Asynchronous flow optimization scheduling method for 5G + TSN fusion network with high schedulability

Abstract

The invention discloses a 5G + TSN fusion network asynchronous flow optimization scheduling method with high schedulability, which comprises the following steps: S1, in a 5G and TSN fusion network, firstly, a controller of a center collects overall information of the network through a network control protocol, the information comprises channel conditions and frequency resources in the 5G network, and the channel conditions and the frequency resources in the 5G network are transmitted to a server; the communication requirements of the network topology, the link rate and the flow of the TSN network are met; and S2, running an asynchronous traffic optimization scheduling model of the 5G + TSN hybrid network through the central controller, performing routing decision and transmission scheduling on the traffic collected in the previous step in the hybrid network to obtain a scheduling result, and minimizing the consumption of network resources while meeting the transmission requirement of hard real-time traffic. According to the invention, the schedulability of the TSN gateway is obviously improved, the application range of an asynchronous deterministic access mechanism is expanded, and end-to-end traffic scheduling is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of deterministic real-time transmission of computer networks, and specifically to a 5G+TSN fused network asynchronous traffic optimization scheduling method with high schedulability. Background Art

[0002] The converged networking of the fifth generation mobile communication technology, referred to as 5G, and the time-sensitive network, referred to as TSN, is an important technology for realizing flexible manufacturing and mobile applications in Industry 4.0. TSN is based on standard Ethernet and introduces a time-triggered transmission mechanism in IEEE802.1Qbv. In industrial control, hard real-time traffic is generally modeled as periodic traffic. This mechanism periodically reserves transmission resources on the transmission link for such periodic data flows, that is, plans the sending time of each data packet on the routing device. Therefore, clock synchronization of the entire network is required. One of the three major technical visions of 5G is to achieve ultra-reliable and low-latency transmission. For this purpose, semi-persistent scheduling is introduced, which is consistent with the time-triggered transmission of TSN. Similarly, periodic transmission resources are reserved for data packets, so it is very suitable for networking with TSN to achieve more flexible data transmission. In the existing traffic scheduling of 5G and TSN integrated networks, the common practice is to estimate the worst latency of traffic in the 5G network and reserve transmission resources in the TSN network according to the 5G worst latency. Then the traffic accesses the TSN network according to the worst latency. This solution has the following two problems: (1) The worst latency is difficult to estimate (2) It may be difficult to meet the end-to-end latency requirements of the traffic according to the worst latency. Treating the traffic accessing TSN from the 5G network as traffic sent by asynchronous terminals and performing asynchronous traffic scheduling is a very promising solution.

[0003] In the prior art, such as the "A transmission deterministic access method for asynchronous terminals" with Chinese publication number: CN202110583698.4, it includes two mechanisms, one is the asynchronous to synchronous mechanism, and the other is the delay adaptation mechanism. Specifically, after the first switch receives the message, it stays at the nearest sending time point before sending it out. The message is transmitted to the edge switch in the synchronization domain according to the time-triggered transmission mechanism of TSN. The edge switch keeps the message for a period of time before sending it to the receiving terminal. The delayed sending time is equal to the difference between the sending period of the data stream and the residence time of the message, so as to eliminate the uncertainty of the end-to-end delay.

[0004] In the prior art, directly applying this deterministic access method of asynchronous terminals to the network scheduling of 5G and TSN will lead to a thorny problem: the scheduling of traffic at the TSN gateway is prone to failure. This is because: (1) the TSN gateway acts as a bridge between the two networks. All traffic entering the TSN network from the 5G network must be aggregated and distributed at the TSN gateway, and the gateway is highly loaded; (2) the asynchronous deterministic mechanism often provides more transmission resources for traffic; these two factors together can easily cause traffic scheduling failure at the gateway. In traditional TSN traffic scheduling, typical methods for improving traffic schedulability can be divided into two types: relaxation in the time domain and relaxation in the space domain; however, introducing the relaxation in the time domain and space domain into the asynchronous deterministic access mechanism will face the following three unresolved problems:

[0005] (1) How to balance the end-to-end delay of traffic and the consumption of network resources: According to the principle of asynchronous deterministic access mechanism, the physical meaning of alignment time is the maximum waiting time for data packets to be transmitted at the gateway. Assuming that the average resource period allocated to a flow before and after the introduction of time domain relaxation is the same, since time domain relaxation makes the transmission time slot non-periodic, its alignment time must be greater than the case without the introduction of time domain relaxation. In order to reduce the alignment time of a flow, we can provide more transmission resources, but the resources left for other flows will also be reduced accordingly. Faced with complex heterogeneous traffic in the network and its delay requirements, how to comprehensively consider the allocation of an appropriate amount of resources to each flow?

