Method, apparatus, device, medium and product for scheduling of time sensitive network based on non-gap window

Abstract

The application discloses a scheduling method and device based on a non-interval window of a time-sensitive network, an equipment, a medium and a product, and relates to the technical field of communication. The method comprises the following steps: obtaining parameters of a flow in the time-sensitive network; randomly generating a plurality of candidate solutions of non-interval windows for scheduling the time-sensitive network based on the parameters of the flow in the time-sensitive network; each candidate solution is generated by decoding each solution vector; each solution vector comprises a plurality of continuous value intervals; each value interval represents an output port; each value interval is divided into a plurality of segments; and the length value of each segment maps the window length of a flow with the same priority; taking the accumulated upper bound of the delay of each output port as the fitness, optimizing the candidate solution through an improved grey wolf algorithm, and taking the optimized optimal solution as the scheduling scheme of the time-sensitive network; and the improved grey wolf algorithm is the grey wolf algorithm after introducing a crossover operator and a mutation operator. The application can effectively reduce the worst end-to-end delay.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular to a scheduling method and device for time-sensitive network based on non-interval window, equipment, medium and product. BACKGROUND

[0002] With the rapid development of industrial automation and intelligent transportation systems, more and more intelligent sensors, actuators and controllers are deployed in high-level autonomous vehicles. Their complex computing environment and increasing requirements for real-time and deterministic information transmission have become a key challenge. Time-sensitive network (TSN) can provide predictable data transmission delay for traditional Ethernet through a series of standardized extensions and improvements, ensuring real-time and high reliability. TSN is considered a key solution to the above problems. Reasonable gate control list (GCL) design is the key to the implementation of time-aware shaping (TAS) mechanism, and the process of generating GCL is called TSN scheduling. TSN scheduling is a hot research topic in the field of TSN. According to the different scheduling objects, the scheduling can be divided into frame-based scheduling and window-based scheduling. Frame-based scheduling usually schedules the transmission time of each flow and maps the transmission time to a window of the gate control list. Window-based scheduling decouples the scheduling from the transmission time of the flow and completes the scheduling through direct gate control list design. This method usually uses network calculus to analyze the worst delay based on the gate control list, so as to measure the transmission effect. It can measure the scheduling constraints from a pessimistic point of view, and as long as the analysis result meets the deadline requirement, its scheduling effect is easier to guarantee in actual deployment. Especially in autonomous driving and other scenarios, the scheduling result based on the worst delay can greatly improve its safety. However, with the increasing requirements for real-time information transmission, new methods are still needed to further reduce the scheduling delay and speed up the solution from the window-based scheduling. SUMMARY

[0003] The purpose of the present application is to provide a scheduling method, device, equipment, medium and product for time-sensitive network based on non-interval window, which can effectively reduce the worst end-to-end delay.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] In a first aspect, the present application provides a scheduling method for time-sensitive network based on non-interval window, comprising:

[0006] obtaining parameters of flows in the time-sensitive network;

[0007] The multiple candidate solutions of non-interval windows for scheduling the time-sensitive network are randomly generated based on parameters of flows in the time-sensitive network, each candidate solution is generated by decoding each solution vector, each solution vector includes multiple continuous value intervals, each value interval represents an output port, and each value interval is divided into multiple segments, and a length value of each segment maps a window length of a flow of a same priority;

[0008] The upper bound of the accumulated delay of each output port is taken as fitness, the candidate solution is optimized by the improved grey wolf algorithm, and the optimized optimal solution is taken as the scheduling scheme of the time-sensitive network; the improved grey wolf algorithm is the grey wolf algorithm after introducing a crossover operator and a mutation operator.

[0009] Optionally, the parameters of the flow include a priority, a period and a size;

[0010] When the multiple candidate solutions of non-interval windows for scheduling the time-sensitive network are randomly generated based on parameters of flows in the time-sensitive network, a constraint condition of the window length is that

[0011] wherein, is a window length of a flow of an mth priority of an hth output port, represents a set of flows of an mth priority transmitted by an hth output port, HP is a hyper period, and a value of the hyper period is a least common multiple of periods of all flows in the hyper period, f i is an ith flow, f i .size is a size of f i , f i .T is a period of f i , T GCL is a gate list period, BW is a bandwidth, is a guard band length;

[0012] The solution vector is represented as:

[0013] wherein, D is a length of the solution vector, H is a number of output ports in the time-sensitive network, |P h | is a number of priorities of an hth output port;

[0014] A formula that a length value of each segment maps a window length of a flow of a priority is:

[0015]

[0016] wherein, is a segment length corresponding to an ith priority of an hth output port, is a segment length corresponding to an mth priority of an hth output port, T GCLis a gating list period;

[0017] The constraint condition includes

[0018] is a gating list period; is a bias of the opening time of the corresponding window relative to the gating list period start time, is a window opening length of the jth priority of the hth output port of the flow, P m is the mth priority, P j is the jth priority, is a bias of the opening time of the first window relative to the gating list period start time.

