Optical burst link conflict control method and system based on priority scheduling

Abstract

The application discloses a method and system for controlling optical burst link conflict based on priority scheduling, and relates to the technical field of optical communication network control. A burst joint scheduling and conflict cost evaluation mechanism based on priority weight is introduced in an OBS core node. The method comprises the following steps: sorting control packets in a sliding time window, and combining bias time and burst length to predict an occupied time interval; using a lightweight data structure for recording idle intervals of each wavelength to construct a candidate gap set; calculating the conflict cost of each candidate wavelength according to priority weight, time overlap relationship and gap change after insertion, selecting the wavelength with the minimum cost, and updating idle interval information in real time. Thus, the priority of high-priority services is guaranteed, and the wavelength resources are finely allocated, so that the burst conflict and discard rate are reduced, and the OBS network throughput and stability are improved in a high-load scenario.

Description

Technical Field

[0001] This application relates to the field of optical communication network control technology, and in particular to a method and system for controlling optical burst link conflicts based on priority scheduling. Background Technology

[0002] Optical Burst Switching (OBS) is a hybrid technology between Optical Circuit Switching (OCS) and Optical Packet Switching (OPS). It aggregates bursts at the edge, rapidly switches wavelengths at the core, and performs control plane processing in the electronic domain. OBS is more efficient than OCS and easier to implement than OPS. As a highly efficient optical communication technology, OBS forwards burst data units in an all-optical manner at the core node, significantly improving network throughput and reducing latency.

[0003] However, OBS core nodes typically lack sufficient all-optical storage capacity, making it impossible to buffer or queue bursts like traditional electronic nodes. When multiple service flows compete for resources on the same node, bursts are easily dropped if the target wavelength is already occupied, as they cannot obtain available channels in time, leading to a significant decrease in network performance as the load increases. Furthermore, with the diversification of service models, the uncertainty of burst arrival times and differences in service levels make link resource allocation more complex, further exacerbating the difficulty of conflict handling.

[0004] Currently, OBS networks generally rely on a control packet-driven fast scheduling process to complete burst link selection. However, due to the large number of bursts and the strong randomness of service arrival, core nodes still face challenges in completing effective scheduling within a very short processing time. Especially in high-load scenarios, traditional scheduling methods cannot simultaneously take into account resource utilization, service priority guarantee, and overall transmission performance, resulting in certain limitations on network throughput and stability. Summary of the Invention

[0005] This application provides a method, system, storage medium, computer program product, and electronic device for controlling optical burst link conflicts based on priority scheduling, which can at least solve the problems of uneven allocation of link resources, burst packet loss, and scheduling delay in OBS networks in the current related technologies.

[0006] In a first aspect, embodiments of this application provide a priority-based optical burst link conflict control method applied to an OBS core node. The method includes: acquiring multiple optical burst control packets arriving within a sliding time window, and sorting them according to arrival time to form a control packet scheduling list; for each optical burst control packet in the control packet scheduling list, parsing the burst length, offset time, destination port, and priority label in the optical burst control packet, mapping the priority label to weights to determine corresponding priority weights, and predicting the occupancy time interval of the burst corresponding to the optical burst control packet on the output link based on the offset time and burst length; and constructing a connection between the optical burst control packet and each optical burst control packet in the control packet scheduling list from the lightweight data structure set for each output link. The system includes a set of candidate gaps where the time intervals of bursts corresponding to burst control packets overlap; a lightweight data structure is used to record the idle intervals of each wavelength on the corresponding output link; for each optical burst control packet in the control packet scheduling list, on each candidate wavelength corresponding to the candidate gap set of the optical burst control packet, the conflict cost of the burst on each candidate wavelength is calculated based on the priority weight of the optical burst control packet, the time overlap with the scheduled bursts, and the gap change caused after insertion into the corresponding candidate wavelength, and the wavelength with the smallest conflict cost is selected as the target wavelength; the scheduling result of each optical burst control packet is determined based on the conflict cost of the target wavelength, and the wavelength idle interval of the corresponding output link is modified based on the scheduling result to update the corresponding lightweight data structure.

[0007] Secondly, embodiments of this application provide a priority-based optical burst link conflict control system deployed on an OBS core node. The system includes: a scheduling list generation unit, used to acquire multiple optical burst control packets arriving within a sliding time window and sort them according to arrival time to form a control packet scheduling list; a burst parsing unit, used to parse the burst length, offset time, destination port, and priority label of each optical burst control packet in the control packet scheduling list, map the priority label to a weight to determine the corresponding priority weight, and predict the occupancy time interval of the burst corresponding to the optical burst control packet on the output link based on the offset time and burst length; and a candidate gap set construction unit, used to construct a gap set corresponding to the control packet scheduling list from the lightweight data structure set for each output link. The table contains a set of candidate gaps where the time intervals of the bursts corresponding to each optical burst control packet overlap. The lightweight data structure records the idle intervals of each wavelength on the corresponding output link. A conflict cost analysis unit calculates the conflict cost of each optical burst control packet in the control packet scheduling list for each candidate wavelength corresponding to the candidate gap set of the optical burst control packet. This calculation is based on the priority weight of the optical burst control packet, its time overlap with already scheduled bursts, and the gap change after insertion into the corresponding candidate wavelength. The unit then selects the wavelength with the lowest conflict cost as the target wavelength. A burst scheduling control unit determines the scheduling result of each optical burst control packet based on the conflict cost of the target wavelength and modifies the wavelength idle intervals of the corresponding output link based on the scheduling result to update the corresponding lightweight data structure.

[0008] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the priority-based scheduling optical burst link conflict control method of any embodiment of this application.

[0009] Fourthly, embodiments of this application provide a storage medium storing a computer program thereon, characterized in that, when the program is executed by a processor, it implements the steps of the priority scheduling-based optical burst link conflict control method of any embodiment of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the priority-based scheduling optical burst link conflict control method of any embodiment of this application.

[0011] The optical burst link conflict control method and system based on priority scheduling provided in this application can achieve at least the following technical effects:

[0012] (1) By combining the priority scheduling mechanism with the conflict cost calculation, the resource allocation efficiency and service guarantee capability of the optical burst switching network under high load environment are significantly improved. Specifically, after the control packets are priority weighted, the scheduling process takes into account the time overlap relationship, weight difference and the gap change after insertion, and accurately selects the wavelength with the minimum conflict cost for allocation, thereby ensuring that high priority bursts obtain priority transmission paths, effectively optimizing the overall network throughput and stability, and avoiding the performance degradation caused by excessive resource competition.

[0013] (2) By utilizing the sorting and conflict cost calculation of optical burst control packets within a sliding time window, the utilization efficiency of wavelength resources is further improved. In traditional scheduling methods, resource allocation is usually affected by the uncertainty of burst arrival time and the randomness of burst traffic, leading to resource conflicts and burst packet loss. However, this scheme, by considering the burst offset time, length, and predicted occupancy time interval during control packet scheduling, can accurately calculate the overlap relationship between bursts and wavelength idle intervals, and select the most suitable wavelength for scheduling based on this. Through this precise scheduling method, resource conflicts can be significantly reduced, burst packet loss can be reduced, and network stability and transmission capacity can be improved.

[0014] (3) By using lightweight data structures, scheduling efficiency is further improved while simplifying resource management and scheduling calculation processes, reducing the storage and computing pressure on core nodes. By recording the wavelength idle interval for each output link and updating the corresponding lightweight data structure, a candidate gap set can be efficiently constructed, and it can be quickly determined whether a burst can be scheduled on that wavelength. Thus, while retaining efficient scheduling capabilities, the calculation speed of scheduling is also greatly improved, which can meet the real-time scheduling requirements under high load environments.

[0015] This technical solution significantly improves the resource allocation efficiency, throughput, and service assurance capabilities of optical burst switching networks under high load and complex service environments by introducing priority scheduling, conflict cost minimization mechanisms, and lightweight data structures. It not only optimizes the burst resource scheduling process, reducing resource conflicts and packet loss, but also enhances network stability and real-time scheduling performance, enabling it to cope with changing network demands and service loads, and ensuring the efficient operation of the OBS network under high load conditions. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating an example of a priority-based scheduling-based optical burst link conflict control method according to an embodiment of this application is shown.

[0018] Figure 2 A flowchart illustrating an example of generating a control packet scheduling list based on a sliding time window according to an embodiment of this application is shown.

[0019] Figure 3 An example operation flowchart is shown, illustrating an embodiment of the present application, for predicting the time interval occupied by each optical burst control packet on the output link.

[0020] Figure 4 A flowchart illustrating an example of using a conflict cost mechanism to allocate target wavelengths for each optical burst control packet according to an embodiment of this application is shown.

[0021] Figure 5 A flowchart illustrating an example of determining the scheduling results of each optical burst control packet using the collision cost of the target wavelength according to an embodiment of this application is shown.

