Intelligent fire-fighting parallelization service function chain deployment method based on energy perception and time delay guarantee

Abstract

The invention provides an intelligent fire-fighting parallelization service function chain deployment method based on energy perception and time delay guarantee. The method comprises the steps that a traditional service function chain is converted into a parallelization service function chain; under the condition that the end-to-end delay requirement is ensured to be met, establishing a target function which aims at minimizing the energy consumption of deploying the parallel service function chain, and establishing an integer linear programming model according to the target function; and solving the integer linear programming model by adopting a heuristic algorithm considering edge node computing resources, link bandwidth resources, a node starting state and a link starting state to obtain an optimal deployment scheme. According to the method, the intelligent fire-fighting service request is abstracted into the service function chain, the service request processing problem is converted into the service function chain deployment problem, and the deployment strategy and the routing strategy of the parallel SFC are dynamically adjusted under the condition that the low-delay requirement is met, so that the average total energy consumption of SFC deployment is minimized, the SFC request acceptance rate is remarkably improved, and the service performance of the service function chain is improved. The method is suitable for intelligent fire-fighting services with high requirements on time delay sensitivity and reliability.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method for deploying a smart firefighting parallel service function chain with energy perception and delay guarantee. Background Art

[0002] In the field of smart fire protection, as 5G mobile communication technology promotes the development of IoT applications, edge computing networks based on network function virtualization (NFV) technology have brought opportunities for smart fire protection. Network function virtualization (NFV) technology separates network functions from dedicated hardware and runs them as virtual network functions (VNFs) on edge servers, thereby enhancing the ability to provide IoT services in edge computing networks. Smart fire protection applications can be implemented through a series of ordered VNFs, called service function chains (SFCs). However, a series of technical problems have arisen during the deployment of service function chains (SFCs).

[0003] (1) In terms of latency, the computing power of IoT devices in traditional firefighting systems is limited. If the computing tasks are offloaded to cloud servers, the long data transmission distance will cause serious transmission delays. In SFC, when the virtual network function (VNF) is run sequentially, each VNF needs to process data in turn, which cannot fully utilize the parallel processing capabilities of edge computing. This results in limited system response speed in time-critical scenarios such as early warning or emergency rescue, making it difficult to meet the timeliness requirements of firefighting applications, thereby affecting rescue efficiency and effectiveness. For example, in the fire warning stage, the data collected by the sensor needs to be quickly analyzed and processed to determine the fire risk, but the sequentially executed SFC may miss the best warning time due to processing delays, seriously affecting the timeliness and effectiveness of fire rescue.

[0004] (2) Compared with cloud servers, edge server resources have inherent disadvantages in computing power and storage capacity. As the application scenarios of smart fire protection continue to expand and the amount of data and computing tasks continue to grow, the large number of SFC deployment requirements has made the resources of edge servers stretched. For example, when processing monitoring data, analysis tasks, and communication control with fire protection equipment in multiple fire protection areas at the same time, limited resources are difficult to support the smooth operation of all SFCs, which may cause some tasks to be delayed or unable to execute.

[0005] (3) In terms of energy consumption, different VNF ​​deployment locations and routing strategies will lead to differences in the working status of network equipment. Unreasonable deployment may cause data to pass through high-energy consumption paths or run on low-efficiency nodes during transmission and processing, increasing unnecessary energy consumption. In the long-term operation process, high energy consumption not only increases operating costs, but may also affect the stability and reliability of the system due to energy supply issues. For example, in some remote areas, the energy supply of fire-fighting facilities, drone terminals, and firefighters' backpack terminals is relatively tight. The high-energy consumption SFC deployment may cause the equipment to be unable to operate continuously and stably, reducing the safety of the fire-fighting system.

[0006] In summary, in the SFC deployment of smart fire protection, problems such as high latency, limited edge server resources and high energy consumption need to be solved urgently, which also shows the importance and necessity of the energy-aware and latency-guaranteed smart fire protection parallel service function chain deployment method proposed in the present invention. Summary of the invention

[0007] In order to solve the problems of high sequential SFC deployment delay, limited edge server computing resources and high network energy consumption, the present invention proposes an energy-aware and delay-guaranteed smart firefighting parallel service function chain deployment method.

[0008] In order to achieve these purposes and other advantages of the present invention, the present invention provides a method for deploying a smart firefighting parallel service function chain with energy perception and delay guarantee, comprising: Step 1: Build an edge computing network to provide computing resources and link bandwidth resources required to deploy the functional service chain; Step 2: Based on the fire service related requirements, the smart fire terminal issues a function service chain call request (i.e., SFC request); the function service chain call request includes: source node, destination node, delay requirement, bandwidth requirement, and parallel unit; Step 3: according to the function service chain call request, establish an objective function with the goal of minimizing the energy consumption of deploying the parallel service function chain, and establish an integer linear programming model according to the objective function; Step 4: Use a heuristic algorithm that considers computing resources, link bandwidth resources, node startup status, and link startup status in the edge computing network to establish an integer linear programming model to solve the objective function and obtain the optimal deployment plan.

