TSN cross-domain traffic semi-distributed scheduling method based on two-color collaborative bat algorithm

Abstract

The invention belongs to the technical field of wireless communication, and particularly relates to a TSN cross-domain traffic semi-distributed scheduling method based on a two-color cooperative bat algorithm, which comprises the following steps: deploying an intra-domain scheduler in each local TSN network in a global TSN network range, and respectively collecting and reporting a traffic transmission demand in the corresponding local TSN network by each intra-domain scheduler; a cross-domain coordinator is arranged in a global TSN network and used for summarizing traffic transmission requirements in the local TSN network reported by each intra-domain scheduler, obtaining a traffic set and issuing the traffic set, after the intra-domain scheduler receives the traffic set, committed time delay is calculated through an intra-domain traffic scheduling method and reported, and after the cross-domain coordinator receives the calculated committed time delay, the cross-domain scheduler sends the committed time delay to the local TSN network. Generating an inter-domain scheduling strategy through an inter-domain routing method and issuing the inter-domain scheduling strategy; and solving by adopting a two-color collaborative bat algorithm to obtain a global optimal solution of traffic scheduling in the local TSN and path selection of the local TSN so as to improve the scheduling efficiency of the global TSN traffic.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communications, and in particular relates to a TSN cross-domain traffic semi-distributed scheduling method based on a two-color collaborative bat algorithm. Background Art

[0002] With the rapid development of Industry 4.0, smart manufacturing and Internet of Things technologies, Time-Sensitive Networking (TSN), as a key technology to achieve efficient, reliable and real-time communication, has been widely used in industrial automation, smart grid, smart transportation and other fields. Through precise time synchronization and traffic scheduling mechanisms, TSN networks ensure low-latency and low-jitter transmission of data packets, meeting the strict requirements of various services for time sensitivity.

[0003] However, in practical applications, TSN networks often face the challenge of cross-domain traffic scheduling. Since most of the currently built TSN networks are in a closed state with no information exchange, it is difficult for each local TSN network to be connected in series, making it difficult to provide end-to-end transmission guarantee for wide-area services. In order to break this bottleneck, TSN cross-domain traffic scheduling technology came into being and became the core technology to solve this problem.

[0004] Among the existing technical solutions, although there have been some attempts to solve the problems of TSN network time synchronization and traffic scheduling, the existing TSN cross-domain traffic scheduling technology mainly faces a key problem, namely how to generate multi-domain fusion gating list configuration information to enhance the globality of intra-domain network scheduling. Solving this problem is crucial to achieving optimized TSN cross-domain traffic scheduling. Summary of the invention

[0005] In view of this, the present invention provides a TSN cross-domain traffic semi-distributed scheduling method based on a two-color collaborative bat algorithm to solve the technical problem that the prior art cannot generate multi-domain fusion gating list configuration information to enhance the globality of intra-domain network scheduling.

[0006] To achieve the above objectives, this application adopts the following scheme:

[0007] A TSN cross-domain traffic semi-distributed scheduling method based on a two-color cooperative bat algorithm includes the following steps:

[0008] S10. Deploy an intra-domain scheduler in each local TSN network within the global TSN network, and each intra-domain scheduler is responsible for the traffic scheduling in the corresponding local TSN network: each intra-domain scheduler collects the traffic transmission requirements in the corresponding local TSN network, and reports the traffic transmission requirements in the local TSN network;

[0009] S20. A cross-domain coordinator is set in the global TSN network, and the cross-domain coordinator is in communication with each of the intra-domain schedulers. The cross-domain coordinator is responsible for path selection of each of the local TSN networks: the cross-domain coordinator summarizes the traffic transmission requirements in the local TSN network reported by each of the intra-domain schedulers, obtains a traffic set, and sends the traffic set to each of the intra-domain schedulers;

[0010] S30. After receiving the traffic set, each intra-domain scheduler calculates the committed delay by the intra-domain traffic scheduling method, and reports the calculated committed delay to the cross-domain coordinator;

[0011] S40. After receiving the calculated committed delay, the cross-domain coordinator generates an inter-domain scheduling strategy through an inter-domain routing selection method, and sends the inter-domain scheduling strategy to each of the intra-domain schedulers, and the intra-domain scheduler performs distributed scheduling between intra-domain nodes and links for the allocated traffic according to the received inter-domain scheduling strategy;

[0012] S50. The traffic scheduling within the local TSN network and the path selection of the local TSN network are jointly optimized through the server within the intra-domain scheduler, and the two-color collaborative bat algorithm is used to solve the problem, so as to obtain the global optimal solution for the traffic scheduling within the local TSN network and the path selection of the local TSN network, so as to improve the scheduling efficiency of the global TSN traffic.

