A multi-path routing method based on time determinism of stochastic time-varying graphs

Abstract

The application relates to a time-deterministic multi-path routing method based on a random time-varying graph, comprising the following steps: constructing a random time-varying graph model; the minimum link bandwidth of a selected path is smaller than or equal to the bandwidth boundary required by a service; according to the bandwidth constraint condition of the path, the random time-varying graph model is first simplified, and the path that does not meet the bandwidth requirement is deleted; the shortest path considering load balancing is obtained according to a load balancing breadth-first search path optimization algorithm; the second shortest path and the path in the set are compared until two paths meeting the optimization target are found; the capacity of each occupied path is updated according to the calculation result, and the path set is returned. The application plans end-to-end transmission multi-paths meeting the differential delay requirement, effectively avoids network congestion, improves the reliability of end-to-end communication, and guarantees the certainty of service transmission delay.

Description

Technical Field

[0001] This application relates to the field of dynamic routing, and more particularly to a multi-path routing method based on the time determinism of random time-varying graphs. Background Technology

[0002] With the development of technology, the traditional best-effort forwarding method of IP networks is insufficient to meet the transmission requirements of services such as "time determinism", "ultra-low packet loss" and "ultra-low jitter". Although there are related technologies to guarantee the QoS attributes of services, there is still a certain gap in meeting the requirements of time determinism. The time determinism and reliability of IP network service transmission still face huge problems and challenges.

[0003] In existing networks, there are many reasons for the loss of service data packets. The main method for ensuring data transmission reliability is retransmission mechanisms, such as Transmission Control Protocol (TCP) and Real-Time Control Protocol (RTCP). The latency caused by service retransmission can severely impact the end-to-end latency performance of time-sensitive services. This is because during service retransmission, both sending back retransmission commands (ACK packets, NAK packets, etc.) and retransmitting the service data packets consume at least one round-trip time (RTT). Furthermore, detecting data packet loss also requires additional latency overhead. All of these factors contribute to uncontrollable end-to-end latency and latency jitter for the service.

[0004] Therefore, to address the above issues, we aim to design a time-deterministic multipath algorithm and protocol based on the theory of stochastic time-varying graphs. This algorithm ensures the time determinism of services while guaranteeing end-to-end reliability through multipath backup transmission of services. Summary of the Invention

[0005] The purpose of this application is to provide a multi-path routing method based on the time determinism of random time-varying graphs, which improves the utilization of network resources and the reliability of end-to-end communication, and ensures the determinism of service transmission delay.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] This application provides a time-deterministic multipath routing method based on a random time-varying graph, including the following steps:

[0008] Step 1: Construct a stochastic time-varying graphical model;

[0009] Step 2: The minimum link bandwidth of the selected path must be less than or equal to the bandwidth required by the service.

[0010] The constraints are summarized as follows:

[0011] Step 3: Based on the bandwidth constraints of the path, first simplify the stochastic time-varying graph model and delete the paths that do not meet the bandwidth requirements;

[0012] Step 4: Find the shortest path from source node S to destination node D that takes load balancing into the set using the load balancing breadth-first search path optimization algorithm;

[0013] Step 5: Based on the shortest path considering load balancing obtained by the breadth-first search path optimization algorithm, and using the idea of ​​the KSP algorithm, solve for the set of deviation paths of the shortest path, and thus select the second shortest path.

[0014] Step 6: Compare the second shortest path with the paths in the set. If two paths that meet the optimization objective are found, exit. If not, add the path to the set and iterate continuously using the KSP method, then compare until two paths that meet the optimization objective are found.

[0015] Step 7: Update the capacity of each occupied path based on the calculation results and return the path set.

[0016] Based on the stochastic time-varying graph model proposed in step 1, and by jointly utilizing multi-dimensional network resources such as node packet processing queues, links, attributes, and time scales, the network is accurately characterized. The specific modeling steps are as follows:

[0017] Consider an integrated space-ground network comprising a set of satellites S and a set of ground stations G, where the satellite set S = {S1, ..., S2} N} contains N satellites, and the set of ground stations G = {G1, ..., G...} P} contains P ground stations, and the set of tasks to be transmitted is M = {M1, ..., M} Q The task contains Q tasks, and snapshots can be used to depict the connectivity between nodes over h time intervals.

[0018] Tasks for one time period can be transmitted in the next time period using the storage within that node, thus jointly utilizing storage resources across multiple time periods. For any satellite S i ∈S, if it belongs to the k-th snapshot, define it as

[0019] Connect two nodes of the same satellite in two adjacent snapshots with a directed link, with the link pointing in the direction of increasing time. The network topology can be represented by a directed graph G = (V, E), where the set of vertices V contains satellite nodes for all time periods, and the set of edges E contains all transmission links within snapshots and storage links between snapshots.