[0009] The deployment method of the present invention is suitable for intelligent fire service networks that are sensitive to latency and have high reliability requirements. It can dynamically adjust deployment and routing strategies while meeting low latency requirements, minimize energy consumption and improve SFC request acceptance rates. The deployment method particularly considers the energy consumption of the parallelized service function chain. The final deployment solution minimizes energy consumption while meeting business needs, especially in the case of large-scale deployment. It can save a lot of energy and significantly improve service endurance on power-constrained intelligent fire terminals (such as drone terminals, firefighter backpack terminals, and remote fire equipment terminals). This is of great significance for emergency situations, long-term stable operation, and reduced operating costs of intelligent fire systems.

[0010] Preferably, the establishment of the integer linear programming model comprises the following steps: 31) Establish system model according to SFC request; 32) Calculating computing resource costs and bandwidth resource costs in the network according to the system model; 33) Calculating the total energy consumption of the network based on the computing resource cost, bandwidth resource cost, and resource conditions of nodes and links in the network; 34) constructing an objective function based on the energy consumption of the network; 35) Complete the establishment of the integer linear programming model.

[0011] Preferably, the system model of step 31) includes an edge computing network and a parallelized service function chain.

[0012] Preferably, the resource conditions of the nodes and links in the network in step 33) include: computing resources of the nodes, bandwidth resources of the links, startup status of the nodes, and startup status of the links.

[0013] Preferably, the total energy consumption of the network in step 33) is: ; Where N represents the physical node in the network, represents the energy consumption of physical node k, L represents the link between nodes, represents the energy consumption of link e; The energy consumption of a physical node is calculated using the following formula: ; In the formula, represents the startup energy consumption of physical node k, represents the peak energy consumption of physical node k, Indicates the computing resources used by physical node k, is the total computing resource of physical node k; Calculate the energy consumption of the link: ; In the formula, in the formula, represents the startup energy consumption of link e, represents the peak energy consumption of link e, Indicates the bandwidth resources used by link e, is the total bandwidth resource of link e; is the total bandwidth resource of link e, which is the maximum bandwidth that the link can provide; Preferably, the resource constraints include: (1) The VNF can be deployed to the node only when the remaining computing resources of the node meet the resource requirements of the VNF: ; (2) A link can be selected only when the remaining bandwidth resources of the link meet the resource requirements of the SFC: ; (3) The end-to-end delay including the selected node must satisfy the end-to-end delay constraint of the parallelized SFC: ; In the formula, The end-to-end delay for deploying parallel SFC consists of processing delay and propagation delay. It is the latency requirement of SFC.

[0014] Preferably, in step 34), the objective function is: ; In the formula, when node k is in the startup state, 1, otherwise 0, when link e is in the startup state is 1 if the value is true, otherwise it is 0.

[0015] Preferably, in step 4, the process of solving the heuristic algorithm considering computing resources, link bandwidth resources, node startup status and link startup status in the edge computing network is as follows: 41) Obtain the edge computing network model, the parallelized SFC request set, and the VNF resource-delay dependency table; 42) Sort the parallel SFC request set in a non-decreasing order according to the size of computing resources required by the parallel SFC; 43) All parallelizable VNFs in a parallelized SFC are regarded as a parallel unit (PU), and each non-parallelizable VNF is regarded as a PU. Each PU of the parallelized SFC adjusts the resource allocation of other VNFs according to the maximum VNF delay in the PU and the VNF resource-delay dependency table; 44) Selecting the physical node with the smallest energy consumption increment on the path with the minimum end-to-end delay for each PU of each parallelized SFC; 45) Under the condition that the end-to-end delay requirement of the parallelized SFC request is met, a sub-path with the smallest energy consumption increment between nodes from the source point to the destination point is selected for the selected physical node, and then each sub-path is combined into an end-to-end path; 46) Finally, the optimal deployment plan for the SFC request is output and mapped to the edge computing network, and the network deployment is updated to show the resource status after the SFC is completed.

[0016] The present invention has at least the following beneficial effects: 1. The traditional sequential SFC deployment method will bring significant processing delays when processing smart firefighting tasks because VNFs need to run sequentially. Firefighting applications are highly sensitive to delays. For example, timely responses are required during early warning or emergency rescue. This high delay may affect the performance of the firefighting system and the rescue effect. This invention converts traditional SFC into parallel SFC and uses parallel processing to effectively reduce processing delays, improve the response speed of the system, and meet the strict requirements of smart firefighting for low latency.

[0017] 2. Compared with cloud servers, edge servers have limited resources and are difficult to support the SFC deployment requirements of all smart fire protection applications. This invention fully considers the limitations of edge node computing resources and link bandwidth resources during the deployment process, and reasonably allocates resources by establishing an integer linear programming model and adopting a heuristic algorithm to ensure the optimal service function chain deployment under limited resource conditions, improve resource utilization efficiency, and enable edge servers to better support smart fire protection applications.