[0006] (2) How to model the alignment time when the time domain and space domain are relaxed: For the original asynchronous deterministic access mechanism, the alignment time is the resource cycle; but after the introduction of relaxation in the time domain and space domain, the alignment time becomes more difficult to calculate; How to model the alignment time of the asynchronous deterministic access mechanism when the time domain and space domain are relaxed?

[0007] (3) How to coordinate 5G and TSN network resource allocation: In a heterogeneous network of 5G and TSN, the end-to-end latency includes both TSN latency and 5G latency. If the resource allocation situation in the 5G network changes, it will also affect the resource scheduling of TSN. For example, if data packets occupy more frequency resources to consume less time domain resources, the 5G latency will be reduced, and the scheduling pressure on the TSN network will be reduced, but more frequency resources will be consumed. In the face of complex traffic in the network, how to coordinate the allocation of 5G and TSN network resources? Summary of the invention

[0008] The purpose of the present invention is to provide a 5G+TSN converged network asynchronous traffic optimization scheduling method with high schedulability to solve the problems raised in the above background technology, namely, how to balance the end-to-end delay of traffic and the consumption of network resources, how to model alignment time under relaxation in the time domain and space domain, and how to coordinate the allocation of 5G and TSN network resources.

[0009] To achieve the above object, the present invention provides the following technical solution: a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability, comprising the following steps:

[0010] S1. In the converged network of 5G and TSN, the central controller first collects the overall information of the network through the network control protocol. This information includes the channel conditions and frequency resources in the 5G network, the network topology and link rate of the TSN network, and the communication requirements of the traffic.

[0011] S2. Run the asynchronous traffic optimization scheduling model of the 5G+TSN hybrid network through the central controller, make routing decisions and transmission scheduling for the traffic collected in the previous step in the hybrid network, obtain scheduling results, and minimize network resource consumption while meeting the transmission requirements of hard real-time traffic;

[0012] S3. Calculate the configuration information of each network device based on the scheduling result obtained in the previous step. The configuration information includes the queue gating list of the TSN switch and the resource time configuration of the gateway and edge switch in the asynchronous access deterministic mechanism.

[0013] S4. In the operation phase, when a data packet arrives at the TSN gateway from the 5G network, the gateway identifies which traffic the data packet belongs to and delivers it to the nearest time slot prepared for the traffic. The gateway adds the waiting time of the data packet to the back of the data packet payload, and then encapsulates the header and trailer in the standard Ethernet way.

[0014] S5. During the operation phase, when the data packet reaches the edge switch through TSN transmission, the residence time is calculated. According to the original asynchronous deterministic access mechanism, the residence time should be the difference between the alignment time and the waiting time at the gateway, and an additional residence time needs to be introduced. Its value is the difference between the maximum TSN delay of all candidates of the flow and the TSN delay of the actual candidate that transmits the data packet.

[0015] Preferably, in step S2, the asynchronous traffic optimization scheduling model of the 5G+TSN hybrid network includes an alignment time model under relaxation conditions in the time domain and space domain, a 5G network resource scheduling model and a TSN network resource scheduling model.

[0016] Preferably, the alignment time model under the relaxed conditions of the time domain and the space domain includes a resource allocation scheme of the asynchronous deterministic access mechanism under the relaxed conditions of the time domain and the space domain and the alignment time model under the scheme;

[0017] The basic resource cycle provided by TSN in the resource allocation scheme is T0. The slot resource set that can provide continuous transmission from the gateway to the destination node for a data packet on a route is called a candidate. The candidate's cycle is T0. The scheduling model needs to decide the number of candidates allocated to each flow to realize the resource allocation of the asynchronous deterministic access mechanism and determine the route taken by each candidate and its scheduling on the route.

