Incremental scheduling method for deterministic service of time-sensitive network

Abstract

The invention particularly relates to an incremental scheduling method for deterministic services of a time-sensitive network, and belongs to the technical field of scheduling of the time-sensitive network. Aiming at the problems of high conflict detection overhead under dynamic access, large search scale of a long-short period mixed scene, reduced schedulability caused by time slot fragments and the like, the method comprises the following steps: constructing a scheduling model, and performing multi-rule sorting and route pre-selection on to-be-accessed services; time granularity scaling and periodic modular domain projection technologies are introduced, multi-link and multi-period conflict detection is converted into bitmap operation in a single period, and a modular domain taboo table is rapidly generated; a conflict-free sending phase is efficiently searched based on bitmap operation, a comprehensive cost function integrating fragment sensing, time delay and load balancing is adopted for multi-dimensional evaluation, and an optimal scheduling scheme is selected. According to the method, the efficiency and the real-time performance of online incremental scheduling can be remarkably improved, the search overhead is reduced, time slot fragments are reduced, and the service scheduling success rate in a high-load scene is improved.

Description

Technical Field

[0001] This invention relates to the field of communication network technology, and specifically to an incremental scheduling method for deterministic services in time-sensitive networks. Background Technology

[0002] With the evolution of Industrial Internet, vehicular networks, and aerospace technologies, Time-Sensitive Networking (TSN), as the next-generation deterministic network standard, is widely used in hard real-time scenarios with strict requirements for latency and jitter. TSN provides deterministic communication capabilities for scenarios such as industrial control, automotive Ethernet, and avionics through time synchronization and gating mechanisms. Time-triggered (TT) services rely on gating control lists to send data within predetermined time windows, requiring deterministic requirements such as periodic constraints, end-to-end latency constraints, and link collision-free constraints.

[0003] In TSN, TT services enjoy the highest priority, and their transmission relies on a gated list (GCL). In practical applications, network environments are not static. For example, scenarios such as adding new equipment to industrial production lines or dynamically upgrading vehicle systems require the TSN controller (CNC) to have incremental scheduling capabilities—that is, to quickly calculate a conflict-free scheduling table for newly arriving TT services without altering existing scheduled traffic. This necessitates that the algorithm balance computational speed (millisecond-level response) and scheduling quality.

[0004] The existing technical solutions are as follows: In their paper "Scheduling Real-Time Communication in IEEE 802.1Qbv Time Sensitive Networks" presented at RTNS 2016, Craciunas, Serna Oliver, Chmelík, and Steiner computed fully deterministic offline scheduling for 802.1Qbv multi-hop switching networks. They identified key functional parameters and derived TSN-specific constraints, employing the SMT solver to compute static offline scheduling to meet end-to-end deterministic latency and jitter constraints. They also discussed the computational time trade-offs under different configurations and optimization objectives. However, this method involves offline global solving. As the number of services, candidate routes, and the size of the supercycle increase, the size of constraints and variables expands rapidly, significantly increasing the feasible solution search space. The solution time then struggles to meet the real-time requirements of dynamically changing scenarios.

[0005] Nayak et al. proposed an incremental scheduling and routing method for new TT flows in Time-Sensitive Networks (TSNs) to address the dynamic changes in TT services. This method, while maintaining existing scheduling results, models the routing and transmission time allocation of new TT services as a unified integer programming problem, and obtains a scheduling scheme that satisfies end-to-end deterministic constraints through a solver. Although this method has good theoretical guarantees in terms of scheduling feasibility and optimality, it relies on an exact optimization model and solver computation. When new services are added, large-scale feasibility solving is still required under multiple links and constraints. As the network size and the number of services increase, the computational complexity rises significantly, resulting in low scheduling efficiency.

[0006] Dürr and Naya proposed the No-wait Packet Scheduling Problem (NW-PSP) in their work "No-wait Packet Scheduling for IEEE Time-sensitive Networks (TSN)" to model TSN scheduling and map it to a no-wait job shop scheduling problem. They presented a heuristic solution method based on tabu search and proposed a schedule compression technique to reduce the number of guard bands in the scheduling, thereby improving scheduling availability and efficiency. The drawback of this approach is that the heuristic search framework requires iterative evaluation of candidate solutions. For multi-hop paths and long-cycle scenarios, the candidate size and the number of evaluations increase significantly with the expansion of the time window, resulting in high computational overhead and making it difficult to meet the requirements for fast online solutions.