[0018] 3. Different VNF ​​deployment and routing methods will lead to different energy consumption. In smart fire protection applications, it is necessary to reduce energy consumption while ensuring system performance. The present invention aims to minimize the energy consumption of deploying parallel service function chains. By accurately calculating the energy consumption of nodes and links in the network and constructing corresponding objective functions and models for solving, the present invention minimizes network energy consumption while meeting the end-to-end delay requirements, reduces the operating cost of the smart fire protection system, and improves the energy efficiency of the system.

[0019] 4. To apply the model to smart fire protection, we need to face the difficulties of modeling the complexity of the problem and the high computational complexity.

[0020] The present invention abstracts the smart fire protection system into an edge computing network and a parallelized service function chain, and represents the edge computing network with an undirected graph, which simplifies the description of the network topology structure; for the service function chain, it is divided into multiple parallel units (PUs), each PU is composed of parallel virtual network functions (VNFs), which reduces the complexity of the problem; at the same time, it focuses on key factors such as edge node computing resources, link bandwidth resources, node startup status and link startup status, and integrates these factors into the constraints and objective function of the integer linear programming model to avoid processing too many complex factors and make the model construction more focused and effective.

[0021] At the same time, a heuristic algorithm is used to solve the integer linear programming model, avoiding the high computational complexity of traditional exact algorithms on large-scale problems; and the service function chain deployment problem is decomposed into multiple stages, such as resource allocation adjustment, node selection, path determination, etc., each stage is solved and optimized separately, reducing the complexity of the overall problem and improving the solution efficiency; the heuristic algorithm has a faster solution speed and can give a feasible deployment plan in a short time, so it can meet the real-time requirements of smart fire protection.

[0022] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a method for deploying a parallelized service function chain of intelligent fire protection with energy perception and delay guarantee according to the present invention; Figure 2 It is a schematic diagram of a smart fire fighting SFC deployment of the present invention; Figure 3 The performance comparison diagram of the method proposed in the present invention with the traditional sequential SFC when deploying different numbers of SFCs is shown. DETAILED DESCRIPTION

[0024] The present invention is further described in detail below in conjunction with embodiments so that those skilled in the art can implement the invention with reference to the description.

[0025] like Figure 2As shown, in one example of the present invention, the smart fire SFC architecture is divided into three layers: the fire IoT layer, the edge layer and the SFC layer. The fire IoT layer collects the physiological data and environmental information of firefighters through various sensor devices. The edge layer includes edge servers, switches and wireless access points, etc. The SFC layer includes environmental monitoring and real-time monitoring of the command center, etc. It has data preprocessing, image analysis, time series data analysis, data encryption, image mapping, firefighter body monitoring, etc. According to the SFC request of the application, the VNF is mapped to the edge layer, and the edge server is used to effectively process the data collected by the fire IoT layer, thereby achieving flexible resource allocation and real-time response to meet the needs of smart fire applications.

[0026] like Figure 1 As shown, a method for deploying a parallelized service function chain of intelligent fire protection with energy perception and delay guarantee of the present invention includes: Step 1: Build an edge computing network to provide computing resources and link bandwidth resources required to deploy the functional service chain; For example, the edge computing network can be modeled as an undirected graph G=(S, L), where S represents the set of all edge servers in the network, and L represents the set of links connecting these edge servers; each edge server (k∈S) is connected to other servers through switches in the network, and its computing power is represented by Ck; for any link e∈L, its characteristics include delay De and bandwidth In addition, the startup status of the node and the startup status of the link will also be recorded, which will affect the subsequent calculation of the energy consumption of the node and the link. In an example, the edge computing network consists of 25 nodes and 40 links. The computing resources of the nodes are [100, 250], the bandwidth resources of the links are [5, 10] Gbps, there are 10 types of VNFs, the length of each SFC request is [4, 8], and the end-to-end delay requirement is [80, 120] ms.

[0027] Step 2: Based on the fire service related requirements, the smart fire terminal issues a function service chain call request (i.e., SFC request); the function service chain call request includes: source node, destination node, delay requirement, bandwidth requirement, and parallel unit; For example, the parallelized SFC request set contains multiple service function chain requests, each of which has its own unique requirements. The parallelized SFC consists of parallel units (PUs), each PU consists of all parallelizable VNFs in the SFC or a single non-parallelizable VNF. Specifically, for each SFC request, it contains the source node, destination node, delay requirement, bandwidth requirement and parallel unit (PU). The parallelized SFC is an ordered set of logically connected PUs, which can be expressed as ; ∈S are the source node and destination node of the i-th SFC request respectively; and They are latency and bandwidth requirements, respectively; represents the ordered set of PUs of parallelized SFC i.

[0028] The above information reflects the requirements of different smart fire service functions, such as collecting data from different sensors (source nodes) and sending the processing results to the designated processing or storage nodes (destination nodes), and there are clear requirements for the latency and bandwidth of data processing, and the service functions can be divided into multiple parallel units to meet different processing tasks. These requests come from different business needs of the smart fire system. For example, one request may be to collect temperature, smoke and gas data at the fire scene and analyze and process it, and another request may be to transmit the processing results to the fire command center or trigger the corresponding fire equipment.

