Dedicated Restricted Target Wake Time for Real-Time Applications in Wi-Fi 7

TL;DR

The paper tackles delivering millisecond-scale QoS for real-time Wi-Fi 7 traffic using Restricted Target Wake Time (R-TWT). It introduces a slotted, batch-arrival analytical model that converts continuous-time dynamics into a tractable queueing framework and uses a two-dimensional Markov chain to estimate the packet delay distribution and loss given R-TWT parameters. The authors validate the model against event-driven simulations, showing high accuracy in predicting delay percentiles and loss, and demonstrate how an access point can optimize SP duration $SP$ and wake period $T$ to maximize system capacity while meeting QoS targets. This work provides a practical, fast tool for R-TWT parameter optimization and paves the way for future integration with mixed access and OFDMA in real deployments.

Abstract

Real-time applications (RTA) tend to play a crucial role in people's everyday life. Such applications are among the key use cases for the next generations of wireless technologies. RTA applications are characterized by strict guaranteed delay requirements (in the order of a few milliseconds). One of the pillars of enabling RTA in next-generation Wi-Fi standards is Restricted Target Wake Time (R-TWT), which provides Wi-Fi stations exclusive channel access within negotiated service periods (SPs). If each RTA data flow uses dedicated SPs for data transmission, they are completely isolated from each other and do not experience any contention. To ensure the satisfaction of RTA QoS requirements while minimizing the channel airtime consumption, it is important to properly select the R-TWT parameters, namely the duration of SPs and the period between SPs. In this paper, we develop a mathematical model that estimates the delay probability distribution and packet loss probability for a given set of network, traffic and R-TWT parameters. Using this model, the access point can select the optimal R-TWT parameters for the given QoS requirements. The high accuracy of the model is proven by means of simulation.

Figures (6)

• Figure 1: System model for $N = 2, R = 3$.
• Figure 2: Definition of Markov chain states.
• Figure 3: Model validation: R-TWT period variation. Figures: (a) average delay, (b) jitter, (c) packet loss probability, (d) 99.9% delay percentile.
• Figure 4: Model validation: R-TWT SP duration variation. Figures: (a) average delay, (b) jitter, (c) packet loss probability, (d) 99.9% delay percentile.
• Figure 5: Model validation: variation of load.