WO2023039965A1

WO2023039965A1 - Cloud-edge computing network computational resource balancing and scheduling method for traffic grooming, and system

Info

Publication number: WO2023039965A1
Application number: PCT/CN2021/123183
Authority: WO
Inventors: 陈伯文; 梁瑞鑫; 王守翠; 刘玲; 郑雯雯; 高明义; 陈虹; 邵卫东; 沈纲祥
Original assignee: 苏州大学
Priority date: 2021-09-17
Filing date: 2021-10-12
Publication date: 2023-03-23
Also published as: CN113742046A

Abstract

The present invention relates to a cloud-edge computing network computational resource balancing and scheduling method for traffic grooming, and a system. An embodiment comprises: obtaining network topology information, initializing a network parameter, and generating a connection request set; determining for each connection request whether a source region and a destination region have nodes with sufficient computational resources to process the connection request; using a k shortest paths algorithm and calculating K candidate paths from a source node to a destination node, performing sorting according to priority levels, and organizing transmission bandwidths of the connection requests; determining in order whether an available spectrum resource on a candidate transmission path selected according to priority level satisfies a spectrum resource requirement of a connection request as well as spectral consistency and spectral continuity conditions for spectrum-flexible optical network transmission; updating link spectrum resource state information and node computational resource information in a central processor, and calculating a connection request blocking rate in the entire network. The present invention facilitates reduction of the blocking rate of a service and improves network spectrum resource utilization.

Description

Cloud edge computing network computing resource balance scheduling method and system for traffic grooming

technical field

The present invention relates to the technical field of cloud-edge computing, in particular to a cloud-edge computing network computing resource balance scheduling method for flow grooming.

Background technique

In recent years, with the rapid development of the Internet of Things (IoT) and the widespread use of big data applications, users' demand for network computing resources has increased dramatically. In addition, technological advances in smartphones, laptops and tablets have enabled the emergence of new demanding services and applications. Although new mobile devices are becoming more and more powerful in terms of central processing units (CPUs), they may not be able to process heavy-duty applications in a short period of time. Cloud computing has powerful computing capabilities. Devices can offload computing resources and transmit computing tasks to remote cloud servers for execution, which can effectively alleviate the problem of large demand for computing resources. However, cloud computing resources are also limited. When the number of computing tasks is too large, the cloud server cannot process the computing tasks achieved in a short time, and some computing tasks cannot be processed in real time, resulting in business congestion. In addition, transmitting computing tasks to cloud servers will bring problems such as high latency, additional transmission energy consumption, and data leakage.

In order to solve the problems of high latency in the process of offloading cloud computing, the European Telecommunications Standards Institute (ETSI) first proposed the concept of Mobile Edge Computing MEC in 2014: a wireless access network close to users that provides IT and cloud computing platform, is considered to be one of the key technologies of the fifth generation mobile communication. The MEC system allows devices to offload computing tasks to network edge nodes, such as base stations, wireless access points, etc., which not only meets the expansion requirements of computing capabilities of terminal devices, but also makes up for the long delay of cloud computing. MEC technology helps to achieve key technical indicators such as ultra-low latency, ultra-high energy efficiency, and ultra-high reliability of 5G services. By deploying cloud computing and cloud storage to the edge of the network, a high-performance, low-latency, and high-bandwidth telecom service environment is provided to accelerate the distribution and download of various content, services, and applications in the network, allowing consumers to enjoy higher Quality web experience.

First, MEC can shorten task execution delay. Mobile application task processing delay includes transmission delay, calculation delay and communication delay. In traditional mobile cloud computing, information needs to go through the wireless access network and backhaul link to reach the cloud server located in the core network. MEC deploys edge servers on the radio access network side, shortening the distance between computing servers and mobile devices. On the one hand, due to the shortened transmission distance, MEC task offloading does not need to go through the backhaul link and core network, which reduces the delay overhead. On the other hand, the computing and processing power of edge servers is much greater than that of mobile devices, which greatly reduces task computing delays. Therefore, MEC has the characteristics of short transmission distance and flat protocol, so that it can meet the ultra-low latency requirements of 5G networks.

Second, MEC can greatly improve network energy efficiency. IoT devices can be widely used in various scenarios such as environmental monitoring, crowd sensing, and intelligent agriculture. However, most of the deployed IoT devices cannot be powered by the power grid. In the case of limited battery energy, MEC shortens the distance between edge servers and mobile devices, greatly saves task offloading and energy consumption for wireless transmission, and prolongs the life of IoT devices. Use cycle.

In addition, MEC can provide higher service reliability. The servers of MEC adopt distributed deployment, and the service scale of a single server is small, so it cannot store too much valuable information. Therefore, compared with the data center of mobile cloud computing, it is less likely to be attacked and can provide more reliable services. At the same time, most mobile edge cloud servers are private clouds, with low risk of information leakage and higher security.