[0029] Step 3: according to the function service chain call request, establish an objective function with the goal of minimizing the energy consumption of deploying the parallel service function chain, and establish an integer linear programming model according to the objective function; Step 4: Use a heuristic algorithm that considers computing resources, link bandwidth resources, node startup status, and link startup status in the edge computing network to establish an integer linear programming model to solve the objective function and obtain the optimal deployment plan.

[0030] The deployment method of the present invention is suitable for intelligent fire service networks that are sensitive to latency and have high reliability requirements. It can dynamically adjust deployment and routing strategies while meeting low latency requirements, minimize energy consumption and improve SFC request acceptance rates. The deployment method particularly considers the energy consumption of the parallelized service function chain. The final deployment solution minimizes energy consumption while meeting business needs, especially in the case of large-scale deployment. It can save a lot of energy and significantly improve service endurance on power-constrained intelligent fire terminals (such as drone terminals, firefighter backpack terminals, and remote fire equipment terminals). This is of great significance for emergency situations, long-term stable operation, and reduced operating costs of intelligent fire systems.

[0031] Furthermore, the establishment of the integer linear programming model includes the following steps: 31) Establish system model according to SFC request; 32) Calculating computing resource costs and bandwidth resource costs in the network according to the system model; 33) Calculating the total energy consumption of the network based on the computing resource cost, bandwidth resource cost, and resource conditions of nodes and links in the network; 34) constructing an objective function based on the energy consumption of the network; 35) Complete the establishment of the integer linear programming model.

[0032] Furthermore, the system model of step 31) includes an edge computing network and a parallelized service function chain. On the one hand, the edge computing network provides the resources and constraints required for the deployment of the service function chain, and its resources and performance characteristics determine how the service function chain can be deployed and operated; on the other hand, the parallelized service function chain is a specific logical organization for realizing the functions of the smart fire protection system. By parallelizing the VNF and deploying it in the edge computing network, functions such as data collection, processing, transmission and decision-making can be realized, and the performance requirements of the system, such as low latency and low energy consumption, must be met. Such a system model comprehensively considers the organization of the network architecture and service functions, and provides a better deployment and operation mode for the smart fire protection system by optimizing the relationship between the two.

[0033] Furthermore, the resource conditions of the nodes and links in the network in step 33) include: computing resources of the nodes, bandwidth resources of the links, startup status of the nodes, and startup status of the links.

[0034] Furthermore, the total energy consumption of the network in step 33) is: ; Where N represents the physical node in the network, represents the energy consumption of physical node k, L represents the link between nodes, represents the energy consumption of link e; The energy consumption of a physical node is calculated using the following formula: ; In the formula, represents the startup energy consumption of physical node k, represents the peak energy consumption of physical node k, Indicates the computing resources used by physical node k, is the total computing resource of physical node k; This formula is used to calculate the energy consumption of physical nodes. It takes into account the startup energy consumption of the node and, based on this, calculates the additional energy consumption relative to the peak energy consumption according to the ratio of the computing resources used by the node to the total computing resources. relatively When the node is running at full capacity ( equal ), its energy consumption is close to the peak energy consumption. In this way, the energy consumption of the node under different load conditions can be calculated more accurately, making the energy consumption calculation more consistent with the actual operating status.

[0035] Calculate the energy consumption of the link: ; In the formula, in the formula, represents the startup energy consumption of link e, represents the peak energy consumption of link e, Indicates the bandwidth resources used by link e, is the total bandwidth resource of link e; is the total bandwidth resource of link e, which is the maximum bandwidth that the link can provide; This formula is used to calculate the energy consumption of the link. It takes into account the startup energy consumption of the link and calculates the additional energy consumption relative to the peak energy consumption based on the ratio of the bandwidth resources used by the link to the total bandwidth resources. relatively When the link is fully utilized ( = ), its energy consumption is close to the peak energy consumption. This formula can accurately reflect the energy consumption of the link under different data transmission volumes, which helps to optimize the energy consumption of the link when considering network performance and deploying service function chains.

[0036] These formulas comprehensively consider the energy consumption of physical nodes and links under different usage conditions. When building a deployment plan for the service function chain, the total energy consumption is calculated. The calculation and analysis can minimize the energy consumption of the network while meeting functional requirements, realize low-energy operation of the system, and take into account the resource usage of nodes and links, ensuring the rational use of resources and performance balance, which is of great significance for optimizing the network deployment of smart fire protection systems.

[0037] Preferably, the resource constraints include: (1) The VNF can be deployed to the node only when the remaining computing resources of the node meet the resource requirements of the VNF: ; This constraint ensures that when deploying virtual network functions (VNFs) to physical nodes, the computing capacity of the nodes will not be exceeded; it guarantees the resource availability of the nodes and avoids node overload due to excessive allocation of resources to VNFs, which affects the performance and stability of the nodes. For example, in a smart fire protection system, if the total computing resources of a node are is 100 units, and the computing resources currently used The resource requirement of a VNF is 80 units. Only when the resource requirement of a VNF is less than or equal to 20 units can it be deployed to the node. Such constraint design helps to reasonably allocate computing resources, enable each node to play its role within its capability, and ensure the normal operation of the system.