[0022] Figure 6 A diagram illustrating the comparison of average loss rates for bursts of different priorities under varying peak loads is provided.

[0023] Figure 7 A comparative illustration of the complexity of three scheduling algorithms in scheduling bursts is shown;

[0024] Figure 8 A structural block diagram of an example of a priority-based scheduling optical burst link conflict control system according to an embodiment of this application is shown. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] It should be noted that in current related technologies, some studies have proposed the LAUC-VF (Latest Available Unused Channel with Void Filling) algorithm for OBS scheduling. This is an unbuffered OBS scheduling algorithm that maintains the start and end times of reserved bursts on each wavelength and scans the gap intervals of all wavelengths when a new burst arrives. It selects the wavelength that minimizes the remaining "start gap" between the end of the new burst and the start of the next burst, thus maximizing the utilization of existing gaps. This makes it superior to the Horizon algorithm, which only records the last occupied time, in terms of utilization and packet loss rate. However, LAUC-VF requires checking all wavelengths and their gap intervals during each scheduling, and its complexity is proportional to the number of wavelengths and gaps, making it unsuitable for real-time hardware implementation.

[0027] In addition, some advanced scheduling algorithms may support priority-differentiated services in the OBS network. Their decision-making is mainly based on queue order, resource availability and simple preemption rules. However, these algorithms are prone to causing long-term starvation of low-priority traffic, increased latency jitter and overall fairness imbalance under high load conditions. At the same time, the hardware implementation is complex and it is difficult to fully balance the risk of conflict and the efficiency of resource utilization in real-time scheduling.

[0028] It should be understood that the above description of the relevant technologies is intended only to help the public better understand the inventive spirit and motivation of this application, and is not intended to limit this application. Furthermore, the technical solutions described in the above-mentioned relevant technologies are not prior art, and may also be undisclosed technical solutions, such as those under research or in the laboratory stage.

[0029] The technical solutions in this application, including the collection, storage, use, processing, transmission, provision, and disclosure of users' personal information, comply with relevant laws and regulations and do not violate public order and good morals.

[0030] Figure 1 A flowchart illustrating an example of a priority-based scheduling optical burst link conflict control method according to an embodiment of this application is shown.

[0031] Regarding the execution entity of the method in this application embodiment, it can be any controller or processor with computing or processing capabilities, such as an optical burst scheduling controller deployed in the OBS core node. This controller performs real-time parsing of burst control packets, rapid retrieval and updating of resource status, and wavelength selection by combining priority weights and conflict cost models, thereby achieving fine-grained scheduling of link resources. This controller can complete joint scheduling decisions for multiple bursts within a very short processing time, ensuring timely protection of high-priority services, while improving wavelength utilization and reducing link conflicts, thus enabling the OBS core node to operate efficiently and stably under high load environments.

[0032] In some examples, it may be integrated into an electronic device or terminal through software, hardware, or a combination of both, and the type of terminal or electronic device may be diverse.

[0033] like Figure 1 As shown, in step S110, multiple optical burst control packets arriving within the sliding time window are obtained, and a control packet scheduling list is formed by sorting them according to their arrival time.

[0034] It should be noted that the OBS core node itself does not have optical domain caching capabilities for burst data, therefore it needs to rely on control packets that arrive in advance as the basis for scheduling. In order to ensure that the scheduling decision has temporal continuity and reflects the actual competition relationship of burst arrivals, control packets are locally batched with a sliding time window as the boundary, and a reliable scheduling sequence is formed based on the actual arrival order.

[0035] Specifically, the OBS core node continuously monitors the control channel and adds control packets arriving within the current window period to the buffer queue. As the window slides, older packets automatically exit the window, maintaining the temporal locality of the scheduling range.

[0036] In some implementations, the system captures all arriving control packets (including burst offset time, destination port, priority label, etc.) within a set sliding time window. The sliding window can adaptively adjust according to node processing capacity and network load, ensuring that the window covers a sufficient number of bursts without causing accumulated scheduling delays. Control packets within the window are sorted according to their recorded arrival times to form a control packet scheduling list. This allows the scheduler to grasp the burst arrival status (not just individual control packets) within a short timeframe, providing an input basis for subsequent multi-burst joint scheduling.

[0037] In step S120, for each optical burst control packet in the control packet scheduling list, the burst length, offset time, destination port and priority label in the optical burst control packet are parsed, the priority label is mapped to a weight to determine the corresponding priority weight, and the occupancy time interval of the corresponding burst on the output link is predicted based on the offset time and burst length.

[0038] By analyzing burst length and offset time, the start and end times of burst arrival at the output link can be predicted, and the corresponding output link can be determined by combining the destination port information, thus obtaining the burst's occupancy time interval on the output link. By mapping priority labels to weight values, a unified metric that can be used for conflict calculation can be generated. The mapping method can adopt linear, multi-level, or exponential weight models to adapt to different operator strategies.

[0039] Based on these parameters, the time interval occupied by the burst on the corresponding output link can be accurately calculated within the electronic domain. An exemplary calculation process is as follows:

[0040] Start time = Control packet reception time + Offset time

[0041] End time = Start time + (Burst length / Link transmission rate)

[0042] This interval will serve as the core basis for collision detection and gap matching.

[0043] It should be understood that the simplified calculation process above is only used as an example to illustrate the meaning of each term. The calculation method can be diverse and more advanced and refined calculation models can be used, which are not limited here.

[0044] By unifying and parameterizing the expression of sudden attributes, the system can accurately predict the time window of sudden occupancy and quantitatively guarantee resources for services of different priorities. This transforms the scheduling decision-making process from qualitative judgment to quantifiable calculation, improving the interpretability and targetability of scheduling behavior.

[0045] In step S130, based on the lightweight data structure set for each output link, a set of candidate gaps is constructed from the lightweight data structure that overlaps with the occupancy time intervals of the bursts corresponding to each optical burst control packet in the control packet scheduling list; the lightweight data structure is used to record the idle intervals of each wavelength on the corresponding output link.

[0046] It should be noted that the OBS core node cannot cache bursts, so it is necessary to find wavelength time slots that can accommodate bursts in advance in order to avoid ineffective scheduling.

[0047] In some implementations, a lightweight data structure is maintained for each output link to record the idle time intervals of each wavelength (e.g., using an ordered linked list or compact array structure) to enable fast retrieval, insertion, and deletion operations.

[0048] More specifically, based on the burst's predicted occupancy time interval, all wavelengths' idle intervals are queried on the corresponding destination output link. For each wavelength, those gap intervals that overlap with the burst's occupancy time interval or can potentially accommodate the burst are identified. These gap intervals that meet scheduling possibilities are formed into a candidate gap set. Thus, through a lightweight data structure, a full search of the entire wavelength resource table is avoided, resource matching efficiency is improved, and link resource status can be maintained in real time.

[0049] In step S140, for each optical burst control packet in the control packet scheduling list, on each candidate wavelength corresponding to the candidate gap set of the optical burst control packet, the conflict cost of the burst on each candidate wavelength is calculated according to the priority weight of the optical burst control packet, the time overlap relationship with the scheduled burst, and the gap change caused after insertion into the corresponding candidate wavelength, and the wavelength with the smallest conflict cost is selected as the target wavelength.

[0050] Here, when multiple candidate wavelengths can accommodate bursts, the multi-dimensional impact is quantified through a conflict cost model by considering comprehensive business priorities, the degree of future resource fragmentation, and overlapping relationships, and the wavelength with the most global benefit is selected for scheduling and allocation.

[0051] Specifically, the cost model comprehensively considers multiple factors, such as burst priority weight, time overlap, and the impact of gap changes. Regarding burst priority weight, bursts with higher service levels receive a higher cost weight for violating their requirements. For time overlap, if the occupied interval of a burst is close to or has a high risk of overlapping with the interval of already scheduled bursts, the cost increases. Regarding the impact of gap changes, if burst insertion may cause an increase in wavelength time fragmentation, its impact on future scheduling efficiency also increases the cost; conversely, if gaps can be tightly filled, the cost decreases.

[0052] By jointly quantifying the scheduling conflict cost through multiple factors, differentiated resource selection is achieved, so that scheduling is no longer limited to the current availability judgment of resources, but also considers the future resource utilization structure, effectively improving the long-term performance of the entire network and the stability under high load scenarios.

[0053] In step S150, the scheduling result of each optical burst control packet is determined according to the collision cost of the target wavelength, and the wavelength idle interval of the corresponding output link is modified based on the scheduling result to update the corresponding lightweight data structure.

[0054] In some implementations, the system adds bursts to the occupied range of a determined target wavelength, then modifies the wavelength free range list (e.g., splitting gaps or reducing boundaries), and synchronizes the update results back to a lightweight data structure. The entire update process can be performed incrementally, avoiding global reconstruction and improving real-time performance.

