF silent state. The ON state represents the data flow of random burst control type services, and the rate is while the OFF state represents no data flow of random burst control type services.

[0105] The probability of the data flow with random burst control type services changing to the data flow without random burst control type services is and the probability of the data flow without random burst control type services changing to the data flow with random burst control type services is Then the random arrival curve of random burst control type services is The boundary function is a free parameter greater than 0, used to represent the randomness of service arrival, and c is the free parameter in the boundary function. represents the slope of the random arrival curve of random burst control type services.

[0106] (4)

[0107] Let the arrival process of random burst control type services be: . is the arrival curve is the boundary function .

[0108] In the embodiments of the present invention, it is assumed that the system transmits data at a service rate and only a single node and a single server are considered. Therefore, it can be obtained that the random service curve of the system is and the boundary function is . and are both positive constants, related to the randomness of the service process, and can be understood as the fluctuation degree of the service process. In random network calculus, represents the upper bound of the service process, represents the lower bound of the service process.

[0109] Introduce the definition of the random service curve:

[0110] For all there is:

[0111] ​​(5)

[0112] is established, it can be said that the stochastic service curve of is ; is the multiplication operation. The probability of data generation outside the service capacity is constrained by the boundary function .

[0113] After the above analysis, the proportion of static reserved resources is set to , and the proportion of dynamic reserved resources is . When there are fixed-period control services, stochastic burst control services, and collection services to be transmitted in the system simultaneously, all three types of services can obtain service capacity. Among them, the stochastic service curve of the fixed-period control service is , the boundary function of the stochastic service curve is , the service process is , the stochastic service curve of the stochastic burst control service is , the boundary function of the stochastic service curve is , the service process is , the stochastic service curve of the stochastic collection service is , the boundary function of the stochastic service curve is , the service process is , which is shown as follows:

[0114] (6)

[0115] (7)

[0116] (8)

[0117] and are both positive constants, related to the randomness of the service process, is the data flow rate of the stochastic burst control service, is the system service rate, is the probability from the data flow with the stochastic burst control service to the data flow without the stochastic burst control service, is the probability from the data flow without the stochastic burst control service to the data flow with the stochastic burst control service.

[0118] Define the stochastic delay boundary as:

[0119] If the system input data flow Subject to a stochastic arrival curve and the system providing a stochastic service curve for the input data stream, for all cases, the upper bound of the stochastic delay of the system is:

[0120] (9)

[0121] where, represents the actual delay of the data stream, is the boundary function of the stochastic arrival curve, is the boundary function of the stochastic service curve, and the function represents and the maximum horizontal distance between them, and also represents the upper bound of the packet delay, which can be written as:

[0122] (10)

[0123] represents the infimum. In mathematics, the infimum is the greatest lower bound of a set, represents the stochastic arrival curve of the data stream, where is an additional delay, represents the stochastic service curve of the data stream, where is a delay.

[0124] Substituting the stochastic arrival curve and the stochastic service curve of the fixed-period control type service into the above stochastic delay boundary formula (9), the delay boundary of the fixed-period control type service can be obtained:

[0125] (11)

[0126] In the formula, the maximum delay that the packets of the fixed-period control type service can tolerate is: , the maximum error rate that the packets of the fixed-period control type service can tolerate , represents the actual delay of the packets of the fixed-period control type service at time.

[0127] Through derivation, the relationship between the proportion of static reserved resources and the delay, error rate, service arrival rate of the control type service, and the system service rate can be obtained:

[0128] (12)

[0129] Substituting the stochastic arrival curve and the stochastic service curve of the stochastic burst control type service into the above stochastic delay boundary formula (9), the delay boundary of the stochastic burst control type service can be obtained:

[0130] (13)

[0131] is a parameter related to the boundary function of the stochastic service curve and the related function .