[0038] (2) A link can be selected only when the remaining bandwidth resources of the link meet the resource requirements of the SFC: ; This constraint ensures that when deploying a service function chain (SFC), the bandwidth resources of the link are sufficient to support data transmission. When selecting a link to transmit data, the used bandwidth cannot exceed the total bandwidth of the link. For example, for a link, its total bandwidth If the current bandwidth is 50Mbps, If the bandwidth requirement of the SFC is 30Mbps, then the link can be used for data transmission only when the bandwidth requirement of the SFC is less than or equal to 20Mbps, thus preventing link congestion and ensuring smooth data transmission.

[0039] (3) The end-to-end delay including the selected node must satisfy the end-to-end delay constraint of the parallelized SFC: ; In the formula, The end-to-end delay for deploying parallel SFC consists of processing delay and propagation delay. It is the latency requirement of SFC.

[0040] This constraint stipulates that the end-to-end latency (including processing latency and propagation latency) of deploying parallelized SFC must meet the latency requirements of SFC; in time-sensitive systems such as smart fire protection, this is crucial to ensure that the entire process from data generation to the final processing result output meets the system's response time requirements. For example, for the fire alarm function, the time from the sensor detecting the data, through a series of VNF processing, to the final alarm must be within the specified time. To ensure that firefighters can receive alarm information in time and avoid missing the best rescue opportunity due to delayed processing.

[0041] Through the above-mentioned resource constraints on nodes and links, we can avoid over-allocation and unreasonable use of resources, solve the problem of unbalanced and wasteful resource allocation, for example, prevent multiple VNFs from being concentrated on a few nodes, resulting in the exhaustion of resources in these nodes and the idleness of other nodes, and avoid the bottleneck problem of data transmission due to bandwidth exhaustion in the link, thereby improving the utilization of the entire network resources and the stability of the system. The constraint on end-to-end delay guarantees the performance of the system, solves the problem of system performance degradation due to excessive processing and transmission delays, and ensures that the system completes data processing and response within the specified time in time-sensitive smart consumption application scenarios, improves the system's response speed and service quality, and avoids system performance deterioration and functional failure due to delays.

[0042] Furthermore, in step 34), the objective function is: ; In the formula, when node k is in the startup state, 1, otherwise 0, when link e is in the startup state is 1 if the value is true, otherwise it is 0.

[0043] The objective function aims to minimize the total energy consumption of the entire network. It is constructed by considering the energy consumption of nodes and links in the edge computing network, and finding a deployment scheme of a service function chain (SFC) so that the network can minimize energy consumption while meeting various service function requirements.

[0044] Furthermore, in step 4, the process of solving the heuristic algorithm considering computing resources, link bandwidth resources, node startup status and link startup status in the edge computing network is as follows: 41) Obtain the edge computing network model, the parallelized SFC request set, and the VNF resource-delay dependency table; Specifically, the edge computing network can be modeled as an undirected graph G=(S, L), where S represents the set of all edge servers in the network, and L represents the set of links connecting these edge servers. (k∈S) is connected to other servers through switches in the network, and its computing power is represented by Ck; for any link e∈L, its characteristics include delay De and bandwidth ; In addition, the startup status of the node and the startup status of the link will also be recorded, which will affect the subsequent calculation of the energy consumption of the node and the link.

[0045] The parallelized SFC request set contains multiple service function chain requests, each of which has its own unique requirements. The parallelized SFC consists of parallel units (PUs), each PU consists of all parallelizable VNFs in the SFC or a single non-parallelizable VNF. Specifically, for each SFC request, it contains the source node, destination node, delay requirement, bandwidth requirement and parallel unit (PU). The parallelized SFC is an ordered set of logically connected PUs, which can be expressed as ; ∈S are the source node and destination node of the i-th SFC request respectively; and They are latency and bandwidth requirements, respectively; The PUs of parallelized SFC i are ordered sets. This information reflects the requirements of different smart fire service functions, such as collecting data from different sensors (source nodes) and sending the processing results to designated processing or storage nodes (destination nodes). At the same time, there are clear requirements for the latency and bandwidth of data processing, and the service functions can be divided into multiple parallel units to meet different processing tasks. These requests come from different business needs of the smart fire protection system. For example, one request may be to collect temperature, smoke and gas data at the fire scene and analyze and process them, and another request may be to transmit the processing results to the fire command center or trigger the corresponding fire protection equipment.

[0046] Get the VNF resource-delay dependency table: The relationship table stores the resource-delay relationship of each VNF, such as [maximum resource, minimum resource, minimum delay, maximum delay]; for each VNF, the required computing resources and the delay under different resource allocations are listed. For example, some VNFs may significantly reduce the delay when more computing resources are allocated, while other VNFs may not significantly reduce the delay after the resources are increased to a certain level. This table provides a basis for subsequent resource allocation to optimize the performance of the service function chain.