[0013] Preferably, the intra-domain traffic scheduling method comprises the following steps:

[0014] S31. Define the local TSN network topology as a directed graph G = (V, E), where V represents a node set, E represents a link set, and the node set V includes an internal node v in and cross-domain node v e , the traffic set is F, each flow f in the traffic set F i It is expressed as:

[0015] f i =(T i ,D i ,L i ,st i ,des i )

[0016] Among them, T i is the period of the flow, D i is the data frame composition of the traffic, L i is the cut-off time of the flow, st i ,des i are the source node and destination node of the traffic respectively;

[0017] S32. The constraints on the order in which each flow is sent include that for the same link, the sending time of the first frame must be non-negative, and each frame must be sent in a sequential order:

[0018]

[0019] in, Denotes that each flow f in the flow set F i The kth frame on link e, Indicates the closing time of GCL. Indicates the door opening time of GCL, i, e, k are all indexes;

[0020] S33. The frame collision constraint includes that within the same transmission time slot, the link resources should be exclusively occupied by the current traffic, otherwise collisions will occur between different traffic flows:

[0021]

[0022] Among them, T i and T j f i and f j The corresponding flow period, m, n are non-negative integers;

[0023] S34. The intra-domain path constraints include that the local TSN network traffic is transmitted in the topology in accordance with the order of links:

[0024]

[0025] Among them, e ab ,e bc ∈E,e ab ,e bc is any adjacent link in the link set E;

[0026] S35. The cut-off time constraints for each flow include that the end-to-end delay of the flow does not exceed its cut-off time:

[0027]

[0028] Among them, L i is the cut-off time of the flow, st i is the source node of the traffic, des i is the destination node of the traffic, Indicates the opening time of GCL. Indicates the closing time of GCL;

[0029] S36. The traffic scheduling objective function within the definition domain, i.e. minimizing the sum of the time required for each flow to reach the destination node, includes:

[0030]

[0031] Among them, f i For each flow in the flow set F', H1 represents the objective function 1.

[0032] Preferably, the inter-domain routing selection method comprises the following steps:

[0033] S41. Assume that the network topology covering multiple local TSN networks is G'=<D,E> , D represents the set of local TSN networks in the network, E represents the link set, the set of traffic reported by all local TSN networks is F', and the optimal transmission path set R between domains for each traffic is calculated. Each flow f in the traffic set F' i It is expressed as:

[0034] f i =(P i ,U i ,L i ,st i ,des i )

[0035] Among them, P i is the lower bound of the delay commitment of the flow in each domain, U i is the upper bound of the delay commitment of each domain to the flow, L i is the deadline of the flow, st i ,des i are the source node and destination node of the flow respectively;

[0036] S42. The inter-domain path constraint includes flow f in each flow in the flow set F' i The path R i Should be st i As the starting point, des i For the end point:

[0037]

[0038] Among them, R is the optimal transmission path set of each flow between domains, R i is the optimal transmission path between domains for each flow, st i is the source node of the stream, des i is the destination node of the flow, d a and d b for any two adjacent domains;

[0039] S43. When the flow cut-off time is constrained, the TSN flow f i The end-to-end delay of is composed of the intra-domain transmission delay and the inter-domain link delay, and the delay does not exceed its specified deadline:

[0040]

[0041] Among them, f i For each flow in the flow set F', is the flow rate f i The lower bound of the delay commitment in the d domain, L i is the cut-off time of the traffic;

[0042] S44. The objective function of the inter-domain traffic routing selection is:

[0043]

[0044] Among them, H2 represents the objective function 2, is the flow rate f i The lower bound of the delay commitment in the d domain is, is the flow rate f i The upper bound of the delay commitment in the d domain.

[0045] Preferably, the two-color collaborative bat algorithm comprises the following steps:

[0046] S51. The M bats are divided into U black bats and W gray bats, wherein the black bats are responsible for intra-domain traffic scheduling and the gray bats are responsible for inter-domain routing selection;

[0047] S52. The initial bat position, discretizing the bat position in a given space to form a set of solutions, the bats are randomly distributed in the space, and the initial position of the bat is:

[0048]

[0049] Among them, the vector represents the initial position of the μ / ωth bat;

[0050] S53. Using real number coding strategy, use vector information to represent each bat individual. Ordinalize by the size of the element Get the priority order of individual bat position information:

[0051]

[0052] S54. In each super period t, each bat performs a foraging process, and updates the speed and position of each bat. The speed and position of the bat are iteratively updated to find the global optimal solution. The speed and position update is divided into three stages: independent update, linkage update, and local search.

[0053] Preferably, the independent update algorithm is:

[0054]

[0055] in, is the flying speed of the μ / ωth black / gray bat at time t, is the flight speed of the μ / ωth black / gray bat at time t-1, rand is the random factor generated by Chebyshev mapping, rand∈(0,1), f μ / ω is the pulse frequency, is the position of the black / gray bat after independent update at time t, ∈ is the mutation weight, C(0,1) is the random number generated by the Cauchy probability distribution with a scale parameter of t=1, is the logarithmically decreasing inertia weight.