[0132] Let the maximum error rate that the stochastic burst control type service can tolerate , then we can get

[0133] (14)

[0134] Under the above conditions, the relationship between the violation probability of the service, the service arrival rate, the delay requirement, the maximum error rate, and the system service rate can be observed, and thus and can be calculated. The violation probability of the service refers to the probability that the performance indicators (such as delay, packet loss rate, etc.) of a certain service fail to meet the predetermined requirements under the given quality of service (QoS) requirements. In other words, it measures the possibility that the service violates its quality of service guarantee within a specific time

[0135] In the embodiments of the present invention, the resource reservation ratio is dynamically calculated according to the probability of the stochastic burst control type service occurring within a period of time. If the occurrence probability is low, the resource reservation ratio will decrease accordingly

[0136] In the embodiments of the present invention, when calculating the slice weight, the resource allocation is guaranteed according to the service level agreement (SLA) of the slice, that is, the entire scheduling process must meet the service level agreement. The slice weights are continuously sorted within the slice, and the common resources are managed through the slice weights and the priorities of the services they need

[0137] Define to represent the network slice weight, whose main function is to represent the size of the slice utility at a specific moment. The network slice weight is expressed as follows

[0138] (15)

[0139] In the formula represents the network slice weight of the transmission time interval , is the weighted coefficient of the network slice, and each network slice has a different weighted coefficient , represents the number of bandwidth blocks required by the slice according to the service level agreement at the initial stage is the average utilization rate of bandwidth resource blocks during the time period from the start to the previous transmission time interval (TTI). is the utilization rate of bandwidth resource blocks from the start to the transmission time interval within the bandwidth resource blocks. is the utilization rate of bandwidth resource blocks from the start to the transmission time interval within the bandwidth resource blocks. is the total amount of bandwidth resource blocks during the transmission process of the entire system. is the transmission time interval the number of bandwidth resource blocks within, according to the service level agreement.

[0140] The specific steps for network slice weight sorting are as follows:

[0141] S1. Initialize an empty network slice queue;

[0142] S2. Calculate the weight of each network slice, sort them in descending order, and add them to the network slice queue;

[0143] S3. When allocating the power services to be processed each time, give priority to selecting the network slice with the largest weight;

[0144] S4. After the power services to be processed are processed, update the weights of each network slice according to Equation (15), and re-sort them in descending order.

[0145] In the embodiment of the present invention, the service layer mainly calculates the priority of the power services to be processed. The service priority calculation mainly performs dynamic priority sorting according to service requirements such as delay and bandwidth within each TTI. Requests with high priority preferentially select network links and slice resources.

[0146] Since the resource requirements and delays of different service requests are different, high priorities are set for control services with large resource requirements.

[0147] Define the service resource requirement of the virtual node in the service layer as:

[0148] (16)

[0149] In the formula, is the service resource requirement of the virtual node, is the virtual node the cpu computing power required for the power service request to be processed, is the th virtual node, is the virtual link the remaining bandwidth resource, represents the virtual link set, represents the virtual node Set

[0150] Define as the waiting time of the power service request to be processed and the longest waiting time The ratio is:

[0151] (17)

[0152] Define the priority of the power service to be processed as:

[0153] (18)

[0154] In the formula, is the duration of the power service request to be processed.

[0155] When is greater than , it means that the delay requirement of the current priority service cannot be met, so the priority is increased.

[0156] It is represented as a binary value and defined as:

[0157] (19)

[0158] When is 0, is greater than , if it is a control service, its priority is raised to the highest to ensure that the delay requirement of the service is met. If the request time of the collection service is too long, its priority is temporarily increased and the request is sent first.

[0159] In the embodiments of the present invention, when the service layer priority is 0, it means the highest priority, followed by the largest value, and finally the smallest value. If the request time of the collection service is too long, its priority is temporarily increased and the request is sent first.

