A Radio Map Approach for Antenna Pattern Calibration in Low-Altitude Air-to-Ground Sensing

Reconstructing antenna radiation patterns typically requires specialized anechoic or reverberation chambers, which are costly and resource-intensive. This paper investigates the feasibility of using outdoor signal strength measurements to achieve the same goal. Current methods relying on controlled environments are less accessible for extensive use, often requiring sophisticated instruments. To address these limitations, we propose a approach that leverages sparse reference signal received power (RSRP) measurements to build a two-dimensional (2D) power angular map (PAM) of the antenna. Assuming horizontal isotropy of the antenna, the PAM is constructed with distances as column indices and elevation angles as row indices, capturing key characteristics of the antenna's radiation pattern. A weighted K-Nearest Neighbors (KNN)-based matrix completion method is employed to impute missing values, followed by Singular Value Decomposition (SVD) to separate antenna gain from path loss. Our experimental results, obtained through simulations and real-world measurements, demonstrate that only 30 elevation angle measurements are needed to achieve a normalized mean squared error (NMSE) of less than 0.01 for the reconstructed antenna pattern of a half-wave dipole antenna. Further validation using a UAV measurement system for the VERT 2450 antenna shows that the reconstructed radiation pattern has a similar distribution to that provided in the VERT 2450 datasheet, confirming the effectiveness of our approach.