[0056] Preferably, the logarithmically decreasing inertia weight It is obtained by the following formula:

[0057]

[0058] in, λ is the inertia weight adjustment factor, which is a constant in the range of (0,1); T max is the maximum number of iterations of bat flight, and a is a constant.

[0059] Preferably, the linkage update includes defining {x μi t ,x ωj t} represents a set of bat positions that achieve the global optimal objective function value at super period t;

[0060] Where i,j = argmin {i,j} F(x μi t ,x ωj t

[0061] Within the constraints, the speed and position of the black bat at time t are updated as:

[0062]

[0063] The speed and position of the gray bat at time t are updated as:

[0064]

[0065] in, is the logarithmically decreasing inertia weight, v represents velocity, and x represents position.

[0066] Preferably, during the local search: if the bat flies to a location greater than {x μ t-1 ,x ω t-1} A better new position, the bat will conduct a local search around the position. During the local search, the random walk method is used, that is, when the respective objective function satisfies H1(x μ t ) μ t-1 ) or H2(x ω t ) <H2(x ω t-1 ), and when the random conditions and constraints are met, they are:

[0067]

[0068] Among them, H1 represents objective function 1, H2 represents objective function 2, and the random variable ε obeys the uniform distribution on [-1,1]. Represents the average sound intensity of the tth generation in the black bat / gray bat population. At each moment, the pulse intensity and pulse emission frequency are updated according to the formula:

[0069]

[0070] is the pulse intensity emitted by the μ / ωth bat in the tth generation, α is the pulse intensity attenuation factor, represents the pulse emission frequency of the μ / ωth bat in the tth generation, and γ represents the adjustment factor of the pulse emission frequency; at the end of each moment, it is determined whether the maximum number of iterations has been reached. If so, the search is stopped and the position and fitness value of the bat corresponding to the global optimal solution are output, otherwise the search continues.

[0071] The technical solution adopted in this application can achieve the following beneficial effects:

[0072] ​In a TSN cross-domain traffic semi-distributed scheduling method based on a two-color collaborative bat algorithm provided by the present invention, an intra-domain scheduler is deployed in each local TSN network within the scope of the global TSN network, and each of the intra-domain schedulers respectively collects the traffic transmission requirements in the corresponding local TSN network, and reports the traffic transmission requirements in the local TSN network; a cross-domain coordinator in the global TSN network is communicated with each of the intra-domain schedulers, and is used to summarize the traffic transmission requirements in the local TSN network reported by each of the intra-domain schedulers, obtain a traffic set, and send the traffic set to each of the intra-domain schedulers. After receiving the calculated committed delay, the cross-domain coordinator generates an inter-domain scheduling strategy through an inter-domain routing selection method, and sends the inter-domain scheduling strategy to each of the intra-domain schedulers. The controller performs distributed scheduling of the allocated traffic between nodes and links in the domain according to the received inter-domain scheduling strategy; the intra-domain traffic scheduling aims to optimize the intra-domain scheduling so that the sum of the total delay time of all flows when arriving at the destination node is minimized, and the inter-domain routing selection aims to optimize load balancing, solves the problem of generating gating list configuration information for multi-domain fusion, improves the globality of intra-domain network scheduling, breaks through the information barrier between multiple domains, and supports flexible domain-level routing; the intra-domain traffic scheduling and inter-domain routing selection are jointly optimized, and the two-color collaborative bat algorithm is used to solve it. By jointly optimizing the large-scale search problem in high-dimensional discrete space, the global optimal solution is obtained. The global optimal solution is used to redistribute and select the intra-domain traffic scheduling and the inter-domain routing selection, which strengthens the globality of intra-domain network scheduling and improves the scheduling efficiency of global TSN traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is the TSN cross-domain traffic semi-distributed scheduling architecture in the present invention.

[0074] Figure 2 This is an interactive flow chart of cross-domain semi-distributed scheduling of TSN traffic in the present invention.

[0075] Figure 3 Schematic diagram of the cross-domain semi-distributed scheduling method of TSN streams based on dual-color collaborative bats in the present invention. DETAILED DESCRIPTION

[0076] In order to facilitate the understanding of the present application, the present application will be described more comprehensively below in conjunction with the accompanying drawings. And the preferred implementation of the present application is given. However, the present application can be implemented in many different forms and is not limited to the implementation described herein. On the contrary, the purpose of providing these implementations is to make the disclosure of the present application more thoroughly and comprehensively understood.