[0018] Preferably, the alignment time model is a calculation of modeling alignment time, and the physical meaning of the alignment time is the maximum waiting time for a data packet to wait for transmission at the gateway. Therefore, a logical link can be abstracted for a candidate of a flow scattered on different egress links at the gateway. On this logical link, the flow f i The offset of the jth candidate at the gateway is represented as Then obtain the time calculation model of the alignment on the logical link;

[0019] Defined as a flow f i The maximum number of candidates that can be provided is n i,max , whose number starts from j=1 to j=n i,max , each candidate has a binary variable h ij ∈{0,1}, represents whether the candidate is selected. In order to ensure the physical meaning of the candidate, it should complete the transmission within the period T0, which can be easily obtained:

[0020]

[0021]

[0022] In order to distinguish each candidate, the relative positions of candidates are constrained:

[0023]

[0024]

[0025] Then, for the flow f i For example, the alignment time T i It can be calculated by the following theorem:

[0026] Theorem 1:

[0027] T i =max(R i ∪S i );

[0028] Two of the sets and The definitions are:

[0029]

[0030]

[0031] Preferably, the transmission resources in the 5G network resource scheduling model are located in two dimensions: time domain and frequency domain. In the 5G resource grid, the basic scheduling unit occupies 1 resource block RB in frequency and occupies a transmission time interval TTI in time domain. One scheduling unit can carry a certain number of bytes for transmission;

[0032] The RB set that can be used by the 5G network is expressed as X = {1, 2, .., k max}, where k max Represents the number of RBs;

[0033] In the 5G network, that is, l∈L 5GS , the resources allocated to a flow are composed of the following decision variables:

[0034]

[0035]

[0036]

[0037] The physical meaning of each decision variable is: The resource period is equal to the flow period, and the resource start time is Duration of the resource Represents the flow f i Whether to occupy the kth RB of 5G link l, Indicates that RB numbered k is allocated to flow f i ;

[0038] The time domain resources are allocated continuously, and the frequency domain resources can be allocated discontinuously. Since the traffic cycles may be different, the lowest common multiple of their cycles is defined as the macro cycle, denoted as T 5GS ,Right now:

[0039] T 5GS =LCM({f i .period|f i∈F}).

[0040] Preferably, the decision variables need to satisfy the following constraints:

[0041] Transmission opportunity constraint: Data packets can only be transmitted at an integer time of TTI, the transmission duration must also be an integer of TTI, and the data packet must be transmitted within one cycle;

[0042] Transmission resource constraints: The resources allocated for a data packet ensure that they are sufficient to transmit the data;

[0043] OFDMA constraints: The first two constraints ensure the validity of resource allocation for a single stream. OFDMA constraints ensure that resources between multiple streams do not conflict, that is, an RB-TTI block can be occupied by at most one stream.

[0044] RB constraint: introducing variables To indicate whether the kth RB of 5G link l is allocated for traffic transmission.

[0045] Preferably, the TSN network resource scheduling model is in the TSN network, i.e., l∈L TSN , the resources allocated to a flow are composed of the following decision variables:

[0046]

[0047]

[0048]

[0049] The meaning of each decision variable is: ij =1(0) represents the flow f i Whether the jth candidate is used, Represents the flow f i Whether the route of the j-th candidate contains link l, This determines the flow f i The transmission status of the jth candidate on link l, where is the period, whose size is T0, Offset is a decision variable, is the transmission duration.

[0050] Preferably, the decision variables need to satisfy the following transmission constraints: routing constraint, frame constraint, resource constraint, alignment time constraint, transmission order constraint, TDMA constraint, frame isolation constraint and end-to-end delay constraint.

[0051] Preferably, in step S2, the meeting of the transmission requirements of hard real-time traffic while minimizing the consumption of network resources includes a 5G network part and a TSN network part;

[0052] The 5G network part minimizes the number of RBs allocated for hard real-time critical traffic. Since the 5G system has only one uplink, we omit the subscript l for the following formula, that is:

[0053]

[0054] The TSN network part minimizes the number of candidates allocated to the entire network, that is:

[0055]

[0056] In summary, by changing the optimization direction and adding the weight factor γ, the final optimization goal is:

[0057] min[γ*obj1+(1-γ)*obj2].

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] (1) Significantly improve the schedulability of TSN gateways: By introducing relaxation in the time and space domains, it is easier to meet the requirement that the transmission time slots allocated by TSN to different flows do not overlap, thereby improving the schedulability at the gateway.

[0060] (2) Expanding the scope of application of the asynchronous deterministic access mechanism: The original asynchronous deterministic access mechanism assumes that periodic resources are provided for traffic. The present invention breaks this assumption, making the scope of application of the asynchronous deterministic access mechanism wider.