[0019] Optionally, the fitness is represented as:

[0020]

[0021] Wherein, O is the upper bound of the delay of each output port, F is a set of flows, f i is the ith flow, represents the maximum end-to-end delay of f i ;

[0022]

[0023] Wherein, is the upper bound of the queuing and transmission delay of the flow with the priority P m at the hth output port, R(f i ) is a set of all output ports on the routing path of f i , des represents the last output port, t prop is the propagation delay, t proc is the processing delay.

[0024] Optionally, the upper bound of the delay of each output port is taken as the fitness, the candidate solution is optimized by the improved grey wolf algorithm, and the optimized optimal solution is taken as the scheduling scheme of the time-sensitive network, and specifically includes:

[0025] Initialize the first iteration number;

[0026] Take each candidate solution as a grey wolf, and calculate the fitness of each grey wolf;

[0027] Update the position of the grey wolf according to the fitness to obtain the current first wolf group;

[0028] Optimize the current first wolf group by using a crossover operator to obtain the current second wolf group;

[0029] The current second wolf group is optimized by using a mutation operator, and a tabu search algorithm is introduced in the mutation process to obtain a current third wolf group.

[0030] It is determined whether the first iteration number reaches a first set iteration number.

[0031] If yes, the solution corresponding to the current optimal gray wolf position is taken as the scheduling scheme of the time-sensitive network.

[0032] If no, the first iteration number is incremented by 1, the current third wolf group is taken as the current first wolf group, the fitness of each gray wolf in the current first wolf group is calculated, and the step of optimizing the current first wolf group by using a crossover operator to obtain a current second wolf group is returned.

[0033] Optionally, the current first wolf group is optimized by using a crossover operator to obtain a current second wolf group, and the method specifically includes the following steps.

[0034] The current first wolf group is traversed, and it is randomly determined whether the gray wolf currently traversed is optimized by using a crossover operator.

[0035] If yes, another gray wolf is randomly selected from the current first wolf group except the gray wolf currently traversed, the gray wolf currently traversed and the other gray wolf are taken as parents to perform crossover to obtain two offspring gray wolves, and the gray wolf with the minimum fitness from the two parent gray wolves and the two offspring gray wolves is taken as the gray wolf after crossover optimization.

[0036] After the traversal of the current first wolf group is completed, the gray wolf after crossover optimization and the wolf group in the current first wolf group that is not optimized by using the crossover operator constitute the current second wolf group.

[0037] Optionally, the current second wolf group is optimized by using a mutation operator, and a tabu search algorithm is introduced in the mutation process to obtain a current third wolf group, and the method specifically includes the following steps.

[0038] Each gray wolf in the current second wolf group is traversed, if the gray wolf currently traversed is a head wolf, the gray wolf is taken as a gray wolf to be mutated; if the gray wolf currently traversed is not a head wolf, it is randomly determined whether the gray wolf is optimized by using a mutation operator, if the random selection is yes, the gray wolf is taken as a gray wolf to be mutated, and if the random selection is no, the gray wolf is not involved in mutation.

[0039] If the gray wolf currently traversed is taken as a gray wolf to be mutated, a second iteration number is initialized.

[0040] The gray wolf currently traversed is mutated by using a mutation operator to obtain a current gray wolf group to be mutated composed of multiple gray wolves after mutation optimization.

[0041] The gray wolf with the minimum fitness in the current gray wolf group to be mutated is taken as a temporary optimal solution.

[0042] determining whether the fitness of the temporary optimal solution is lower than the fitness of the historical optimal solution;

[0043] if yes, storing the temporary optimal solution into the tabu table, and updating the historical optimal solution as the temporary optimal solution;

[0044] if no, determining whether the temporary optimal solution is stored in the tabu table, if yes, removing the grey wolf corresponding to the temporary optimal solution from the current to-be-mutated wolf group, and returning to the step of taking the grey wolf with the minimum fitness in the current to-be-mutated wolf group as the temporary optimal solution; if no, storing the temporary optimal solution into the tabu table;

[0045] determining whether the second iteration number reaches the second set iteration number;

[0046] if no, increasing the second iteration number by 1, mutating the temporary optimal solution by using a mutation operator to obtain a current to-be-mutated wolf group composed of a plurality of mutated and optimized grey wolves, and returning to the step of taking the grey wolf with the minimum fitness in the current to-be-mutated wolf group as the temporary optimal solution;

[0047] if yes, initializing the second iteration number, and traversing the following grey wolves in the current second wolf group until the traversal of the grey wolves in the current second wolf group is completed to obtain a current third wolf group.

[0048] In a second aspect, the present application provides a scheduling device for a time-sensitive network based on a non-interval window, the scheduling device comprising:

[0049] a parameter acquisition module for a flow in the time-sensitive network, configured to obtain parameters of a flow in the time-sensitive network;

[0050] a candidate set generation module, configured to randomly generate a plurality of candidate solutions of non-interval windows for scheduling the time-sensitive network; each candidate solution is generated by decoding each solution vector, and each solution vector comprises a plurality of continuous value intervals, each value interval representing an output port, and each value interval being divided into a plurality of segments, and the length value of each segment mapping a window length of a flow of the same priority;

[0051] a candidate set optimization module, configured to take the cumulative upper bound of the delay of each output port as the fitness, optimize the candidate solution by using an improved grey wolf algorithm, and take the optimized optimal solution as the scheduling scheme of the time-sensitive network; the improved grey wolf algorithm is a grey wolf algorithm after introducing a crossover operator and a mutation operator.