[0047] 42) Sort the parallel SFC request set in a non-decreasing order according to the size of computing resources required by the parallel SFC; Specifically, for each parallelized SFC request, the total amount of computing resources required needs to be evaluated. For example, the resource requirements of the VNFs in each parallel unit (PU) contained in it are accumulated to determine the computing resource requirements of the entire SFC request. For example, for each VNF in the PU, the required computing resources are calculated according to the VNF resource-delay dependency table, and then the computing resource requirements of all PUs are added to obtain the total computing resource requirements.

[0048] All parallelized SFC requests are sorted in non-decreasing order according to the total computing resource requirements. The sorting algorithm can use a simple comparison sorting algorithm, such as bubble sort, insertion sort, or a more efficient sorting algorithm, such as quick sort or merge sort. After sorting, SFC requests with smaller resource requirements are placed in the front, and those with larger resource requirements are placed in the back. The purpose of this is to give priority to requests with smaller resource requirements during subsequent resource allocation, so as to more effectively utilize the limited resources in the network and avoid local optimal solutions for resource allocation.

[0049] 43) All parallelizable VNFs in a parallelized SFC are regarded as a parallel unit (PU), and each non-parallelizable VNF is regarded as a PU. Each PU of the parallelized SFC adjusts the resource allocation of other VNFs according to the maximum VNF delay in the PU and the VNF resource-delay dependency table; Specifically, for each parallelized SFC request, each VNF contained in it is checked. VNFs are divided into different PUs according to their parallelism. For VNFs that can be executed simultaneously, they are grouped into one PU because they can be processed in parallel to improve processing efficiency; for VNFs that cannot be executed simultaneously, they are treated as a single PU.

[0050] For example, in a smart fire protection system, VNFs that collect data from different sensors may be able to execute in parallel because there is no dependency between them and can be divided into one PU; while VNFs for data processing and data transmission may not be able to proceed at the same time and need to be separate PUs.

[0051] For each VNF in a PU, resource allocation is adjusted based on the maximum VNF latency in the PU and the VNF resource-latency dependency table. First, find the VNF with the maximum latency in the PU. Then, for other VNFs in the PU, adjust their resource allocation based on their information in the VNF resource-latency dependency table while meeting the performance requirements of the entire PU. Assuming that the VNF with the maximum latency has a higher latency, more resources may need to be allocated to other VNFs to reduce their processing latency, making the processing time of the entire PU more balanced and improving overall performance. At the same time, the computing resource constraints of the nodes must be considered. , ensuring that the adjusted resource allocation does not exceed the total computing resources of the node, avoiding resource conflicts.

[0052] 44) Selecting the physical node with the smallest energy consumption increment on the path with the minimum end-to-end delay for each PU of each parallelized SFC; Specifically, for each PU, use a path search algorithm (such as the Dijkstra algorithm or the Floyd-Warshall algorithm) to find all possible paths from its source node to its destination node. For each path, calculate its end-to-end delay D p and the energy consumption increment Δ caused by deploying PU on this path .

[0053] End-to-end delay D p The calculation includes the processing delay of each node on the path and the propagation delay of the link. The processing delay depends on the computational load of the node and the processing time of the VNF, and the propagation delay depends on the delay properties of the link. D e Energy consumption increment Δ The calculation of is based on the energy consumption formula of nodes and links, taking into account the startup status of nodes and links and resource usage. For example, for a path s i d i , calculate the total processing delay and total propagation delay of the path, and at the same time calculate the increase in energy consumption caused by the use of node computing resources and link bandwidth resources when PU is deployed on the path.

[0054] Among all possible paths, choose the end-to-end delay D p If there are multiple paths with the same minimum end-to-end delay, the path with the smallest energy consumption increment Δ The node corresponding to the smallest path is the physical node where the PU is to be deployed.

[0055] This selection method ensures that while meeting low latency requirements, energy consumption is minimized, achieving a balance between performance and energy consumption to meet the response time and energy efficiency requirements of the smart fire protection system.

[0056] 45) Under the condition that the end-to-end delay requirement of the parallelized SFC request is met, a sub-path with the smallest energy consumption increment between nodes from the source point to the destination point is selected for the selected physical node, and then each sub-path is combined into an end-to-end path; Specifically, the path consisting of the selected physical node sequence is divided into multiple sub-paths. For each sub-path, its energy consumption increment Δ is calculated. For example, for the path , which can be divided intos i and d i etc., and calculate their energy consumption increments.

[0057] The energy consumption increment calculation of the sub-path needs to consider the bandwidth usage of the link and the resource usage of the node, as well as the startup status of the node and the link, and is calculated according to the energy consumption formula.

[0058] While meeting the end-to-end latency requirements of the entire SFC request Under the premise of , select the sub-path combination that minimizes the energy consumption increment of the overall path. You can try different sub-path combinations, and find the optimal sub-path combination by evaluating the energy consumption increment and total delay of different combinations, and finally combine these sub-paths into an end-to-end path.

[0059] Doing so can further optimize the energy consumption of the path, achieve more refined energy consumption control while meeting system performance requirements, and improve the energy efficiency of the system.

[0060] 46) Finally, the optimal deployment plan for the SFC request is output and mapped to the edge computing network, and the network deployment is updated to show the resource status after the SFC is completed.