Finally, the software-defined network technology based on the OpenFlow extension protocol has also developed rapidly in recent years. It is designed for software-defined edge server and switch networking problems and the establishment of end-to-end business data transmission required by application services. Through the mechanism of open flow extension protocol of software-defined network, it has the ability of intelligent integrated communication transmission and exchange, forms unified control and management function, dynamic changes of real-time scheduling strategy, and enhances the rapid response speed of edge computing network service request access. Programming networking technology improves the flexibility and transmission efficiency of edge computing networking.

At present, in the research of cloud computing and edge computing, there are business offloading methods that only consider optimizing the selection of computing resources and business offloading methods that only consider optimizing the allocation of link spectrum resources. However, these two allocation methods are only for an optimized A resource allocation method designed for the purpose. The optimal allocation method of computing resources can effectively reduce the computing delay of services, but it cannot improve the ability of network links to transmit services, and the service transmission delay can still be further optimized; the optimal allocation method of link spectrum resources can improve the speed of node processing services , but the computing resource utilization of each edge node in the network can still be further optimized.

Contents of the invention

Therefore, the technical problem to be solved by the present invention is to overcome the problem of high service blocking rate and low utilization rate of network spectrum resources in the prior art, so as to provide a method that makes the service blocking rate as low as possible and improves the utilization rate of network spectrum resources A cloud-edge computing network computing resource balance scheduling method and system for traffic grooming.

In order to solve the above-mentioned technical problems, a cloud-edge computing network computing resource balance scheduling method for flow grooming according to the present invention includes the following steps: Step S1: Obtain network topology information, initialize network parameters, and generate a set of connection request sets; Step S2: For each connection request, determine whether the source area and the destination area have computing nodes with sufficient computing resources to handle the connection request. If not, the establishment of the connection request fails; if so, select the server with the most idle computing resources in the corresponding area as the source of the connection request node and destination node; step S3: use the K shortest path algorithm to calculate K candidate paths from the source node to the destination node, and arrange them according to priority, arrange the transmission bandwidth of the connection request, and obtain the transmission bandwidth Spectrum resources are required; step S4: sequentially judge whether the idle spectrum resources on the candidate transmission path selected according to the priority meet the spectrum resource requirements of the connection request and the spectrum consistency and spectrum continuity conditions of the spectrum flexible optical network transmission, if satisfied, select The shortest path is used as the transmission path of the connection request; if not satisfied, continue to judge whether the K candidate paths are completed, if not, return to step S3, if completed, the establishment of the connection request fails; step S5: update the link in the central controller Spectrum resource status information and node computing resource information to calculate the blocking rate of connection requests in the entire network.

In one embodiment of the present invention, initializing network parameters includes initializing computing resources of cloud servers and edge servers in the network, initializing spectrum flexible optical networks, and initializing state information stored inside an OpenFlow-based central controller.

In one embodiment of the present invention, the server with the most idle computing resources in the corresponding area is selected by the central controller according to the stored real-time information on the computing resource occupancy of each edge server in the source area and the destination area respectively. .

In an embodiment of the present invention, when arranging by priority, they are arranged in ascending order of distance from small to large, and the smaller the path distance, the higher the priority.

In an embodiment of the present invention, the spectrum resource requirement of a connection request is jointly determined by the bandwidth requirement and the line rate used for traffic grooming to determine the spectrum resource required for transmission of each connection request.

In one embodiment of the present invention, before updating the link spectrum resource status information and spectrum resource information in the central controller, after the connection request successfully establishes a working path and allocates spectrum resources, the calculation of each edge computing server is recorded in the central controller Resources and the status information of the spectrum resources of each link are updated.

In one embodiment of the present invention, when updating link spectrum resource status information and node computing resource information in the central controller, after the current connection request is transmitted, the occupied spectrum resources are released, and the link spectrum resources in the central controller are updated. status information; at the same time, after the edge computing server finishes processing the connection request, it releases the computing resources occupied by the edge computing server, and updates the computing resource status information of each edge computing server stored in the central controller.

The present invention also provides a cloud-edge computing network computing resource balance scheduling system for flow grooming, including: a network topology initialization module, used to obtain network topology information, initialize network parameters, and generate a set of connection requests; an edge computing server selection module , for each connection request, it is used to determine whether the source area and the destination area have enough computing resources to process the computing nodes of the connection request, if not, the establishment of the connection request fails; if so, select the server with the most idle computing resources in the corresponding area as the connection The source node and the destination node of the request; the working path calculation module is used to calculate the K candidate paths from the source node to the destination node with the K shortest path algorithm, and arrange them according to the priority, and transfer the connection request Broadband sorting to obtain the spectrum resources required for transmission; the spectrum resource allocation module is used to sequentially determine whether the idle spectrum resources on the candidate transmission path selected according to the priority meet the spectrum resource requirements of the connection request and the spectrum of the spectrum flexible optical network transmission is consistent and spectrum continuity conditions, if satisfied, select the shortest path as the transmission path of the connection request; if not satisfied, continue to judge whether the K candidate paths are completed, if not completed, return to the working path calculation module, if completed, the connection request The establishment fails; the network resource information update module is used to update the link spectrum resource status information and node computing resource information in the central controller, and calculate the connection request blocking rate in the entire network.