[0052] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the steps of the scheduling method for a time-sensitive network based on a non-interval window according to any one of the above.

[0053] In a fourth aspect, the present application provides a computer readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the steps of the scheduling method for time-sensitive network based on non-interval window according to any one of the preceding embodiments.

[0054] In a fifth aspect, the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the scheduling method for time-sensitive network based on non-interval window according to any one of the preceding embodiments.

[0055] According to the specific embodiments provided by the present application, the following technical effects are disclosed:

[0056] The present application provides a scheduling method, device, equipment, medium and product for time-sensitive network based on non-interval window, based on the parameters of the flow in the time-sensitive network, a plurality of candidate solutions of non-interval window for scheduling the time-sensitive network are randomly generated; the window switching times of the compressed gating list are reduced by non-interval design, the deployment difficulty is reduced, the upper bound of the delay of each output port is accumulated as the fitness, the improved grey wolf algorithm is used to optimize the candidate solution, and the optimized optimal solution is used as the scheduling scheme of the time-sensitive network, so as to realize the delay optimization and effectively reduce the worst end-to-end delay. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0058] Figure 1 An application environment diagram of a scheduling method for time-sensitive network based on non-interval window according to an embodiment of the present application;

[0059] Figure 2 A flowchart of a scheduling method for time-sensitive network based on non-interval window according to an embodiment of the present application;

[0060] Figure 3 A framework of a scheduling method for time-sensitive network based on non-interval window according to an embodiment of the present application;

[0061] Figure 4 An interval window gating list diagram according to an embodiment of the present application;

[0062] Figure 5 A non-interval window gating list diagram according to an embodiment of the present application;

[0063] Figure 6 An encoding and decoding process diagram provided by an embodiment of the present application is shown in the following table.

[0064] Figure 7 An improved gray wolf algorithm flow diagram provided by an embodiment of the present application is shown in the following table.

[0065] Figure 8 A structure diagram of a computer device provided by an embodiment of the present application is shown in the following table. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0067] The above purposes, features and advantages of the present application can be more obvious and easy to understand. The present application will be described in further detail below with reference to the drawings and specific embodiments.

[0068] The scheduling method of the time-sensitive network based on the non-interval window provided by the embodiments of the present application can be applied to, for example, Figure 1The application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data required by the server 104 to process. The data storage system can be set separately, or integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the parameters of the flow in the time sensitive network to the server 104, and after the server 104 receives the parameters of the flow in the time sensitive network, the server 104 obtains the parameters of the flow in the time sensitive network; based on the parameters of the flow in the time sensitive network, a plurality of candidate solutions of non-interval window scheduling for the time sensitive network are randomly generated; each candidate solution is generated by decoding each solution vector, and each solution vector includes a plurality of continuous value intervals, each value interval represents an output port, and each value interval is divided into a plurality of segments, and the length value of each segment maps the window length of a flow of the same priority; the upper bound of the delay of each output port is accumulated as the fitness, and the improved grey wolf algorithm is used to optimize the candidate solution, and the optimized optimal solution is used as the scheduling scheme of the time sensitive network; the improved grey wolf algorithm is the grey wolf algorithm after introducing the crossover operator and the mutation operator. The server 104 can feed back the obtained scheduling scheme of the time sensitive network to the terminal 102. In addition, in some embodiments, the scheduling method of the time sensitive network based on the non-interval window can also be implemented by the server 104 or the terminal 102 alone, such as the terminal 102 directly scheduling the time sensitive network based on the non-interval window for the parameters of the flow in the time sensitive network, or the server 104 obtaining the parameters of the flow in the time sensitive network from the data storage system and scheduling the time sensitive network based on the non-interval window for the parameters of the flow in the time sensitive network.

[0069] Among them, the terminal 102 can be, but not limited to, various desktop computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices, and the Internet of Things devices can be smart speakers, smart televisions, smart air conditioners, smart vehicle devices, etc. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers, and can also be a cloud server.

[0070] In an exemplary embodiment, as Figure 2 shown, a scheduling method of a time sensitive network based on a non-interval window is provided, which is executed by a computer device, specifically by a terminal or a server, etc. Computer device alone, or by a terminal and a server together, in the embodiment of the application, taking the server 104 in Figure 1 The steps 201 to 208 are described as follows. Among them:

[0071] Step 201: Obtain the parameters of the flow in the time sensitive network.

[0072] Step 202: Randomly generate a plurality of candidate solutions of non-interval windows for scheduling the time-sensitive network based on parameters of flows in the time-sensitive network; each candidate solution is generated by decoding each solution vector, and each solution vector includes a plurality of continuous value intervals, each value interval representing an output port, and each value interval is divided into a plurality of segments, and the length value of each segment maps the window length of a flow of the same priority.

[0073] Step 203: Taking the accumulated upper bound of the delay of each output port as the fitness, optimizing the candidate solution by the improved grey wolf algorithm, and taking the optimized optimal solution as the scheduling scheme of the time-sensitive network; the improved grey wolf algorithm is the grey wolf algorithm after introducing the crossover operator and the mutation operator.