[0055] Specifically, after selecting the target wavelength, the OBS core node inserts the burst's occupied time interval into the existing idle interval structure for that wavelength. Simultaneously, it modifies, splits, or deletes occupied idle interval segments and updates the lightweight data structure to reflect the new wavelength resource status in real time. Then, it uses the final scheduling result to drive the optical switch configuration, ensuring that the burst switches to the designated wavelength on time.

[0056] Through the embodiments of this application, an operational chain is adopted sequentially, including priority mapping, accurate occupancy prediction, candidate gap set construction, conflict-cost-based scheduling optimization, and lightweight resource status updates. This enables the OBS core node to maintain efficient and stable burst scheduling capabilities under high load conditions. This not only improves the burst scheduling success rate and reduces burst loss, but also significantly enhances wavelength resource utilization and overall network throughput performance, ensuring the quality of service for different service levels during complex burst contention.

[0057] Figure 2 A flowchart illustrating an example of generating a control packet scheduling list based on a sliding time window according to an embodiment of this application is shown.

[0058] like Figure 2 As shown, in step S210, the length of the sliding time window is dynamically calculated based on the current link load estimate.

[0059] It should be noted that the OBS core node does not have optical caching capabilities, so scheduling relies on the early arrival of control packets. However, a fixed window length cannot adapt to real-time changes in sudden business events: if the window is too large, it will cause scheduling delays; if the window is too small, it will not be able to fully capture sudden conflict relationships.

[0060] In this embodiment, the sliding time window is dynamically adjusted according to changes in link load, enabling the scheduling strategy to adapt to different service densities.

[0061] Equation (1)

[0062] in, The length of the sliding time window. This is the estimated link load. This is an empirical coefficient. and These are the preset minimum and maximum window lengths, respectively.

[0063] Here, the link load estimate Information can be acquired from multiple sources, including but not limited to burst arrival rates, occupancy rates of each wavelength, fragmentation of gaps, and instantaneous growth trends of burst queuing requests, all aggregated in real time by the control plane. After continuously monitoring these quantitative characteristics, core nodes can estimate the current link load level using methods such as moving averages, exponential weighting, predictive models, or statistical estimation based on historical data.

[0064] The busier the link, the shorter the window, in order to reduce aggregation latency. In equation (1), when the network load increases, the window is adjusted according to... The relationship between the control packets is automatically shortened, increasing the scheduling frequency; when the load decreases, the window lengthens, allowing more control packets to be aggregated to improve conflict detection.

[0065] In step S220, arriving optical burst control packets are collected based on a sliding time window and arranged in order of their arrival time to form a control packet scheduling list.

[0066] Specifically, in the window The system continuously collects control packets, adding packets containing the following information to a buffer: offset time, burst length, destination port, priority tag, etc. After collection, a control packet scheduling list is generated from morning to night based on the actual arrival timestamps. The sorting logic must strictly follow the arrival order of the control plane, rather than the service attributes of the bursts themselves, to ensure that conflict judgments accurately reflect physical layer contention.

[0067] In step S230, if the number of optical burst control packets within the sliding time window is less than a preset threshold, the scheduling mechanism is switched from the aggregation scheduling mechanism of the sliding time window to online scheduling. The arriving optical burst control packets are scheduled in real time to reduce scheduling delay and avoid additional waiting caused by window aggregation under low arrival density conditions.

[0068] Sliding window aggregation scheduling is suitable for medium-to-high load scenarios, but under low load conditions, the number of control packets in the window is insufficient. Continuing to wait for the window to close will incur additional scheduling delays, and since there is almost no competition between bursts, aggregation scheduling actually reduces the overall response time. Therefore, it is necessary to switch to online scheduling mode under low arrival density to achieve real-time scheduling. The aforementioned preset threshold can be pre-set or adaptively adjusted, for example, dynamically adjusted based on historical average burst arrival rates or online prediction models.

[0069] In online scheduling mode, control packets do not need to wait for the window to close; each arriving control packet immediately enters the scheduling process. Under low traffic density, the probability of conflicts is low, so real-time scheduling does not degrade scheduling quality. This significantly reduces scheduling waiting time under low load conditions, avoids meaningless delays caused by window aggregation, and improves response speed and link utilization efficiency in low-traffic scenarios.

[0070] This application's embodiments employ a dual-mode adaptive mechanism combining high-load aggregation scheduling and low-load switching online scheduling. Under medium-to-high load conditions, a dynamic window ensures optimal timeliness and aggregation in scheduling across different loads. Under low load conditions, an online scheduling switching mechanism reduces latency at low density and improves overall scheduling flexibility.

[0071] Figure 3 An example operation flowchart is shown, illustrating an embodiment of the present application, for predicting the time interval occupied by each optical burst control packet on the output link.

[0072] like Figure 3 As shown, in step S310, for each optical burst control packet in the control packet scheduling list, the priority label of the optical burst control packet is converted into a priority weight to characterize the scheduling importance of different priority bursts.

[0073] In high-load scenarios, OBS core nodes face limited wavelength resources and extremely valuable scheduling opportunities, necessitating a priority mechanism to ensure that critical service bursts are prioritized. However, raw priority labels are typically linearly differentiated, failing to fully reflect the true scheduling differences between high-priority (e.g., latency-sensitive services) and low-priority bursts. Therefore, priority labels need to be mapped to highly discriminative priority weights, giving high-priority bursts a significant advantage in conflict handling and wavelength selection.

[0074] Specifically, the priority weights are calculated using an exponential mapping method to amplify the weight differences between high and low priorities:

[0075] Equation (2)

[0076] In the formula, Indicates the first Priority label for each optical burst control packet Set to take discrete values An integer, where 1 represents the highest priority. Indicates the lowest priority; Indicates the first Each burst control packet corresponds to a burst priority weight. This represents the cardinality of the exponential mapping.

[0077] In equation (2), The exponential form amplifies the differences in priority levels many times over. For example, when When the weight difference between two adjacent priority levels reaches a doubling relationship, high-priority bursts can be clearly distinguished in conflict cost calculation. Therefore, converting linear priority labels into exponentially discriminative labels significantly enhances the scheduling advantage of high-priority bursts in resource contention.

[0078] In step S320, based on the current scheduling time, the offset time recorded by the optical burst control packet, the burst length, and the cumulative processing delay experienced by the optical burst control packet before reaching the node, and in combination with the link transmission rate, the start time and end time of the occupancy of the optical burst control packet on the output link are calculated, thereby determining the occupancy time interval of the burst corresponding to the optical burst control packet on the output link.

[0079] In this embodiment, the actual start and end time of the burst arriving at the output link is accurately calculated by combining the burst's own parameters (bias, length), scheduling time information, and the cumulative processing delay experienced by the burst before entering this node.

[0080] Equation (3)

[0081] Equation (4)

[0082] In the formula, and They represent the first The start and end times of the occupancy of the burst on the target link corresponding to each optical burst control packet. Indicates the current time of scheduling execution. Indicates the first The transmission bias time carried by each optical burst control packet; Indicates the first The transmission path of an optical burst control packet from the source node to the current node through all intermediate nodes. Representing a path The middle node in This represents the total number of intermediate nodes in the transmission path. Indicates the first The processing delay of optical burst control packets by each intermediate node; Indicates the first Burst length of each optical burst control packet This indicates the transmission rate of the output link.

[0083] In some implementations, the transmission path The routing identifier information carried in the control packet or dynamically resolved by the control plane based on the overall network topology can be obtained. For example, the complete intermediate hop sequence from the source node to the current node can be reconstructed by consulting the local topology table. Simultaneously, the processing delay of each intermediate node along the path can be obtained by parsing the cumulative processing delay field carried in the control packet at previous nodes (e.g., recorded in the extended header). This reflects the actual processing latency experienced at different nodes.

[0084] In equation (3), when calculating the start time, the following is introduced: This effectively compensates for the loss of offset time in actual transmission, making the burst arrival time prediction closer to reality. In particular, OBS control packets experience processing delays (such as parsing, forwarding, queuing delays, etc.) at each intermediate node. These delays affect the effectiveness of offset time. Ignoring them will cause deviations in burst arrival time prediction, leading to inaccurate collision judgment.

[0085] Therefore, by introducing path accumulation processing delay compensation, the misjudgment of conflicts caused by actual transmission delays is effectively avoided. In particular, in large-scale networks, the delay differences of intermediate nodes are significant. The compensation mechanism avoids the accumulation of scheduling errors and improves the stability of overall network scheduling.

[0086] Regarding the details of constructing the candidate gap set, in some examples of embodiments of this application, the lightweight data structure adopts a double-ended queue sorted by start time, and maintains an occupied interval list for each wavelength of each output link. Record the time intervals occupied by currently reserved bursts on that wavelength, and sort these intervals in order of their start times; among them, Indicates link The first A list of occupied intervals for each wavelength, including the reserved time intervals for bursts on that wavelength.