[0061] (3) End-to-end traffic scheduling: The present invention models end-to-end transmission, comprehensively considers the network and resource characteristics of 5G and TSN, the latency experienced by traffic in 5G and TSN, and leaves as many resources as possible for low-priority traffic while meeting traffic transmission requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 Schematic diagram of the time-triggered transmission process in an ideal 5G and TSN converged network;

[0063] Figure 2 A schematic diagram of a deterministic access solution for asynchronous terminals;

[0064] Figure 3 Schematic diagram of scheduling failure for two flows on a link;

[0065] Figure 4 It is a schematic diagram of relaxation in time domain;

[0066] Figure 5 It is a schematic diagram of relaxation in the space domain;

[0067] Figure 6 Schematic diagram of asynchronous deterministic access mechanism for applying time domain and space domain relaxation;

[0068] Figure 7 This is a schematic diagram of the alignment time calculation on the logical link;

[0069] Figure 8 This is a schematic diagram of 5G network transmission resources;

[0070] Fig. 9 The present invention is a schematic diagram of an implementation method of a 5G+TSN asynchronous traffic optimization scheduling method with high schedulability. DETAILED DESCRIPTION

[0071] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] See also Figure 1 and Figure 2 The present invention provides a technical solution: a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability, comprising the following steps:

[0073] Step 1: In the converged network of 5G and TSN, the central controller first collects the overall information of the network through the network control protocol. This information includes the channel conditions and frequency resources in the 5G network, the network topology and link rate of the TSN network, and the communication requirements of the traffic;

[0074] Step 2: Run the asynchronous traffic optimization scheduling model of the 5G+TSN hybrid network through the central controller, make routing decisions and transmission scheduling for the traffic collected in the previous step in the hybrid network, obtain scheduling results, and minimize network resource consumption while meeting the transmission requirements of hard real-time traffic;

[0075] Among them, the asynchronous traffic optimization scheduling model of the 5G+TSN hybrid network includes the alignment time model under relaxation conditions in the time domain and space domain, the 5G network resource scheduling model and the TSN network resource scheduling model;

[0076] Specifically, the alignment time model under the relaxation conditions of the time domain and space domain includes the resource allocation scheme of the asynchronous deterministic access mechanism under the relaxation conditions of the time domain and space domain and the alignment time model under the scheme; the basic resource cycle provided by TSN in the resource allocation scheme is T0, and the slot resource set that can provide continuous transmission from the gateway to the destination node for a data packet on a route is called a candidate, and the candidate's cycle is T0. The scheduling model needs to decide the number of candidates allocated to each flow to realize the resource allocation of the asynchronous deterministic access mechanism, and determine the route taken by each candidate and its scheduling on the route;

[0077] Among them, the alignment time model is the calculation of modeling alignment time. The physical meaning of alignment time is the maximum waiting time for a data packet to wait for transmission at the gateway. Therefore, a logical link can be abstracted for the candidates of a flow scattered on different egress links at the gateway. On this logical link, the flow f i The offset of the jth candidate at the gateway is represented as Then obtain the time calculation model of the alignment on the logical link;

[0078] Defined as a flow f i The maximum number of candidates that can be provided is n i,max , whose number starts from j=1 to j=n i,max , each candidate has a binary variable h ij ∈{0,1}, represents whether the candidate is selected. In order to ensure the physical meaning of the candidate, it should complete the transmission within the period T0, which can be easily obtained:

[0079]

[0080]

[0081] In order to distinguish each candidate, the relative positions of candidates are constrained:

[0082]

[0083]

[0084] Then, for the flow f i For example, the alignment time T i It can be calculated by the following theorem:

[0085] Theorem 1:

[0086] Ti =max(R i ∪S i );

[0087] Two of the sets and The definitions are:

[0088]

[0089]

[0090] prove:

[0091] After abstracting the logical link, we Figure 7 As a heuristic example to prove the theorem. According to T i The meaning is the maximum interval between adjacent candidates, so,

[0092] T i =max(Δ1,Δ2,Δ3);