[0160] The specific steps for sorting the service priorities are as follows:

[0161] A1. Initialize an empty queue of power services to be processed;

[0162] A2. In each iteration, select the service with priority of 0 and add it to the queue of power services to be processed for processing. If there is no power service to be processed with priority 0, then select the largest power service to be processed, and finally the smallest power service to be processed, and remove it from the queue of power services to be processed after processing;

[0163] A3. After each iteration, update the priority of the power service to be processed according to Equation (18). , and then return to A2, considering the current resource allocation situation and other possible influencing factors.

[0164] A4. Termination condition: When all the power services to be processed have been processed, the iteration ends.

[0165] In the embodiment of the present invention, the network layer considers the load balancing of the physical network and the virtual link delay factor, constructs a reliability penalty function based on the virtual link delay and the variance of the physical link load ratio, and outputs an optimal network slice link configuration result. In the embodiment of the present invention, the penalty function is inversely proportional to the link performance. The larger the penalty function value, the worse the link performance. The virtual link delay mainly includes four parts: transmission delay, queuing delay, propagation delay, and processing delay.

[0166] Among them, the transmission delay is the ratio of the data packet size to its transmission rate in the link; the propagation delay is the ratio of the data packet size to the link bandwidth capacity; the queuing delay depends on the link bandwidth utilization and the propagation delay; the processing delay mainly depends on the node CPU resource utilization and the total processing delay of each data packet at this node.

[0167] Define the reliability penalty function as follows:

[0168] (20)

[0169] Among them, represents the reliability penalty function, is the weight value, is the virtual link delay, that is:

[0170] (21)

[0171] represents the variance of the physical link load ratio,

[0172] (22)

[0173] In the formula, is the physical link in the network slice, is the set of physical links, is a binary value, representing the mapping index variable, that is =1 indicates that the physical link is successfully mapped to the virtual link, =0 indicates that the physical link is not successfully mapped. is the physical link The load ratio represents the average value of the load ratios of all physical links.

[0174] The physical link load ratio is defined as:

[0175] (23)

[0176] In the formula, is the physical link The total bandwidth capacity of is the physical link The remaining bandwidth resource of

[0177] During the network slicing service process, the propagation delay must meet the maximum allowable delay of the service request. Therefore, the propagation delay constraint can be expressed as:

[0178] (24)

[0179] is the maximum allowable delay of the service request.

[0180] The link bandwidth limit is expressed as:

[0181] (25)

[0182] In the formula represents the virtual node The bandwidth requirement of the service request is a binary value represents the virtual node of the service request Successfully mapped to the physical link.

[0183] In the embodiment of the present invention, by introducing the virtual link propagation delay constraint and the physical link load balancing constraint, on the premise of ensuring network reliability, the optimal link with a smaller delay is selected as much as possible.

[0184] Based on the above constraints, the optimal network slicing link configuration problem can be expressed as:

[0185] (26)

[0186] The embodiment of the present invention uses a genetic algorithm to solve the optimal network slicing link configuration problem. The specific steps are as follows:

[0187] B1. Let the fitness be the reciprocal of the penalty function Calculate the fitness of each physical link Then the link selection probability is ; is the randomly generated population length.

[0188] B2. Selection: Use the roulette wheel selection method for selection. The larger it is, the greater the probability that the link is selected.

[0189] B3. Crossover: Use the single-point crossover method for crossover operation.

[0190] B4. Mutation: Use the inversion mutation operator for mutation operation. Randomly select two points in the parent generation, and then invert the content between the two points.

[0191] B5. Finally, obtain the optimal link configuration result of the network slice.

[0192] Based on the above network slice dynamic resource scheduling system with multi-service differentiation, another embodiment of the present invention provides a network slice dynamic resource scheduling method. Refer to Figure 1 , including:

[0193] The slicing layer calculates the network slice resource reservation ratio according to the power service type, service delay, and service arrival rate of the service layer, and calculates and sorts the weights of the network slices.

[0194] The service layer calculates and sorts the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request delay.

[0195] Allocate the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed.

[0196] The network layer performs network slice link configuration with the physical link load balancing as the constraint and the minimum virtual link delay of the network slice as the goal, and outputs the optimal link configuration result of the network slice to complete the resource allocation for the power services to be processed.

