Enhancing WiFi RSS fingerprint positioning accuracy: lobe-forming in radiation pattern enabled by an air-gap

We propose a physical arrangement consisting of a 2.4 GHz WiFi radio, and high attenuation surfaces with an air-gap in-between, to delimit a fingerprinting region. We show by simulations, and field measurements, that such arrangement allows to form a characteristic lobe in the radiation pattern after the gap, which is characterized by an abrupt change in the Received Signal Strengths (RSSs) along the direction parallel to the surfaces. From measured RSS samples, we construct, to aid our analysis, an equation that approximates the radiation pattern as a continuous RSS radio map, using symbolic regression. Finally, we observe the positioning performance of this arrangement, performing RSS fingerprint pattern matching with a Neural Network. Results are compared to the positioning accuracy that would be obtained by using an omnidirectional radiation pattern antenna. We conclude that, the abrupt change in RSSs obtained with this arrangement, translates into a better positioning accuracy, close and along a direction parallel to the surfaces.