Optimizing Scheduling in Wireless TSN Utilizing Genetic Algorithms

Time-sensitive networking (TSN) is proposed to support deterministic communication for industrial automation use cases. To harvest the wireless communication flexibility, TSN features have been extended to the wireless domain as well. One of the key TSN features is the ability to assign transmission schedules to different traffic flows in the network with the aim of reducing time slot access delay on each network node. In the wireless domain, this becomes even more challenging due to the shared medium, lower reliability, and slower transmission rates compared to wired systems, reducing the available time resources. In this paper we look at utilizing genetic algorithms to support scheduling of traffic flows from different wireless end devices in a shared schedule. Two optimization functions are defined. The first optimization is based on minimizing the overall shared air time between different end nodes, while the second optimization is based on maximizing time slots that can be used without any interference. Both optimizations aim to reduce the collision probability. With these initial results, we identify the best parameters for genetic algorithms and examine the initial population's impact on overall performance. We show that a fully randomized initial population does not achieve the highest fitness value, even after several generations.