NIST's Industrial Wireless Systems Team Publishes Paper on Analytical Evaluation for Software-based TSN in Industrial Wi-Fi Networks

NIST's Industrial Wireless Systems team’s paper, titled “An Analytical Evaluation for Software-based TSN in Industrial Wi-Fi Networks,” was accepted to the 7^th IEEE International Conference on Industrial Cyber-Physical Systems (ICPS 2024). The conference is sponsored by the IEEE Industrial Electronics Society (IES). The conference will be held in May 2024 at St. Louis, MO, where the paper will be presented by Mohamed Kashef (Hany).

Time-sensitive networking (TSN) is an emerging topic for advancing wireless networking for industrial applications. Wireless TSN (WTSN) is the wireless counterpart of the wired TSN, which faces the challenges associated with wireless networking, such as reliability, stability, and interference-related effects. One aspect of WTSN is the traffic scheduling of the time-critical (TC) traffic streams to coexist with the best-effort (BE) traffic over the same Wi-Fi network. Traffic scheduling can be achieved by implementing the IEEE 802.1Qbv protocol that defines priority queues for traffic forwarding, based on their quality of service (QoS) requirements.

One direct software-based implementation of IEEE 802.1Qbv was achieved by using the Qdisc token bucket algorithm with a gated input at Linux-based devices equipped with Wi-Fi interfaces. This implementation enables the mapping of traffic streams to queues at the network stack (using the Linux Qdisc features) and controls the queues based on the requirements of the traffic streams as shown in the above diagram. In the paper, the authors introduce a numerical analysis of this software-based WTSN implementation to provide a tool to measure the ability of utilizing already deployed legacy Wi-Fi chipsets to achieve WTSN latency requirements in a practical way. 

The analysis presented in the paper demonstrates the impact of the Token Bucket algorithm parameters on latency performance. Two main challenges are studied. First, the utilization of the wireless channel, due to applying a schedule, is affected by resource reservation. Second, the protected window of the schedule for the TC stream may have overflow from the BE traffic window because the TSN schedule is implemented before the Wi-Fi chipsets. Overall, this work identifies the advantages and limitations of the current software-based WTSN implementations and provides a tool for application-based tuning of the Qdisc parameters to achieve a desired level of performance.