| Keywords [eng] |
GNSS, L1, L5, L1+L5, simulation, record, replay, GPS, GALILEO, BEIDOU, GLONASS, įrašymas, atkūrimas |
| Abstract [eng] |
The widespread adoption of Internet of Things (IoT) technologies and other electronic devices that need to know their global position has increased the need for Global Navigation Satellite System (GNSS) receiver testing. This usually is done using GNSS simulators, however the cost of GNSS testing equipment is very high and IoT manufacturer find it hard to justify. To solve this problem, software-defined radio (SDR) can be used, however, currently developed solutions generate or record - replay only GPS L1 C/A signals, uses SDR equipment, the price of which is comparable to the price of commercial solutions, and usually these solutions are not portable. In other words, there is currently no SDR-based GNSS L1+L5 test equipment that is suitable for IoT manufacturers, is affordable and portable. Since the need for GNSS testing has increased and there is no available SDR-based testing equipment, the goal of this work is to create a GNSS L1+L5 record--replay simulator, which can simulate L1 and L5 signals, and is affordable and portable. The device created during this work is based on HackRF One SDR modules that are synchronized in terms of time, phase and temperature. The simulator was tested with GPS, GALILEO, BEIDOU and GLONASS satellites, as well as Airoha AG3335, Quectel LG69T AA and u-blox NEO-M9N GNSS receivers. Testing has shown that thermal stabilization of discrete SDRs and synchronization equipment during record and replay is imperative to capture and maintain stable relation between numbers of visible satellites and carrier to noise ratios of L1 and L5 signals as well as mitigated sudden disappearances of L5 signals during playback. Also, data losses during long term sustained recording operations can be prevented only if computer is configured as a real-time machine. Analysis of recordings using circular statistics allowed to statistically characterize GNSS signals. Recorded GNSS signals share many similarities with complex white noise and have phase distributions, which can be well modeled using von Mises distribution. Phase distribution can be influenced by the amplitude clipping of the input signal, but appropriate clipping mitigates phase distribution distortions caused by quadrature sampling of SDRs. Generic SDR based simulation approach puts large load on SDR host computer, therefore, ways to increase system stability were explored. One of explored optimizations was recording of GNSS signals using lower than necessary sampling rate. The results show that any GNSS L1 signal can be successfully recorded using sampling rate as low as 4 MHz. In case of L5, any GNSS L5 signal can be recorded using sampling rate of 6 MHz. Regardless of absolute lower limits of sampling rate, it is imperative, that all recorded signals would use the same sampling rate, when L1 and L5 are being recorded simultaneously. If this condition is not satisfied, GNSS receivers do not recognize L5 signals, only L1. Multiple sampling rates are only viable, when signals are being recorded in the same band. Results obtained during dynamic testing show close overlap of all performance metrics like trajectory, speed, number of visible satellites or carrier to noise ratio when comparing direct reception signals with signals produced by SDRs. This implies that from point of view of GNSS receiver signal origin is indistinguishable and records produced by proposed GNSS L1+L5 simulator can be used for equipment testing purposes. Overall, results produced in this work show that the goal of this master thesis is achieved. |