[0007] Patent application CN117792893A proposes an online scalable configuration method and system for dynamic changes in time-sensitive networks. It employs an efficient inter-flow collision detection mechanism using multi-level flow packets and reduces collision detection complexity by utilizing flow period and offset correlation analysis. Simultaneously, it adopts an online incremental strategy of decoupling routing and scheduling, and offline pre-routing, to improve online computation speed and reduce sensitivity to calculations related to time-lapse and link slot occupancy, thereby supporting online configuration and rapid response for large-scale data flows. While this scheme emphasizes accelerated collision detection and online incremental processing, it does not consider the segmentation of large idle windows after the insertion of new services. Therefore, under high load or long-cycle scenarios, the efficiency of searching for feasible transmission times and subsequent schedulability may still decrease.

[0008] Furthermore, in the engineering implementation of TSN, there exists a category of technical solutions that focus on optimizing hardware execution efficiency. For example, US Patent US12301471B2 discloses a hardware mechanism for configuring resources in a time-aware network. Through dual-memory switching, DMA prefetching, and register state management, it solves the read latency and switching interruption problems during the physical layer execution of gated queues. However, this type of solution has inherent limitations: the technology only focuses on how to efficiently execute existing scheduling tables, and its prefetching and caching mechanisms are based on the physical movement of known GCL entries. It does not have the ability to generate scheduling tables, meaning it cannot, under complex network constraints, calculate and decide which service should be allocated to which time slot. When faced with the dynamic access of new services, if there is a lack of efficient scheduling calculation algorithms to generate new GCLs, the hardware acceleration mechanism will be unable to function due to a lack of effective input data.

[0009] However, existing technologies have the following drawbacks: First, insufficient real-time performance: High conflict detection overhead under dynamic access leads to low scheduling efficiency. When there are a large number of TT services, all services need to be rescheduled, which significantly increases the time required to generate a new scheduling table, making it difficult for the network to update in a timely manner and failing to meet the real-time requirements in dynamically changing scenarios.

[0010] Second, the search space is enormous: in scenarios with mixed long and short cycles, the large size of the candidate search leads to excessively high online solution time and resource consumption. Heuristic methods typically require checking each link for conflicts one by one on a nanosecond timeline, and the search space expands significantly when dealing with multi-hop paths and multiple candidate routes.

[0011] Third, severe resource fragmentation: The lack of assessment and constraints on link time slot occupancy patterns easily leads to time slot fragmentation, resulting in decreased schedulability. Blindly inserting new flows into the earliest available positions often truncates originally continuous large blocks of time slots, leaving invalid fragments smaller than the minimum frame length, leaving no place for subsequent large-cycle traffic.

[0012] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0013] This invention provides an incremental scheduling method for deterministic services in time-sensitive networks, aiming to solve the following technical problems: First, address the issue of high conflict detection overhead leading to low scheduling efficiency under dynamic access. Specifically, in existing time-sensitive network incremental scheduling, as the network's scheduled traffic increases, new service access requires conflict detection and feasibility assessment under multi-link and multi-cycle constraints. This results in detection time increasing linearly or superlinearly with the scale of the service, leading to poor dynamic response capabilities.

[0014] Second, address the issue of excessively large candidate search size in scenarios involving a mix of long and short cycle traffic. Existing scheduling methods, when handling scenarios with a mix of long and short cycle traffic, require searching and matching available time slots within a relatively long common expansion time window. The number of candidate time slots and the number of combination verifications increase significantly with the expansion of the time window, leading to excessively high scheduling solution time and resource overhead, and making online solutions difficult.

[0015] Third, address the issue of decreased schedulability caused by time slot fragmentation. Traditional greedy scheduling algorithms lack assessment and constraints on link time slot occupancy, easily creating numerous unusable micro-slots within the link time slots. This leads to scheduling failures due to the inability of subsequent high-bandwidth, long-duration, or long-period critical traffic to obtain continuous available time slots.

