A Time-sensitive Traffic Scheduling Algorithm for Industrial Network of Power Grid

Industrial Internet communications in power scenarios are highly time-sensitive and require low network latency and high stability. Therefore, in order to ensure the smooth operation of the automated production line, we propose a time-sensitive traffic scheduling (TsTS) algorithm combined with SRv6 technology based on the application background of industrial and civil converged networks. The method is: through the previous hop node sensing of network congestion Link congestion, and then adjust the detour path for time-sensitive traffic through Dijkstra’s algorithm. Moreover, in order to effectively reduce the link overhead of SRH and prevent bandwidth waste caused by large IPv6 addresses, we also propose a node list compression algorithm for detour paths. The core method is to put only necessary nodes into SRH, that is, Sensitive traffic can be forwarded with the help of ordinary routing. Experiments show that this method performs well in terms of delay and link utilization.


Introduction
Industrial networks connect various devices, PCs, systems, and offices across different spaces to transfer data on a large scale.They enable effective communication between elements in an industrial business [1].As a key network infrastructure that meets the development of industrial intelligence, the Industrial Internet for power scenarios has strong time-sensitive characteristics [2], requiring low network latency and high stability to ensure the smooth operation of automated production lines.However, due to construction costs, its actual deployment scope is often very limited.Therefore, building an industrial Internet platform that is fully integrated and deeply integrated and supports widearea applications, it is very necessary, that is, building a time-sensitive network locally (within the industrial park) [2], with the help of the existing Internet in the middle.However, due to the bursty nature of existing Internet traffic, hierarchical scheduling of industrial communication traffic and ordinary civilian traffic is imperative in converged networks.Among them, the service quality assurance of real-time, bandwidth, jitter and other aspects of high-priority traffic in industrial communications is one of the key and difficult issues.To address this problem, a common practice is to use QoS scheduling within nodes to try to ensure the quality of service (such as queue scheduling within routing and switching equipment) for high-priority traffic (such as time-sensitive traffic) [3].However, this local scheduling strategy has limited effect.The reasons are as follows: when implementing the local QoS policy, the computational burden introduced by the scheduling overhead may actually increase a certain amount of delay; moreover, the queue scheduling "ceiling" of a single node is low, even if high-priority services are placed at the top Priority queues may also be unable to meet the higher latency and bandwidth requirements in industrial applications.Therefore, the academic community aims to optimize the entire network.For example, Li [4] and others proposed a time-sensitive flow management method based on the flow retention protocol, an important component of the TSN protocol.By using the flow management method, it can significantly reduce Average response delay; Wu et al. [5] proposed a time-sensitive network (TSN) network communication delay optimization scheme based on the IEEE 802.1Qbv protocol.This scheme distinguishes network data types with different delay requirements and proposes a TAS-EWRR scheduling algorithm for the data link layer of the OSI model realizes the optimization of the entire network delay; Ma et al. [6] proposed a cost-effective delay-bounded topology construction method, which uses a delay-bound topology construction method.Scheduling.Constrained priority connections to expand local networks.This article attempts to solve the problem from the perspective of seeking a higher priority path.However, this path scheduling is not simple.As we all know, the existing Internet has not completely transitioned to the SD-WAN architecture and is still a purely distributed architecture.If the route update method is used to complete the above path scheduling, that is, trying to use multi-path routing to "expense" high-priority services, it will introduce greater pressure on the forwarding plane of the router (as of 2022, IPv4 prefixes have exceeded one million Article [7]).In addition, this method may also cause route flapping, which is not conducive to topology stability.Therefore, this paper proposes a time-sensitive traffic scheduling (TsTS) algorithm based on SRv6 for converged networks (that is, the scenario where industrial network islands for power scenarios are interconnected through the civil Internet).The basic idea is: only after the congested previous-hop node senses congestion, it first The Dijkstra algorithm calculates the detour path (based on OSPF topology); then, in order to effectively reduce the link overhead of SRH (large IPv6 addresses cause bandwidth waste), a detour path node list compression (DPNLC) on the detour path is proposed, which can use ordinary For routing, only necessary nodes are put into SRH, and the algorithm can effectively prevent self-loops.

Dijkstra Algorithm
Dijkstra The algorithm, named after Dutch computer scientist Edsger W. Dijkstra, is a graph search algorithm used to find the shortest path from the starting node to all other nodes in the weighted graph.It works by iteratively exploring the nodes in the graph from the initial node and constantly updating the shortest path [8] to each node.Specifically, it maintains a prioritized cohort or a similar data structure to effectively select the next node to be explored based on the currently known shortest path.During each iteration, the algorithm selects the unexplored node with the smallest known shortest path, if the shorter path is found, updates the shortest path to its adjacent nodes, and marks the current node as accessed.This process continues until all reachable nodes are visited, and their shortest paths have been identified.Dijkstra Algorithms are often used in various applications, such as finding the shortest path of packets in a computer network, for route optimization in transportation and logistics, and various other areas crucial to finding the shortest path in the network.The TsTS algorithm in this paper is improved based on this algorithm.Please refer to part A of Section 3 for detailed steps.