[0020] For the task Since the task release time and deadline are already given, the task's latency requirements can be characterized by adding a task requirement chain.

[0021] Specifically, during the task release period, an auxiliary source node M′1 is inserted and connected to the source node, serving as the release link for task M1. Similarly, during the task deadline period, an auxiliary destination node is inserted. It is connected to the target node and serves as the task termination link.

[0022] The breadth-first search path optimization algorithm for load balancing satisfies both user service requests and network QoS. It then sets a dynamic traffic load balancing control threshold based on the traditional breadth-first search algorithm. When the network detects that the link traffic load balancing parameters exceed the real-time dynamic traffic threshold, it further optimizes path allocation, improving network resource utilization. The specific steps can be modeled as follows:

[0023] The formula for defining traffic load balancing parameters is as follows:

[0024] r k (t) represents the utilization rate of active link k when the t-th task is accessed; This represents the average utilization rate of currently active links; n is the number of active links; t is the number of access tasks.

[0025] Initialize Φ to an empty set, then execute a loop to calculate and store the data transfer path between the user and the terminal server.

[0026] Within the algorithm's inner loop, to prevent V and E in the path topology from being overwritten and losing path information when calculating new paths, two temporary variable sets, V1 and E1, are initialized to V and E respectively.

[0027] Remove links and nodes that do not meet the forwarding conditions.

[0028] Find the process between the source node and the destination node, starting from the auxiliary source node M. ′ 1. Perform BFS. If this is the first task, store its path directly; otherwise, record the Source as the access point. If the link traffic load parameter from the Source to the next node is less than... If the next node is marked as Source, then the path is temporarily blocked; otherwise, perform BFS again from Source until the auxiliary destination node M is found. ′ 1 ′ If the final set of nodes used is obtained, the weights will be updated for the next round of search.

[0029] The optimization objective mentioned in step 6 is expressed as:

[0030] |Delay(P)-Delay(P ′ )|≤ε

[0031] That is, while ensuring the business latency limit, minimize the latency between two paths, where Delay(P) is the latency of path P. ′ ) Path P ′ The time delay.

[0032] Compared with the prior art, the beneficial effects of this application are:

[0033] 1. This invention proposes a stochastic time-varying graph model, which jointly characterizes the stochastic characteristics of link, storage, and time resources caused by the randomness of services, and characterizes the correlation between storage and link resources. This provides a model basis for improving the utilization of link resources and ensuring the time determinism of services, and realizes the joint planning of routing and service scheduling.

[0034] 2. This invention proposes a time-deterministic multipath routing algorithm. This algorithm comprehensively considers various QoS index constraints that affect service latency in the network, plans end-to-end transmission multipaths that meet differential latency requirements, effectively avoids network congestion, improves the reliability of end-to-end communication, and ensures the determinism of service transmission latency. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0037] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. It should be noted that similar reference numerals and letters in the following drawings indicate similar items; therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0038] The terms “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0039] The following is combined Figure 1 This invention introduces a time-deterministic multipath dynamic routing method based on a random time-varying graph, comprising the following specific steps:

[0040] Step 1: Construct a stochastic time-varying graphical model

[0041] Step 2: Consider an integrated space-ground network comprising a satellite set S and a ground station set G, where the satellite set S = {S1, ..., S2} N} contains N satellites, and the set of ground stations G = {G1, ..., G...} P} contains P ground stations, and the set of tasks to be transmitted is M = {M1, ..., M} Q The dataset contains Q tasks, and snapshots can be used to depict the connectivity between nodes over h time periods.

[0042] To more clearly illustrate the graphical model representation process of the topology, a simplified 4-UAV network is constructed, such as... Figure 1 As shown. First, the time period T of the drone network is divided into 3-hour segments, keeping the network topology unchanged within each segment. The drone network has different topologies within each 3-hour segment, and the topology of each snapshot is characterized by nodes and links respectively.

[0043] Naturally, the same drone has two nodes in two adjacent snapshots, connected by a directed link. Given drone S3, the nodes in its first snapshot can be... With the nodes in the second snapshot Connect them to obtain their storage links.

[0044] This representation method supports store-and-carry forwarding mode, such as... Data will be transmitted to during the second time period. By caching data to Finally, the data can be transmitted in the third time period.