[0016] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0017] According to a first aspect of the present invention, an incremental scheduling method for deterministic services in a time-sensitive network is provided, the method comprising: Step 1: Construct a Time-Sensitive Network (TSN) scheduling model: Step 1a: Define the network topology as a directed graph and number all nodes and links in the network; Step 1b: Initialize network parameters, including switch processing latency, link propagation latency, and link bandwidth; Step 1b: Obtain the routing parameters for all scheduled time-triggered TT services; Step 1d: Obtain the period parameters of all TT services to be scheduled, and calculate the least common multiple of all periods as the super-period; Step 2: Preprocessing of new business flows: Step 2a: Obtain the period, transmission delay, and optional route list for all newly added TT services, sort all newly added TT services according to the multi-rule method, and sort the optional routes for each newly added service by the number of hops; Step 2b: Calculate the scaling of the period for the newly added TT service based on the preset time granularity; Step 3: Construct the module domain taboo table: For each candidate route of a newly added TT service, perform the following operations: Step 3a: Initialize a zero-based bitmap with a length equal to the scaled period; Step 3b: Traverse each hop link on the candidate route, calculate the cumulative offset of the data packet to the hop link, and obtain the time interval of all scheduled services on the link; Step 3c: Based on the cumulative offset and the transmission delay of the new service, perform offset backward calculation and leftward expansion processing on each occupied time interval to obtain the conflict time interval in the reference frame of the source end transmission time; Step 3d: Discretize the conflict time interval into time slot index intervals according to the time granularity, and take the modulus of the scaled period for the time slot index interval, project it into the modulus domain, and set the corresponding modulus domain index position in the bitmap; after traversing all hop links, obtain the modulus domain taboo table corresponding to the candidate route; Step 4: Rapid Search and Multi-Dimensional Cost Assessment Step 4a: In the tabu table of the modal domain, quickly search all indices with a bit value of zero using bitmap operations to send phases as candidate collision-free source ends; Step 4b: For each candidate collision-free transmission phase, restore it to the supercycle time axis to generate the candidate's start transmission time, and evaluate it using a comprehensive cost function, which includes at least fragment awareness cost, latency cost, and link load balancing cost. Step 5: Scheduling Decisions and Applications Step 5a: Select the scheme with the lowest overall cost from all candidate start times of the newly added TT service as the scheduling result; Step 5b: Update the time occupancy status of each link in the network based on the scheduling results; Step 6: Repeat steps 2 to 5 until all newly added TT services have been scheduled. Finally, generate the transmission timetable for each system and the gating list for each port of the switch based on the scheduling results of all services.

[0018] In some exemplary embodiments, the sorting of all newly added TT services according to a multi-rule method in step 2a specifically includes: Obtain the attribute parameters for each newly added TT service, including: whether it is a multicast service identifier, period, maximum allowed end-to-end latency, and single frame transmission latency; Sort in descending order based on whether it is a multicast service identifier as the first sorting key; If the first sorting key is the same, the maximum allowable end-to-end delay is used as the second sorting key, and the data is sorted in ascending order. If the first and second sorting keys are the same, use the cycle as the third sorting key and sort in ascending order; If the first, second, and third sorting keys are all the same, the single-frame transmission delay is used as the fourth sorting key, and the data is sorted in descending order.

[0019] In some exemplary embodiments, the root in step 2b performs a scaling calculation on the period of the newly added TT service using the following formula:

[0020] Where Period is the original period for new TT services. For time granularity, This is the scaled period.

[0021] In some exemplary embodiments, the offset backward calculation and leftward expansion processing in step 3c specifically includes: For an occupied time interval [L, R) of the h-th hop link on a candidate route, its conflict time interval [RawStart, RawEnd) relative to the source transmission time is calculated as follows:

[0022] in, To reduce the single-frame transmission latency of the newly added TT service, This is the cumulative offset when the data packet arrives at the h-th hop link.

[0023] In some exemplary embodiments, the step 4a of quickly searching all indices with a bit value of zero using bitmap operations specifically involves: starting from index 0, iteratively calling the next zero-bit search operator. This continues until the returned index exceeds the range of the scaled period SP, where BitSet is a modal tabu bitmap. For the currently found feasible phase index, This indicates that the next bit with a value of 0 is searched to the right of the given starting index in the bitmap.

