CN-122002590-A - Dynamic data packet scheduling method and system

CN122002590ACN 122002590 ACN122002590 ACN 122002590ACN-122002590-A

Abstract

The invention provides a dynamic data packet scheduling method and a system, and belongs to the field of data processing. According to the invention, a wireless state sensing, real-time flow sensing and TSN time constraint analysis mechanism is introduced at the DS-TT side, and closed-loop control is formed by dynamic scheduling decision and scheduling execution feedback, so that time sensing cooperative scheduling of a cross-5G network and a time sensitive network TSN is realized. The method can reduce the influence of wireless fluctuation on the time delay of the high-priority data packet, reduce scheduling conflict, improve the utilization efficiency of resources, ensure that the TSN service of the key time-sensitive network finishes forwarding within a specified time window, and is suitable for deterministic communication scenes such as intelligent manufacturing, rail transit, electric power automation and the like.

Inventors

WANG XIAOBO
LIN SIYU
YANG SHAN
LI JIAJUN
WANG ZHIHUI
YE HAINA

Assignees

北京交通大学

Dates

Publication Date: 20260508
Application Date: 20260212

Claims (9)

1. The dynamic data packet scheduling method is characterized by comprising the following steps: S1, after data packets from a 5G network side and a time sensitive network TSN side are accessed in a unified way, traffic state snapshots are formed through traffic flow classification, queue length statistics and queuing delay statistics; S2, acquiring dynamic information of a wireless link towards a 5G network side, converting the dynamic information into wireless risk factors directly acting on scheduling, binding the wireless risk factors with a traffic state snapshot to form a wireless part of a combined input frame, and executing time alignment processing, wherein the wireless risk factors simultaneously carry sampling time stamps and validity periods; S3, after binding and time alignment are completed, acquiring the periodicity and window boundary of the time sensitive network TSN service, and outputting conflict early warning information by calculating the residual deadline and constraint level; S4, fusing a traffic state snapshot and a wireless risk factor with a time sensitive network TSN time constraint, generating a complete strategy of weight, quota, dequeue sequence, transmission plan in a window and conflict avoidance, and performing parameter self-adaption and threshold hysteresis control based on the complete strategy, wherein the time sensitive network TSN time constraint comprises a period, a window, a deadline deadline and conflict early warning; s5, based on a threshold hysteresis control result, optimizing and guaranteeing the TSN service of the key time sensitive network in the window according to the currently effective executable strategy; S6, based on the optimization guarantee processing result, executing data forwarding and gate control transmission; And S7, counting the data forwarding and gating transmission results to form a feedback report required by closed-loop scheduling, and performing parameter self-adaption or triggering decision-making to complete scheduling of the dynamic data packet.
2. The dynamic data packet scheduling method according to claim 1, wherein the step S1 comprises the steps of: Acquiring data packets from a 5G network side and a time sensitive network TSN side; Classifying and identifying the data packet according to a preset service identification rule, mapping the data packet to a queue group, updating statistics with the granularity of a queue/service flow, and storing the statistics result to a state sharing and caching area, wherein the statistics comprise queue length, average and split queuing delay estimation, arrival rate, burst degree, discard count, rearrangement count and key flow duty ratio; Based on statistics, when a scheduling period boundary or a triggering event threshold is reached, solidifying the current statistics into a flow state snapshot, wherein the flow state snapshot comprises various queue congestion levels, queuing delay estimation, critical flow quantity and duty ratio, a burst index and a snapshot time stamp T_snap; after the traffic state Snapshot is generated, writing a Snapshot serial number Snapshot_ID and a Snapshot time stamp T_snap, and setting a bit stream Snapshot effective mark; After the flow Snapshot effective mark is read by the scheduling controller SC, the Snapshot serial number Snapshot_ID is locked as the decision input of the round, and meanwhile, the read-write version of the corresponding Snapshot is frozen to complete the formation of the flow state Snapshot, wherein if the new flow state Snapshot is generated before the completion of S2 in S1, the flow state Snapshot only enters a queue to be processed, the current locked flow state Snapshot is not replaced, and the consistency of the decision period is ensured.