[0074] Based on the parameters of flows in the time-sensitive network, the present application randomly generates a plurality of candidate solutions of non-interval windows for scheduling the time-sensitive network; the window switching times of the compressed gating list are reduced by the non-interval design, the deployment difficulty is reduced, the accumulated upper bound of the delay of each output port is taken as the fitness, the candidate solution is optimized by the improved grey wolf algorithm, the optimized optimal solution is taken as the scheduling scheme of the time-sensitive network, the delay optimization is realized, and the worst end-to-end delay is effectively reduced.

[0075] In an exemplary embodiment, the time-sensitive network is a vehicle-mounted time-sensitive network, such as Figure 3 As shown, the scheduling method of the time-sensitive network based on the non-interval window specifically includes five parts: the first part is the input of the scheduling environment of the vehicle-mounted time-sensitive network environment; the second part is the encoding and decoding method based on the non-interval window, which is used to generate the candidate solution and is the basis of combining the delay analysis theory with the scheduling method; the third part is to solve the worst delay according to the decoded gating list parameters, and the method involved is the worst delay analysis theory based on network delay; the fourth part is the improved grey wolf algorithm search with the worst delay as the fitness, and the coded solution is repeatedly iterated and searched based on the fitness to find a better solution; and the last fifth part is to output the optimal solution as the scheduling result after the algorithm converges.

[0076] The present application does not limit the specific network environment of the time-sensitive network, but needs to clearly define the parameters of the flow, i.e., the routing path, priority, period, and size of the flow.

[0077] The present application is based on the encoding and decoding of non-gap windows, and the encoding and decoding process is a process of randomly generating multiple candidate solutions of non-gap windows for scheduling time-sensitive networks. According to the worst-case delay analysis theory based on network calculus, the worst-case end-to-end delay bound is affected by the window opening time and the mutual position of each priority window in the same gate list cycle. Longer windows and smaller maximum waiting time help to reduce the worst-case end-to-end delay and improve the transmission performance of the network. Therefore, it is necessary to optimize the window size and position of the gate list, and the encoding of the solution optimized by the grey wolf algorithm needs to reflect the window size and position.

[0078] In determining the gate list cycle T GCL After that, the window design mainly includes the window opening length and the window opening time There needs to be enough length to ensure that all flows of this priority can be transmitted within the super cycle, and the constraint condition of the window opening length is:

[0079] Wherein, is the window opening length of the mth priority flow of the hth output port, represents the set of mth priority flows transmitted by the hth output port, HP is the super cycle, and the value of the super cycle is the least common multiple of the periods of all flows in the super cycle, f i is the ith flow, f i .size is the size of f i , f i .T is the period of f i , T GCL is the gate list cycle, BW is the bandwidth, is the guard band length, is the maximum length of the flow of P m .

[0080] The window opening time is the offset of the window opening time relative to the start time of the cycle, and the value range is represented as:

[0081] The windows of different priorities of the same output port do not overlap with each other, which not only reduces the design complexity, but also reduces the mutual interference between the windows and improves the utilization rate of the bandwidth. Under this premise, a gap GCL design method is to take and as design variables at the same time, as shown in Figure 4 The number of design variables for each port is 2x|P h|}. Meanwhile, since it cannot be guaranteed that the previous window is closed and the next window is opened at the same time, the maximum number of GCLs for each output port is also 2x|P h |In theory, this design allows windows to be arranged at will. However, the large variable space makes the search difficult, and the gap between adjacent windows causes a waste of bandwidth. To solve the above problems, the present application proposes a gapless design method, as shown in Figure 5 When designing GCL, it is defined that there is no gap between adjacent windows, that is, the previous window is closed and the next window is opened at the same time, and the windows are arranged in priority order. The design variable of this gapless method is only The number of design variables is only |P h |The search space is effectively compressed. The number of GCLs for each output port is fixed as |P h |The fewer number of GCLs simplifies the deployment process. Compared with the gap design, although the gapless design may limit the window offset freedom, it can fully utilize the network resources and often ensure greater window opening time.

[0082] The essence of the encoding and decoding process of the gapless GCL design of the present application is to map the optimization variable solution set to the window length combination of GCL. The grey wolf algorithm calculates the optimization target value represented by the solution based on the GCL mapped by the solution using network calculus theory, and finally obtains the optimal solution of the search through iteration.

[0083] Figure 6 An example of an encoding and decoding process is given, h1, h2, h3 are the priority numbers of three output ports, h1, h2, h3 are the priority numbers of three output ports, h1, h2, h3 are the priority numbers of three output ports, 00 h1, h2, h3 are the priority numbers of three output ports, 01 h1, h2, h3 are the priority numbers of three output ports, 02 h1, h2, h3 are the priority numbers of three output ports, 03 h1, h2, h3 are the priority numbers of three output ports, h1, h2, h3 are the priority numbers of three output ports, h1, h2, h3 are the priority numbers of three output ports, h h1, h2, h3 are the priority numbers of three output ports,

[0084] The solution vector is represented as: where D is the length of the solution vector, H is the number of output ports in the time-sensitive network, |P h | is the priority number of the hth output port.