[0087] In some implementations, the system on each output link Each wavelength Maintain a list of occupied intervals sorted in ascending order by start time. Each of them This represents the time interval occupied by a scheduled burst on this wavelength, with all intervals arranged in order. The data is sorted in ascending order (starting at time), and update operations are performed using a double-ended queue (Deque), keeping the complexity of insertion and deletion operations low. Therefore, OBS scheduling can determine interval conflicts based on time order, significantly improving the efficiency of interval queries and insertions through a time-ordered data structure, ensuring real-time performance under high load.

[0088] Then, based on the list of occupied intervals The idle time interval is calculated by using the gaps between adjacent occupied time intervals, and then a set of idle intervals for that wavelength is constructed. ;in, Indicates link Upper A set of idle intervals for each wavelength.

[0089] More specifically, for any adjacent intervals in the occupied interval list and If satisfied Then there is a usable free interval (gap) between the two. The system traverses the entire List all gaps in turn to construct a set of free intervals. .

[0090] Therefore, by using the set of idle intervals to provide an accurate description of the future schedulable time period for each wavelength, the scheduling system can find all potential accommodation windows, providing a reliable basis for burst insertion.

[0091] Furthermore, for each optical burst control packet in the control packet scheduling list, the overlap relationship between the occupied time interval of the optical burst control packet and the set of idle intervals of each output link is calculated. The idle intervals that overlap with the occupied time interval of the optical burst control packet are selected, and then a set of candidate slots is constructed for the optical burst control packet on each output link.

[0092] More specifically, for each optical burst control packet in the control packet scheduling list, the system will occupy the specified interval. Set of idle intervals for each wavelength of each output link Compare them one by one, and determine the candidate free intervals that can be accommodated based on the following conditions. and .

[0093] If a free interval satisfies the above inclusion relationship, then it is considered as the first... A control packet in the link wavelength Add the candidate gaps above to the candidate gap set:

[0094] Equation (5)

[0095] This application's embodiments employ a time-sorted double-ended queue to maintain occupied intervals, allowing free intervals to be obtained through a single sequential traversal. This reduces complex sorting and search operations, improving real-time scheduling capabilities under high load scenarios. Furthermore, by accurately comparing burst occupied intervals with free intervals, not only can wavelengths that can accommodate bursts be found, but also erroneous conflict judgments caused by inconsistent states or insufficient calculations are avoided, improving scheduling reliability. Additionally, the gap calculation covers all unused time periods on the wavelength, ensuring a complete candidate gap set without omitting any schedulable gaps, providing a complete search space for subsequent conflict cost selection.

[0096] Figure 4 A flowchart illustrating an example of using a conflict cost mechanism to allocate target wavelengths for each optical burst control packet according to an embodiment of this application is shown.

[0097] like Figure 4 As shown, in step S410, the occupancy time interval of the burst corresponding to the optical burst control packet is compared with the occupancy time interval of the scheduled bursts on each candidate wavelength, the time overlap length is calculated and the conflict coverage rate is determined.

[0098] Here, the scheduler first obtains the target occupancy time interval of the current optical burst control packet, and compares this interval with the occupancy time intervals of the scheduled bursts on each candidate wavelength. During the comparison, the time overlap length of the two is calculated through interval intersection operation, and the total burst length of the optical burst control packet is used as the normalization benchmark to obtain the collision coverage rate.

[0099] Equation (5)

[0100] In the formula, Indicates the first The individual burst control package and the first The coverage rate of conflicts between dispatched emergencies Indicates the first The individual burst control package and the first The time overlap length of each scheduled emergency. Indicates the first The burst length of each optical burst control packet.

[0101] In equation (5), the index It can quantitatively reflect the potential conflict intensity between the current burst and the existing burst in the time dimension at a certain candidate wavelength. The greater the conflict coverage, the higher the potential conflict risk, thus realizing a fine-grained time overlap measurement method.

[0102] In step S420, a priority weight coupling function is introduced. When the priority weight of the burst corresponding to the optical burst control packet is much greater than the priority weight of the scheduled burst, the priority weight coupling function approaches its maximum value, so as to reflect that the conflict cost of high priority bursts to low priority bursts is higher.

[0103] To ensure that scheduling results balance conflict avoidance and service priority assurance, a priority weight coupling function is introduced. This function establishes a nonlinear mapping based on the difference in priority weights between the current burst and the already scheduled bursts:

[0104] Equation (6)

[0105] In the formula, This represents the priority weight coupling function. Indicates the first Each burst control packet corresponds to a burst priority weight. Indicates the first Priority weights for each scheduled emergency.

[0106] In equation (6), when Significantly greater than hour, Approaching the maximum value indicates that a higher penalty should be applied when a high-priority service conflicts with a low-priority burst; when the two are close or a low-priority burst is inserted into a gap of a high-priority burst, the function value tends to be smaller.

[0107] Through the weighted coupling function, the scheduler can dynamically differentiate the conflict costs between high and low priority services, prioritizing critical services (such as control signaling and high time-sensitive data) under resource constraints, thereby achieving differentiated processing of service levels in the global scheduling strategy.

[0108] In step S430, the gap size of the idle interval generated after the optical burst control packet is inserted into the corresponding candidate wavelength is obtained, and the conflict cost is calculated by combining the calculation results of the conflict coverage rate and the priority weight coupling function.

[0109] Here, we further consider the impact of a sudden insertion of a candidate wavelength on the structure of the wavelength's free interval.

[0110] Equation (7)

[0111] In the formula, Indicates the first Each optical burst control packet corresponds to a burst at a specific wavelength. The cost of conflict Indicates wavelength The above has already scheduled the sudden collection; For the weighting factor; Indicates the insertion of the first After a burst of light, at a wavelength The size of the gap in the free space generated above.

[0112] In equation (7), the size of the new free interval formed after the burst insertion is analyzed. To assess the fragmentation of wavelength resources: if the insertion location results in an excessively small idle interval, it will reduce the schedulability of subsequent bursts, and the conflict cost should be increased; if the insertion creates a larger and reusable idle interval, the conflict cost is relatively reduced. Subsequently, the new idle interval cost is compared with the aforementioned conflict coverage rate. and priority weight coupling function By performing a weighted combination, we obtain the comprehensive conflict cost. .

[0113] Therefore, while avoiding time conflicts, the structure of the wavelength idle interval is optimized, enabling the scheduler to maximize wavelength resource utilization while ensuring scheduling success rate, effectively suppressing idle interval fragmentation, and improving overall link throughput.

[0114] In step S440, during the scheduling process, the candidate wavelength with the lowest conflict cost is selected as the target wavelength of the corresponding optical burst control packet.

[0115] After calculating the conflict cost for all candidate wavelengths, the wavelength with the minimum conflict cost is selected as the target wavelength for the optical burst control packet. The combined costs of time conflict, service priority, and resource fragmentation are incorporated into a unified optimization objective, ensuring that each scheduling decision is locally optimal and collectively constitutes the optimal global resource allocation.

[0116] Thus, the final wavelength selection strategy achieves a balance between maximizing resource utilization, prioritizing high-quality services, and minimizing sudden conflicts, enabling the OBS core node to maintain strong robustness and low packet loss rate under high load conditions, effectively improving the overall service quality of the network.

[0117] In one example of the embodiments of this application, the target wavelength with the lowest cost can be used directly for allocation. However, there may be some problems. That is, if the lowest overall conflict cost is used as the sole decision-making criterion, high-priority bursts may still be arranged on wavelengths with relatively high conflict levels, thereby causing the scheduling of critical services to be hindered.

[0118] The specific reason is that although the priority impact has been partially reflected in the aforementioned conflict cost through the priority weight coupling function, the target wavelength is still a locally optimal result obtained by uniformly comparing the costs among all bursts. In OBS scenarios with large differences in service levels, slight cost differences may mask the need of high-priority services for a lower conflict environment, resulting in insufficient priority protection for critical services, or even the optimal wavelength being "squeezed out" by low-priority bursts.

[0119] In view of this, Figure 5 A flowchart illustrating an example of determining the scheduling results of individual optical burst control packets using the collision cost of the target wavelength, according to an embodiment of this application, is shown.

[0120] like Figure 5 As shown, in step S510, the collision cost of the burst corresponding to the optical burst control packet at the target wavelength is calculated. If the minimum conflict cost Less than the priority for optical burst control packets The corresponding conflict cost threshold If so, then the target wavelength will be selected for scheduling.

[0121] In some implementations, the minimum collision cost of the target wavelength is... Priority for this optical burst control packet Mapped conflict cost threshold Compare. If If the value is less than the threshold, the target wavelength is considered to have acceptable scheduling quality and can be scheduled directly on that wavelength.

