TSN (Time-Sensitive Networking) is promised to play an important role in industrial automation since more and more industrial environments need the support of time-sensitive networking technology to ensure real-time traffic transmission. In IEEE 802.1Qbv of TSN, the GCL (Gate Control List) calculation problem in TAS (Time Aware Shaper) is very critical. We refer to the time it takes for all real-time traffic to complete a cycle of transmission as Flowspan. The objective of a typical GCL calculation approach is to minimize Flowspan using tabu search. However, this method tends to a local optimal solution when the number of streams increases, resulting in a final solution that differs significantly from the global optimal solution. To address this problem, we proposed GAPER, a novel GCL calculation approach based on genetic algorithm with parent elite reservation in this paper. We conducted extensive experiments on DetNet. The experimental results show that our proposed GAPER has significant performance comparing with tabu search. In particular, when the number of flows reaches 20, the Flowspan derived by GAPER is 36.43% lower than the traditional method.
Paper
Full text
Efficient Traffic Scheduling Using Genetic Algorithm in Time-Sensitive Networking
Semantic Scholar · Computer Science · 2022
Abstract
TSN (Time-Sensitive Networking) is promised to play an important role in industrial automation since more and more industrial environments need the support of time-sensitive networking technology to ensure real-time traffic transmission. In IEEE 802.1Qbv of TSN, the GCL (Gate Control List) calculation problem in TAS (Time Aware Shaper) is very critical. We refer to the time it takes for all real-time traffic to complete a cycle of transmission as Flowspan. The objective of a typical GCL calculation approach is to minimize Flowspan using tabu search. However, this method tends to a local optimal solution when the number of streams increases, resulting in a final solution that differs significantly from the global optimal solution. To address this problem, we proposed GAPER, a novel GCL calculation approach based on genetic algorithm with parent elite reservation in this paper. We conducted extensive experiments on DetNet. The experimental results show that our proposed GAPER has significant performance comparing with tabu search. In particular, when the number of flows reaches 20, the Flowspan derived by GAPER is 36.43% lower than the traditional method.