[0024] In some exemplary embodiments, the total cost function TotalCost in step 4b is calculated as follows: in, The cost of fragment perception is calculated based on the size of the idle gap on both sides of the candidate scheduling position; The latency cost is calculated based on the ratio of the actual end-to-end transmission latency to the maximum allowable latency and the compactness score of the transmission time. The link load balancing cost is calculated based on the load rate of each link on the candidate routing path; , , These are the corresponding weighting factors.

[0025] In some exemplary embodiments, the generation of the gating list for each port of the switch in step 6 specifically includes: for the transmission link corresponding to each port, traversing all occupied time slot intervals within the supercycle; for the first time slot interval, if its left endpoint is not 0, adding a gating entry indicating a closed state; for each occupied time slot interval, adding a gating entry indicating an open state; between two consecutive occupied time slot intervals, if there is an idle gap, adding a gating entry indicating a closed state; finally, if the right endpoint of the last occupied time slot interval is less than the supercycle, supplementing a gating entry indicating a closed state so that the gating list cyclically covers the entire supercycle.

[0026] The incremental scheduling method for deterministic services in time-sensitive networks provided by embodiments of the present invention introduces time granularity scaling technology to compress the search space, and utilizes service cycle modal domain projection technology to reduce the dimensionality of ultra-periodic conflict detection to single-service cycle bitmap operations, achieving ultra-fast conflict detection. Simultaneously, it designs a comprehensive cost function incorporating fragment awareness, latency, and load, proactively avoiding scheduling schemes that generate minor fragments. This improves the scheduling success rate of newly added TT services in dynamic access scenarios, while also considering latency and load balancing. Compared with existing technologies, it has the following beneficial effects: First, it improves the online scheduling efficiency and real-time response capability in dynamic access scenarios. Addressing the issue that existing technologies require conflict detection item by item under multi-hop paths and multi-cycle constraints during dynamic access, leading to a significant increase in detection overhead with the growth of service scale, this invention employs a service cycle modal domain projection mechanism. This reduces conflict detection within a super-cycle range to a single service cycle space and represents link time slot occupancy and conflict constraints using a bitmap. This transforms conflict determination from interval enumeration and multiple traversals into bit-parallel computation, significantly reducing the overhead of conflict detection and feasibility assessment, and improving the computational efficiency and dynamic response capability of online incremental scheduling.

[0027] Second, this invention reduces the candidate search scale and online solution overhead in scenarios with mixed long and short cycles. Addressing the problem that existing methods require candidate slot search and combination verification within a relatively long common unfolding time window under conditions of mixed long and short cycles and multi-hop paths, leading to search space expansion and excessive time and resource overhead, this invention introduces a time granularity scaling technique to compress the time axis, and combines heuristic candidate selection and pruning strategies to reduce the number of candidate positions and combination verifications, thereby reducing online solution complexity and improving solution efficiency in mixed-cycle business scenarios.

[0028] Third, it alleviates time slot fragmentation and improves schedulability and success rate under high load conditions. Addressing the problem that existing incremental insertion strategies lack assessment and constraints on link time slot occupancy patterns, easily fragmenting continuous idle windows and generating numerous small, unreusable fragments, leading to difficulties in scheduling subsequent critical services, this invention designs a comprehensive cost function incorporating fragmentation awareness, latency, and link load. During the selection of feasible insertion points, it evaluates the degree of fragmentation of the inserted slots, prioritizing scheduling locations that minimize disruption to continuous idle windows. This reduces fragmentation generation, protects time slot continuity, improves the scheduling success rate of new deterministic services under high load scenarios, and balances end-to-end latency and load balancing.

[0029] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0030] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0031] Figure 1 This is a schematic diagram illustrating the implementation process of the present invention; Figure 2 This is a schematic diagram of network topology modeling in step 1a of the present invention; Figure 3 This is a schematic diagram of the taboo table generation in step 3 of the present invention (assuming a timeout of 4ms). Figure 4 The algorithm's performance is shown in the test scenario of a mixed long and short cycle business (Baseline is the algorithm that iterates through each idle time window). Detailed Implementation