[0122] The conflict cost thresholds for different priorities differ, with higher priority thresholds being lower than those for lower priorities. For example, for high-priority bursts, the allowed conflict cost threshold is typically set lower to ensure stricter conflict avoidance protection during scheduling; while for low-priority bursts, a relatively higher conflict cost threshold can be set to allow scheduling to be completed even under acceptable conflict conditions. By assigning differentiated conflict cost thresholds to different priorities, the system can improve the overall flexibility of link resource utilization while ensuring the scheduling quality of critical services.

[0123] It should be noted that the minimum conflict cost not only reflects the degree of time conflict between the burst and the scheduled burst, but also comprehensively reflects the priority difference between the two and the impact of the burst insertion on the wavelength idle interval structure. Therefore, this cost represents the locally optimal scheduling choice under the current network state. Subsequently, this minimum conflict cost is compared with the conflict cost threshold corresponding to the burst priority to determine whether the locally optimal choice can meet the QoS protection requirements of the burst service level, effectively preventing high-priority bursts from being allocated to wavelengths with intense competition or large conflict accumulation due to low cost differences, thereby achieving priority-differentiated resource protection.

[0124] In step S520, if the minimum conflict cost is greater than the conflict cost threshold... If so, we will try to optimize scheduling by adjusting the bias delay and re-evaluate the conflict cost.

[0125] More specifically, under the constraint of the upper limit of the allowed delay of the optical burst control packet, the transmission offset time of the burst corresponding to the optical burst control packet is iteratively increased. This allows for the recalculation of the occupied time interval and the estimation of new conflict costs.

[0126] In some implementations, when the minimum collision cost exceeds the collision cost threshold corresponding to the priority, it indicates a high risk of collision when inserting that wavelength under the existing initial offset time. To avoid packet loss or major collisions caused by forced scheduling, an offset delay adjustment mechanism is introduced. This involves gradually increasing the transmission offset time of the optical burst control packet, for example, by incrementing it according to a preset time step. After each increase in offset time, the system recalculates the occupancy time interval of the burst and recalculates the time overlap length, gap size, and overall collision cost with the scheduled bursts on the wavelength.

[0127] By incrementally adjusting the bias, the scheduling system has the opportunity to insert bursts into more relaxed or regular idle intervals, thereby reducing the degree of conflict.

[0128] In step S531, if the re-estimated conflict cost is lower than the conflict cost threshold... Then, the increased transmit bias time and target wavelength are used for scheduling.

[0129] More specifically, after completing one or more rounds of offset time adjustment, the scheduler compares the re-estimated conflict cost with the threshold: if the re-estimated conflict cost is reduced to below the threshold, it is considered that the offset adjustment has generated an effective scheduling window, and the updated transmission offset time and target wavelength can be used for final scheduling.

[0130] In the embodiments of this application, by fine-tuning the offset time, without changing the size of the burst data or introducing additional wavelength switching overhead, a better scheduling opportunity can be found without compromising service QoS, significantly reducing the cost of conflict and improving resource utilization.

[0131] Furthermore, by combining dynamic bias adjustment with threshold determination, real-time optimization of conflict costs can be achieved while ensuring the schedulability of emergencies. At the same time, it provides stronger conflict suppression guarantees for high-priority emergencies, so that the scheduling strategy not only pursues the minimum cost, but also takes into account the needs of business level, and achieves better global resource allocation.

[0132] In step S533, if the re-estimated conflict cost is not lower than the conflict cost threshold... It identifies whether the priority of the optical burst control packet belongs to the high-priority group.

[0133] Here, when the recalculated conflict cost is still not lower than the conflict cost threshold corresponding to the priority of the optical burst control packet, the system further identifies whether the priority of the burst control packet belongs to the high priority group in order to determine whether a stronger resource guarantee mechanism needs to be triggered.

[0134] If the priority of a burst control packet is not found to belong to a high-priority group, it indicates that the burst belongs to a normal or low-priority service. Its scheduling only needs to follow the regular cost matching strategy and does not need to occupy already scheduled resources. Therefore, the burst control packet can be directly marked as a scheduling failure or postponed to a subsequent scheduling cycle, thereby avoiding additional disturbance to the overall resource structure.

[0135] In step S540, when the priority of the optical burst control packet... When a burst belongs to a high-priority group, obtain the time interval of each scheduled burst that belongs to a low-priority burst group.

[0136] When an optical burst control packet belongs to a high-priority group, the system initiates a resource pre-allocation and conflict resolution mechanism for critical services. To this end, the system first retrieves the currently scheduled burst set, then filters out all bursts belonging to low-priority burst groups and extracts their respective occupancy time intervals. By establishing a candidate low-priority occupancy interval set, alternative resource carriers can be provided for subsequent conflict resolution steps, allowing available scheduling windows to be freed up for high-priority bursts when necessary.

[0137] In step S550, when it is detected that the time interval of the first low-priority burst and the time interval of the corresponding burst of the optical burst control packet exceed a preset time overlap threshold, the scheduling for the first low-priority burst is removed, and the first low-priority burst is put back to the end of the control packet scheduling list, thereby providing pre-occupied resources for high-priority bursts.

[0138] Here, time overlap detection is performed on the time interval occupied by each low-priority burst and the corresponding time interval occupied by the high-priority burst control packet. When a time overlap exceeding a preset threshold is detected between the time interval occupied by the first low-priority burst and the time interval occupied by the target high-priority burst, it indicates that the low-priority burst directly blocks the timely scheduling of high-priority services. At this time, the system removes the scheduling result of the low-priority burst on the current wavelength and adds the low-priority burst back to the end of the control packet scheduling list, allowing it to participate in scheduling again in subsequent windows. Through partial rollback and resource release mechanisms, available time intervals can be promptly built for high-priority bursts, thereby ensuring that critical services can still obtain reliable priority protection capabilities under high-load scenarios.

[0139] Based on the target wavelength selection strategy determined by minimizing conflict costs, this application further constructs a comprehensive scheduling mechanism that integrates elements such as priority group identification, conflict cost threshold constraint, offset time flexible adjustment, and low-priority scheduling rollback. This mechanism is used to dynamically allocate wavelength resources during the scheduling process and ensure that critical services obtain higher scheduling reliability.

[0140] When the minimum conflict cost is insufficient to meet the priority threshold of a burst, the system first attempts to optimize the burst's occupied range on the output link by adjusting the offset time within the allowable delay range, so that it can avoid local high-conflict sections and thus form a schedulable window that meets the priority requirements. If the adjustment still fails to meet the requirements, the system further identifies whether the burst belongs to a high-priority group and performs overlap detection on the scheduled low-priority bursts. By selectively canceling low-priority scheduling items with severe time overlap, the system releases the necessary wavelength occupancy space for high-priority bursts, thereby achieving the resource preemption capability of critical services.

[0141] This allows the OBS core node to maintain global cost optimization capabilities while dynamically adjusting resource allocation based on service priorities. This provides explicit priority protection for critical bursts, preventing them from being blocked by low-priority services occupying locally optimal wavelengths. Simultaneously, through bias adjustments and limited-range local scheduling rollback, the system avoids a complete reconstruction of the overall scheduling state, ensuring high real-time performance and stability in scheduling decisions. This achieves a multi-objective coordinated scheduling effect that balances priority reliability, conflict suppression capabilities, and resource utilization efficiency even under high load scenarios.

[0142] In some examples of embodiments of this application, the required conflict control strength varies for different priority bursts, and fixed thresholds are difficult to adapt to dynamically changing network load and service hierarchy structures over a long period. Therefore, in this embodiment, an adaptive threshold update mechanism based on historical conflict cost sequences is introduced, enabling the conflict cost threshold to be dynamically adjusted according to real-time environmental changes during the scheduling process, thereby improving the stability and accuracy of priority determination.

[0143] In some implementations, during the scheduling process, a historical list of conflict costs is maintained for each priority level. The expected value and standard deviation of the conflict cost for the corresponding priority are calculated by using an exponentially weighted moving average.

[0144] More specifically, during actual scheduling execution, a historical conflict cost list is maintained for each priority level. This list records the conflict costs calculated for each priority level in previous scheduling iterations. As scheduling progresses, the conflict cost sequence in the historical list gradually accumulates, reflecting long-term trends in current network load, burst arrival density, and priority resource contention. Based on this historical sequence, the system calculates the expected conflict cost for each priority level using an Exponentially Weighted Moving Average (EWMA) method. Standard deviation of conflict costs with corresponding priorities .

[0145] Equation (8)

[0146] Equation (9)

[0147] In the formula, Indicates the corresponding priority The The cost of conflict in this scheduling and Each corresponds to a priority Expected value and standard deviation; For the weighting coefficients, such that and The updates are both smooth and can respond promptly to changes in the environment of sudden conflicts.