SRv6 Technology
SRv6, also known as Segment Routing IPv6, is a next-generation network technology that leverages the IPv6 data plane to implement source routing and network programming functions [9].It is an extension of the original Segment Routing (SR) technology, which uses the data plane.In SRv6, the data plane has been enhanced to carry routing instructions within the IPv6 packet itself, thereby achieving finegrained control and flexible flow control of the network.SRv6 has a wide range of applications, including traffic engineering, service function chaining, network slicing, etc.It is a versatile and adaptable technology that meets the requirements of modern networks.As shown in Fig. 1, the IPv6 Segment Routing Header (SRH) is an extension header used in IPv6 messages.It is used to support the implementation of Segment Routing (SR) in IPv6 networks.SRH allows explicit instructions for packet forwarding and processing to be encoded within the packet itself, with packets carrying specific paths in packet headers, providing flexibility for path manipulation, network programming, and traffic engineering in IPv6 networks.support, so that the forwarding path of packets in the network can be accurately controlled.Overall, SRv6 technology represents a major advancement in the networking field, enhancing the programmability and scalability of IPv6 networks.The scheduling of time-sensitive traffic in this article is also implemented based on this technology.

Methodology
In response to the time-sensitive traffic scheduling needs in the context of industrial and civil converged network applications for power scenarios, this article proposes a time-sensitive traffic scheduling algorithm based on SRv6 segment routing technology.Next, this algorithm will be introduced in detail.

Time-sensitive Traffic Scheduling Algorithm
The basic idea of the time-sensitive traffic scheduling algorithm is: when the previous hop node of network congestion senses link congestion, it adjusts the detour path of the time-sensitive traffic and compresses the node list in the path to facilitate placement in SRH.Subsequent The node only performs SRv6 forwarding.As shown in Fig. 2, the algorithm is divided into 5 steps: Step 1, when the pre-congestion node senses link congestion, it determines the post-congestion node (i.e. the egress node) of the congested link, and adds the current node of the congested link to The cost is changed to infinity; Step 2, the pre-congestion node uses the Dijkstra algorithm to calculate all the detour paths between the congested links based on the OSPF topology, sorts them according to their cost, selects the detour path with the smallest cost, and then divides the congested links into The cost of is changed to the original value; Step 3, the node list on the detour path is compressed using the DPNLC algorithm (see Part B of Section 3 for details on the compression algorithm); Step 4, the node before congestion will Detour path node compression list, post-congestion node and destination node are stored in SRH (since this algorithm only ensures that time-sensitive traffic bypasses congested links, it is necessary to store post-congestion nodes and destination nodes in SRH to ensure that time-sensitive traffic remains Take the original optimal path based on OSPF topology calculation); Step 5, time-sensitive traffic packets are forwarded according to the SRH and one-dimensional routing table ; Step 6, if congestion occurs on the current detour path, return step 1 and recalculate the new detour path.The advantage of this algorithm is that only necessary nodes are placed into the SRH, time-sensitive traffic forwarding can be achieved with the help of ordinary routing, and the algorithm can effectively prevent self-loops.