[0032] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0033] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0034] The implementation idea of ​​this invention is as follows: 1) For dynamic access of new TT services, first sort the set of TT services to be accessed according to multiple rules to determine the scheduling priority; 2) For each new TT service, sort its candidate routes in ascending order of hop count and prioritize the shortest path; 3) For each candidate route, extract the occupied time interval of each link on the route, and discretize the time axis into time slots with fixed granularity to compress the search space; 4) Perform offset back-reasoning and left expansion on each hop occupied interval to obtain the conflict interval, project it onto the service cycle domain and construct a tabu table of the full path domain using a bitmap; 5) Quickly search and locate feasible phases in the tabu table by clearing the bitmap, and enumerate the same phase instances within the supercycle to generate candidate start times; 6) First verify the end-to-end maximum delay constraint for the candidate start times, and then evaluate the candidate solutions with a comprehensive cost function that includes fragmentation cost, delay cost and link load cost; 7) Select the scheme with the minimum comprehensive cost from all candidate routes and start times to achieve rapid incremental scheduling of new TT services.

[0035] refer to Figure 1 As shown, the present invention may specifically include the following steps: Step 1; Construct a Time-Sensitive Network (TSN) scheduling model: Step 1a: Define the network topology as a directed graph and number all nodes and links in the network; Step 1b: Initialize the processing latency of switches in the network Link propagation delay Link bandwidth ; Step 1c: Obtain the routing parameters of all scheduled TT services, and define the routes of all scheduled TT services according to the link numbers in Step 1a; Step 1d: Obtain the period parameters of all TT streams to be scheduled, calculate the least common multiple of the periods of all TT streams, and denote it as the supercycle period. ; Step 2: Preprocessing of new business flows: Step 2a: Obtain the period of all new flows Transmission delay (Including frame length and physical layer overhead) and an optional route list, sort all new service flows according to a multi-rule method, and sort the optional routes for each new service; Step 2b: Calculate the scaled period based on the time granularity; Step 3: Construct the module domain taboo table: For each candidate route of the new flow, construct a length of A bitmap is used to identify which time slots on this route are unavailable. The specific steps are as follows: Step 3a: Initialize the all-zero bitmap, size is .

[0036] Step 3b: Traverse each hop link on the route : Calculate cumulative offset This means the data packet has arrived at the first... The relative time of the jump; Obtain all existing flow intervals on this link. ; Step 3c: Offset and Leftward Expansion: To transform the conflict detection between intervals into the detection between points and bitmaps, the conflict intervals are expanded to the left to obtain the original conflict intervals. ; Step 3d: Coordinate Discretization and Modular Projection: Convert nanosecond-level coordinates to time slot indexes , ; for scaling period Modulo operation projects the global time onto a single cycle, thereby forming a taboo set of the h-th hop link pair source phase in the bitmap; Step 4: Rapid search and multi-dimensional cost assessment.

[0037] The optimal sending time is found using the constructed taboo list. The specific steps are as follows: Step 4a: Bitwise operation fast search: Search the tabu table for all indices with a value of 0 (i.e., conflict-free transmission times).

[0038] Step 4b: For each feasible transmission time, calculate the multidimensional cost TotalCost: Step 5: Scheduling Decisions and Applications Step 5a: Compare the TotalCost of all candidate sending times in the current TT stream and select the scheme with the minimum TotalCost; Step 5b: Does a TT stream exist to be scheduled? If it does, proceed to step 2a; otherwise, proceed to step 6. Step 6: Generate the transmission timetable for each system and the gating list for each port of the switch based on the scheduling results.

[0039] The steps in this exemplary embodiment will now be described in more detail with reference to the accompanying drawings and embodiments.

[0040] In step 2a, the process of sorting all services to be scheduled using multiple rules is as follows: 2a1) Obtain the attribute parameters of each time-triggered service in the set of services to be scheduled, including at least: whether it is a multicast service (isMulticast identifier), period. Maximum permissible end-to-end latency (maxLatency) and single-frame transmission latency. ; 2a2) Using isMulticast as the first sorting key, sort in descending order to give priority to multicast services over unicast services; 2a3) If the first sorting key is the same, use the maxLatency as the second sorting key and sort in ascending order so that services with smaller maximum allowable end-to-end latency take priority over services with larger maximum allowable end-to-end latency. 2a4) If the first and second sorting keys are the same, then... As the third sorting key, it sorts in ascending order, prioritizing businesses with shorter cycles over those with longer cycles; 2a5) If the first, second, and third sorting keys are all the same, then... As the third sorting key, it sorts in descending order, giving priority to services with longer single-frame transmission latency than services with shorter single-frame transmission latency; 2a6) Output the sorted sequence of services to be scheduled, and execute the subsequent candidate route search and start transmission time determination steps in the order of the sequence.