[0148] about The explanation is in the first section. In this scheduling, when a certain priority is After the optical burst control packet completes the target wavelength selection, the comprehensive collision cost of the optical burst control packet at the selected target wavelength is denoted as... The comprehensive conflict cost is obtained from the conflict cost calculation model. That is, at the target wavelength, the time overlap length and conflict coverage between it and all scheduled bursts are statistically analyzed, and a weighted sum is obtained by combining the priority weight coupling function and the size of the idle interval gap generated after insertion. The single scalar result is regarded as the priority in this scheduling. The cost of conflict.

[0149] In equation (8), the system updates the expected value using an exponentially weighted moving average. : Current number Conflict cost observed in this scheduling By weight Including the expected value from the previous moment. By weight Retained. This shows that newer conflict cost samples have a greater impact on expected updates, while earlier samples gradually decay with iterations, causing... It can reflect the average conflict level of this priority more promptly, without being overly biased by older samples.

[0150] In equation (9), the standard deviation is... It also employs a similar exponential weighting approach to account for the deviation of the current conflict cost from the expected value. Assign weights This is used to measure the impact of the latest sampling point on overall volatility. If the conflict cost has changed significantly recently, the bias term contributes more, increasing the standard deviation; if the conflict cost tends to stabilize, the standard deviation gradually decreases over time, reflecting the strength of volatility of different priority incidents in the current scheduling environment.

[0151] Based on expected value and standard deviation The conflict cost threshold is adaptively set as follows:

[0152] Equation (10)

[0153] In the formula, For sensitivity parameters, For corresponding priority The conflict cost threshold.

[0154] Here, the sensitivity parameter This is used to adjust the responsiveness of the threshold to fluctuations in conflict costs, and can be flexibly configured according to business models or priority strategies. When When the value is large, the contribution of the standard deviation is amplified, and the threshold increases with environmental fluctuations, thereby strengthening the protection of high-priority outbreaks during periods of conflict instability; when When the threshold is smaller, it is closer to the average level, which helps to improve the efficiency of wavelength resource utilization.

[0155] Through the aforementioned adaptive update mechanism, the conflict cost threshold no longer depends on a fixed configuration, but can be automatically adjusted according to the average conflict level and fluctuation range of the current priority outbreak.

[0156] As a further preferred implementation, the pre-occupancy ratio of pre-occupied low-priority bursts is monitored.

[0157] Specifically, a separate pre-scheduled statistics module is maintained for each priority burst, used to record the number of times a low-priority burst is forcibly rolled back or rescheduled due to a conflict with a high-priority burst within a sliding time window. For the same priority... The system has a window length. Internal statistics on the total number of low-priority bursts The number of times it was pre-booked And calculate the pre-occupancy ratio in real time:

[0158] Equation (11)

[0159] This ratio allows the system to quantify the pressure exerted by high-priority bursts on low-priority bursts in the current scheduling environment.

[0160] Furthermore, if the pre-occupancy ratio exceeds a preset threshold, the sensitivity coefficient will be adjusted. This is to increase the conflict cost threshold for low-priority bursts, thereby lowering the scheduling threshold for low-priority bursts.

[0161] In some implementations, when the pre-occupancy ratio calculated by the system... Exceeding the preset ratio threshold At that time, it is considered that the current network is in a state of "low priority being excessively squeezed". It can be parameterized according to the service SLA, bandwidth guarantee policies of different priorities and link resource tension, for example, the value range is 0.1~0.3.

[0162] Upon detecting excessive preemption of a low-priority burst, the system will automatically raise the conflict cost threshold for that priority level. Sensitivity coefficient This changes the threshold calculation method to:

[0163] Equation (12)

[0164] In the formula, This is the adjusted sensitivity coefficient, its value compared to the original coefficient. The larger threshold enhances the response of the threshold to fluctuations in conflict costs. and Each corresponds to a priority The expected value and standard deviation are calculated. As the sensitivity coefficient increases, the threshold controlled by the standard deviation term rises, increasing the "acceptable conflict cost ceiling" for low-priority bursts. This means that the system will only trigger a rollback or rescheduling of low-priority bursts when the conflict cost is significantly higher than the recent average. This reduces the sensitivity of low-priority bursts to triggering conflicts, expanding their scheduling space from a scheduling strategy perspective to reduce the probability of them being squeezed out by high-priority bursts.

[0165] In this application embodiment, an abnormal scenario is identified in which a large amount of high-priority bursts are continuously occupied, causing low-priority bursts to not get scheduling opportunities for a long time, thereby improving the sensitivity parameter. This sets a low-priority conflict cost threshold. After the improvement, low-priority bursts will no longer frequently back off during scheduling due to minor conflicts or tolerable overlaps.

[0166] Furthermore, due to the use of in step S510 The threshold serves as the criterion for determining whether to directly allow scheduling. The smaller the threshold, the stricter the scheduling decision; conversely, the larger the threshold, the easier it is to complete scheduling directly on the current wavelength. Therefore, for high-priority services requiring strict conflict control, generally... Set it to a relatively small size, so that Only slightly above the long-term average level of conflict Therefore, direct scheduling is only permitted when the cost of conflict is low.

[0167] Specifically, when a conflict cost is detected to be too high, the system consciously "does not allow direct passage" but instead triggers subsequent bias time adjustments or low-priority rollback mechanisms to further find wavelength resources with lower conflict rates and a "cleaner" environment for high-priority contingencies. In this sense, "not allowing passage prematurely" is actually a proactive guarantee of execution reliability.

[0168] For low-priority services, when it is necessary to relax the scheduling threshold or alleviate the risk of starvation, the priority can be appropriately increased. ,make Follow By allowing the system to float upwards, it can accept a certain range of conflict costs and relax the passing conditions to some extent, making it easier to complete scheduling directly under an acceptable level of conflict, thereby avoiding long-term backlog of low-priority bursts in high-load scenarios.

[0169] For example, at a low-sudden-crisis conflict cost threshold After the increase, the cost of low-instance conflicts is more likely to fall below the new threshold in the threshold determination. The system allows low-priority bursts to directly occupy wavelength resources to complete scheduling, effectively preventing low-priority bursts from being frequently rolled back due to minor conflicts and accumulating continuously.

[0170] This effectively avoids the starvation of low-priority services caused by frequent arrivals of high-priority bursts, enabling dynamic compensation for resource competition among services of different priorities. Ultimately, without weakening the priority guarantee mechanism, the system expands the acceptable conflict range for low-priority bursts, ensuring that low-priority bursts still have reasonable transmission opportunities in highly competitive scenarios, thereby improving the overall service fairness and scheduling performance of the link.

[0171] Through the aforementioned adaptive update mechanism, in scenarios with bursty or intense wavelength competition, the frequency of high-priority wavelengths can be appropriately reduced. Or directly lower This makes scheduling decisions more stringent, thus preventing them from being assigned to wavelengths with excessive conflict. Simultaneously, combined with the subsequent pre-occupancy ratio statistical mechanism, the pre-occupancy ratio of low-priority wavelengths that have been consistently suppressed can be appropriately increased. This makes low-priority bursts Raising and relaxing scheduling thresholds allows scheduling to be completed within acceptable conflict levels. When the load is stable or resources are sufficient, the thresholds for each priority level can be converged to [a specific value]. In the vicinity, reduce the waste of resources caused by overprotection, thereby achieving a dynamic balance between priority protection and resource utilization efficiency.

[0172] To verify the effectiveness of the priority scheduling link conflict control method proposed in this application, we designed an OBS network topology with 10 nodes. Each link provides 8 data channels and 1 control channel, with a transmission rate of 1 Gb / s. The initial offset time between control packets and bursts is set to 2 time slots (20 µs). The burst length follows a Pareto distribution. The simulation considers three priorities (Class-1, Class-2, and Class-3), with weights assigned according to... Mapping is performed. The load is based on a total average rate of 6 Gb / s, and the burst arrival density is adjusted by varying the peak rate between 8 and 26 Gb / s. We compare the proposed algorithm with G-LAUC-VF and LAUC-VF (without priority support) in terms of metrics including the loss probability of bursts at each priority and the average number of gap checks required for scheduling.

[0173] 1) Simulation method

[0174] The simulation employs discrete event modeling, where each burst control packet enters its corresponding window for scheduling upon arrival at the core node. For G-LAUC-VF, three queues are established for the three burst types, and LAUC-VF scheduling is invoked in priority order. For LAUC-VF, priority is disregarded, and scheduling is based on arrival order. The algorithm presented in this paper uses sliding window aggregation to calculate the cost function and select the wavelength.

[0175] Running 5x10 under each load configuration 6 Each sudden event records the loss rate and average number of searches for each priority event.