In one embodiment of the present invention, it also includes a network resource release module and a blocking rate calculation module, wherein the network resource release module is used to release the spectrum resources occupied by the working path after the connection request is successfully transmitted, and at the same time, After the connection request is processed by the corresponding computing node, the computing resources of the server processing the user request are released, and finally, the information of the working path established by the connection request is cleared; the blocking rate calculation module is used for all connection requests in the network After sending, calculate the overall service blocking rate, where the number of unsuccessfully established connection requests includes the number of connection request blocking due to insufficient computing resources of the source node or destination node and the number of connection request blocking due to insufficient link spectrum resources on the transmission path quantity.

In one embodiment of the present invention, it also includes a central controller module and a judgment and early warning module, the central controller module is used to complete network initialization, connection request edge computing server selection, transmission path calculation, spectrum resource allocation, Computing resource update, resource release, and network congestion rate calculation status monitoring functions; the judgment and early warning module is used to implement the coordination function between each module, and the judgment and early warning function of whether each module is successfully established to complete the entire network topology The goal of reducing the traffic congestion rate.

The above technical solution of the present invention has the following advantages compared with the prior art:

The cloud-edge computing network computing resource balance scheduling method and system for traffic grooming described in the present invention mainly aim at how to balance computing resources in the cloud-edge computing network, reduce the service blocking rate and improve the utilization rate of network spectrum resources, and propose a traffic grooming-based method. Computing resource balancing method and system.

For each connection request, select the server with the most idle computing resources in the source area and the destination area as the source node and the destination node according to its computing resource requirements. Since the server with the most idle computing resources is selected in the corresponding area according to the computing resources required by the connection request as the source node or the destination node, the computing resources of the servers in each area in the network are balanced.

The K shortest path algorithm is used to calculate the candidate path between the source node and the destination node. Use the method of traffic grooming to organize the bandwidth of the connection request and obtain the spectrum resources required for the transmission of the connection request.

For the K candidate paths in order of priority from high to low, use the spectrum allocation algorithm of the first hit to allocate spectrum resources for the path. If the two constraints of spectrum consistency and spectrum continuity are satisfied at the same time, this path is selected as the connection The requested work path. Then update the status of network computing resources and spectrum resources in real time. In the process of traffic grooming, select the appropriate line rate to organize the transmission bandwidth of the connection request. For the connection request whose bandwidth requirement is less than the capacity of an optical channel, the established optical channel is given priority to use, so as to make full use of the idle bandwidth of each optical channel. Improve the utilization of network spectrum resources.

By selecting the computing node with the highest idleness for each connection request and using the method of traffic grooming to sort out the transmission bandwidth, it can effectively balance the computing resource occupancy of each area of the cloud edge computing network and provide spectrum resource utilization. In the case of reasonable allocation of resources Reduce service congestion rate as much as possible.

The OpenFlow-based central controller needs to ensure the update speed of the stored network computing resources and link spectrum resources to be able to ensure the timeliness and accuracy of the network resource information required when multiple consecutive connection requests find paths and allocate spectrum resources.

Description of drawings

In order to make the content of the present invention more easily understood, the present invention will be described in further detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein

Fig. 1 is a flowchart of a method for balanced scheduling of cloud-side computing network computing resources according to the present invention;

Fig. 2 is a network diagram of the balanced dispatching of computing resources in the cloud-edge computing network based on traffic grooming in the present invention;

Fig. 3 is a flowchart of the cloud-edge computing network computing resource balance scheduling system for traffic grooming according to the present invention.

Detailed ways

Embodiment one

As shown in FIG. 1 , this embodiment provides a cloud-edge computing network computing resource balance scheduling method for traffic grooming, including the following steps: Step S1: Obtain network topology information, initialize network parameters, and generate a set of connection request sets; Step S2 : For each connection request, determine whether the source area and the destination area have computing nodes with sufficient computing resources to process the connection request. If not, the establishment of the connection request fails; if yes, select the server with the most idle computing resources in the corresponding area as the connection request Source node and destination node; Step S3: Calculate and obtain K candidate paths from the source node to the destination node with the K shortest path algorithm, and arrange them according to priority, sort out the transmission bandwidth of the connection request, and obtain the transmission bandwidth Spectrum resource required; Step S4: sequentially judge whether the idle spectrum resource on the candidate transmission path selected according to the priority meets the spectrum resource requirement of the connection request and the spectrum consistency and spectrum continuity conditions of spectrum flexible optical network transmission, if satisfied, Select the shortest path as the transmission path of the connection request; if not satisfied, continue to judge whether the K candidate paths are completed, if not, return to step S3, if completed, the establishment of the connection request fails; step S5: update the chain in the central controller State information of road spectrum resources and node computing resource information to calculate the blocking rate of connection requests in the entire network.