[0197] In the embodiment of the present invention, allocating the service to be processed to the network slice according to the network slice weight and the priority of the service to be processed means that the network slice with the largest weight is preferentially selected each time the power service to be processed is allocated. It should be noted that after the power service to be processed is processed, the weights of each network slice are updated, and the descending order sorting is performed again; and the priorities of the power services to be processed are recalculated, and then the power services to be processed are added to the queue according to the priorities.

[0198] It should be noted that considering the existence of random burst control services in the slice layer, a resource reservation method based on random network calculus is adopted to dynamically reserve resources for random burst control services. The reserved resources are divided into two parts: static reserved resources and dynamic reserved resources. The static reserved resources are used by fixed-cycle control services, and the dynamic reserved resources are used by random burst control services. If there are no random burst control services, the dynamic reserved resources are used by collection services, and the remaining resources in the network are supplied to collection services.

[0199] It should be noted that when the service priority is 0, it indicates the highest priority, followed by the largest priority value, and finally the smallest priority value. If the request time of the collection service is too long, its priority is temporarily increased to send the request first.

[0200] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0201] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0202] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0203] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the steps in the process Figure 1 one process or a plurality of processes and / or blocks Figure 1 steps for the functions specified in one block or a plurality of blocks.

[0204] 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 them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention. Any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A network slice dynamic scheduling system based on power service differences, characterized in that, Including: A slice layer, configured to calculate the network slice resource reservation ratio according to the power service type, service latency, and service arrival rate of the service layer, and calculate and sort the weights of the network slices; A service layer, configured to calculate and sort the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request latency, and allocate the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed; A network layer, configured to perform network slice link configuration with physical link load balancing as a constraint and the minimum virtual link latency of the network slice as a goal, output the optimal network slice link configuration result, and complete the resource allocation for the power services to be processed. The specific implementation process is as follows: Determine the optimal network slice link configuration problem, expressed as: ； ； ； And the following constraints need to be satisfied: ； ； Among them, represents minimization, is the reliability penalty function, is the weight value, is the virtual link delay, is the variance of the physical link load ratio, is the physical link in the network slice, is the set of physical links, is the virtual link, represents the set of virtual links, represents the virtual node set, is the transmission delay, is the propagation delay, is the queuing delay, is the processing delay, is a binary value, representing the mapping metric variable, when it means the physical link is successfully mapped to the virtual link, when it means the physical link mapping is unsuccessful, is the physical link load ratio, represents the average value of all physical link load ratios, is the maximum allowable delay of the service request, is the virtual node bandwidth requirement of the service request, is a binary value, represents the virtual node of the service request is successfully mapped to the physical link, is the remaining bandwidth resource of the physical link ; Use a genetic algorithm to solve the optimal network slice link configuration problem to obtain the optimal network slice link configuration result.

2. The network slice dynamic scheduling system based on power service differences according to claim 1, wherein The power service types include control services and acquisition services; The control services include fixed-period control services and random burst control services.

3. A network slice dynamic scheduling system based on power service differences according to claim 2, characterized in that, Specifically, the slice layer is used for Dividing the reserved resources of the slice layer into static reserved resources and dynamic reserved resources; The static reserved resources are used by the fixed-period control services, and the dynamic reserved resources are used by the random burst control services. If there are no random burst control services, the dynamic reserved resources are used by the acquisition services, and the remaining resources other than the static reserved resources and the dynamic reserved resources are used by the acquisition services.