[0077] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0078] This embodiment provides a TSN cross-domain traffic semi-distributed scheduling method based on a two-color cooperative bat algorithm, including the following steps:

[0079] S10. Deploy an intra-domain scheduler in each local TSN network within the global TSN network, and each intra-domain scheduler is responsible for the traffic scheduling in the corresponding local TSN network: each intra-domain scheduler collects the traffic transmission requirements in the corresponding local TSN network, and reports the traffic transmission requirements in the local TSN network;

[0080] S20. A cross-domain coordinator is set in the global TSN network, and the cross-domain coordinator is in communication with each of the intra-domain schedulers. The cross-domain coordinator is responsible for path selection of each of the local TSN networks: the cross-domain coordinator summarizes the traffic transmission requirements in the local TSN network reported by each of the intra-domain schedulers, obtains a traffic set, and sends the traffic set to each of the intra-domain schedulers;

[0081] S30. After receiving the traffic set, each intra-domain scheduler calculates the committed delay by the intra-domain traffic scheduling method, and reports the calculated committed delay to the cross-domain coordinator;

[0082] S40. After receiving the calculated committed delay, the cross-domain coordinator generates an inter-domain scheduling strategy through an inter-domain routing selection method, and sends the inter-domain scheduling strategy to each of the intra-domain schedulers, and the intra-domain scheduler performs distributed scheduling between intra-domain nodes and links for the allocated traffic according to the received inter-domain scheduling strategy;

[0083] S50. The traffic scheduling within the local TSN network and the path selection of the local TSN network are jointly optimized through the server within the intra-domain scheduler, and the two-color collaborative bat algorithm is used to solve the problem, so as to obtain the global optimal solution for the traffic scheduling within the local TSN network and the path selection of the local TSN network, so as to improve the scheduling efficiency of the global TSN traffic.

[0084] In this embodiment, Figure 1A semi-distributed scheduling architecture for TSN cross-domain traffic is proposed. Within the global TSN network, it is assumed that an intra-domain scheduler is deployed in each local TSN network. At the same time, there is a cross-domain coordinator in the network that can communicate with each intra-domain scheduler. The intra-domain scheduler is responsible for the traffic scheduling in the corresponding local TSN network, which includes: calculating the path planning and GCL (Gate Control List) gating scheduling of the carried traffic to ensure that the traffic arrives on demand and optimize the overall transmission delay; the cross-domain coordinator further considers the cross-domain link overhead.

[0085] Figure 2 The interaction process between the intra-domain scheduler and the cross-domain coordinator in S10 to S40 of this embodiment is given. Each of the intra-domain schedulers collects the traffic transmission requirements in the corresponding local TSN network respectively, and reports the traffic transmission requirements in the local TSN network. The cross-domain coordinator summarizes the traffic transmission requirements in the local TSN network reported by each of the intra-domain schedulers to obtain the global traffic transmission requirements, that is, the traffic set, and sends the traffic set to each of the intra-domain schedulers; after each of the intra-domain schedulers receives the traffic set, it uses the intra-domain traffic coordination The committed delay is calculated by a method, where the committed delay is specifically the committed delay of the traffic set, and the committed delay includes a lower bound and an upper bound of the delay generated by priority scheduling or post-scheduling of each traffic. The calculated committed delay is reported to the cross-domain coordinator. After receiving the calculated committed delay, the cross-domain coordinator generates an inter-domain scheduling policy through an inter-domain routing selection method, and sends the inter-domain scheduling policy to each of the intra-domain schedulers. The intra-domain schedulers perform distributed scheduling on the allocated traffic between intra-domain nodes and links according to the received inter-domain scheduling policy.

[0086] Since the upper and lower bounds of the committed delay obtained by the intra-domain traffic scheduling method will affect the inter-domain routing selection, and the inter-domain routing selection will also affect the scheduling of intra-domain traffic, under a semi-distributed architecture, when the two are solved as independent problems, the optimality of the global traffic scheduling will be difficult to guarantee. In order to maximize the scheduling efficiency of global TSN traffic, this embodiment uses a two-color collaborative bat algorithm to jointly optimize the traffic scheduling method and the inter-domain routing selection method.

[0087] Specifically, the intra-domain traffic scheduling method includes the following steps:

[0088] S31. Intra-domain traffic scheduling must ensure that the end-to-end transmission time of each TSN flow does not exceed its deadline. At the same time, the total traffic scheduling time of the entire TSN network must be as short as possible, so that the overall GCL (Gate Control List) scheduling table is as compact as possible. The intra-domain TSN network topology with TSN switches as nodes is abstracted into a directed graph G = (V, E), where V represents the node set and E represents the link set. V contains two types of nodes: internal nodes and cross-domain nodes, which are represented by v respectively. in and v e Indicates that node v i to v j The link between them is represented by e ij The set of all flows in this domain is F, where each flow f i Represented as a quintuple f i =(T i ,D i ,L i ,st i ,des i ), where T i is the period of the flow, D i The data frame of the stream is composed of L i is the deadline of the flow, st i ,des i They are the source node and destination node of the flow respectively. When transmitting in the TSN domain network, in order to ensure that the data flow can be delivered in a timely and accurate manner, the following constraints must be met:

[0089] S32. Stream sending order constraint: For the same link, the sending time of the first frame must be non-negative, and each frame must be sent in sequence:

[0090]

[0091] S33. Frame collision constraint: In the same transmission time slot, the link resources should be exclusively occupied by the current flow, otherwise collisions will occur between different flows. i and T j f i and f j The corresponding flow period, m, n are non-negative integers:

[0092]

[0093] S34. Intra-domain path constraint: TSN flows are transmitted in the topology in accordance with the order of links. ab ,e bc ∈E, we have:

[0094]

[0095] S35. Flow deadline constraint: The end-to-end delay of a flow should not exceed its deadline.