The cloud-edge computing network computing resource balance scheduling method for traffic grooming described in this embodiment determines whether the source area and the destination area have enough computing resources to process the computing nodes of the connection request. If not, the connection request establishment fails; if there is, select the corresponding The server with the most idle computing resources in the area is used as the source node and the destination node of the connection request. According to the computing resource requirements of the connection request, the edge server with the most idle computing resources can be selected for the connection request in the source area and the destination area as the source node and destination node. At the same time, during the connection request transmission process, according to the transmission bandwidth requirements of each connection request, the method of traffic grooming is used to realize the sharing of optical fiber spectrum resources, thereby improving the utilization of network spectrum resources and reducing the blocking rate of connection requests; The path path algorithm calculates K candidate paths from the source node to the destination node, and arranges them according to priority, which is beneficial to make the service blocking rate as low as possible, and at the same time improve the utilization rate of network spectrum resources; Whether the free spectrum resources on the candidate transmission path selected by priority meet the spectrum resource requirements of the connection request and the spectrum consistency and spectrum continuity conditions of spectrum flexible optical network transmission, comprehensively considering each edge server in each edge area of the cloud edge computing network Computing resources and the spectrum resource usage of each fiber link, and selecting the optimal computing node and spectrum resource allocation method for each connection request, which is conducive to rational planning of network resource scheduling in edge areas and cloud areas to minimize The connection request blocking rate and the utilization rate of network spectrum resources are improved, thereby improving the service quality of the network.

In the step S1, initializing the network parameters includes initializing the computing resources of the cloud server and the edge server in the network, initializing the spectrum flexible optical network, and initializing the state information stored inside the central controller based on OpenFlow.

Specifically, the computing resources of the cloud server and the edge server in the network are initialized, the spectrum flexible optical network is initialized, and the state information stored inside the OpenFlow-based central controller is initialized. In the cloud-edge computing network G(CR,N _e ,N _c ,L,S,C), CR={CR1,CR2,...,CR _|CR| } represents a group of connection requests, and |CR| is the number of connection requests; N _e ＝{n _e1 ,n _e2 ,…,n _|Ne| } means a group of edge nodes, |N _e | is the number of edge nodes; N _c ＝{n _c1 ,n _c2 ,…,n _|Nc| } means A group of cloud nodes, |N _c | is the number of cloud nodes; L={l ₁ ,l ₂ ,…,l _|L| } is the set of optical fiber links in the cloud edge computing network, |L| is the optical fiber link in the network topology The total number of channels; S={s ₁ ,s ₂ ,…,s _|S| } represents the set of spectrum slots in the fiber link, |S| is the total number of spectrum slots in the fiber link; C represents the central controller, all The switch shares a controller, and the controller can monitor the computing resources of each server connected to the switch and the spectrum resource occupancy of each link in real time.

In the step S2, the server with the most idle computing resources in the corresponding area is selected by the central controller according to the stored real-time information on the computing resource occupancy of each edge server in the source area and the destination area respectively. Select the server with the most idle computing resources.

Specifically, a set of connection request sets CR is generated, each connection request CR(s,d,BR,C _s ,C _d )∈CR, where s represents the source node, d represents the destination node, and s and d need to be determined by the source area R _s and the computing resource occupancy in the destination area R _d are determined, R _s and R _d are randomly generated, BR represents the bandwidth requirement of the connection request, C _s represents the computing resources required by the connection request at the source node, and C _d represents the connection request Computational resources required at the destination node.

For each connection request CR(s,d,BR,C _s ,C _d ), assuming that its source region R _s and destination region R _d have been randomly generated, the central controller first calculates the resource occupancy according to the stored edge servers The real-time information of the situation selects the server with the most idle computing resources in the source area and the destination area as the source node and the destination node of the connection request. If the computing resources of the source node or the destination node cannot meet the computing resource requirements of the connection request, the connection request will be blocked. .

In the step S3, when arranging by priority, the distance is arranged in ascending order from small to large, and the smaller the path distance, the higher the priority.

According to the K shortest path algorithm in the network topology, K candidate paths from the source node to the destination node are obtained. And in ascending order of distance from small to large, that is, the smaller the path distance, the higher the priority, that is, the higher the priority.