[0085] solutions of windows belonging to the same output port are considered as a group of solutions for decoding together, are decoded and

[0086] The formula of mapping the length value of each segment to the window length of a stream with a priority is:

[0087]

[0088] wherein, is the segment length corresponding to the i-th priority of the h-th output port, is the segment length corresponding to the m-th priority of the h-th output port, T GCL is the gating list period;

[0089] The constraint condition of and The constraint condition guarantees that all windows on one port can cover the whole period. The decoding and encoding method is only related to the value of each bit in the solution vector, and is not affected by the iteration process of the grey wolf algorithm.

[0090] is the offset of the opening time of the corresponding window relative to the start time of the gating list period, is the window length of the stream with the j-th priority of the h-th output port, P m is the m-th priority, P j is the j-th priority, is the offset of the opening time of the first window relative to the start time of the gating list period.

[0091] The worst-case delay analysis method of the present application adopts network calculus theory. After constructing the arrival curve and the service curve based on the gating list, the upper bound of the queuing and transmission delay of the stream with the priority P m at the output port h is expressed as: wherein, and are the arrival curve and the service curve of the stream at the output port h, respectively, and h(·) is the maximum horizontal distance of the two curves.

[0092] By accumulating the delay upper bound of each output port on the routing path, the worst-case end-to-end delay of the stream f i can be obtained and expressed as:

[0093] wherein, is the priority P mthe upper bound of queuing and transmission delay of the flow at the hth output port, des represents the last output port, t prop is the propagation delay, t proc is the processing delay. |R(f i ) | represents the number of output ports in the set.

[0094] Since the propagation and processing delays can be regarded as constants, in the optimization design, the present application only focuses on the queuing and transmission delay part related to the gating list. Therefore, for the gating list scheme of the entire solution, the smaller the fitness is, the smaller the delay of the entire scheduling scheme is, and the better the scheduling effect is.

[0095] The fitness is represented as:

[0096] wherein O is the accumulated upper bound of the delay of each output port, F is the set of flows, f i is the ith flow, R(f i ) is the set of all output ports on the routing path of f i , D fi represents the maximum end-to-end delay of f i .

[0097] Through the decoding and encoding methods described above, each candidate solution can be mapped to a gating list scheme and its worst delay can be obtained through network calculus. The grey wolf algorithm can change the numerical value of each bit of the solution to generate different gating list schemes, so as to realize the search of a better scheduling solution. As the network scale expands, the dimension of the calculated solution will be higher and higher, and the search space will be larger and larger, so the grey wolf algorithm needs to be improved accordingly to improve its search ability. The improvement of the present application mainly consists of the following two aspects, and the overall process is as shown in Figure 7 .

[0098] The step 203 specifically comprises:

[0099] Step 2031: initialize the first iteration number.

[0100] Step 2032: take each candidate solution as a grey wolf, and calculate the fitness of each grey wolf.

[0101] Step 2033: update the position of the grey wolf according to the fitness to obtain the current first wolf pack.

[0102] Step 2034: optimize the current first wolf pack by using a crossover operator to obtain the current second wolf pack.

[0103] Step 2035: optimize the current second wolf pack by using a mutation operator, and introduce a tabu search algorithm in the mutation process to obtain the current third wolf pack.

[0104] Step 2036: judging whether the first iteration number reaches the first set iteration number.

[0105] If step 2036 is yes, step 2037 is executed.

[0106] Step 2037: taking the solution corresponding to the current optimal gray wolf position as the scheduling scheme of the time-sensitive network.

[0107] If step 2036 is no, step 2038 is executed.

[0108] Step 2038: adding 1 to the first iteration number, taking the current third wolf group as the current first wolf group, calculating the fitness of each gray wolf in the current first wolf group, and returning to step of obtaining the current second wolf group by optimizing the current first wolf group using the crossover operator.

[0109] In step 2033, three gray wolves with the optimal fitness are selected as the head wolves, the search parameters of the gray wolf algorithm are updated, and the positions of other gray wolves are updated according to the head wolves, that is, other solutions are updated.

[0110] (1) Crossover operator based on genetic algorithm: genetic algorithm uses two parent partial structures to replace and recombine to generate new individuals through crossover operation, which greatly improves the global search ability. Inspired by this, we introduce the crossover operator into the gray wolf algorithm, and let the solution have a certain probability P c Continue to cross. The present application chooses the precedential order crossover (POX) which can well inherit the excellent features of the parent generation while ensuring the feasibility of the offspring. Another parent individual required for crossing is obtained by random selection. The offspring selection uses the elite selection strategy to compare the fitness of the two offspring and the two parents, and the optimal individual is selected to enter the subsequent iteration.

[0111] In step 2034, it specifically includes:

[0112] Traverse the current first wolf group, and randomly select whether the currently traversed gray wolf is optimized using the crossover operator, which specifically includes: if P i1 >P c , the currently traversed gray wolf is optimized using the crossover operator, and if P i1 ≤P c , the currently traversed gray wolf is not optimized using the crossover operator; P i1 is a random value, and P c is a fixed value.