[0093] Where Δ1 and Δ2 represent the intervals between different and adjacent candidates within a cycle, which are included in the set R i ; and Δ3 represents the interval between the last candidate and the first candidate in the next cycle, included in the set S i In this way, T i is Figure 7 The maximum interval between adjacent candidates in is the alignment time. In general, in order to find the maximum interval between adjacent candidates, for the set R i , the right half of each element of the product is non-negative if and only if the pth and qth candidates exist and there is no candidate between them (that is, h ip =h iq =1, ), the product term of the left half of the element is 1. In this case, the element is non-negative and represents the interval between two existing, different and adjacent candidates. If there is another candidate between p and q, then the element is negative. Similarly, for the set S i , the right half of the product of the elements is easily obtained to be a positive number if and only if the pth and qth candidates exist, and there is no candidate before p and after q (that is, h ip =h iq =1, ), the left half of the element is 1, the element is positive, and represents the interval between the last candidate in the cycle and the first candidate in the next cycle. If there are other candidates before p and after q, the value of the element is negative; p can be equal to q to deal with the situation where only one candidate is assigned to a stream; since at least one candidate must be provided for a stream, the set There must be only one positive element. Finally, since T i =max(R i ∪S i ), so T i It must be a positive number, which means that T i is the maximum interval between adjacent candidates. The proof is complete.

[0094] Specifically, the transmission resources in the 5G network resource scheduling model are located in two dimensions: time domain and frequency domain. In the 5G resource grid, the basic scheduling unit occupies one resource block RB in frequency and one transmission time interval TTI in time domain. One scheduling unit can carry a certain number of bytes for transmission;

[0095] The RB set that can be used by the 5G network is expressed as X = {1, 2, .., k max}, where k max Represents the number of RBs;

[0096] In the 5G network, that is, l∈L 5GS , the resources allocated to a flow are composed of the following decision variables:

[0097]

[0098]

[0099]

[0100] The physical meaning of each decision variable is: The resource period is equal to the flow period, and the resource start time is Duration of the resource Represents the flow f i Whether to occupy the kth RB of 5G link l, Indicates that RB numbered k is allocated to flow f i ;

[0101] Time domain resources are allocated continuously, and frequency domain resources can be allocated discontinuously. Since the traffic cycles may be different, the lowest common multiple of their cycles is defined as the macrocycle, denoted by T 5GS ,Right now:

[0102] T 5GS =LCM({f i .period|f i ∈F}).

[0103] Among them, the decision variables need to satisfy the following constraints:

[0104] (1) Transmission opportunity constraint: Data packets can only be transmitted at an integer time of TTI, the transmission duration must also be an integer of TTI, and the data packet must be transmitted within one cycle; therefore:

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111]

[0112] Among them, the variable integer The meaning is flow f i The TTI number at which transmission starts on 5G link l, Represents the number of TTIs sent continuously;

[0113] (2) Transmission resource constraint: The resources allocated to a data packet ensure that they are sufficient to transmit the data. Represents f i The number of bytes that can be transmitted in one TTI on the kth RB on 5G link l is thus:

[0114]

[0115]

[0116]

[0117]

[0118] The above two constraints ensure that the allocated resources are just enough to transmit data;

[0119] (3) OFDMA constraints: The first two constraints ensure the validity of resource allocation for a single stream. The OFDMA constraint ensures that resources between multiple streams do not conflict, that is, an RB-TTI block can be occupied by at most one stream. Therefore:

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128] in, It means that when the i-th stream and the j-th stream are both transmitted through the k-th RB of the 5G link l, the time domain resources of the two streams are constrained not to overlap;

[0129] (4) RB constraint: introducing variables To indicate whether the kth RB of 5G link l is allocated to traffic transmission. Specifically, if any traffic is allocated to the kth RB, the indicator variable is set to 1, so:

[0130]

[0131]

[0132] Specifically, the TSN network resource scheduling model is in the TSN network, that is, l∈L TSN , the resources allocated to a flow are composed of the following decision variables:

[0133]

[0134]

[0135]

[0136] The meaning of each decision variable is: ij =1(0) represents the flow f i Whether the jth candidate is used, Represents the flow f iWhether the route of the j-th candidate contains link l, This determines the flow f i The transmission status of the jth candidate on link l, where is the period (whose size is T0, which is a priori knowledge), Offset is a decision variable, is the transmission duration;

[0137] Among them, the decision variables need to satisfy the transmission constraints: routing constraints, frame constraints, resource constraints, alignment time constraints, transmission order constraints, TDMA constraints, frame isolation constraints and end-to-end delay constraints;

[0138] (1) Route constraints: Routes need to be scheduled for each candidate that is selected. When a candidate is not selected, no route is assigned. A route is assigned only when a candidate is selected. Therefore, h ij The value of is modeled; in the flow f i When using the jth candidate (i.e. h ij =1), we need to allocate routes, and we only consider simple loop-free routes. The following three constraints respectively characterize the characteristics and loop-free constraints of the sending node (TSN gateway in the scenario considered by this invention) and the intermediate forwarding node in TSN: The gateway node is used as the flow f i For the sending node, only one output link is activated (link l is activated, i.e. ), flow f i The intermediate links of the route only bear the receiving and forwarding functions, so the number of activated input links is equal to the number of activated output links. In order to limit the loop, we limit a node to have at most one link activated.