According to the bandwidth requirement BR of each connection request, it can be channeled to different optical channels with different line rates, and the optical channels with less than full capacity can continue to be used for subsequent connection requests. The spectrum resources required for each connection request transmission are determined by its bandwidth requirements and the line rate used for traffic grooming. Among them, the spectrum resources required for each optical channel are established, and the spectrum allocation algorithm of the first hit is adopted. According to all links on the path Generate a spectrum resource table according to the spectrum resource status of the road, and start to search for available spectrum slots from the end with the smaller number. If an available spectrum gap is found, the spectrum resource allocation is performed and the spectrum resource status is updated; if no spectrum resource allocation is found, the spectrum resource allocation fails. Assuming that there are M connection requests sharing one optical channel, and the selected line rate is LR, the following constraints should be met:

In the step S4, after finding out the candidate transmission path of the connection request, the central controller judges whether the idle spectrum resources on the paths satisfy the spectrum of the connection request according to the priorities of the selected K paths from high to low. Resource requirements and spectrum consistency and spectrum continuity conditions for spectrum flexible optical network transmission, select the shortest path that satisfies each condition as the transmission path of the connection request; if none of the K paths can satisfy all the conditions at the same time, the connection request is blocked.

The spectrum resource requirement of a connection request is jointly determined by its bandwidth requirement and the line rate used for traffic grooming. The spectrum resource required for transmission of each connection request is determined.

Determine in turn whether the idle spectrum resource on the candidate transmission path selected according to the priority meets the spectrum resource requirement of the connection request and the spectrum consistency and spectrum continuity conditions of spectrum flexible optical network transmission, if not satisfied, return to step S3, according to the formula ( 4) Calculate the connection request blocking rate BP in the entire network, where F _CRc represents the number of connection request blocks caused by insufficient computing resources of the edge computing server, and F _CR1 represents the number of connection requests blocked due to the inability of the spectrum resources on the K candidate transmission paths to meet the transmission requirements. The number of blocked connection requests.

BP=(F _CRl +F _CRc )/|CR| (4)

In the step S5, before updating the link spectrum resource status information and node computing resource information in the central controller, after the connection request successfully establishes a working path and allocates spectrum resources, record the computing resources of each edge computing server in the central controller and The status information of the spectrum resource of each link is updated.

Specifically, after the connection request CR(s,d,BR,C _s ,C _d ) successfully establishes a working path and allocates spectrum resources, the central controller records the calculation resources of each edge computing server and the spectrum resources of each link Status information is updated.

When updating the link spectrum resource status information and node computing resource information in the central controller, after the current connection request is transmitted, release the occupied spectrum resource and update the link spectrum resource status information in the central controller; at the same time, the edge computing server processes After the connection request is completed, the computing resources of the edge computing servers occupied are released, and the computing resource status information of each edge computing server stored in the central controller is updated.

Specifically, after the current connection request is transmitted, the occupied spectrum resource is released, and the state information of the link spectrum resource in the central controller is updated; at the same time, after the edge computing server finishes processing the connection request, the occupied computing resource of the edge computing server is released , updating the computing resource status information of each edge computing server stored in the central controller, so as to provide it for subsequent service requests.

The following is a detailed description of the cloud edge computing network computing resource balance scheduling method for traffic grooming in combination with the accompanying drawings:

As shown in Figure 2, this network includes 1 cloud area and 3 edge areas. The cloud area is composed of 3 interconnected switches, and each cloud server is connected to the switch; the edge area is composed of 3 interconnected switches. , each of which has an edge server connected to the switch; all switches in the cloud area and the edge area share a central controller. The cloud area and the edge area are connected with each other through base station transmission. Assume that each edge server contains 50 computing units, the cloud server contains 1000 computing units, each link in the network contains 100 spectrum slots, and the spectrum width occupied by each spectrum slot is 12.5GHz. The source of the connection request Areas, destination areas, bandwidth requirements, and computing resource requirements are randomly generated.

First, initialize the spectrum flexible optical network G(CR,N _e ,N _c ,L,S,C), including connection requests, base stations, switches, computing resources of each server, spectrum resources of each link segment, central control Status information stored in the memory. The connection request is represented by CR(s, d, BR, C _s , C _d ), R _s represents the source area of the connection request, R _d represents the destination area of the connection request, BR represents the transmission bandwidth required by the connection request, and C _s represents Computing resources required by the connection request in the source area, C _d represents the computing resources required by the connection request in the destination area. Assume that three pending connection requests CR1(s,d,40G,2,30), CR2(s,d,30G,3,9), CR3(s,d,10G, 4, 12), the source areas of the three connection requests are edge area 1, and the destination areas are cloud area, edge area 2, and edge area 3 respectively.