[0113] If yes, randomly select another gray wolf from the current first wolf pack except the currently traversed gray wolf, cross the currently traversed gray wolf and the other gray wolf as parents to obtain two offspring gray wolves. The gray wolf with the smallest fitness among the two parent gray wolves and the two offspring gray wolves is used as the gray wolf after cross optimization.

[0114] After the traversal of the current first wolf pack is completed, the gray wolves after crossover optimization and the wolf packs in the current first wolf pack that are not optimized using the crossover operator constitute the current second wolf pack.

[0115] (2) Mutation operator based on taboo search: This application also introduces a mutation operator and a taboo search algorithm in the mutation process to improve the local development capability of the gray wolf algorithm. Its basic implementation process is as follows: first, generate candidate solutions, take the best individual after crossover as the initial solution, generate 50 domain solutions by randomly exchanging points, and select 30 of them with the best fitness as candidate solutions; then update the taboo table, select the best candidate solution S and compare it with the historical optimal solution S*, and add it to the taboo table if it is better than the historical optimal. The taboo table stores the changed points and does not participate in the subsequent iterations to avoid repeated roundabout searches. If it is not better than the historical optimal and has appeared in the taboo table, the solution is deleted and the next candidate solution is determined. Otherwise, the solution is added to the taboo table and used as the initial solution for the next iteration. Finally, the individual iteration 50 times is used as the convergence criterion, and the historical optimal solution is output as the optimization result. In order to speed up the solution, we stipulate that the three leading wolves in each iteration of the gray wolf algorithm must perform taboo search, while the other wolves have only a lower probability of performing taboo search.

[0116] Step 2035 specifically includes:

[0117] Traverse each gray wolf in the current second wolf pack. If the gray wolf currently traversed is the alpha wolf, then the gray wolf is used as the gray wolf to be mutated; if the gray wolf currently traversed is not the alpha wolf, randomly select whether to use the mutation operator for optimization. If the random selection is yes, then the gray wolf is used as the gray wolf to be mutated. If the random selection is no, then the gray wolf is not involved in the mutation. Specifically, if P i2 >P s The currently traversed gray wolf is the gray wolf to be mutated. If P i2 ≤P s The currently traversed gray wolf is not involved in the mutation; P i2 is a random value, P s is a fixed value.

[0118] If the currently traversed gray wolf is the gray wolf to be mutated, initialize the second iteration number;

[0119] The current traversed gray wolf is mutated by using a mutation operator to obtain a plurality of mutated and optimized gray wolves, and the current to-be-mutated wolf group is formed by the plurality of mutated and optimized gray wolves;

[0120] The gray wolf with the minimum fitness in the current to-be-mutated wolf group is taken as a temporary optimal solution;

[0121] It is judged whether the fitness of the temporary optimal solution is lower than the fitness of the historical optimal solution;

[0122] If yes, the temporary optimal solution is stored in the tabu table, and the historical optimal solution is updated to the temporary optimal solution;

[0123] If no, it is judged whether the temporary optimal solution is stored in the tabu table, if the temporary optimal solution is stored in the tabu table, the gray wolf corresponding to the temporary optimal solution is removed from the current to-be-mutated wolf group, and the step of taking the gray wolf with the minimum fitness in the current to-be-mutated wolf group as the temporary optimal solution is returned; if the temporary optimal solution is not stored in the tabu table, the temporary optimal solution is stored in the tabu table;

[0124] It is judged whether the second iteration number reaches a second set iteration number;

[0125] If no, the second iteration number is increased by 1, the temporary optimal solution is mutated by using a mutation operator to obtain a plurality of mutated and optimized gray wolves, and the current to-be-mutated wolf group is formed by the plurality of mutated and optimized gray wolves, and the step of taking the gray wolf with the minimum fitness in the current to-be-mutated wolf group as the temporary optimal solution is returned;

[0126] If yes, the second iteration number is initialized, and the following gray wolf in the current second wolf group is traversed until the traversal of the gray wolf in the current second wolf group ends, and the current third wolf group is obtained. The current third wolf group includes the gray wolf after the mutation and optimization of each to-be-mutated gray wolf (the temporary optimal solution of the last iteration) and the gray wolf not participating in the mutation.

[0127] The method provided in the application is based on network calculus analysis of the worst delay, reduces the worst delay by optimizing window parameters, and improves the scheduling success rate. Meanwhile, the gap-free design is adopted to compress the window switching number of the gating list, and the deployment difficulty is reduced. In order to further improve the global and local optimization effect, the gray wolf algorithm is improved, and a genetic operator and a mutation operator are introduced.

[0128] Based on the same inventive concept, the embodiments of the application also provide a gap-free window-based time-sensitive network scheduling device for implementing the gap-free window-based time-sensitive network scheduling method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, and therefore the specific limitations in one or more gap-free window-based time-sensitive network scheduling device embodiments provided below can be referred to the limitations of the gap-free window-based time-sensitive network scheduling method in the foregoing, which will not be described herein again.

[0129] In one example embodiment, the application provides a scheduling device based on non-interval window of time-sensitive network, which comprises:

[0130] A parameter obtaining module for obtaining parameters of a flow in a time-sensitive network.