3. The dynamic data packet scheduling method according to claim 2, wherein S2 comprises the steps of: The method comprises the steps of collecting dynamic information of a wireless link towards a 5G network side, wherein the dynamic information comprises link quality level, scheduling rhythm information, wireless side cache/queue backlog and end-to-end wireless transmission delay statistics of near N periods; Performing sliding window filtering and threshold hysteresis judgment on the dynamic information index; Based on the judging result, converting the dynamic information into a wireless risk factor R directly acting on debugging, wherein the wireless risk factor R carries a sampling time stamp T_wireless and an effective period at the same time; Binding the wireless risk factor R with the Snapshot sequence number Snapshot_ID to form a wireless part Snapshot_ID, T_snap, R, T_wireless of the combined input frame; The scheduling controller SC takes the Snapshot sequence number Snapshot_ID as a reference, collects the wireless state of the same period and executes time alignment verification; if |T_wireless And triggering the supplementary acquisition when T_snap|exceeds a threshold, wherein the supplementary acquisition is failed by using an upper period wireless risk factor R and marking an R_along state, if the wireless interface is abnormal, the scheduling controller SC enters a degradation state to generate an available risk factor R, and after binding and time alignment are completed, the method enters S3.
4. The dynamic data packet scheduling method according to claim 1, wherein the step S3 comprises the steps of: The method comprises the steps of reading a key parameter set of each time sensitive network TSN flow, wherein the key parameter set comprises a period T, a gate control list GCL or equivalent window boundaries W_start to W_end, a maximum occupiable time slot/bandwidth of each period and a service level and priority constraint rule, wherein W_start represents a window starting time when the time sensitive network TSN flow is allowed to start transmitting data in one scheduling period, and W_end represents a window ending time when the time sensitive network TSN flow must finish transmitting in one scheduling period; Constructing a mapping relation between a Stream ID of a Stream identifier of the time sensitive network TSN and a local queue/service Stream identifier, and associating the time sensitive network TSN data packet identified in the S1 to a corresponding time constraint entry; Based on the mapping result, calculating an effective window boundary in the current scheduling period by utilizing a local unified time reference, calculating a remaining deadline deadline _domain for each TSN data packet of the time sensitive network to be scheduled, and outputting a constraint level for a queue level; The method comprises the steps of calculating window pressure indexes, namely judging whether total bytes of a key packet to be sent can be completed in a window under the residual capacity of the current window, outputting conflict early warning information if a plurality of key flow windows are insufficient or overlap competition exists, wherein after outputting (T, W_start/W_end, deadline _remain, constraint level and conflict early warning), packaging the window pressure indexes, together with a flow state Snapshot and a wireless risk factor R, by a scheduling controller SC to form a decision Input frame, uniformly binding the decision Input frame with the same Snapshot sequence number Snapshot_ID and a decision version number input_Ver, and carrying out consistency check based on packaging results, wherein if the fact that a time sensitive network TSN is found to be lack in configuration, the window is not computable, a time reference is abnormal or mapping is lacked, setting a constraint abnormality mark and adopting a safety strategy, and after the consistency check is completed, freezing the decision version number input_Ver by the scheduling controller SC, and generating a strategy capable of tracing specific Input versions.
5. The dynamic data packet scheduling method according to claim 4, wherein S4 comprises the steps of: utilizing a flow state, a wireless risk factor R, a resource state and a time sensitive network TSN time constraint corresponding to a scheduling controller SC convergence decision version number input_Ver; Based on the convergence result, calculating comprehensive dispatching priority P by taking the queue/service flow as granularity; Based on the comprehensive scheduling priority P, outputting a complete strategy of weight, quota, dequeue sequence, transmission plan in a window and conflict avoidance; applying a threshold value and hysteresis rules to the complete strategy to generate an executable strategy policy_Ver; Writing an executable strategy policy_ver into a strategy double buffer area, wherein a standby strategy area is written and attached (input_ver, policy_ver, effective boundary moment), and then a new strategy available mark is issued through atomic switching; based on the writing result, switching to a new executable strategy policy_ver only when the effective boundary is detected, and finishing threshold hysteresis control.
6. The method for scheduling dynamic data packets according to claim 5, wherein S5 is specifically: The method comprises the steps of carrying out differentiated scheduling according to window states and non-window states based on a current executable strategy policy_Ver, wherein when the method is processed in a time sensitive network TSN window, a key time sensitive network TSN queue which allows transmission in the window is preferentially selected, and sequencing is carried out according to the small size of a residual deadline deadline _domain in combination with transmission calculation; at the end of each scheduling slot/cycle, the selected packet or transmit descriptor, carrying the current executable Policy policy_ver, queue service statistics, and exception flags, goes to S6.