[0176] 2) Simulation Results and Analysis

[0177] Figure 6 This diagram illustrates the comparison of average loss rates for different priority bursts under varying peak loads. Specifically, Figure 6 The paper shows the average loss rate comparison between the priority conflict control algorithm proposed in this application and the baseline algorithm G-LAUC-VF on Class-1 (high priority) and Class-3 (low priority) bursts at different peak rates.

[0178] As can be seen, in the low load range (approximately 8-12 Gb / s), the curves are quite similar, and the difference in loss rate between the two algorithms at different priorities is not significant. As the peak rate continues to increase, the loss rate of the baseline algorithm on Class-1 high-priority bursts rises rapidly with the load, while the Class-1 loss rate of the algorithm of this invention remains at a significantly lower level, approximately half that of the baseline algorithm. This indicates that by assessing the conflict risk in advance in the cost function and combining it with a dynamic bias adjustment mechanism, the probability of loss of high-priority services under high load scenarios can be significantly reduced.

[0179] Meanwhile, in Class-3 low-priority bursts, the loss rate of the algorithm in this application is slightly higher than that of the baseline algorithm, reflecting the design trade-off of consciously sacrificing the performance of some low-priority services in order to enhance the guarantee of high-priority services when link resources are tight. However, its overall loss rate is still within an acceptable range.

[0180] comprehensive Figure 6 It can be seen that, compared with traditional scheduling strategies that do not distinguish between service levels or only adopt static priority order, the algorithm in this paper, which is based on cost function and priority awareness, significantly improves the service quality of high priority bursts under the same load conditions. At the same time, by making appropriate concessions to low priority bursts, it achieves a balance between overall network loss performance and critical service assurance capabilities.

[0181] Figure 7This diagram illustrates a comparison of the complexity of three scheduling algorithms when scheduling bursts. The vertical axis represents the average number of channel gaps that need to be checked for each burst, which characterizes the scheduling complexity.

[0182] like Figure 7 As shown, the LAUC-VF algorithm requires traversing all available slots when selecting wavelengths for bursts, with an average of approximately 120 checks. G-LAUC-VF, by introducing a multi-priority queue, slightly reduces invalid checks, lowering the average number of checks to approximately 100, but still requires traversing a large search space. In contrast, the algorithm proposed in this paper constructs a candidate slot set for each burst and prioritizes candidate wavelengths using a conflict cost function, enabling most bursts to be scheduled by checking only about 60 slots, with an average number of checks approximately half that of LAUC-VF.

[0183] Therefore, while ensuring scheduling quality, the scheduling algorithm proposed in this invention significantly reduces the computational overhead of single-burst scheduling and is more suitable for deployment and application in OBS core nodes with high load and strict real-time requirements.

[0184] Furthermore, this application introduces a scheduling algorithm with a conflict comparison mechanism. The conflict cost threshold in the scheduling algorithm is not simply increased or decreased based on the load size, but rather adaptively calibrated based on the statistical results of conflict costs for each priority level. For high-priority bursts, when a sustained increase or intensified fluctuation in their conflict cost level is detected, the system reduces the corresponding sensitivity parameter or lowers the conflict threshold, making it more difficult for high-priority bursts to be "directly allowed" on wavelengths with higher conflict levels. This triggers bias adjustment and low-priority cancellation mechanisms more frequently, ensuring that they prioritize occupying wavelength resources with lower conflict levels. For low-priority bursts, when a persistently high pre-occupancy ratio and a starvation trend are detected, the system appropriately increases the corresponding priority threshold, further relaxing the scheduling threshold for them at acceptable conflict levels, preventing frequent rollbacks due to minor conflicts. Combined with a dynamic priority weight design based on exponential mapping, this application can significantly alleviate the problem of low-priority services being suppressed or even nearly completely discarded in fixed-weight priority scheduling under high-load scenarios. This ensures stronger protection for high-priority bursts while maintaining stable scheduling opportunities for low-priority bursts without backlog.

[0185] To address the link conflict control problem in unbuffered OBS networks, this application proposes a priority-based scheduling control method. This method introduces a sliding window aggregation control packet at the core node, significantly reducing scheduling complexity through priority mapping and a candidate gap set. It constructs a priority-aware conflict cost function, enabling scheduling decisions to simultaneously consider weight coupling between bursts and gap utilization. Through dynamic bias adjustment, pre-occupancy, and adaptive conflict zone learning, it achieves fairness in ensuring high-priority flow while also considering low-priority flow under high load.

[0186] Simulation results show that the proposed algorithm outperforms traditional algorithms in both high-optimal burst loss rate and scheduling complexity, demonstrating high practical value and hardware feasibility. Due to its flexible structure, the algorithm can be further integrated with mechanisms such as multi-path routing and adjustable fiber delay lines to expand its application in large-scale high-speed optical networks.

[0187] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of combined actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Secondly, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application. In the above embodiments, the descriptions of each embodiment have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0188] Figure 8 A structural block diagram of an example of a priority-based scheduling optical burst link conflict control system according to an embodiment of this application is shown, which is deployed on an OBS core node.

[0189] like Figure 8 As shown, the optical burst link conflict control system 800 based on priority scheduling includes a scheduling list generation unit 810, a burst parsing unit 820, a candidate gap set construction unit 830, a conflict cost analysis unit 840, and a burst scheduling control unit 850.

[0190] The scheduling list generation unit 810 is used to obtain multiple optical burst control packets arriving within the sliding time window and sort them according to their arrival time to form a control packet scheduling list.

[0191] The burst parsing unit 820 is used to parse the burst length, offset time, destination port and priority label in each optical burst control packet in the control packet scheduling list, map the priority label to a weight to determine the corresponding priority weight, and predict the occupancy time interval of the burst corresponding to the optical burst control packet on the output link based on the offset time and burst length.

[0192] The candidate gap set construction unit 830 is used to construct a candidate gap set from the lightweight data structure set for each output link, which overlaps with the occupancy time interval of the bursts corresponding to each optical burst control packet in the control packet scheduling list; the lightweight data structure is used to record the idle interval of each wavelength on the corresponding output link.

[0193] The conflict cost analysis unit 840 is used to calculate the conflict cost of each optical burst control packet in the control packet scheduling list on each candidate wavelength corresponding to the candidate gap set of the optical burst control packet, based on the priority weight of the optical burst control packet, the time overlap relationship with the scheduled burst, and the gap change caused after insertion into the corresponding candidate wavelength, and select the wavelength with the minimum conflict cost as the target wavelength.

[0194] The burst scheduling control unit 850 is used to determine the scheduling result of each optical burst control packet according to the collision cost of the target wavelength, and modify the wavelength idle interval of the corresponding output link based on the scheduling result to update the corresponding lightweight data structure.

[0195] In some embodiments, this application provides a non-volatile computer-readable storage medium storing one or more programs including execution instructions. The execution instructions can be read and executed by an electronic device (including but not limited to a computer, server, or network device) to perform the steps of any of the priority scheduling-based optical burst link conflict control methods described above.

[0196] In some embodiments, this application also provides a computer program product, the computer program product including a computer program stored on a non-volatile computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of any of the above-described priority-based optical burst link conflict control methods.

[0197] In some embodiments, this application also provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform steps of a priority-based optical burst link conflict control method.

[0198] The above-described product can perform the methods provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects for performing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in the embodiments of this application.

[0199] The electronic devices in this application can exist in various forms, including but not limited to: mobile communication devices, ultra-mobile personal computer devices, portable entertainment devices, or other airborne electronic devices with data interaction functions.

[0200] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0201] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0202] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A priority-based scheduling-based optical burst link conflict control method applied to an OBS core node, characterized in that, The method comprises: acquiring a plurality of optical burst control packets arriving within a sliding time window, and forming a control packet scheduling list according to the arrival time; for each optical burst control packet in the control packet scheduling list, analyzing the burst length, offset time, destination port and priority label in the optical burst control packet, mapping the priority label to a weight to determine the corresponding priority weight, and predicting the occupation time interval of the burst corresponding to the optical burst control packet on the output link according to the offset time and the burst length; based on a lightweight data structure set for each output link, constructing a candidate gap set in which the occupation time intervals of the bursts corresponding to each optical burst control packet in the control packet scheduling list have an overlapping relationship from the lightweight data structure; the lightweight data structure is used to record the idle interval of each wavelength on the corresponding output link; for each optical burst control packet in the control packet scheduling list, on each candidate wavelength corresponding to the candidate gap set of the optical burst control packet, calculating the conflict cost of the burst on each candidate wavelength according to the priority weight of the optical burst control packet, the time overlapping relationship with the scheduled burst, and the change of the gap generated after being inserted in the corresponding candidate wavelength, and selecting the wavelength with the minimum conflict cost as the target wavelength; determining the scheduling result of each optical burst control packet according to the conflict cost of the target wavelength, and modifying the wavelength idle interval of the corresponding output link based on the scheduling result to update the corresponding lightweight data structure.