[0045] Step 3: Tasks for one time period can be transmitted in the next time period using the storage within that node, thus jointly utilizing storage resources across multiple time periods. For any satellite S i ∈S, if it belongs to the k-th snapshot, define it as

[0046] Step 4: Connect the two nodes of the same satellite in two adjacent snapshots with a directed link, with the direction pointing in the direction of increasing time. The network topology can be represented by a directed graph G = (V, E), where the set of vertices V contains satellite nodes for all time periods, and the set of edges E contains all transmission links within snapshots and storage links between snapshots.

[0047] Step 5: For the task Since the task release time and deadline are given, the task's latency requirements can be characterized by adding a task requirement link.

[0048] Specifically, an auxiliary source node M is inserted during the task release period. ′ 1 is connected to the source node, serving as the release link for task M1. Similarly, an auxiliary destination node M is inserted during the task's deadline period. ′ 1 ’ It is connected to the target node and serves as the task termination link.

[0049] exist Figure 1 In the example, task M1 is released during the first hour and ends during the second hour, with its source node being S1 and its destination node being S4.

[0050] Then an auxiliary source node M can be inserted in the first hour segment. ′ 1 and S1 1 Connected, serving as the release link for task M1. Similarly, an auxiliary destination node M can be inserted in the second hour segment. ′ 1 ’ and Connected, serving as the link for task termination.

[0051] Therefore, the time-guaranteed path finding problem for task M1 can be transformed into finding the source node M. ′ 1 to destination node M ′ 1 ’ The path.

[0052] Step 5: To ensure the time determinism of time-sensitive services, the latency and bandwidth of the link must meet the latency and bandwidth requirements of the time-sensitive services. In other words, the latency of the selected path must be less than or equal to the latency limit required by the service; the minimum link bandwidth of the selected path must be less than or equal to the bandwidth limit required by the service.

[0053] The constraints of this algorithm are summarized as follows:

[0054] Where T bounded B represents the maximum latency limit that the business can tolerate; bounded Indicates the upper bound of the bandwidth required by the service; B P(s,d) This represents the minimum bandwidth of each link in path P(s,d).

[0055] Step 6: Based on the bandwidth constraints of the path, first simplify the random time-varying graph model and delete the paths that do not meet the bandwidth requirements.

[0056] Step 7: Initialize Φ to an empty set, and then execute a loop to calculate and store the data transfer path between the user and the terminal server.

[0057] Step 8: In the inner loop of the algorithm, in order to avoid the loss of path information due to the overwriting of V and E in the path topology when calculating the new path, the two temporary variable sets V1 and E1 are initialized to V and E respectively.

[0058] Step 9: Remove links and nodes that do not meet the forwarding conditions.

[0059] Step 10: Locate the process between the source node and the destination node, starting from the auxiliary source node M. ′ 1. Perform BFS. If this is the first task, store its path directly; otherwise, mark the Source as the access point. If the link traffic load parameter from the Source to the next node is less than φ, mark the next node as the Source; otherwise, temporarily block this path. Re-perform BFS from the Source until the secondary destination node M is found. ′ 1 ′ If the final set of nodes used is obtained, the weights will be updated for the next round of search.

[0060] Step 11: Find the shortest path from source node S to destination node D that takes load balancing into the set using the load balancing breadth-first search path optimization algorithm described in steps 7-10.

[0061] In a stochastic time-varying graph model, the load balancing breadth-first search path optimization algorithm is used to find paths. It can be optimized based on the resource sharing constraint of the delay queue, that is, after a service is hosted at a certain node, the service must be sent to the next hop node and cannot continue to be hosted.

[0062] Therefore, when the load balancing breadth-first search path optimization algorithm finds the next hop, if the previous hop of the current node is itself, then the next hop must be another node; if the previous hop of the current node is not itself, then the next hop can be any node.

[0063] Step 12: Based on the shortest path considering load balancing obtained by the load balancing breadth-first search path optimization algorithm, and using the idea of ​​the KSP algorithm, solve for the set of deviation paths of the shortest path, thereby selecting the second shortest path.

[0064] Step 13: Compare the second shortest path with the paths in the set. If two paths that meet the optimization objective are found, exit. If not, add the path to the set and iterate continuously using the KSP method, then compare until two paths that meet the optimization objective are found.

[0065] Step 14: Update the capacity of each occupied path based on the calculation results and return the path set.

[0066] Compared to existing technologies, this invention proposes a time-deterministic multi-path dynamic routing method based on stochastic time-varying graphs. The stochastic time-varying graph model jointly characterizes the stochastic characteristics of link, storage, and time resources caused by service randomness, and also characterizes the correlation between storage and link resources. This provides a model foundation for improving link resource utilization and ensuring the time determinism of services, enabling joint planning of routing and service scheduling. The time-deterministic multi-path routing algorithm comprehensively considers various QoS constraints affecting service latency in the network, planning end-to-end transmission multi-paths that meet differential latency requirements, effectively avoiding network congestion, improving the reliability of end-to-end communication, and ensuring the determinism of service transmission latency.