4. The network slice dynamic scheduling system based on power service differences according to claim 3, wherein The slice layer calculates the network slice resource reservation ratio according to the power service type, service latency, and service arrival rate of the service layer, expressed as: ； ； Among them, is the reservation ratio of static reserved resources, is the reservation ratio of dynamic reserved resources, is the average rate of data arrival for fixed-period control services, represents the maximum delay that fixed-period control service data packets can tolerate, is the service rate of the scheduling system, is the maximum error rate that fixed-period control service data packets can tolerate, is the maximum error rate that random burst control service data packets can tolerate, represents the slope of the random arrival curve of random burst control services; Expressed as: ； Expressed as: ； Expressed as: ； Among them, and are both normal constants related to the randomness of the service process, represents a variable in the boundary function of the random arrival curve, is a free parameter greater than 0 related to the randomness of the arrival of random burst control type services, is a parameter related to the random service curve of random burst control type services, represents time.

5. The network slice dynamic scheduling system based on power service differences according to claim 4, wherein The slice layer calculates and sorts the weights of the network slices. Specifically, Initialize an empty network slice queue; Calculate the weights of each network slice, sort them in descending order, and add them to the network slice queue; When allocating the power services to be processed each time, preferentially select the network slice with the largest weight; After the power services to be processed are processed, update the weights of each network slice and re-sort them in descending order; Calculate the weights of the network slices in the following way: ； Wherein: represents the transmission time interval of the network slice weight, is the weighting coefficient of the network slice, and each network slice has a different weighting coefficient, represents the utilization rate of the bandwidth resource block from the start to the transmission time interval within, is the utilization rate of the bandwidth resource block from the start to the transmission time interval within.

6. The network slice dynamic scheduling system based on power service differences according to claim 5, wherein The service layer calculates and sorts the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request latency. Specifically, Calculate the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request latency; Initialize an empty queue of power services to be processed; In each iteration, select the power service to be processed with a priority of 0 and add it to the queue of power services to be processed for processing. If there is no power service to be processed with a priority of 0, then select the power service to be processed with the largest priority, and so on until the power service to be processed with the smallest priority, and remove it from the queue of power services to be processed after processing; After each step of iteration, recalculate the priorities of the power services to be processed, and then add them to the queue of power services to be processed according to the priorities; When all the power services to be processed are completed, the iteration ends; Calculate the priorities of the power services to be processed in the following way: ； Among them, is the priority of the power service to be processed, is the service resource demand of the virtual node, is the duration of the power service request to be processed, is the ratio of the waiting time to the maximum waiting time of the power service request to be processed, is a binary value; Expressed as: ； Among them, is a virtual node The CPU computing power required for the power service request to be processed, is the th virtual node, is a virtual link The remaining bandwidth resource; the virtual node is an intelligent terminal, a base station, and a control center in the distribution network; Expressed as: ； Among them, is the waiting time for the power service request to be processed, is the maximum waiting time for the power service request to be processed.

7. A network slice dynamic scheduling method based on power service differences, characterized in that, Implemented by using the network slice dynamic scheduling system based on power service differences according to any one of claims 1 to 6, the method includes: The slice layer calculates the network slice resource reservation ratio according to the power service type, service delay and service arrival rate of the service layer, and calculates and sorts the weights of the network slices; The service layer calculates and sorts the priorities of the power services to be processed according to the resource requirements of different power service requests and the service request delay; Allocate the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed; The network layer configures the network slice links with the physical link load balancing as the constraint and the minimum virtual link delay of the network slice as the goal, and outputs the optimal link configuration result of the network slice to complete the resource allocation for the power services to be processed.

8. A network slice dynamic scheduling method based on power service differences according to claim 7, characterized in that The step of allocating the power services to be processed to the network slices according to the network slice weights and the priorities of the power services to be processed includes: When allocating the power services to be processed each time, preferentially select the power services to be processed with a priority of 0 to join the queue of power services to be processed. If there are no power services to be processed with a priority of 0, then select the power service to be processed with the highest priority, and so on until the power service to be processed with the lowest priority; Allocate the selected power service to be processed to the network slice with the highest weight.

9. A network slice dynamic scheduling method based on power service differences according to claim 8, characterized in that After the power services to be processed are allocated and processed, update the weights of each network slice and re-sort them; And recalculate the priorities of the power services to be processed, and then join the queue of power services to be processed according to the priorities.

Images