[0139]

[0140]

[0141]

[0142]

[0143]

[0144]

[0145] In addition, in the flow i When the jth candidate is not used (i.e. h ij=0), we hope that for this candidate, all the input and output links of the node will not be activated by this candidate, that is,

[0146]

[0147] (2) Frame Constraint: Each candidate needs to have a non-negative offset on each link, and for modeling convenience, the transmission must be completed in period T0. In addition, if a candidate does not pass through link l, its offset should be 0. Therefore:

[0148]

[0149] l[1]! = f i .dst→

[0150]

[0151]

[0152]

[0153]

[0154]

[0155] We do not schedule the last hop (i.e. l[1]≠f i .dst, the same below);

[0156] It is worth noting that we have already obtained in the routing constraints And here we have Therefore, we can draw a conclusion: if a candidate of a flow is not used, all links will not be activated by the candidate and the offset on the link will be 0. If it is used, the offset may not be 0 only on the activated link;

[0157] (3) Resource constraint: In TSN, sufficient resources need to be allocated for a flow, that is, the number of candidates provided should not be less than the number of data packets appearing in T0. Therefore:

[0158]

[0159]

[0160] (4) Alignment time constraint: The purpose of this constraint is to abstract a logical link at the gateway from candidates of a flow scattered throughout the network, that is, to obtain the variable When a candidate is not used, there is no need to obtain the value. When it is used, according to the previous constraints, only one gateway output link is activated and the other links are not activated (offset is 0), then you need to use To capture its offset at the gateway, since we do not know on which egress link the candidate is transmitted, we need to accumulate the offsets on all egress links. Therefore:

[0161]

[0162] (h ij ==1)→

[0163]

[0164] (5) Transmission order constraint: A switch in the network can only transmit data packets on the next link after it has completely received the data packet on the previous link. Let ldelay l represents the propagation delay of link l. Since we do not know which output or input link of a node is activated, we need to accumulate all input or output links (the same below), so:

[0165]

[0166] v≠sw i →

[0167]

[0168]

[0169] When the flow i When the jth candidate is not selected (i.e. h ij =0), both sides of the inequality are 0, so it is naturally established. ij =1), if the route does not pass through node v, it is also true; when passing through node v, this constraint represents the transmission order constraint of the transmission time of the previous and next links;

[0170] (6) TDMA constraint: On a TSN link, reserved transmission time slots cannot overlap with each other and must follow TDMA constraints. Note that constraints are only required when two candidates pass through the same link (i.e. ),therefore:

[0171]

[0172]

[0173]

[0174]

[0175] (((r≠s)∨(u≠v))∧

[0176] (l[1]≠f r .dst)∧(l[1]!=f s .dst)∧

[0177]

[0178]

[0179]

[0180] (7) Frame isolation constraint: Due to the use of the asynchronous deterministic access mechanism, the transmission resources provided for a flow are excessive. In order to ensure the determinism of queue queuing in the network, each transmission instance needs to be isolated in the queue. Therefore:

[0181]

[0182]

[0183]

[0184]

[0185] (((r≠s)∨(u≠v))∧

[0186] (l[0]≠GW)∧(l[1]!=dst r )∧(l[1]!=dst s )∧

[0187]

[0188]

[0189]

[0190] Where l[0]≠GW means that link l must have a preceding link;

[0191] (8) End-to-end delay constraint: Let the processing delay of the base station be T proc , then the delay experienced by the 5G system is:

[0192]

[0193]

[0194] According to the deterministic asynchronous access mechanism, the latency of each candidate in TSN can be calculated according to the following formula:

[0195]

[0196]

[0197]

[0198]

[0199] in Indicates the interval between the time when a data packet is sent to the edge switch and the time when it is sent to the gateway. is the transmission and propagation delay of the last hop.