[0131] A candidate set generating module for randomly generating a plurality of candidate solutions of non-interval windows for scheduling the time-sensitive network; each candidate solution is generated by decoding each solution vector, and each solution vector comprises a plurality of continuous value intervals, each value interval representing an output port, and each value interval being divided into a plurality of segments, and the length value of each segment mapping a window length of a flow of the same priority.

[0132] A candidate set optimizing module for taking the accumulated upper bound of the delay of each output port as fitness, optimizing the candidate solution through an improved grey wolf algorithm, and taking the optimized optimal solution as a scheduling scheme of the time-sensitive network; the improved grey wolf algorithm is a grey wolf algorithm after introducing a crossover operator and a mutation operator.

[0133] In one example embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram thereof can be as shown in Figure 8 The computer device comprises a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store scheduling data of a time-sensitive network based on non-interval windows. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement a scheduling method based on non-interval windows of a time-sensitive network.

[0134] Those skilled in the art can understand that, Figure 8The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in the above method embodiments.

[0135] In an exemplary embodiment, a computer readable storage medium is provided, storing a computer program, which is executed by a processor to implement the steps in the above scheduling method embodiments of the time sensitive network based on the non-interval window.

[0136] In an exemplary embodiment, a computer program product is provided, including a computer program, which is executed by a processor to implement the steps in the above scheduling method embodiments of the time sensitive network based on the non-interval window.

[0137] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0138] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, databases or other media used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc.

[0139] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a data processing logic of a programmable logic device, etc., without being limited thereto.

[0140] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the present application.

[0141] The principles and implementation modes of the present application are described by applying specific examples in the present application. The above-mentioned embodiments are only used to help understand the method and its core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A scheduling method for a time-sensitive network based on a non-interval window, characterized in that: The scheduling method based on a time-sensitive network without an interval window includes: Obtaining parameters of flows in a time-sensitive network; Based on the parameters of the flow in the time-sensitive network, multiple candidate solutions for gapless windows for scheduling the time-sensitive network are randomly generated. Each candidate solution is generated by decoding each solution vector. Each solution vector includes multiple continuous value intervals. Each value interval represents an output port. Each value interval is divided into multiple segments. The length of each segment maps the window length of the flow with the same priority. The accumulated upper bound of the delay of each output port is used as fitness, and the candidate solution is optimized by using the improved gray wolf algorithm, and the optimized optimal solution is used as the scheduling scheme of the time-sensitive network; the improved gray wolf algorithm is the gray wolf algorithm after introducing the crossover operator and the mutation operator; Flow parameters include priority, period, and size; When randomly generating multiple candidate solutions for scheduling time-sensitive networks with no gap windows based on the parameters of the flows in the time-sensitive network, the constraints on the window length are as follows: in, is the window length of the mth priority flow at the hth output port, It represents the set of flows of the mth priority transmitted by the hth output port, HP is the super period, and the value of the super period is The least common multiple of the periods of all flows in , f i is the i-th flow, f i .size is f i The size of f i .T is f i The period, T GCL is the gate list period, BW is the bandwidth, is the guard band length; The solution vector is expressed as: Where D is the length of the solution vector, H is the number of output ports in the time-sensitive network, |P h | is the priority number of the h-th output port; The formula for mapping the length of each fragment to the window length of a priority stream is: in, is the fragment length corresponding to the i-th priority of the h-th output port, is the length of the fragment corresponding to the mth priority of the hth output port, T GCL is the gate list period; The constraints include and for The corresponding window opening time is offset from the start time of the gate list period, is the window length of the j-th priority flow at the h-th output port, P m is the mth priority, P j is the jth priority, The opening time of the first window is offset from the start time of the gate list period.

2. The scheduling method based on time-sensitive network without interval window according to claim 1 is characterized in that: The fitness is expressed as: Among them, O is the cumulative upper bound of the delay of each output port, F is the set of flows, f i is the i-th stream, represents f i Maximum end-to-end delay; in, The priority is P m The upper bound of the queuing and transmission delay of the flow at the hth output port, R(f i ) is f i The set of all output ports on the routing path, des represents the last output port, t prop is the propagation delay, t proc To process delay.

3. The scheduling method based on time-sensitive network without gap window according to claim 1, characterized in that: The accumulated upper bound of the delay of each output port is used as the fitness, and the candidate solutions are optimized using the improved gray wolf algorithm. The optimized optimal solution is used as the scheduling scheme for the time-sensitive network. Specifically, it includes: Initialize the first iteration number; Taking each candidate solution as a gray wolf, calculating the fitness of each gray wolf; Update the position of the gray wolf according to each fitness to obtain the current first wolf pack; The crossover operator is used to optimize the current first wolf pack to obtain the current second wolf pack; The mutation operator is used to optimize the current second wolf pack, and the taboo search algorithm is introduced in the mutation process to obtain the current third wolf pack; Determine whether the first number of iterations reaches a first set number of iterations; If so, the solution corresponding to the current optimal gray wolf position is used as the scheduling solution for the time-sensitive network; If not, the first iteration number is increased by 1, the current third wolf pack is used as the current first wolf pack, the fitness of each gray wolf in the current first wolf pack is calculated, and the process returns to the step of using the crossover operator to optimize the current first wolf pack to obtain the current second wolf pack.