7. The dynamic data packet scheduling method according to claim 6, wherein S6 comprises the steps of: Acquiring a transmission descriptor and a current executable policy_ver; When each sending period starts, based on the window state and non-window state judged by the equivalent window boundary W_start to W_end and the current time reference, carrying out window consistency check on a to-be-sent queue one by one, if the to-be-sent queue belongs to a key time sensitive network TSN stream which is required to be sent in a window, only allowing sending in the window, if the predicted sending is to cross the tail end of the window, carrying out advanced sending, temporary storage to the next window or triggering degradation processing according to a current executable strategy policy_Ver, and sending according to the current executable strategy policy_Ver quota and a shaping result for a common service stream; recording the execution evidence in the process of completing the deterministic forwarding to the TSN side of the time sensitive network and the forwarding/returning to the 5G side; summarizing the execution evidence to form a key performance index KPI set of the period, binding the key performance index KPI set of the period with a decision version number Input_Ver, a current executable strategy policy_Ver and a window period number cycle_ID, and generating a feedback report packet for closed-loop optimization; Based on generating a feedback report packet, when the package boundary is reached, encapsulating the execution Record in the boundary into a Record packet, atomically writing the Record packet into a buffer zone, and positioning a Record packet ready mark, wherein the Record packet comprises a packet identifier Record record_ID, a window cycle_ID, a decision version number input_Ver, a current executable strategy policy_Ver and a key state abstract; When the record packet ready flag is detected, a consistency check is performed, and the check is passed, the process proceeds to S7.
8. The dynamic data packet scheduling method according to claim 7, wherein S7 comprises the steps of: Collecting key performance indexes KPI and system resource states by taking the queue/service flow/port/window period as granularity; binding the acquired key performance index KPI with the current round executable strategy policy_Ver, the decision version number input_Ver and the window period number cycle_ID to form a feedback report packet; Based on the feedback report packet, if the TSN flow of the key time sensitive network has super window/super period or the window utilization rate is abnormally high and frequently collides, the strategy mismatch or insufficient resources are judged, the rearrangement needs to be triggered, if the key service is stable and hits the window and the congestion is controllable, the fine adjustment optimization is judged, the weight/quota/shaping rate increment and hysteresis threshold adjustment strategy are output, and the dispatching of the dynamic data packet is completed.
9. A dynamic data packet scheduling system for performing the dynamic data packet scheduling method of any one of claims 1-8, comprising: The real-time flow sensing module is used for carrying out service flow identification and statistics on data packets from the 5G network side and the time sensitive network TSN side to form a flow state snapshot; the wireless state sensing module is used for acquiring dynamic information of a wireless link towards a 5G network side, converting the dynamic information into wireless risk factors directly acting on scheduling, binding the wireless risk factors with the traffic state snapshot to form a wireless part of a combined input frame, and executing time alignment processing; The system state monitoring module is used for continuously monitoring the resources and congestion states of the DS-TT; The TSN time constraint module is used for acquiring the periodicity and window boundary of the time sensitive network TSN service and outputting conflict early warning information by calculating the residual deadline and constraint level; The scheduling controller SC is used for fusing traffic state snapshots, wireless risk factors and time-sensitive network TSN time constraints to generate a complete strategy of weight, quota, dequeue sequence, intra-window transmission plan and conflict avoidance; The strategy adjustment module is used for carrying out parameter self-adaption and threshold hysteresis control on the complete strategy; the queue scheduling module is used for optimizing and guaranteeing the TSN business of the key time sensitive network in the window according to the currently effective executable strategy based on the threshold hysteresis control result; the data forwarding module is used for executing data forwarding and gate control sending based on the optimization guarantee processing result; and the scheduling execution and feedback module is used for counting the data forwarding and gating sending results to form a feedback report required by closed-loop scheduling, and performing parameter self-adaption or triggering decision-making to complete the scheduling of the dynamic data packet.