[0096]

[0097] S36. Obtaining Objective 1: The intra-domain traffic scheduling objective function is defined as the sum of the time required for each flow to reach the destination node:

[0098]

[0099] The intra-domain traffic scheduling will obtain the GCL solution for each intra-domain traffic and the corresponding delay of each traffic when the solution is adopted;

[0100] In the above S32 to S36, Represents the flow f i In the kth frame on link e, Indicates the opening time of GCL. Indicates the closing time of GCL. In the domain, the link propagation delay can be regarded as 0, and because of the use of time slot conflict-free scheduling, the queuing delay is also 0, so the one-hop delay is equal to the processing delay in the switch plus the transmission delay of the outbound port.

[0101] Specifically, the inter-domain routing selection method includes the following steps:

[0102] S41. The goal of inter-domain traffic routing selection is to make reasonable routing decisions for each flow at the entire network level based on the flow transmission demand information reported by the intra-domain scheduler and the delay commitment for each flow. Suppose the network topology covering multiple TSN domains is G'=<D,E> Where D represents the set of TSN domains in the network, and E represents the set of cross-domain links connecting the domains. The set of traffic reported by all TSN domains is F', where each flow f i Represented as a quintuple f i =(P i ,U i ,L i ,st i ,des i ), where P i is the lower bound of the delay commitment for the traffic in each domain, U i is the upper bound of the delay commitment of each domain to the flow, L i is the deadline of the flow, st i ,des iare the source node and destination node of the flow respectively. Assume that the cross-domain scheduler does not know the resource details of each domain, but only makes decisions based on part of it. Therefore, at this stage, the detailed features such as the traffic cycle and data frame composition are no longer of concern. The goal of this stage is to use the algorithm to input the directed graph G'=<D,E> and traffic set F', calculate the optimal transmission path set R of each traffic between domains, and obtain the optimal traffic success scheduling rate and average delay. Between domains, the link propagation delay cannot be considered as 0. Assume that the link delay between domains is Among them, d a and d b is an adjacent domain, the flow f i The lower bound of the delay commitment in the d domain is The upper bound of the delay commitment is Delay Commitment Lower Bound Defined as d-domain priority scheduling f i That is, assuming that the traffic set F in the domain contains only a single f i When , the delay value output by the intra-domain traffic scheduling method; the upper bound of the delay commitment is It is defined as: domain d first schedules all other flows in the flow set F within the domain, and finally schedules a single flow f i When , the delay value output by the intra-domain traffic scheduling method.

[0103] S42. Inter-domain path constraints: TSN flow f i The path R i Should be st i As the starting point, des i The endpoints are connected at the network level and there is no loop:

[0104]

[0105] S43. Flow deadline constraint: TSN flow f i The end-to-end delay of is composed of the intra-domain transmission delay and the inter-domain link delay. The delay should not exceed its specified deadline. Within the constraints of this stage, the intra-domain transmission delay only considers the optimal case, that is, the case where the lower bound of the delay commitment is achieved:

[0106]

[0107] S44. Obtaining Objective 2: This embodiment sets the objective function of inter-domain traffic routing selection from the perspective of load balancing:

[0108]

[0109] Inter-domain routing selection will result in a domain-level routing solution for global traffic in the entire TSN network.

[0110] The upper and lower bounds of the promised latency obtained by the intra-domain traffic scheduling method will affect the inter-domain routing selection. At the same time, the inter-domain routing selection will also affect the scheduling of intra-domain traffic. In a semi-distributed architecture, when the two are solved as independent problems, the optimality of global traffic scheduling will be difficult to guarantee. In order to maximize the scheduling efficiency of global TSN traffic, please refer to Figure 3 The present invention provides the following embodiment to select the above-mentioned target 1 and the above-mentioned target 2 for joint optimization using the two-color cooperative bat algorithm:

[0111] In this embodiment, assume that there are M bats, divided into black and gray. In each super period t (equal to the least common multiple of all traffic periods), black bats are responsible for intra-domain traffic scheduling, and there are U black bats in total; gray bats are responsible for inter-domain routing selection, and there are W gray bats in total.

[0112] The positions of black / gray bats represent feasible solutions to the corresponding problems. During initialization, the positions of bats are discretized in a given space to form a set of solutions. Bats are randomly distributed in the space, as shown below:

[0113]

[0114] where the vector Represents the initial position of the μ / ωth bat.