Second, according to the amount of idle computing resources of each server in each area, select the server with the most idle computing resources in the source area and the destination area as the source node and destination node of the connection request. The number next to each server in Figure 2 is the current computing resource usage. According to this calculation node selection principle, the final three connection requests are CR1(1,8,40G,2,30), CR2(1,4,30G, 3,9), CR3 (1,6,10G,4,12).

Thirdly, use the K shortest path algorithm to create K paths between node 1 to node 8, node 1 to node 4, and node 1 to node 6, and arrange them in ascending order of distance from small to large, that is, the smaller the path distance , the higher the priority. When a high-priority path is blocked on a certain link, the lower-priority paths are sequentially selected for spectrum resource allocation until resources are allocated successfully or all paths are blocked.

Fourth, use the 40Gbps line rate of dual polarization quadrature phase shift keying (DP-QPSK) to conduct traffic grooming for the bandwidth requested by each connection. Under this line rate condition, the spectrum width occupied by each optical channel is 25GHz. According to the transmission bandwidth requirements of each connection request, CR1 uses one optical channel alone, and CR2 and CR3 can share one optical channel, so CR1 needs 2 spectrum slots, and CR2 and CR3 need 2 spectrum slots in total.

Fifth, according to the priority order of the K candidate paths from high to low, the spectrum allocation algorithm of the first hit is adopted, and the spectrum resources are allocated according to the constraints of spectrum consistency and spectrum continuity. According to the link spectrum resource status in the network, all three connection requests can successfully allocate spectrum resources, and the working paths of CR1, CR2, and CR3 are path ①(1-2-7-8), path ②(1-2-7 -9-3-4), path ③ (1-2-7-9-10-5-6).

Finally, after the spectrum resources are allocated, the connection request is successfully established, and the computing resources and spectrum resource status are updated. The computing resource occupancy of node 1, node 4, node 6, and node 8 are updated to 39, 34, 32, and 920 respectively, and are updated in real time The computing resources of each node of the entire network topology and the spectrum resource status information of each link stored in the central controller are used to calculate the blocking rate of connection requests that fail to be successfully established using formula (4).

Embodiment two

Based on the same inventive concept, this embodiment provides a cloud-edge computing network computing resource balance scheduling system for traffic grooming. The problem-solving principle is similar to the traffic grooming cloud-edge computing network computing resource balance scheduling method. No longer.

As shown in FIG. 3 , this embodiment provides a cloud edge computing network computing resource balance scheduling system for traffic grooming, including:

The network topology initialization module is used to obtain network topology information, initialize network parameters, and generate a set of connection request sets;

The edge computing server selection module is used for each connection request to determine whether the source area and the destination area have enough computing resources to process the computing nodes of the connection request. If not, the establishment of the connection request fails; if there is, select the idle computing resources in the corresponding area The most servers are used as the source node and destination node of the connection request;

The working path calculation module is used to calculate K candidate paths from the source node to the destination node with the K shortest path algorithm, and arrange them according to priority;

The spectrum resource allocation module is used to sequentially determine whether the idle spectrum resources on the candidate transmission path selected according to the priority meet the spectrum resource requirements of the connection request and the spectrum consistency and spectrum continuity conditions of the spectrum flexible optical network transmission. If satisfied, select The shortest path is used as the transmission path of the connection request; if not satisfied, continue to judge whether the K candidate paths are completed, if not, return to the working path calculation module, if completed, the connection request establishment fails;

The network resource information update module is used to update link spectrum resource status information and node computing resource information in the central controller, and calculate the connection request blocking rate in the entire network.

In the network topology initialization module, in the network topology G (CR, N _e , N _c , L, S, C), the number of servers in the cloud area and edge area, the number of base stations, and the number of switches are configured, and the edge server and cloud server , network topology information, network link spectrum resources, and an OpenFlow-based central controller for initialization.

In the edge computing server selection module, a connection request generation module is also included, which is used to generate a set of connection requests according to user requests, configure the number of connection requests, the source area and destination area of each connection request generated, and the connection request at the source Information such as the size of computing resources required by the area and destination area, and the amount of bandwidth required for connection request transmission.

In the edge computing server selection module, the central controller, according to the size of the idle computing resources of each computing node in the network topology, and on the premise of satisfying the computing resources required by the connection request, assigns each connection request in its source area and destination area. Select the node with the largest free computing resources in the corresponding area as the source node and destination node. If the server with the largest free computing resources in the corresponding area cannot satisfy the computing resources required by the connection request in this area, the connection request will be blocked.

In the working path calculation module, according to the connection request CR (s, d, BR, C _s , C _d ), the source node and the destination node obtained by the calculation node selection module, the K shortest path algorithm is used to calculate the connection request from The K candidate paths from the source node to the destination node are arranged in ascending order of distance, that is, the smaller the path distance, the higher the priority.

