Evaluation of Sparse Acoustic Array Geometries for the Application in Indoor Localization

Abstract

Angle-of-arrival (AoA) estimation technology, with its potential advantages, emerges as an intriguing choice for indoor localization. Notably, it holds the promise of reducing installation costs. In contrast to time-of-flight (ToF)/time-difference-of-arrival (TDoA) based systems, AoA-based approaches require a reduced number of nodes for effective localization. This characteristic establishes a tradeoff between installation costs and the complexity of hardware and software. Moreover, the appeal of acoustic localization is further heightened by its capacity to provide cost-effective hardware solutions while maintaining a high degree of accuracy. Consequently, acoustic AoA estimation technology stands out as a feasible and compelling option in the field of indoor localization. Sparse arrays additionally have the ability to estimate the direction-of-arrival (DoA) of more sources than available sensors by placing sensors in a specific geometry. In this contribution, we introduce a measurement platform designed to evaluate various sparse array geometries experimentally. The acoustic microphone array comprises 64 microphones arranged in an 8×8 grid, following an uniform rectangular array (URA) configuration, with a grid spacing of 8.255 mm. This configuration achieves a spatial Nyquist frequency of approximately 20.8 kHz in the acoustic domain at room temperature. Notably, the array exhibits a mean spherical error of 1.26 $^{\circ }$ when excluding higher elevation angles. The platform allows for masking sensors to simulate sparse array configurations. We assess four array geometries through simulations and experimental data, identifying the open-box and nested array geometries as robust candidates. In addition, we demonstrate the array's capability to concurrently estimate the directions of three emitting sources using experimental data, employing waveforms consisting of orthogonal codes.