[0115] Considering the characteristics of the heuristic algorithm, this embodiment adopts a real number encoding strategy and uses vector information to represent each individual bat. The specific dimension of the solution depends on the problem itself. Before the bat algorithm solves the TSN traffic scheduling problem, a suitable encoding mechanism must be established to convert the location information of individual bats into a traffic scheduling sequence with a certain topological structure, which is expressed as follows:

[0116] First, the vector Ordinalize by the size of the element

[0117]

[0118] The possible values ​​are from {1,2,…,N} and they are not repeated.

[0119] Will Specifies the priority order of the next hop node selected by the intra-domain scheduler or inter-domain coordinator for the next flow to be scheduled. i , the ordered set of nodes directly connected to it is cw i =[…,n α0 ,…,n β0 ,…,n γ0 ,…]. Given a TSN stream f j, connect the nodes according to f j Destination address des j The minimum distance of the order from small to large is d i,j =[…,n α ,…,n β ,…,n γ ,…]. Combine the two ordered sets and select the item with the smallest sum of ordinals in the two ordered sets as the given TSN flow f j The next hop node (excluding the previous hop node).

[0120] After randomly initializing the bat's position, the bat begins to forage for food by echolocation. In each super period t, the foraging process of each bat, i.e., speed and position update, is divided into three stages: independent update, linkage update, and local search:

[0121] Independent Updates: For black / gray bats, the proposed algorithm gives independent updates of velocity and position as follows:

[0122] f μ / ω =f μ / ωmin +(f μ / ωmax -f μ / ωmin )rand

[0123]

[0124] In the above formula, is the flight speed of the μ / ωth black / gray bat at time t and time t-1, and the dimension is The dimension is the same as that of the random factor generated by the Chebyshev map, rand∈0,1), and is continuously updated with use, so the pulse frequency f μ / ω It is random and satisfies f μ / ωmin ≤f μ / ω ≤f μ / ωmax , is the position of the black / gray bat after independent update at time t, ∈ is the mutation weight, and C(0,1 is the random number generated by the Cauchy probability distribution with a proportional parameter t=1, which makes the algorithm mutate to a greater extent in the early stage and to a lesser extent in the later stage, making the algorithm based on convergence ability and avoiding falling into local optimality. is the logarithmically decreasing inertia weight, expressed as:

[0125]

[0126] in λ is the inertia weight adjustment factor, which is a constant in the range of (0,1); T max is the maximum number of iterations of bat flight; a is a constant used to adjust the update speed, which can generally be set to 1. With the number of iterations, the accuracy decreases rapidly in the early stage and slowly in the later stage, so that the accuracy can be improved through fine search after quickly reaching the optimal range.

[0127] Linked update: Considering that inter-domain traffic scheduling and cross-domain routing selection are not completely independent, at the end of each super period t, the positions and speeds of black bats and gray bats will be linked and updated according to the objective function:

[0128] Define {x μi t ,x ωj t} represents a set of bat positions that achieve the global optimal objective function value at super period t. Where i, j = argmin {i,j} F(x μi t ,x ωj t ). F(x μi t ,x ωj t ) is summarized by each intra-domain scheduler to the cross-domain coordinator.

[0129] Within the constraints, the speed and position of the black bat at time t are updated as:

[0130]

[0131] The speed and position of the gray bat at time t are updated as:

[0132]

[0133] Local search: Considering its own objective function, if the bat flies to a location greater than {x μ t-1 ,x ω t-1} A better new position, the bat will conduct a local search around the position. In the local search, the random walk method is used. That is, when the respective objective function satisfies H1(x μ t ) μ t-1 ) or H2(x ω t ) <H2(x ω t-1 ), and when the random conditions and constraints are met, they are:

[0134]

[0135] Among them, the random variable ε follows a uniform distribution on [-1,1], ​Represents the average sound intensity of the tth generation in the black bat / gray bat population. At each moment (each generation), the pulse intensity and pulse emission frequency are updated according to the formula:

[0136]

[0137] is the pulse intensity emitted by the μ / ωth bat in the tth generation, α is the pulse intensity attenuation factor, satisfying represents the pulse emission frequency of the μ / ωth bat in the tth generation, γ represents the adjustment factor of the pulse emission frequency, satisfying 0≤γ≤1.

[0138] At the end of each moment, it is determined whether the end condition is reached, and the end condition is whether the maximum number of iterations is reached. If so, the search is stopped and the position and fitness value of the bat corresponding to the global optimal solution are output, otherwise the search continues.