[0067] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A time deterministic multi-path routing method based on a random time- varying graph, characterized in that, Comprising the following steps: Step 1: Constructing a random time-varying graph model; Step 2: The minimum link bandwidth of the selected path is greater than or equal to the bandwidth required by the service, and the delay of the selected path is less than or equal to the delay required by the service, The constraints are summarized as follows: , wherein denotes the maximum delay bound that the traffic can tolerate; denotes the bandwidth upper bound required by the traffic; denotes the path the minimum bandwidth of each link, Step 3: According to the bandwidth constraint condition of the path, first simplify the random time-varying graph model, and delete the path that does not meet the bandwidth requirement; Step 4: Find a shortest path considering load balancing from source node S to destination node D through load balancing breadth-first search path optimization algorithm, and add it to the set; Step 5: According to the shortest path considering load balancing obtained by the load balancing breadth-first search path optimization algorithm, with the help of the idea of KSP algorithm, the deviation path set of the shortest path is solved, and the second shortest path can be selected; Step 6: Compare the second shortest path with the path in the set. If two paths meeting the optimization target are found, exit. If not, add the path to the set, and then compare through KSP method iteration, until two paths meeting the optimization target are found; Step 7: Update the capacity of each occupied path according to the calculation result, and return the path set.

2. The time deterministic multi-path routing method based on random time- varying graph according to claim 1, wherein, Based on the random time-varying graph model proposed in step 1, the node data packet processing queue, link, attribute and time scale and other multi-dimensional network resources are used to accurately describe the network, and the specific modeling steps are as follows: Consider an integrated network containing a set of satellites S and a set of ground stations G, where the set of satellites S contains N satellites, the set of ground stations G contains P ground stations, and the set of tasks to be transmitted T contains Q tasks. Consider an integrated network containing a set of satellites S and a set of ground stations G, where the set of satellites S contains N satellites, the set of ground stations G contains P ground stations, and the set of tasks to be transmitted T contains Q tasks. Consider an integrated network containing a set of satellites S and a set of ground stations G, where the set of satellites S contains N satellites, the set of ground stations G contains P ground stations, and the set of tasks to be transmitted T contains Q tasks. Consider an integrated network containing The tasks of one time period can be transmitted by means of the storage within the node in the next time period, so as to jointly utilize the storage resources of multiple time periods, for any satellite If it belongs to the kth snapshot, it is defined as , The two nodes of the same satellite in two adjacent snapshots are connected by a directed link, and the direction points to the direction of time increase, and the network topology can be represented by a directed graph , wherein the point set contains all satellite nodes of all time periods, and the edge set contains all transmission links within the snapshots and all storage links between the snapshots, For task Since its task release time and deadline have been given, the delay requirement of the task can be characterized by adding a task requirement link, Specifically, an auxiliary source node is inserted at the task release period connected to the source node as the task release link, similarly, an auxiliary destination node is inserted at the task deadline period connected to the target node as the task end link.

3. The method of claim 1, wherein, The load balancing breadth-first search path optimization algorithm can guarantee to meet the user service request and meet the network QoS, and then set a dynamic flow load balancing control threshold based on the traditional breadth-first search algorithm. When the network detects that the link flow load balancing parameter exceeds the real-time dynamic flow threshold, the path allocation is optimized again, which improves the utilization rate of network resources. The specific steps can be modeled as: The traffic load balancing parameter formula is defined as , denotes the utilization of the active link k at the tth task access; denotes the average utilization of the currently active link; n is the number of active links; t is the number of access tasks, Initialize Φ as an empty set, and then execute a loop to calculate and store the data transmission path between the user and the terminal server, In the inner loop of the algorithm, in order to avoid path topology in and When calculating the new path, the path information is lost by covering, two temporary variable sets and are initialized to and , Remove the links and nodes that do not meet the forwarding conditions, Finding the process between source node and destination node from auxiliary source node BFS, if this is the first task, directly store its path, otherwise mark the source as access point; if the link flow load parameter from source to next node is less than , mark the next node as source, otherwise temporarily shield this path; re-BFS from source until auxiliary destination node is searched , output the final node set used, and update the weight for the next round of search.

4. The time deterministic multi-path routing method based on random time- varying graph of claim 1, wherein, The optimization target mentioned in step 6 is expressed as: ， That is, in the case of guaranteeing the service delay limit, the delay between the two paths is as small as possible, wherein the delay of path P, the path the delay of path P.

Images