Abstract:

Objectives: With the arrival of the industry 4.0 era, the industrial production control system is becoming more and more intelligent, which puts forward higher requirements for real-time and deterministic transmission of information. Time-Sensitive Networking (TSN) has been introduced into industrial network because of its good compatibility and low delay jitter. To realize efficient transmission of mixed traffic in industrial network with the help of TSN, we explored a new traffic scheduling mechanism in TSN.

Methods: Based on architecture of the centralized software defined network (SDN), we designed a method to determine the minimum scheduling slot of the network and adjusted sampling period of Scheduled Traffic (ST) based on the minimum slot. By reducing occupation of transmission bandwidth by ST traffic, more transmission resources were reserved for Stream Reservation (SR) traffic to improve network schedulability. Furthermore, for SR traffic, a parity mapping scheme was proposed. When SR flow was not schedulable, a flow offset planning (FOP) algorithm was designed to offset injection time of SR flow, which can further improve schedulability of the network by improving utilization of system resources.

Results: To verify performance of core algorithm of HSTC mechanism, we built an experimental platform and compared performance of our mechanism with existing mechanisms from the perspectives of ST traffic bandwidth occupation, traffic scheduling priority, SR traffic mapping, injection slot selection, etc. In the experiment, the maximum transmission unit (MTU) of network was 1500B, the maximum buffer size of single queue in switch was 6MTU, and link rate was 1000 Mbit/s. The maximum sampling period and packet length of each ST flow were randomly selected from the set {0.6, 0.8, 1, 1.2, 1.6}ms, {0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1}kB. Minimum sampling period of ST flow was set to 0.1 ms. And the deadline (DDL) of ST flow was its actual sampling period. Sampling period and packet length of SR flow were randomly selected from set {4, 6, 8, 10, 12, 16, 20}ms, {1.5, 2, 2.5, 3, 3.5, 4, 4.5}kB. DDL of each SR flow was a random integer value within range of $[0.5\times T_{\text{j}}^R,T_{\text{j}}^R]$. Main experimental results are showed as follows: ①Adjusting sampling period of ST traffic: if sampling period was not adjusted, when the number of flows reached 12, transmitting ST traffic occupied more than 80% of network bandwidth. However, after adjusting sampling period by our scheme, the bandwidth occupation of ST traffic was greatly reduced. ②Traffic scheduling priority: weighted ranking scheme proposed in this paper had the best performance. Performance ranking of other schemes was the maximum packet length first, the shortest deadline first, and the minimum sampling period first. Compared with the sub optimal scheme, weighted ranking scheme can improve scheduling success rate by up to 0.52. ③SR traffic mapping: as the number of SR flows increased, impact of parity mapping scheme on improving scheduling success rate became more significant. The maximum difference between scheduling success rate of our scheme and the scheme which mapped according to DDL was 20%. ④Compared with the suboptimal random slot injection scheme, the slot sequencing scheme in HSTC can increase network scheduling success rate by up to 0.77, and the limit bandwidth utilization can reach 88%.⑤Overall performance comparison: we compared with two mechanisms which were proposed in two representative existing literatures under the same simulation parameter configuration. The experimental results showed that HSTC mechanism realized dual optimization of reducing solution complexity and improving scheduling performance.

Conclusions: How to make full use of TSN's accurate flow scheduling capability to provide certainty and real-time guarantee for production control system is still a research focus of TSN. Therefore, we proposed a mixed traffic scheduling mechanism called HSTC, which combined two existing schemes of Time-Aware Shaper (TAS) and Cyclic Queuing and Forwarding (CQF) and formulated different scheduling strategies for time-sensitive traffic and large bandwidth traffic according to their characteristic. The experimental results showed that HSTC mechanism significantly improved network schedulability by improving system resource utilization, and it realized efficient scheduling of mixed traffic. Existing network schedule schemes for TSN are mostly based on off-line scheduling scenarios. However, in actual industrial network, there is still a small number of burst traffic triggered by events. Burst traffic has no fixed parameters but has an important impact on the normal operation of the system. Therefore, how to improve our current mechanism to support mixed transmission of burst traffic at the same time is our next research direction.

Key words: time-sensitive networking, time-aware shaper, cyclic queuing and forwarding, traffic scheduling