Detour Path Node List Compression Algorithm
The DPNLC algorithm mentioned in Part A of Section 3 needs to be discussed in two situations.Case 1: When the number of detour path nodes is 1 (excluding nodes before and after congestion), the compression list of detour path nodes is determined to be detour nodes.Case 2: When the number of detour nodes is greater than 1 (excluding nodes before and after congestion), in order to reduce SRH overhead, compression is required.The compression steps are as follows: Step 1, the nodes before congestion calculate the detour path in list based on the OSPF topology.The shortest path from the node to the destination node is determined to determine the node after demarcation of the detour path, that is, the first demarcation point on the detour path to avoid link self-loop; Step 2, the pre-congestion node calculates the distance from this node to the detour based on the OSPF topology.The shortest path of the node after the detour path is demarcated; Step 3, determine whether the shortest path from the current node to the detour path demarcation node passes through a congested link, and determine the detour path node compression list.Give examples to illustrate the two situations of detouring.As shown in Fig. 3, the detour path is F, and the number of detour path nodes is 1.If situation 1 is met, the detour path node compression sequence is F. As shown in Fig. 4 and Fig. 5, the detour path K→L→M→N, and the number of nodes in the detour path is greater than 1, conforms to the second situation and needs to be compressed.It is known that node H before congestion is calculated based on OSPF topology: ① The shortest path from node K to D2 is K→H→I→J→D2, which does not bypass the link {H, I}, causing self-loops in nodes H and I; ② Calculate the shortest path from node L to D2 as L→K→H→I→J→D2, which does not bypass the link {H, I}, causing self-loops in nodes H and I; ③ Calculate the shortest path from node M to D2 The path is M→N→I→J→D2, successfully bypassing link {H, I}.Therefore, the node after demarcation of the detour path is M. Next, determine the detour path node compression sequence.
 If the shortest path from this node to the node after demarcation of the detour path does not pass through the congested link, the compression sequence of the detour path node is one node, that is, the node after demarcation of the detour path.As shown in Fig. 4, the node H before congestion is based on OSPF topology calculates the shortest path from node H to node M as H→K→L→M.This path does not pass through the congested link {H, I}, so the detour path node compression sequence is M.  If the shortest path from this node to the node after the detour path demarcation passes through a congested link, the detour path node compression sequence is two nodes, that is, the node before and after the detour path demarcation, as shown in Fig. 5, the node before congestion H is based on OSPF topology, and the shortest path from node H to node M is calculated as H→I→N→M.This path passes through the congested link {H, M}, so the detour path node compression sequence is L, M.   The advantage of this algorithm is that no matter how many nodes are in the detour path, the integrated detour compression sequence will not exceed two nodes at most.Although sometimes the sequence compression rate is stable, it will not be compressed when the detour path is very long and the number of nodes is large.The efficiency is high and can significantly reduce SRH overhead.

Experiment and Result Analysis
In the experiment, we chose the classic scheme Dijkstra's algorithm as the main comparison object.We first evaluated the scheduling performance of TsTS and compared its latency with Dijkstra's algorithm.Then, we changed the link status and verified the effectiveness of the TsTS algorithm again by comparing it with the Dijkstra algorithm.The key indicators of the experiment are defined as follows: Definition 1.Time Delay: The time required for a data packet to be transmitted from source to destination.Definition 2. Link utilization: The amount of data transmitted by the link compared to its maximum capacity.

Time Delay
We established a network topology with 100 nodes through the simulation tool Mininet, and designed different numbers of detour nodes by changing the shape of the network topology, and tested the relationship between the number of detour nodes and delay, as shown in Fig. 6.As the detour path grows, the delay becomes larger and larger, which is in line with the basic forwarding principles of network traffic.Moreover, when the maximum number of nodes tested is 30, the delay of the detour path is only about 8ms, and the overall scheduling effect is good.Next, we evaluated the impact of topology size (number of network links) on the algorithm runtime.We use different topologies (the number of network links varies from 10 to 100), and set 30 timetriggered flows for transmission scheduling, and then measure the runtime under different sizes of network topologies.Fig. 7 illustrates the impact of network topologies of different sizes on running time.We can observe that the running time of both algorithms increases linearly with the topology size.The Dijkstra algorithm is slightly longer than the TsTS algorithm, but for slightly larger networks topology, which shows that the TsTS algorithm is more effective in solving larger-scale network scheduling problems and is more suitable for power scenarios industrial Internet environments.

Link Utilization
Finally, we verify the effectiveness of the TsTS algorithm from the perspective of link utilization.We create detour conditions by changing the network topology state.As shown in Fig. 8, we inject a large amount of ordinary traffic into a link on the original path in the network topology at time 4, creating link congestion.At this time, we can see that in the network topology running the TsTS algorithm , the link utilization on the original path showed a downward trend after a brief surge, while the link utilization on the detour path showed an upward trend, indicating that time-sensitive traffic is scheduled to the detour path to avoid congested links path, on the other hand, looking at the network topology running Dijkstra's algorithm, the link utilization on the original path did not change significantly in a short period of time after the surge.The above is the experimental part.

Conclusions
Considering the time-sensitive traffic scheduling requirements in industrial and civil converged network applications, we propose a time-sensitive traffic scheduling algorithm and a detour node sequence compression algorithm that combines SRv6 segment routing technology.The basic method is: with the help of ordinary routing, only placing key nodes in the SRH can achieve stack forwarding of timesensitive traffic.Experimental results show that compared with the classic Dijkstra algorithm, this algorithm has significant optimization effects and is more suitable for industrial networks in power scenarios.

Figure 6 .
Figure 6.The relationship between detour path and delay.

Figure 7 .
Figure 7.The impact of network topology size on time delay.

Figure 8 .
Figure 8. Link utilization comparison when congestion occurs.