Estimating Regions of Wireless Coexistence with Gaussian Process Surrogate Models

Simultaneous coexistence of multiple wireless communications systems sharing the same spectrum is critical for the success of modern and future communications. We develop a technique for estimating regions of wireless coexistence (RWC) – the transmission configurations of each of the wireless systems which permit coexistence – based on measurements of key performance indicators (KPIs) of those systems. In this article we focus on two-way coexistence tests, which aim to determine the impact each of the communications systems have on each other. The new technique is based on a Gaussian process surrogate model of the unknown transmission-configuration-to- KPI functions. We introduce a sequential design of experiments based on this surrogate model which is designed to reduce the number of measurements necessary to reach a highly-accurate estimate of a RWC. On an illustrative example, this technique reduces the average number of required measurements by over 40% compared to a baseline experimental design. Similar results are achieved for a measurement-informed simulation based on a coexistence test between an Bluetooth Low Energy device and an IEEE 802.11n Wi-Fi devices.