[0061] Specifically, each PU of each SFC request is mapped to the selected physical node and path to form the final deployment plan. This plan specifies which node each VNF runs on and how data is transmitted between nodes through links to ensure that the functions of the service function chain are realized. For example, for an SFC request for smart fire protection, it is clearly stated that the VNF for temperature sensor data collection is deployed on node , the VNF for smoke sensor data collection is also deployed on the node The VNFs that process the results are deployed on the nodes. And through the link Transmit to the destination node wait.

[0062] After deploying the SFC request to the network, the network resource status needs to be updated. For each node used, update its used computing resources , so that it increases the corresponding resource usage; for each link used , update its used bandwidth resources , causing it to increase the corresponding bandwidth usage.

[0063] At the same time, according to the energy consumption formula of nodes and links, the energy consumption of nodes and links is updated, taking into account the startup status of nodes and links, accurately reflecting the resource usage and energy consumption. The updated resource status will provide accurate resource information for subsequent SFC request deployment, ensuring the reasonable allocation of resources and the stable operation of the system.

[0064] Through the above detailed steps, the heuristic algorithm comprehensively considers network resources, SFC requests, VNF performance and resource relationships. While meeting the requirements of the smart fire protection system for low latency and low energy consumption, it gradually finds the optimal deployment plan for each SFC request and updates network resources, thereby achieving efficient, low-energy and reliable operation of the system, and providing a complete and practical solution for the deployment of the service function chain of smart fire protection.

[0065] Effect comparison Figure 3 The performance of the energy-aware and delay-guaranteed smart firefighting parallel service function chain deployment method (DGEAP) of the present invention was evaluated.

[0066] Non-Energy-Aware Scheme (NES): The main difference between this scheme and the proposed DGEAP scheme is that it does not consider the impact of energy consumption. It includes the following steps: 1. Obtain the edge computing network model, parallelized SFC request set, and VNF resource-delay dependency table; 2. Sort the parallel SFC request set in a non-decreasing order according to the size of computing resources required by the parallel SFC; 3. All parallelizable VNFs in a parallelized SFC are regarded as a parallel unit (PU), and each non-parallelizable VNF is regarded as a PU. Each PU of the parallelized SFC adjusts the resource allocation of other VNFs according to the maximum VNF delay in the PU and the VNF resource-delay dependency table; 4. Select a physical node on the path with the minimum end-to-end delay for each PU of each parallelized SFC; 5. To ensure that the end-to-end latency requirement of the parallelized SFC request is met, select a sub-path from the source to the destination node for the selected physical node, and then combine each sub-path into an end-to-end path; 6. Finally, the deployment plan of the SFC request is output and mapped to the edge computing network, and the resource status after the network deployment is updated to complete the SFC.

[0067] Energy-aware sequential SFC deployment scheme (EASP): The difference between this scheme and the proposed scheme is that it adopts the sequential SFC deployment method. It includes the following steps: 1. Obtain the edge computing network model and SFC request set; 2. Sort the SFC request set in a non-decreasing order according to the computing resource size required by the SFC; 3. For each VNF of each SFC, select the physical node with the smallest energy consumption increment on the path with the minimum end-to-end delay; 4. Ensure that the end-to-end delay requirement of the SFC request is met, select the sub-path with the smallest energy consumption increment between the nodes from the source point to the destination for the selected physical node, and then combine each sub-path into an end-to-end path; 5. Finally, the deployment plan requested by the SFC is output and mapped to the edge computing network, and the resource status after the network deployment is updated to complete the SFC.

[0068] Non-Energy-Aware Sequential SFC Deployment Scheme (NEASP): This scheme differs from the proposed scheme in that it uses the sequential SFC deployment method and does not consider the impact of energy consumption. It includes the following steps: 1. Obtain the edge computing network model and SFC request set; 2. Sort the SFC request set in a non-decreasing order according to the computing resource size required by the SFC; 3. Select a physical node on the path with the minimum end-to-end delay for each VNF of each SFC; 4. To ensure that the end-to-end delay requirement of the SFC request is met, select a sub-path from the source to the destination node for the selected physical node, and then combine each sub-path into an end-to-end path; 5. Finally, the deployment plan requested by the SFC is output and mapped to the edge computing network, and the resource status after the network deployment is updated to complete the SFC.

[0069] When the number of SFC requests is small, the sequential SFC deployment (EASP) and parallel SFC deployment (DGEAP) using energy-aware solutions are significantly better than non-energy-aware solutions (NEASP and NES) in terms of energy consumption. This illustrates the effectiveness of the energy-aware solution. In the case of a large number of SFC requests, the total energy consumption of the sequential SFC deployment is slightly lower than that of the parallel SFC deployment. This is because the acceptance rate of the sequential SFC deployment is lower than that of the parallel SFC deployment. In terms of average end-to-end delay and acceptance rate, the parallel SFC solution is always better than the sequential SFC deployment solution. This advantage is mainly due to the fact that the parallelized SFC deployment solution of the present invention utilizes the flexible resource allocation of the VNF resource-delay dependency model to achieve efficient resource utilization, thereby saving resources and accommodating more requests. In addition, the parallel execution of specific VNFs can effectively reduce delays, thereby improving overall performance.

