Session A3b: Next-Generation Satellite Navigation Technology

Development of a Robust Positioning Method for Unknown Radio Sources Using Satellite Constellations
Hirofumi Fukushima, Toshihiro Ito, and Yuki Takabayashi, Information Technology R&D Center, Mitsubishi Electric Corporation
Location: Beacon A

With the rapid expansion of information networks using satellite communications, there is a growing need to locate radio sources that interfere with communications satellites and unknown radio sources for the purpose of information gathering. In recent years, the need for satellite constellations with a large number of satellites has been gaining momentum worldwide, and the demand for unknown radio source localization is expected to increase. When an unknown radio source is a communication wave source for satellite communication, the location of the radio source is calculated by receiving the signal of one radio source via multiple satellites, correlating the signals, and then using the time difference of the signals (Time Difference Of Arrival : TDOA) and frequency difference of the signals (Frequency Difference Of Arrival : FDOA). However, if the unknown radio source was a radar source that output pulse waves, such as those used in pulsed Doppler radar, multiple peaks would be generated after correlation processing, resulting in multiple TDOAs and FDOAs. This would cause ambiguities in the positioning process, making it difficult to determine the target's position. One possible solution is to install an array antenna on the satellite to eliminate ambiguity by using the Direction of Arrival (DOA) information of the target, but this requires the satellite to be equipped with an array antenna, which increases the scale and cost of the H/W and takes time to manufacture.
In this presentation, we propose a method to robustly locate unknown radio sources regardless of the positional relationship between the satellite and the unknown source, while eliminating ambiguity caused by radar waves using signals received by a LEO equipped with a single antenna. Details of the proposed method are described below.
First of all, the proposed method calculates a two-dimensional histogram of the results of positioning process, including ambiguities, acquired over multiple time periods. If the calculated positioning results are ambiguous, they change with time as the satellite moves, while if they are true target origins, they become immobile, and target origin positioning results are dominantly accumulated when the two-dimensional histogram is calculated. The differences in the positioning results of these are used to remove ambiguity. As theoretical support for this result, we also present a formulation of the time-varying amount of ambiguity. Using this first feature, ambiguity can be eliminated by only a single antenna and location can be determined regardless of the type of unknown radio source.
As a second feature, the proposed method evaluates the error ellipse of the positioning results with ambiguity removed. In this study, three satellites are assumed, and in that case, the positioning results can be calculated for each of the three cases using each pair of satellites, including ambiguity. If all of the ambiguity removal results are used in the position determination, errors may occur in the position determination results due to the influence of the points calculated by a pair of satellites with a large DOP (Dilution of Precision). As a countermeasure, the proposed method calculates the equation of the error ellipse in which the positioning results calculated for each pair of satellites exist and estimates the position of the unknown radio source by extracting only the positioning results that exist in the common part of the equation of each error ellipse. Such a process can greatly reduce the influence of the target point calculated by a pair of satellites with a large DOP.
The effectiveness of the proposed method was evaluated by numerical simulation. The target of the localization was assumed to be a radar source moving at a speed similar to that of a ship (60km/h). Even when the target is moving, the influence is more dominant for LEOs moving at relatively high speeds, and we confirmed that the proposed method can remove ambiguity and extract only target-derived location points by exploiting the difference in temporal characteristics between true target points and ambiguity. The results of Monte Carlo simulations confirm that it is possible to identify moving unknown radar wave sources with an accuracy of about 2 km-RMSE. Although this accuracy is greater than that of self-position estimation using GNSS, it is sufficient for wide-area monitoring of unknown radio sources over vast areas of ocean, for example. Furthermore, by combining the evaluation with the error ellipse, we confirmed that the accuracy is less than 1/10 that of the case where only the ambiguity removal process is applied, and the accuracy improves to about 100m-RMSE.