[0041] In step 2b, the method for calculating the scaled period based on the time granularity is as follows:

[0042] Among them, when When the calculation result is 0, let =1 is used to avoid the unavailability caused by a bitmap length of 0; all subsequent bitmap operations are performed within this value. Within the scope, Time granularity; In step 3c, the implementation process of offset backward pushing and leftward expansion is as follows: 3c1) Let the single-frame transmission delay of the newly added TT service on each hop link be... The link propagation delay is The switch processing latency is Assume the candidate routes include Hop links, in the direction from source to destination, the first hop link denoted as ,in ; 3c2) For the first hop link Calculate the cumulative offset of the hop relative to the time the packet was sent from the source. The calculation method is as follows:

[0043] The cumulative offset is used to characterize the elapsed time since the source's initial transmission time. After the hop transmission, it corresponds to the first The time offset of the actual start time of the hop link's occupancy; 3c3) Get the first hop link The set of occupied intervals for already scheduled services Any interval occupied is denoted as a half-open interval. (Left closed, right open) indicates that the link is in The space is already occupied; 3c4) Perform offset back-calculation processing on any occupied interval [L,R) to obtain its equivalent occupied interval relative to the source packet sending time reference. The calculation method is as follows:

[0044]

[0045] The offset back-calculation is used to uniformly map the multi-link occupancy constraints to the same source end packet transmission time reference system; 3c5) Based on the completed offset reverse calculation, to ensure that the newly added TT service is in the first... hop link The sending window does not overlap with an already occupied window. The equivalent occupied interval is then extended to the left by an amount equal to the transmission delay of the new TT service. , obtain the conflict interval :

[0046]

[0047] in This indicates that all factors leading to new business are in the [number]th [period]. hop link The set of initial transmission times of the source ends where the collision occurred; 3c6) The conflict interval By time granularity Convert to a time slot index range, and calculate using both the floor-down division method and the floor-up strategy:

[0048]

[0049] 3c7) For the obtained slot index range Perform a modal domain projection so that it falls within the scaling period range. The projection method is as follows:

[0050]

[0051] Thus forming the first in the bitmap hop link A set of taboos regarding the timing of transmission from the source end; 3c8) Perform the above processing on all hop links on the route, and combine the taboo sets obtained from each hop to form a full path modal domain taboo table, which is used for fast search at subsequent conflict-free transmission times.

[0052] In step 4a, the bitwise operation fast search method is as follows: 4a1) Scaling the period length Build length is The taboo bitmap BitSet contains a bit value of 1 indicating that the transmission time index is the index of the forbidden transmission time, and a bit value of 0 indicating that the phase is the index of the feasible transmission time; the taboo bitmap is obtained by setting the bits of the union of the conflict intervals of each link of the TT service after projection into the modal domain; 4a2) Using the bitmap's zero-bit positioning operation in The method involves quickly enumerating all feasible transmission times within a given range. Specifically, it starts from index 0 and calls the next zero-based search operator to obtain the index of the first feasible transmission time. and iteratively execute

[0053] Until the returned index exceeds up to the range; where This means searching for the next bit with a value of 0 to the right from a given starting index in the bitmap. Its output is directly the set of feasible transmission times t, thus avoiding slot-by-slot comparison and improving the speed of transmission time search. 4a3) For any feasible transmission time index s, restore it to the original time on the timeline. The restoration method is as follows:

[0054] in This refers to the time granularity.