2. The method of claim 1, wherein, The method comprises: dynamically calculating the length of the sliding time window according to the current link load estimation value: ， wherein, is a length of a sliding time window, is a link load estimate value, is an empirical coefficient, and are a preset minimum window length and a preset maximum window length, respectively. collecting the optical burst control packets arriving based on the sliding time window, and arranging them in order of their arrival time to form a control packet scheduling list; in the case where the number of optical burst control packets in the sliding time window is detected to be less than a preset number threshold, switching from the aggregated scheduling mechanism of the sliding time window to online scheduling, and performing immediate scheduling on the arriving optical burst control packets to reduce scheduling delay and avoid additional waiting caused by window aggregation under low arrival density conditions.

3. The method of claim 1, wherein, The method comprises: for each optical burst control packet in the control packet scheduling list, converting the priority label of the optical burst control packet into a priority weight representing the scheduling importance of bursts of different priorities; the priority weight is calculated by exponential mapping to amplify the weight difference between high and low priorities: ， wherein denotes the priority tag of the th optical burst control packet, is set to an integer taking discrete values wherein 1 denotes the highest priority, denotes the lowest priority; denotes the priority weight of the th optical burst control packet corresponding to the burst, denotes the exponent mapping base; The occupation start time and the occupation end time of the optical burst control packet on the output link are calculated based on the current scheduling time, the offset time recorded by the optical burst control packet, the burst length, the cumulative processing delay experienced by the optical burst control packet before arriving at the node, and the link transmission rate, so as to determine the occupation time interval of the burst corresponding to the optical burst control packet on the output link: ， ， In the formula, and respectively represent the start time and the end time of the burst of the th optical burst control packet on the target link, represents the current scheduling execution time, represents the transmission bias time carried by the th optical burst control packet; represents the transmission path of the th optical burst control packet through all intermediate nodes from the source node to the current node, represents the intermediate node in the path , represents the total number of intermediate nodes of the transmission path, represents the processing delay of the optical burst control packet by the th intermediate node; represents the burst length of the th optical burst control packet, represents the transmission rate of the output link.

4. The method of claim 1, wherein, The candidate gap set in which the occupation time intervals of the bursts corresponding to the optical burst control packets in the control packet scheduling list have an overlapping relationship is constructed from the lightweight data structure set for each output link, and the candidate gap set includes: The lightweight data structure adopts a double-ended queue sorted by start time, and an occupied interval list is maintained on each wavelength of each output link , recording the time interval occupied by the currently reserved burst on the wavelength, and the intervals are sorted in order of start time; wherein, represents the occupied interval list of the th wavelength on the link , containing the occupied time interval of the reserved burst on the wavelength; According to the list of occupied areas The idle time interval is calculated by using the gaps between adjacent occupied time intervals, and then a set of idle intervals for that wavelength is constructed. ;in, Indicates link Upper A set of idle intervals for each wavelength; For each optical burst control packet in the control packet scheduling list, the overlapping relationship between the occupation time interval of the optical burst control packet and the idle interval set of each output link is calculated, the idle intervals overlapping with the occupation time interval of the optical burst control packet are screened out, and then a candidate gap set is constructed for the optical burst control packet on each output link.

5. The method of claim 1, wherein, The conflict cost of the burst on each candidate wavelength is calculated according to the priority weight of the optical burst control packet, the time overlapping relationship with the scheduled burst, and the gap change after being inserted in the corresponding candidate wavelength, and the wavelength with the minimum conflict cost is selected as the target wavelength for each optical burst control packet in the control packet scheduling list, and the method includes: The occupation time interval of the burst corresponding to the optical burst control packet is compared with the occupation time interval of the scheduled burst on each candidate wavelength, the time overlapping length is calculated, and the conflict coverage rate is determined: ， wherein denotes the collision coverage between the th optical burst control packet and the th scheduled burst, denotes the time overlap length between the th optical burst control packet and the th scheduled burst. The priority weight coupling function is introduced, and when the priority weight of the burst corresponding to the optical burst control packet is much greater than the priority weight of the scheduled burst, the priority weight coupling function approaches the maximum value, so as to reflect that the conflict cost of the high-priority burst to the low-priority burst is higher: ， wherein represents a priority weight coupling function, represents the priority weight of the th optical burst control packet corresponding to the burst, represents the priority weight of the th scheduled burst; The gap size of the idle interval generated after the optical burst control packet is inserted in the corresponding candidate wavelength is obtained, and the conflict cost is calculated in combination with the calculation results of the conflict coverage rate and the priority weight coupling function: ， In the formula, denotes the cost of the conflict of the burst corresponding to the th optical burst control packet on the wavelength , denotes the set of bursts already scheduled on the wavelength ; is a weighting coefficient; denotes the gap size of the idle interval generated on the wavelength after inserting the th optical burst. In the scheduling process, the candidate wavelength with the minimum conflict cost is selected as the target wavelength of the corresponding optical burst control packet.

6. The method of claim 1, wherein, The scheduling result of each optical burst control packet is determined according to the conflict cost of the target wavelength, and the method includes: Calculate the collision cost of the burst corresponding to the optical burst control packet at the target wavelength. If the minimum conflict cost Less than the priority for optical burst control packets The corresponding conflict cost threshold If the target wavelength is selected for scheduling, then the conflict cost threshold corresponding to the higher priority is less than the conflict cost threshold corresponding to the lower priority. If the minimum conflict cost is greater than the conflict cost threshold then attempt to optimize the schedule by biasing delay adjustments, including in particular: iteratively increase a transmission offset time of a burst corresponding to the optical burst control packet under a constraint of an upper limit of a delay allowed by the optical burst control packet and further recalculate the time interval of occupation and estimate a new conflict cost If the re-estimated collision cost is lower than the collision cost threshold then schedule using the increased transmit bias time and the target wavelength.

7. The method of claim 6, wherein, if the re-estimated conflict cost is not lower than a conflict cost threshold the method further comprises: When the priority of the optical burst control packet When it belongs to a high-priority group, obtain the time interval of each scheduled burst that belongs to a low-priority burst group; When it is detected that the occupation time interval of the first low-priority burst and the occupation time interval of the burst corresponding to the optical burst control packet have a time overlapping length exceeding a preset time overlapping threshold, the scheduling for the first low-priority burst is removed, and the first low-priority burst is put back to the end of the control packet scheduling list, so as to provide pre-occupied resources for high-priority bursts.

8. The method according to claim 6 or 7, characterized in that, The conflict cost threshold is adaptively updated in the following way: In the scheduling process, a history list of conflict cost is maintained for each priority and the expected value of the conflict cost for the respective priority is computed by an exponentially weighted moving average and the standard deviation : ， ， wherein represents the conflict cost in the first dispatch of the th priority, is a weighting coefficient, and are the mean and standard deviation, respectively, of the th priority. According to the expected value and the standard deviation , the adaptive setting conflict cost threshold is: ， wherein is a sensitivity parameter, is a conflict cost threshold corresponding to a priority .

9. The method of claim 8, wherein, The method further includes: The pre-occupied proportion of the low-priority burst is monitored. If the pre-occupation ratio exceeds a preset ratio threshold, the sensitivity coefficient is adjusted to increase the conflict cost threshold of the low-priority burst, so as to reduce the scheduling threshold of the low-priority burst.

10. A priority-based scheduling based optical burst link collision control system deployed in an OBS core node, characterized in that, The system includes: A scheduling list generation unit is configured to obtain a plurality of optical burst control packets arriving within a sliding time window, and form a control packet scheduling list according to the arrival time; a burst analysis unit configured to analyze, for each optical burst control packet in the control packet scheduling list, a burst length, an offset time, a destination port and a priority tag in the optical burst control packet, map the priority tag to a weight to determine a corresponding priority weight, and predict an occupied time interval of a burst corresponding to the optical burst control packet on an output link according to the offset time and the burst length; a candidate gap set construction unit configured to construct, based on a lightweight data structure set for each output link, a candidate gap set corresponding to the occupied time interval of the burst of each optical burst control packet in the control packet scheduling list, the lightweight data structure being configured to record an idle interval of each wavelength on the corresponding output link; a conflict cost analysis unit configured to, for each optical burst control packet in the control packet scheduling list, calculate a conflict cost of the burst on each candidate wavelength corresponding to the candidate gap set of the optical burst control packet according to the priority weight of the optical burst control packet, a time overlapping relationship with a scheduled burst, and a gap change caused by insertion on the corresponding candidate wavelength, and select a wavelength with a minimum conflict cost as a target wavelength; a burst scheduling control unit configured to determine a scheduling result of each optical burst control packet according to the conflict cost of the target wavelength, and modify a wavelength idle interval of the corresponding output link based on the scheduling result to update the corresponding lightweight data structure.

Images