In the spectrum resource allocation module, according to the transmission bandwidth requirement BR of the connection request CR (s, d, BR, C _s , C _d ), an appropriate line rate is selected to channel the bandwidth of each connection request, and the transmission bandwidth of the connection request is obtained. Spectrum resources needed. In the obtained K candidate paths, according to the order of priority from high to low, search for the spectrum resources required to meet the connection request in the path. If the dual constraints of spectrum continuity and spectrum consistency are satisfied at the same time, the spectrum can be allocated successfully. resources; if the dual constraints of spectrum continuity and spectrum consistency cannot be satisfied at the same time, the next candidate path is judged. If none of the spectrum resources of the K candidate paths satisfies the condition, the current connection request is blocked.

In the network resource information updating module, when a working path of a connection request CR(s, d, BR, C _s , C _d ) is successfully established, the computing resources of the source node and the destination node should be occupied by the connection request Update in real time; at the same time, the spectrum resource of each link on the working path of the connection request should also be updated according to the size of the spectrum resource occupied by the transmission of the current connection request. And update it to the central controller information list in real time.

It also includes a network resource release module and a blocking rate calculation module, wherein the network resource release module is used to release the spectrum resources occupied by the working path after the connection request is successfully transmitted, and at the same time, after the connection request is processed by the corresponding computing node Finally, the computing resources of the server processing the user request are released, and finally, the information of the working path established by the connection request is cleared; the blocking rate calculation module is used to calculate the overall business congestion after all the connection requests in the network are sent. The number of unsuccessfully established connection requests includes the number of connection request blocking caused by insufficient computing resources of the source node or destination node and the number of connection request blocking caused by insufficient link spectrum resources on the transmission path.

Specifically, in the network resource release module, after the connection request is successfully transmitted, the spectrum resources occupied by the working path are released. At the same time, after the connection request is processed by the corresponding computing node, the computing resource of the server processing the user request is released. Finally, the information of the working path established by the connection request is cleared.

In the blocking rate calculation module, after all connection requests in the network are sent, the overall service blocking rate is calculated according to formula (4), wherein the number of connection requests that have not been successfully established includes those caused by insufficient computing resources of the source node or the destination node The number of connection requests blocked and the number of connection requests blocked due to insufficient link spectrum resources on the transmission path.

It also includes a central controller module and a judgment and early warning module. The central controller module is used to complete network initialization, connection request edge computing server selection, transmission path calculation, spectrum resource allocation, computing resource update, resource release, and network congestion. The status monitoring function of rate calculation; the judgment and early warning module is used to implement the coordination function between each module, and the judgment and early warning function of whether each module is successfully established, so as to complete the goal of reducing the traffic blocking rate in the entire network topology.

In the central controller module, it mainly completes the state monitoring functions of network initialization, connection request edge computing server selection, transmission path calculation, spectrum resource allocation, computing resource update, resource release, and network blocking rate calculation, so as to realize the status monitoring functions in all Minimize the blocking rate of the network during connection request transmission.

In the judgment and early warning module, the coordination function between each module is executed, and the judgment and early warning function of whether each module is successfully established, and the goal of reducing the service blocking rate in the entire network topology is completed.

Those skilled in the art should understand that the embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowcharts and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each procedure and/or block in the flowchart and/or block diagram, and a combination of procedures and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions may be provided to a general purpose computer, special purpose computer, embedded processor, or processor of other programmable data processing equipment to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing equipment produce a An apparatus for realizing the functions specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions The device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby The instructions provide steps for implementing the functions specified in the flow chart or blocks of the flowchart and/or the block or blocks of the block diagrams.

Apparently, the above-mentioned embodiments are only examples for clear description, and are not intended to limit the implementation. For those of ordinary skill in the art, on the basis of the above description, other changes or changes in various forms can also be made. It is not necessary and impossible to exhaustively list all the implementation manners here. And the obvious changes or changes derived therefrom are still within the scope of protection of the present invention.

Claims

A cloud edge computing network computing resource balance scheduling method for flow grooming, characterized in that it includes the following steps:

Step S1: Obtain network topology information, initialize network parameters, and generate a set of connection request sets;

Step S2: For each connection request, determine whether the source area and the destination area have computing nodes with sufficient computing resources to process the connection request. If not, the establishment of the connection request fails; if yes, select the server with the most idle computing resources in the corresponding area as the connection The source node and destination node of the request;

Step S3: Calculate and obtain K candidate paths from the source node to the destination node by using the K shortest path algorithm, and arrange them according to priority, sort out the transmission bandwidth of the connection request, and obtain the spectrum resources required for transmission;

