Session B6: Emerging PNT Consumer Applications

GNSS Multipath Mitigation by MUSIC-Based DoA Estimation Using a Multi-COTS-Receiver and Array Antenna System
Hiromune Namie, Dai Nagafuchi, Takemitsu Fujita, Kohsuke Uchiyama & Kanato Osawa, National Defense Academy, Ministry of Defense
Location: Prince David Room
Date/Time: Thursday, Apr. 16, 3:17 p.m.

In this study, we aims to estimate the direction of arrival (DoA) of satellite signals using the MUSIC (MUltiple SIgnal Classification? algorithm. As a preliminary step toward improving multipath errors in GNSS, we use data from a combined system of a GNSS array antenna and commercial-off-the-shelf (COTS) GNSS receivers.
Conventionally, the receivers used for array antenna processing are specialized units that could output high-speed digitized IF signals. In contrast, in this research, we attempt to achieve our goal by using observation data, such as carrier phase data, output from COTS GNSS receivers.
In our previous research, we first built a composite system that housed a seven-element array antenna and seven COTS GNSS receivers. The seven integrated receivers were not hardware-time-synchronized, which made direct MUSIC processing impossible.
To address this, we leveraged the known geometric relationship of the seven antenna elements, which were arranged at the vertices and center of a regular hexagon. From the real observation data (e.g., carrier phase) acquired using a single pair of— the central GNSS antenna (our reference antenna) and one receiver, —we computationally generated simulated carrier phase data for the other six pairs. We then performed MUSIC processing and estimated the DoA of satellite signals. From the results, we predicted that the DoA could be estimated by applying MUSIC processing of observation data from COTS receivers.
In this study, we implemented a new calibration method. The signal from the reference antenna was split and input into all seven receivers within the composite system, not just its paired receivers. We then corrected the real observation data for each of the seven pairs by using the property that observation data from satellites in the zenith direction should have a phase difference of zero. This was an attempt to create a situation equivalent to having all seven integrated receivers software-time-synchronized.
First, we performed seven simultaneous sets of single-point positioning with the reference antenna and each of the seven receivers. We then corrected and removed the internal clock errors (clock biases) of each receiver on the basis of the clock differences.
Next, we used the corrected real observation data for a signal from a single satellite in the zenith direction. Since the phase difference of the observation data for all seven pairs should be zero for this signal, we corrected the differences on the basis of the actual phase shifts that occurred.
Finally, we performed seven simultaneous sets of carrier phase positioning with the reference antenna and each of the seven receivers. Because for all seven pairs, the same position for the reference antenna should be calculated, we corrected the delay from the reference antenna to each of the seven receivers by ensuring that the position coordinates were consistent across all seven combinations.
By executing these correction processes, we estimated the DoA of satellite signals using only real observed data. We applied MUSIC processing to the corrected real observation data from each of the seven pairs and confirmed that the estimation was indeed possible.