[0070] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and the embodiments. They can be applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily realized.

Claims

1. A method for deploying a parallelized service function chain of smart fire protection with energy perception and delay guarantee, characterized in that: include: Step 1: Build an edge computing network to provide computing resources and link bandwidth resources required to deploy the functional service chain; Step 2: Based on the fire service related requirements, the smart fire terminal issues a function service chain call request; The function service chain call request includes: source node, destination node, delay requirement, bandwidth requirement and parallel unit; Step 3: according to the function service chain call request, establish an objective function with the goal of minimizing the energy consumption of deploying the parallel service function chain, and establish an integer linear programming model according to the objective function; Step 4: Use a heuristic algorithm that considers computing resources, link bandwidth resources, node startup status, and link startup status in the edge computing network to establish an integer linear programming model to solve the objective function and obtain the optimal deployment plan.

2. According to the energy-aware and delay-guaranteed smart firefighting parallel service function chain deployment method of claim 1, it is characterized in that: The edge computing network is modeled as an undirected graph, which contains a set of edge servers and a set of links to the edge servers.

3. According to claim 1, the method for deploying a parallelized service function chain of smart fire protection with energy perception and delay guarantee is characterized in that: The establishment of the integer linear programming model comprises the following steps: 31) Establish system model according to SFC request; 32) Calculating computing resource costs and bandwidth resource costs in the network according to the system model; 33) Calculating the total energy consumption of the network based on the computing resource cost, bandwidth resource cost, and resource conditions of nodes and links in the network; 34) constructing an objective function based on the energy consumption of the network; 35) Complete the establishment of the integer linear programming model.

4. The energy-aware and delay-guaranteed smart firefighting parallel service function chain deployment method according to claim 3 is characterized in that: The system model of step 31) includes an edge computing network and a parallelized service function chain.

5. The energy-aware and delay-guaranteed smart firefighting parallel service function chain deployment method according to claim 3 is characterized in that: The resource conditions of the nodes and links in the network in step 33) include: computing resources of the nodes, bandwidth resources of the links, startup status of the nodes, and startup status of the links.

6. The method for deploying a parallelized service function chain of smart fire protection with energy perception and delay guarantee according to claim 5 is characterized in that: The total energy consumption of the network in step 33) is: ； Where N represents the physical node in the network, represents the energy consumption of physical node k, L represents the link between nodes, represents the energy consumption of link e; The energy consumption of a physical node is calculated using the following formula: ； In the formula, represents the startup energy consumption of physical node k, represents the peak energy consumption of physical node k, Indicates the computing resources used by physical node k, is the total computing resource of physical node k; Calculate the energy consumption of the link: ； In the formula, in the formula, represents the startup energy consumption of link e, represents the peak energy consumption of link e, Indicates the bandwidth resources used by link e, is the total bandwidth resource of link e; is the total bandwidth resource of link e, which is the maximum bandwidth that the link can provide; Resource constraints include: (1) The VNF can be deployed to the node only when the remaining computing resources of the node meet the resource requirements of the VNF: ; (2) A link can be selected only when the remaining bandwidth resources of the link meet the resource requirements of the SFC: ； (3) The end-to-end delay including the selected node must satisfy the end-to-end delay constraint of the parallelized SFC: ； In the formula, The end-to-end delay for deploying parallel SFC consists of processing delay and propagation delay. It is the latency requirement of SFC.

7. The method for deploying a parallelized service function chain of smart fire protection with energy perception and delay guarantee according to claim 6 is characterized in that: In the step 34), the objective function is: ； In the formula, when node k is in the startup state, 1, otherwise 0, when link e is in the startup state is 1 if the value is true, otherwise it is 0.

8. The method for deploying a parallelized service function chain of smart fire protection with energy perception and delay guarantee according to claim 1 or 7, characterized in that: In step 4, the process of solving the heuristic algorithm considering the computing resources, link bandwidth resources, node startup status and link startup status in the edge computing network is as follows: 41) Obtain the edge computing network model, the parallelized SFC request set, and the VNF resource-delay dependency table; 42) Sort the parallel SFC request set in a non-decreasing order according to the size of computing resources required by the parallel SFC; 43) All parallelizable VNFs in a parallelized SFC are regarded as a parallel unit (PU), and each non-parallelizable VNF is regarded as a PU. Each PU of the parallelized SFC adjusts the resource allocation of other VNFs according to the maximum VNF delay in the PU and the VNF resource-delay dependency table; 44) Selecting the physical node with the smallest energy consumption increment on the path with the minimum end-to-end delay for each PU of each parallelized SFC; 45) Under the condition that the end-to-end delay requirement of the parallelized SFC request is met, a sub-path with the smallest energy consumption increment between nodes from the source point to the destination point is selected for the selected physical node, and then each sub-path is combined into an end-to-end path; 46) Finally, the optimal deployment plan for the SFC request is output and mapped to the edge computing network, and the network deployment is updated to show the resource status after the SFC is completed.

Images