Step S4: Determine in turn whether the idle spectrum resources on the candidate transmission path selected according to the priority meet the spectrum resource requirements of the connection request and the spectrum consistency and spectrum continuity conditions of the spectrum flexible optical network transmission. If so, select the shortest path as the connection The requested transmission path; if not satisfied, continue to judge whether the K candidate paths are completed, if not, return to step S3, if completed, the establishment of the connection request fails;

Step S5: Update link spectrum resource status information and node computing resource information in the central controller, and calculate the connection request blocking rate in the entire network.
The cloud-edge computing network computing resource balance scheduling method for traffic grooming according to claim 1, characterized in that: initializing network parameters includes initializing computing resources of cloud servers and edge servers in the network, initializing spectrum flexible optical networks, and The state information stored inside the OpenFlow-based central controller is initialized.
The cloud-edge computing network computing resource balance scheduling method for flow grooming according to claim 1, characterized in that: selecting the server with the most idle computing resources in the corresponding area is performed by the central controller in real time according to the stored computing resource occupancy of each edge server The information selects the server with the most idle computing resources in the source area and the destination area respectively.
The cloud-edge computing network computing resource balance scheduling method for traffic grooming according to claim 1, wherein when the priority is arranged, the distance is arranged in ascending order from small to large, and the smaller the path distance, the higher the priority.
The cloud-edge computing network computing resource balance scheduling method for traffic grooming according to claim 1, characterized in that: the spectrum resource requirement of a connection request is jointly determined by the bandwidth requirement and the line rate used for traffic grooming to transmit each connection request Spectrum resources needed.
The cloud-edge computing network computing resource balance scheduling method for traffic grooming according to claim 1, characterized in that: before updating the link spectrum resource status information and node computing resource information in the central controller, the connection request successfully establishes a working path and allocates After obtaining the spectrum resources, the central controller records the computing resources of each edge computing server and the state information of the spectrum resources of each link to be updated.
According to claim 1 or 6, the cloud-edge computing network computing resource balance scheduling method for flow grooming is characterized in that: when updating the link spectrum resource status information and node computing resource information in the central controller, after the current connection request is transmitted , to release the occupied spectrum resources, and update the state information of the link spectrum resources in the central controller; at the same time, after the edge computing server finishes processing the connection request, release the occupied computing resources of the edge computing server, and update the edge computing resources stored in the central controller. Computing server computing resource status information.
A cloud-edge computing network computing resource balance scheduling system for flow grooming, characterized in that it includes:

The network topology initialization module is used to obtain network topology information, initialize network parameters, and generate a set of connection request sets;

The edge computing server selection module is used for each connection request to determine whether the source area and the destination area have enough computing resources to process the computing nodes of the connection request. If not, the establishment of the connection request fails; if there is, select the idle computing resources in the corresponding area The most servers are used as the source node and destination node of the connection request;

The working path calculation module is used to use the K shortest path algorithm to calculate K candidate paths from the source node to the destination node, and arrange them according to priority, sort out the transmission bandwidth of the connection request, and obtain the required transmission bandwidth. Spectrum resources;

The spectrum resource allocation module is used to sequentially determine whether the idle spectrum resources on the candidate transmission path selected according to the priority meet the spectrum resource requirements of the connection request and the spectrum consistency and spectrum continuity conditions of the spectrum flexible optical network transmission. If satisfied, select The shortest path is used as the transmission path of the connection request; if not satisfied, continue to judge whether the K candidate paths are completed, if not, return to the working path calculation module, if completed, the connection request establishment fails;

The network resource information update module is used to update link spectrum resource status information and node computing resource information in the central controller, and calculate the connection request blocking rate in the entire network.
The cloud-edge computing network computing resource balance scheduling system for traffic grooming according to claim 8, characterized in that: it also includes a network resource release module and a blocking rate calculation module, wherein the network resource release module is used for successful transmission of the connection request Finally, release the spectrum resources occupied by the working path. At the same time, after the connection request is processed by the corresponding computing node, release the computing resources of the server processing the user request. Finally, clear the information of the working path established by the connection request ; The blocking rate calculation module is used to calculate the overall service blocking rate after all connection requests in the network are sent, wherein the number of connection requests that are not successfully established includes the number of connection request blocks caused by insufficient computing resources of the source node or the destination node and the number of connection request blocks caused by insufficient link spectrum resources on the transmission path.
The cloud-edge computing network computing resource balance scheduling system for traffic grooming according to claim 8, characterized in that: it also includes a central controller module and a judgment and early warning module, and the central controller module is used to complete the initialization of the network, The status monitoring functions of connection request edge computing server selection, transmission path calculation, spectrum resource allocation, computing resource update, resource release, and network congestion rate calculation; the judgment and early warning module is used to perform the coordination function between various modules, and Whether a module has successfully established the judgment and early warning function, and completed the goal of reducing the service blocking rate in the entire network topology.