[0200] Since the transmissions of different candidates assigned to a flow may have different delays in TSN, in order to keep the transmissions of all candidates used by the flow with consistent TSN delay, we distinguish the data packets from different candidates on the edge switch and retain them for some time on top of the residence time of the asynchronous deterministic access mechanism, so that the theoretical transmission time of the candidate in TSN is consistent with the TSN delay of the candidate with the largest TSN delay. Therefore, the TSN delay experienced by a flow should be the maximum value of the TSN delays of all candidates, so the end-to-end delay should meet the delay requirements of the flow, that is:

[0201]

[0202]

[0203]

[0204] Specifically, meeting the transmission requirements of hard real-time traffic while minimizing the consumption of network resources includes the 5G network part and the TSN network part;

[0205] For the 5G network, we minimize the number of RBs allocated for hard real-time critical traffic. Since the 5G system has only one uplink, we omit the subscript l for the following formula, namely:

[0206]

[0207] In the TSN network part, the number of candidates allocated to the entire network is minimized, that is:

[0208]

[0209] In summary, by changing the optimization direction and adding the weight factor γ, the final optimization goal is:

[0210] min[γ*obj1+(1-γ)*obj2].

[0211] Step 3: Calculate the configuration information of each network device based on the scheduling results obtained in the previous step. The configuration information includes the queue gating list of the TSN switch and the resource time configuration of the gateway and edge switch in the asynchronous access deterministic mechanism.

[0212] Step 4: In the operation phase, when a data packet arrives at the TSN gateway from the 5G network, the gateway identifies which traffic the data packet belongs to and delivers it to the nearest time slot prepared for the traffic. The gateway adds the waiting time of the data packet to the end of the data packet payload, and then encapsulates the header and trailer in the standard Ethernet way.

[0213] Step 5: In the operation phase, when the data packet reaches the edge switch through TSN transmission, the residence time is calculated. According to the original asynchronous deterministic access mechanism, the residence time should be the difference between the alignment time and the waiting time at the gateway. In addition, an additional residence time needs to be introduced, and its value is the difference between the maximum TSN delay of all candidates of the flow and the TSN delay of the actual candidate that transmits the data packet.

[0214] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A highly schedulable 5G+TSN fusion network asynchronous traffic optimization scheduling method, characterized in that: The following steps are involved: S1. In the converged network of 5G and TSN, the central controller first collects the overall information of the network through the network control protocol. This information includes the channel conditions and frequency resources in the 5G network, the network topology and link rate of the TSN network, and the communication requirements of the traffic. S2. Run the asynchronous traffic optimization scheduling model of the 5G+TSN hybrid network through the central controller, make routing decisions and transmission scheduling for the traffic collected in the previous step in the hybrid network, obtain scheduling results, and minimize network resource consumption while meeting the transmission requirements of hard real-time traffic; S3. Calculate the configuration information of each network device based on the scheduling result obtained in the previous step. The configuration information includes the queue gating list of the TSN switch and the resource time configuration of the gateway and edge switch in the asynchronous access deterministic mechanism. S4. In the operation phase, when a data packet arrives at the TSN gateway from the 5G network, the gateway identifies which traffic the data packet belongs to and delivers it to the nearest time slot prepared for the traffic. The gateway adds the waiting time of the data packet to the back of the data packet payload, and then encapsulates the header and trailer in the standard Ethernet way. S5. During the operation phase, when the data packet reaches the edge switch through TSN transmission, the residence time is calculated. According to the original asynchronous deterministic access mechanism, the residence time should be the difference between the alignment time and the waiting time at the gateway, and an additional residence time needs to be introduced. Its value is the difference between the maximum TSN delay of all candidates of the flow and the TSN delay of the actual candidate that transmits the data packet.

2. According to claim 1, a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability is characterized in that: In step S2, the asynchronous traffic optimization scheduling model of the 5G+TSN hybrid network includes an alignment time model under relaxation conditions in the time domain and space domain, a 5G network resource scheduling model, and a TSN network resource scheduling model.

3. According to claim 2, a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability is characterized in that: The alignment time model under the relaxation conditions of the time domain and the space domain includes a resource allocation scheme of the asynchronous deterministic access mechanism under the relaxation conditions of the time domain and the space domain and an alignment time model under the scheme; The basic resource cycle provided by TSN in the resource allocation scheme is T0. The slot resource set that can provide continuous transmission from the gateway to the destination node for a data packet on a route is called a candidate. The candidate's cycle is T0. The scheduling model needs to decide the number of candidates allocated to each flow to realize the resource allocation of the asynchronous deterministic access mechanism and determine the route taken by each candidate and its scheduling on the route.