[0055] In step 4b, the multidimensional cost TotalCost is calculated as follows: 4b1) Calculate the cost of fragment perception Based on the constructed module domain taboo table, locate the starting time slot index of the current traffic to be scheduled within the module domain. and end gap index Forward search domain taboo table The most recently set bit is 1, and the left free gap is calculated. Backward search The most recently set bit is then used to calculate the right-side free gap. ;like or If the fragment size is less than the preset minimum fragment threshold, then the fragment perception cost is... Apply a penalty value; simultaneously, incur a fragment perception cost. It is inversely proportional to the total free space on the left and right sides; 4b2) Calculate the basic latency cost Calculate the ratio of the actual end-to-end physical transmission delay of the current traffic to the maximum allowable delay, denoted as the routing quality score; calculate the ratio of the start time of the current traffic transmission to the supercycle period (MC), denoted as the compactness score; and perform a weighted sum of the routing quality score and the compactness score to obtain the base delay cost. ; 4b3) Calculate the link load balancing cost : Iterate through each hop of the current traffic routing path. Get the load rate of the current link. and the load rate of the next hop link ; right and The single-hop load cost is obtained by weighting the data, and then the single-hop load costs of all links on the path are summed to obtain the total link load balancing cost. ; 4b4) Calculate the total cost: Based on the preset fragment weight factor Delay weighting factor and load weighting factor The total cost is calculated using the following formula:

[0056] In step 6, the method for generating the transmission timetable for each terminal system and the gating list for each port of the switch based on the scheduling results is as follows: 1) Iterate through the transmission links corresponding to all end system ports in the network and the transmission links corresponding to each port of the switch one by one; 2) Based on the currently traversed link, initialize the gating list of the corresponding port of the link to be empty, select the first time slot interval of the current link, and determine whether the left endpoint of the time slot interval is 0. If the left endpoint of the time slot interval is 0, set the gating duration to the length of the time slot interval and add the gating record "OCCCCCC" and the gating duration to the gating list; otherwise, first set the gating duration to the left endpoint of the time slot interval, add the gating record "COOOOOOO" and the gating duration to the gating list, and then set the gating duration to the length of the time slot interval and add the gating record "OCCCCCC" and the gating duration to the gating list. 3) Determine if the current time slot interval is the last time slot interval of the current link. If not, select the next time slot interval and proceed to step 4); otherwise, proceed to step 5). 4) First, set the gating duration to the difference between the left endpoint of the current time slot interval and the right endpoint of the previous time slot interval, add the gating record "COOOOOOO" and the gating duration to the gating list, then set the gating duration to the length of the time slot interval, add the gating record "OCCCCCC" and the gating duration to the gating list, and execute 3). 5) Calculate the difference between the supercycle HC and the current time slot interval. If the difference is not 0, first set the gating duration to the difference, and add the gating record "COOOOOOO" and the gating duration to the gating list. 6) The gating list for the current port has been generated.

[0057] like Figure 4 As shown, the performance comparison between the proposed algorithm and the baseline algorithm under different scales of TT services is presented. Regarding scheduling success rate, the proposed algorithm exhibits stronger robustness as the number of service flows increases, with a success rate significantly higher than the baseline algorithm. More significantly, in terms of solution time, the average time of the baseline algorithm increases significantly with the traffic scale, while the time increase of the proposed algorithm is extremely gradual, remaining at a low level. This fully demonstrates that the proposed algorithm possesses good computational efficiency and scalability under different scale scenarios.

[0058] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0059] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0060] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is defined only by the appended claims.

