Dynamic Distributed Threshold Control for Spatial Reuse in IEEE 802.11ax

The IEEE 802.11ax standard (Wi-Fi 6), among other features, adopts a feature called spatial reuse, where new transmissions can be carried out in presence of ongoing, interfering transmissions from nodes in an overlapping basic service set (OBSS). Specifically, a node can adjust its threshold for detecting the interference, by setting a parameter called OBSS Power-Detect level
$(\text{OBSS}_{-}\text{PD}_{\text{level}})$
. When a node hears an ongoing transmission from an OBSS node, if its received signal strength indicator (RSSI) is below the
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
, the node is said to have a spatial reuse opportunity. The node can transmit at a limited transmit power (TX_PWR) during the spatial reuse opportunity. The feasible values of
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
and TX_PWR must satisfy certain constraints laid out by the IEEE 802.11ax standard. In this work, we propose an algorithm that first obtains
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
thresholds for maximizing the number of spatial reuse opportunities, and then selects the one that minimizes the packet error rate among these
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
thresholds. The trade-off involved is the following: setting
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
to a high value increases the number of spatial reuse opportunities, but necessitates transmissions to be at lower transmit power (due to the constraint specified by the standard) resulting in higher packet error rates, and vice versa. The proposed algorithm dynamically varies
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
based on packet error rates. Via simulations, we show that the proposed dynamic algorithm performs better (in terms of achieving a higher throughput and a lower packet error rate) than a naive method which adopts a constant
$\text{OBSS}_{-}\text{PD}_{\text{level}}$
threshold and the case when the spatial reuse is not adopted. When the spatial reuse is implemented using the proposed algorithm, we also explore the performance of different traffic models served using QoS queues having different priorities and transmission parameters.