4. According to claim 3, a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability is characterized in that: The alignment time model is used to calculate the alignment time. The physical meaning of the alignment time is the maximum waiting time for a data packet to be transmitted at the gateway. Therefore, a logical link can be abstracted for the candidates of a flow scattered on different egress links at the gateway. On this logical link, the flow f i The offset of the jth candidate at the gateway is represented as Then obtain the time calculation model of the alignment on the logical link; Defined as a flow f i The maximum number of candidates that can be provided is n i,max , whose number starts from j=1 to j=n i,max , each candidate has a binary variable h ij ∈{0,1}, represents whether the candidate is selected. In order to ensure the physical meaning of the candidate, it should complete the transmission within the period T0, which can be easily obtained: In order to distinguish each candidate, the relative positions of candidates are constrained: Then, for the flow f i For example, the alignment time T i It can be calculated by the following theorem: Theorem 1: T i =max(R i ∪S i ); The two sets R i and S i The definitions are:

5. According to claim 2, a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability is characterized in that: In the 5G network resource scheduling model, transmission resources are located in two dimensions: time domain and frequency domain. In the 5G resource grid, the basic scheduling unit occupies one resource block RB in frequency and one transmission time interval TTI in time domain. One scheduling unit can carry a certain number of bytes for transmission. The RB set that can be used by the 5G network is expressed as X = {1, 2, .., k max }, where k max Represents the number of RBs; In the 5G network, that is, l∈L 5GS , the resources allocated to a flow are composed of the following decision variables: The physical meaning of each decision variable is: The resource period is equal to the flow period, and the resource start time is Duration of the resource Represents the flow f i Whether to occupy the kth RB of 5G link l, Indicates that RB numbered k is allocated to flow f i ; The time domain resources are allocated continuously, and the frequency domain resources can be allocated discontinuously. Since the traffic cycles may be different, the lowest common multiple of their cycles is defined as the macro cycle, denoted as T 5GS ,Right now: T 5GS =LCM({f i .period|f i ∈F})。 6. According to claim 5, a method for optimizing and scheduling heterogeneous traffic in a 5G+TSN fusion network with high schedulability is characterized in that: The decision variables need to satisfy the following constraints: Transmission opportunity constraint: Data packets can only be transmitted at an integer time of TTI, the transmission duration must also be an integer of TTI, and the data packet must be transmitted within one cycle; Transmission resource constraints: The resources allocated for a data packet ensure that they are sufficient to transmit the data; OFDMA constraints: The first two constraints ensure the validity of resource allocation for a single stream. OFDMA constraints ensure that resources between multiple streams do not conflict, that is, an RB-TTI block can be occupied by at most one stream. RB constraint: introducing variables To indicate whether the kth RB of 5G link l is allocated for traffic transmission.

7. According to claim 2, a 5G+TSN fusion network asynchronous traffic optimization scheduling method with high schedulability is characterized in that: The TSN network resource scheduling model is in the TSN network, i.e., l∈L TSN , the resources allocated to a flow are composed of the following decision variables: The meaning of each decision variable is: ij =1(0) represents the flow f i Whether the jth candidate is used, Represents the flow f i Whether the route of the j-th candidate contains link l, This determines the flow f i The transmission status of the jth candidate on link l, where is the period, whose size is T0, Offset is a decision variable, is the transmission duration.

8. The method for optimizing and scheduling asynchronous traffic in a 5G+TSN converged network with high schedulability according to claim 7 is characterized in that: The decision variables need to satisfy the following transmission constraints: routing constraint, frame constraint, resource constraint, alignment time constraint, transmission order constraint, TDMA constraint, frame isolation constraint and end-to-end delay constraint.

9. The method for optimizing and scheduling asynchronous traffic in a 5G+TSN converged network with high schedulability according to claim 1 is characterized in that: In step S2, the method of satisfying the transmission requirements of hard real-time traffic while minimizing the consumption of network resources includes a 5G network part and a TSN network part; The 5G network part minimizes the number of RBs allocated for hard real-time critical traffic. Since the 5G system has only one uplink, we omit the subscript l for the following formula, that is: The TSN network part minimizes the number of candidates allocated to the entire network, that is: In summary, by changing the optimization direction and adding the weight factor γ, the final optimization goal is: min[γ*obj1+(1-γ)*obj2].

Images