Claims

1. An incremental scheduling method for deterministic services in time-sensitive networks, characterized in that, The method includes: Step 1: Construct a Time-Sensitive Network (TSN) scheduling model: Step 1a: Define the network topology as a directed graph and number all nodes and links in the network; Step 1b: Initialize network parameters, including switch processing latency, link propagation latency, and link bandwidth; Step 1b: Obtain the routing parameters for all scheduled time-triggered TT services; Step 1d: Obtain the period parameters of all TT services to be scheduled, and calculate the least common multiple of all periods as the super-period; Step 2: Preprocessing of new business flows: Step 2a: Obtain the period, transmission delay, and optional route list for all newly added TT services, sort all newly added TT services according to the multi-rule method, and sort the optional routes for each newly added service by the number of hops; Step 2b: Calculate the scaling of the period for the newly added TT service based on the preset time granularity; Step 3: Construct the module domain taboo table: For each candidate route of a newly added TT service, perform the following operations: Step 3a: Initialize a zero-based bitmap with a length equal to the scaled period; Step 3b: Traverse each hop link on the candidate route, calculate the cumulative offset of the data packet to the hop link, and obtain the time interval of all scheduled services on the link; Step 3c: Based on the cumulative offset and the transmission delay of the new service, perform offset backward calculation and leftward expansion processing on each occupied time interval to obtain the conflict time interval in the reference frame of the source end transmission time; Step 3d: Discretize the conflict time interval into time slot index intervals according to the time granularity, and take the modulus of the scaled period for the time slot index interval, project it into the modulus domain, and set the corresponding modulus domain index position in the bitmap; after traversing all hop links, obtain the modulus domain taboo table corresponding to the candidate route; Step 4: Rapid Search and Multi-Dimensional Cost Assessment Step 4a: In the tabu table of the modal domain, quickly search all indices with a bit value of zero using bitmap operations to send phases as candidate collision-free source ends; Step 4b: For each candidate collision-free transmission phase, restore it to the supercycle time axis to generate the candidate's start transmission time, and evaluate it using a comprehensive cost function, which includes at least fragment awareness cost, latency cost, and link load balancing cost. Step 5: Scheduling Decisions and Applications Step 5a: Select the scheme with the lowest overall cost from all candidate start times of the newly added TT service as the scheduling result; Step 5b: Update the time occupancy status of each link in the network based on the scheduling results; Step 6: Repeat steps 2 to 5 until all newly added TT services have been scheduled. Finally, generate the transmission timetable for each system and the gating list for each port of the switch based on the scheduling results of all services.

2. The method according to claim 1, characterized in that, Step 2a, which involves sorting all newly added TT services according to a multi-rule method, specifically includes: Obtain the attribute parameters for each newly added TT service, including: whether it is a multicast service identifier, period, maximum allowed end-to-end latency, and single frame transmission latency; Sort in descending order based on whether it is a multicast service identifier as the first sorting key; If the first sorting key is the same, the maximum allowable end-to-end delay is used as the second sorting key, and the data is sorted in ascending order. If the first and second sorting keys are the same, use the cycle as the third sorting key and sort in ascending order; If the first, second, and third sorting keys are all the same, the single-frame transmission delay is used as the fourth sorting key, and the data is sorted in descending order.

3. The method according to claim 1, characterized in that, In step 2b, the root is used to scale the period of the newly added TT service, using the following formula: Where Period is the original period for new TT services. For time granularity, This is the scaled period.

4. The method according to claim 1, characterized in that, The offset backward calculation and leftward expansion process described in step 3c are as follows: For an occupied time interval [L, R) of the h-th hop link on a candidate route, its conflict time interval [RawStart, RawEnd) relative to the source transmission time is calculated as follows: in, To reduce the single-frame transmission latency of the newly added TT service, This is the cumulative offset when the data packet arrives at the h-th hop link.

5. The method according to claim 1, characterized in that, The step 4a described above, which involves quickly searching for all indices with a value of zero using bitmap operations, specifically involves iteratively calling the next zero-bit search operator, starting from index 0. This continues until the returned index exceeds the range of the scaled period SP, where BitSet is a modal tabu bitmap. For the currently found feasible phase index, This indicates that the next bit with a value of 0 is searched to the right of the given starting index in the bitmap.

6. The method according to claim 1, characterized in that, The calculation method for the total cost function TotalCost mentioned in step 4b is as follows: in, The cost of fragment perception is calculated based on the size of the idle gap on both sides of the candidate scheduling position; The latency cost is calculated based on the ratio of the actual end-to-end transmission latency to the maximum allowable latency and the compactness score of the transmission time. The link load balancing cost is calculated based on the load rate of each link on the candidate routing path; , , These are the corresponding weighting factors.

7. The method according to claim 1, characterized in that, Step 6, which involves generating the gating list for each port of the switch, specifically includes: for each port's corresponding transmission link, traversing all occupied time slot intervals within the supercycle; for the first time slot interval, if its left endpoint is not 0, adding a gating entry indicating a closed state; for each occupied time slot interval, adding a gating entry indicating an open state; between two consecutive occupied time slot intervals, if there is an idle gap, adding a gating entry indicating a closed state; finally, if the right endpoint of the last occupied time slot interval is less than the supercycle, adding a gating entry indicating a closed state to make the gating list cyclically cover the entire supercycle.

Images