Description

Dynamic data packet scheduling method and system Technical Field The invention belongs to the field of data processing, and particularly relates to a dynamic data packet scheduling method and system. Background With the continuous promotion of high reliability, low Time delay and deterministic communication demands in the fields of industrial internet, intelligent manufacturing, rail transit, electric power automation and the like, time-sensitive network (Time-SENSITIVE NETWORKING, TSN) technology is becoming a core support technology of industrial ethernet. The time sensitive network TSN realizes the certainty guarantee of the key business flow through time-aware scheduling, flow shaping and accurate time service mechanisms. Meanwhile, the fifth generation mobile communication system is widely introduced into an industrial scene by virtue of large bandwidth, low time delay and flexible deployment, and is used for realizing wireless access and cross-domain interconnection. In order to realize deep fusion of a 5G network and a time sensitive network TSN, 3GPP introduces a 5G-TSN fusion architecture, wherein a distributed synchronous transmission terminal (Distributed TSN Translator, DS-TT) is used as a key network node and is responsible for mapping, forwarding and scheduling data packets between the 5G network and the time sensitive network TSN. In a 5G TSN fusion network, DS-TT needs to face multiple types of data streams from a wireless side and a wired side at the same time, so that the requirements of a time sensitive network TSN on strict time delay upper bound and deterministic forwarding are met, and the dynamic time delay and jitter characteristics of the 5G wireless network, which are introduced by factors such as channel condition change, scheduling strategy adjustment and the like, are adapted. Therefore, the packet scheduling capability of the DS-TT directly affects the overall deterministic communication performance of the 5G-TSN converged network. The disadvantages of the prior art are as follows: The static scheduling algorithm has insufficient adaptability to wireless channel variation, namely the existing DS-TT adopts a static scheduling or fixed priority scheduling algorithm, and is difficult to dynamically adjust according to the quality, load fluctuation and scheduling policy variation of the 5G wireless channel. When the wireless environment changes, the queuing waiting time of the high-priority data packet may be increased, and the problem of overrun of end-to-end delay occurs. The existing scheduling scheme is usually optimized only for a single network side, and the cooperative scheduling mechanism between the 5G network and the time sensitive network TSN is not established, so that scheduling rhythms at two sides are inconsistent. When multiple kinds of services are concurrent, scheduling conflict is easy to generate, and the overall utilization efficiency of network resources is reduced. The scheduling mechanism is simple, lacks real-time traffic sensing capability, has a relatively simple structure, generally relies on static configuration parameters to perform forwarding control, does not sense and feed back the real-time traffic state, queue congestion condition and data packet arrival characteristics, and is difficult to perform fine scheduling according to service dynamic changes. The time constraint characteristic of the time sensitive network TSN service is not fully considered, wherein the service usually has definite transmission period, time window and deadline requirements, the conventional DS-TT scheduling mechanism is not fully combined with TSN time-aware scheduling information, the data packet cannot be subjected to differentiated processing according to the time constraint, and the deterministic transmission performance of the key service is easily affected. Disclosure of Invention Aiming at the defects in the prior art, the dynamic data packet scheduling method and the dynamic data packet scheduling system provided by the invention solve the problems that the existing DS-TT static scheduling algorithm is difficult to adapt to 5G wireless channel change, 5G and TSN cross-network cooperative scheduling is lacking, a scheduling mechanism is simple, real-time traffic perception is not supported, time-sensitive network TSN service time constraint is not fully considered and the like. In order to achieve the above purpose, the technical scheme adopted by the invention is that a dynamic data packet scheduling method comprises the following steps: S1, after data packets from a 5G network side and a time sensitive network TSN side are accessed in a unified way, traffic state snapshots are formed through traffic flow classification, queue length statistics and queuing delay statistics; S2, acquiring dynamic information of a wireless link towards a 5G network side, converting the dynamic information into wireless risk factors directly acting on scheduling, binding the wireles