[0139] For example: When operating, first enter: global TSN network G', each TSN network domain {G 1 ,G 2 ,…}, global traffic set F', traffic sets in each domain {F 1 ,F 2 ,…}, the population size of the two-color bat U,W, the initial value of the chaotic sequence ψ0, the maximum pulse intensity Maximum pulse frequency Pulse frequency upper limit f μ / ωmax , lower limit f μ / ωmin , sound intensity attenuation coefficient α, frequency adjustment coefficient γ, maximum number of iterations T max , inertia weight adjustment factor λ, mutation weight∈;

[0140] Final output: optimal intra-domain traffic scheduling and inter-domain routing selection results The specific steps are as follows:

[0141] Step 1: Initialization: Generate initial positions for μ=1,2,…,U,ω=1,2,…,W According to the initial position of the bat, calculate the global objective function value at t = 0 Get the optimal solution at the current moment

[0142] Step 2: In the independent update phase, adjust the frequency of the black bat μ and update the speed and position;

[0143] Step 3: Independent update phase, adjust the frequency of the gray bat ω, and update the speed and position;

[0144] Step 4: In the linkage update phase, according to the optimal Update the speed and position of all bats;

[0145] Step 5: In the local search phase, new solutions are generated through random walks

[0146] Step 6: Execute steps 2 and 4;

[0147] Step 7: Local search phase, generating new solutions through random walks

[0148] Step 8: Execute steps 3 and 4;

[0149] Step 9: renew

[0150] Step 10: If H1(x μ t ) μ t-1 )&&Satisfy the constraints and set the optimal solution to

[0151] Step 11: If H2(x ω t ) <H2(x ω t-1 )&&Satisfy the constraints and set the optimal solution to

[0152] Step 12: If the constraint is not met, execute step 2.

[0153] The global optimal solution can be used: and The intra-domain traffic scheduling and the inter-domain routing selection are reallocated and selected, thereby strengthening the globality of intra-domain network scheduling and improving the scheduling efficiency of global TSN traffic.

[0154] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed by the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.​

Claims

1. A semi-distributed scheduling method for TSN cross-domain traffic based on a two-color cooperative bat algorithm, characterized in that: The following steps are involved: S10. Deploy an intra-domain scheduler in each local TSN network within the global TSN network, and each intra-domain scheduler is responsible for the traffic scheduling in the corresponding local TSN network: each intra-domain scheduler collects the traffic transmission requirements in the corresponding local TSN network, and reports the traffic transmission requirements in the local TSN network; S20. A cross-domain coordinator is set in the global TSN network, and the cross-domain coordinator is in communication with each of the intra-domain schedulers. The cross-domain coordinator is responsible for path selection of each of the local TSN networks: the cross-domain coordinator summarizes the traffic transmission requirements in the local TSN network reported by each of the intra-domain schedulers, obtains a traffic set, and sends the traffic set to each of the intra-domain schedulers; S30. After receiving the traffic set, each intra-domain scheduler calculates the committed delay by the intra-domain traffic scheduling method, and reports the calculated committed delay to the cross-domain coordinator; S40. After receiving the calculated committed delay, the cross-domain coordinator generates an inter-domain scheduling strategy through an inter-domain routing selection method, and sends the inter-domain scheduling strategy to each of the intra-domain schedulers, and the intra-domain scheduler performs distributed scheduling between intra-domain nodes and links for the allocated traffic according to the received inter-domain scheduling strategy; S50. The traffic scheduling within the local TSN network and the path selection of the local TSN network are jointly optimized through the server within the intra-domain scheduler, and the two-color collaborative bat algorithm is used to solve the problem, so as to obtain the global optimal solution for the traffic scheduling within the local TSN network and the path selection of the local TSN network, so as to improve the scheduling efficiency of the global TSN traffic.

2. According to claim 1, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color collaborative bat algorithm is characterized in that: The intra-domain traffic scheduling method comprises the following steps: S31. Define the local TSN network topology as a directed graph G = (V, E), where V represents a node set, E represents a link set, and the node set V includes an internal node v in and cross-domain node v e , the traffic set is F, each flow f in the traffic set F i It is expressed as: f i =(T i ,D i ,L i ,st i ,des i ) Among them, T i is the period of the flow, D i is the data frame composition of the traffic, L i is the cut-off time of the flow, st i ,des i are the source node and destination node of the traffic respectively; S32. The constraints on the order in which each flow is sent include that for the same link, the sending time of the first frame must be non-negative, and each frame must be sent in a sequential order: in, Denotes that each flow f in the flow set F i The kth frame on link e, Indicates the closing time of GCL. Indicates the door opening time of GCL, i, e, k are all indexes; S33. The frame collision constraint includes that within the same transmission time slot, the link resources should be exclusively occupied by the current traffic, otherwise collisions will occur between different traffic flows: Among them, T i and T j f i and f j The corresponding flow period, m, n are non-negative integers; S34. The intra-domain path constraints include that the local TSN network traffic is transmitted in the topology in accordance with the order of links: Among them, e ab ,e bc ∈E,e ab ,e bc is any adjacent link in the link set E; S35. The cut-off time constraints for each flow include that the end-to-end delay of the flow does not exceed its cut-off time: Among them, L i is the cut-off time of the flow, st i is the source node of the traffic, des i is the destination node of the traffic, Indicates the opening time of GCL. Indicates the closing time of GCL; S36. The traffic scheduling objective function within the definition domain, i.e. minimizing the sum of the time required for each flow to reach the destination node, includes: Among them, f i For each flow in the flow set F', H1 represents the objective function 1.