4. The scheduling method based on time-sensitive network without gap window according to claim 3, characterized in that: The crossover operator is used to optimize the current first wolf pack to obtain the current second wolf pack, specifically including: Traverse the current first wolf pack and randomly select whether to use the crossover operator for optimization for the currently traversed gray wolf; If yes, randomly select another gray wolf from the current first wolf pack except the currently traversed gray wolf, cross the currently traversed gray wolf and the other gray wolf as the parent generation to obtain two gray wolves of the offspring generation. The gray wolf with the smallest fitness among the two parent gray wolves and the two offspring gray wolves is used as the gray wolf after cross optimization; After the traversal of the current first wolf pack is completed, the gray wolves after crossover optimization and the wolf packs in the current first wolf pack that are not optimized using the crossover operator constitute the current second wolf pack.

5. The scheduling method based on time-sensitive network without gap window according to claim 3, characterized in that: The mutation operator is used to optimize the current second wolf pack, and the taboo search algorithm is introduced in the mutation process to obtain the current third wolf pack. Specifically, the following steps are performed: Traverse each gray wolf in the current second wolf pack. If the gray wolf currently traversed is the alpha wolf, then use it as the gray wolf to be mutated. If the gray wolf currently traversed is not the alpha wolf, randomly choose whether to use the mutation operator for optimization. If the random choice is yes, then use it as the gray wolf to be mutated. If the random choice is no, then the gray wolf will not participate in the mutation. If the currently traversed gray wolf is the gray wolf to be mutated, initialize the second iteration number; The mutation operator is used to mutate the currently traversed gray wolves, and the current wolf group to be mutated is formed by multiple optimized gray wolves. The gray wolf with the lowest fitness among the current wolf pack to be mutated is taken as the temporary optimal solution; Determine whether the fitness of the temporary optimal solution is lower than the fitness of the historical optimal solution; If so, the temporary optimal solution is stored in the taboo table, and the historical optimal solution is updated to the temporary optimal solution; If not, determining whether the temporary optimal solution is stored in the taboo table; if so, removing the gray wolf corresponding to the temporary optimal solution from the current wolf pack to be mutated, and returning to the step of using the gray wolf with the smallest fitness in the current wolf pack to be mutated as the temporary optimal solution; if not, storing the temporary optimal solution in the taboo table; Determine whether the second iteration number reaches the second set iteration number; If not, the second iteration number is increased by 1, and the temporary optimal solution is mutated using the mutation operator to obtain the current wolf pack to be mutated, which is composed of multiple mutated optimized gray wolves. The step of returning to the step of taking the gray wolf with the smallest fitness in the current wolf pack to be mutated as the temporary optimal solution is returned. If so, initialize the second iteration number, traverse the gray wolves in the current second wolf pack, until the traversal of the gray wolves in the current second wolf pack is completed, and obtain the current third wolf pack.

6. A scheduling device based on a time-sensitive network without an interval window, characterized in that: The scheduling device based on a time-sensitive network without an interval window includes: A parameter acquisition module for a flow in a time-sensitive network, used to obtain parameters of a flow in a time-sensitive network; A candidate set generation module is used to randomly generate multiple candidate solutions for gapless windows for scheduling time-sensitive networks. Each candidate solution is generated by decoding each solution vector. Each solution vector includes multiple continuous value intervals, each value interval represents an output port, and each value interval is divided into multiple segments. The length of each segment maps the window length of flows with the same priority. A candidate set optimization module is used to use the accumulated upper bounds of the delays of each output port as fitness, optimize the candidate solutions using an improved gray wolf algorithm, and use the optimized optimal solution as the scheduling scheme for the time-sensitive network; the improved gray wolf algorithm is a gray wolf algorithm that introduces a crossover operator and a mutation operator; Flow parameters include priority, period, and size; When randomly generating multiple candidate solutions for scheduling time-sensitive networks with no gap windows based on the parameters of the flows in the time-sensitive network, the constraints on the window length are as follows: in, is the window length of the mth priority flow at the hth output port, It represents the set of flows of the mth priority transmitted by the hth output port, HP is the super period, and the value of the super period is The least common multiple of the periods of all flows in , f i is the i-th flow, f i .size is f i The size of f i .T is f i The period, T GCL is the gate list period, BW is the bandwidth, is the guard band length; The solution vector is expressed as: Where D is the length of the solution vector, H is the number of output ports in the time-sensitive network, |P h | is the priority number of the h-th output port; The formula for mapping the length of each fragment to the window length of a priority stream is: in, is the fragment length corresponding to the i-th priority of the h-th output port, is the length of the fragment corresponding to the mth priority of the hth output port, T GCL is the gate list period; The constraints include and for The corresponding window opening time is offset from the start time of the gate list period, is the window length of the j-th priority flow at the h-th output port, P m is the mth priority, P j is the jth priority, The opening time of the first window is offset from the start time of the gate list period.

7. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the time-sensitive network scheduling method based on gapless windows according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the scheduling method based on a time-sensitive network without an interval window according to any one of claims 1 to 5 is implemented.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the scheduling method based on a time-sensitive network without an interval window according to any one of claims 1 to 5 is implemented.

Images