3. According to claim 2, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color collaborative bat algorithm is characterized in that: The inter-domain routing selection method comprises the following steps: S41. Assume that the network topology covering multiple local TSN networks is G'=<D,E> , D represents the set of local TSN networks in the network, E represents the link set, the set of traffic reported by all local TSN networks is F', and the optimal transmission path set R between domains for each traffic is calculated. Each flow f in the traffic set F' i It is expressed as: f i =(P i ,U i ,L i ,st i ,des i ) Among them, P i is the lower bound of the delay commitment of the flow in each domain, U i is the upper bound of the delay commitment of each domain to the flow, L i is the deadline of the flow, st i ,des i are the source node and destination node of the flow respectively; S42. The inter-domain path constraint includes flow f in each flow in the flow set F' i The path R i Should be st i As the starting point, des i For the end point: Among them, R is the optimal transmission path set of each flow between domains, R i is the optimal transmission path between domains for each flow, st i is the source node of the stream, des i is the destination node of the flow, d a and d b for any two adjacent domains; S43. When the flow cut-off time is constrained, the TSN flow f i The end-to-end delay of is composed of the intra-domain transmission delay and the inter-domain link delay, and the delay does not exceed its specified deadline: Among them, f i For each flow in the flow set F', is the flow rate f i The lower bound of the delay commitment in the d domain, L i is the cut-off time of the traffic; S44. The objective function of the inter-domain traffic routing selection is: Among them, H2 represents the objective function 2, is the flow rate f i The lower bound of the delay commitment in the d domain is, is the flow rate f i The upper bound of the delay commitment in the d domain.

4. According to claim 3, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color collaborative bat algorithm is characterized in that: The two-color cooperative bat algorithm comprises the following steps: S51. The M bats are divided into U black bats and W gray bats, wherein the black bats are responsible for intra-domain traffic scheduling and the gray bats are responsible for inter-domain routing selection; S52. The initial bat position, discretizing the bat position in a given space to form a set of solutions, the bats are randomly distributed in the space, and the initial position of the bat is: Among them, the vector represents the initial position of the μ / ωth bat; S53. Using real number coding strategy, use vector information to represent each bat individual. Ordinalize by the size of the element Get the priority order of individual bat position information: S54. In each super period t, each bat performs a foraging process, and updates the speed and position of each bat. The speed and position of the bat are iteratively updated to find the global optimal solution. The speed and position update is divided into three stages: independent update, linkage update, and local search.

5. According to claim 4, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color cooperative bat algorithm is characterized in that: The independent update algorithm is: in, is the flying speed of the μ / ωth black / gray bat at time t, is the flight speed of the μ / ωth black / gray bat at time t-1, rand is the random factor generated by Chebyshev mapping, rand∈0,1), f μ / ω is the pulse frequency, is the position of the black / gray bat after independent update at time t, ∈ is the mutation weight, C(0,1) is the random number generated by the Cauchy probability distribution with a scale parameter of t=1, is the logarithmically decreasing inertia weight.

6. According to claim 4, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color cooperative bat algorithm is characterized in that: The logarithmically decreasing inertia weight It is obtained by the following formula: in, λ is the inertia weight adjustment factor, which is a constant in the range of (0,1); T max is the maximum number of iterations of bat flight, and a is a constant.

7. According to claim 4, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color cooperative bat algorithm is characterized in that: The linkage update includes defining {x μi t ,x ωj t } represents a set of bat positions that achieve the global optimal objective function value at super period t; Where i,j = argmin {i,j} F(x μi t ,x ωj t ) Within the constraints, the speed and position of the black bat at time t are updated as: The speed and position of the gray bat at time t are updated as: in, is the logarithmically decreasing inertia weight, v represents velocity, and x represents position.

8. According to claim 4, the TSN cross-domain traffic semi-distributed scheduling method based on the two-color cooperative bat algorithm is characterized in that: During the local search: if the bat flies to a location greater than {x μ t-1 ,x ω t-1 } A better new position, the bat will conduct a local search around the position. During the local search, the random walk method is used, that is, when the respective objective function satisfies H1(x μ t ) μ t-1 ) or H2(x ω t ) <H2(x ω t-1 ), and when the random conditions and constraints are met, they are:​ Among them, H1 represents objective function 1, H2 represents objective function 2, and the random variable ε obeys the uniform distribution on [-1,1]. Represents the average sound intensity of the tth generation in the black bat / gray bat population. At each moment, the pulse intensity and pulse emission frequency are updated according to the formula: is the pulse intensity emitted by the μ / ωth bat in the tth generation, α is the pulse intensity attenuation factor, represents the pulse emission frequency of the μ / ωth bat in the tth generation, and γ represents the adjustment factor of the pulse emission frequency; at the end of each moment, it is determined whether the maximum number of iterations has been reached. If so, the search is stopped and the position and fitness value of the bat corresponding to the global optimal solution are output, otherwise the search continues.

Images