The Multi-Sensor Navigation Analysis Tool (MuSNAT) – Architecture, LiDAR, GPU/CPU GNSS Signal Processing

Abstract

In this paper we summarize enhancements to the Global Navigation Satellite System (GNSS)-software receiver ipexSR developed at the Universitat der Bundeswehr Munchen since 2004. The ipexSR was conceived as a multi-GNSS, multi-frequency software receiver and achieved very soon all-in-view real-time tracking capability with a reasonable geodetic quality of the produced raw data. Like other software receiver developments, it has been used in numerous GNSS research projects always allowing to do an indepth analysis of every GNSS signal processing aspect. However, during the last two years, it became obvious that several architectural changes are necessary to improve the usability and scope of this analysis tool. In current high-end receivers, GNSS signal processing is deeply fused with other sensor processing pipelines. The various processes influence each other, and a timely synchronized data analysis is necessary to gain in-depth insight when developing new algorithms. Therefore, Inertial and Light Detection and Ranging (LiDAR) processing has been included (to some extend also camera) in our software package and it was consequently renamed MuSNAT (MUlti-Sensor Navigation Analysis Tool). GNSS signal tracking has been reworked and verified and supports a fast and a precision mode on the Central Processing Unit (CPU). Signal processing on the Graphics Processing Unit (GPU) has been added and this paper discusses the chosen approach and limitations when porting the precision mode to a GPU. Standard GNSS-signals, AltBOC-signals and via the flexible GPU-language also new GNSS-signal candidates are easily supported. LiDAR processing currently targets relative navigation making use of the point-cloud library PCL. The integration filter merges GNSS, inertial and LiDAR data and is an extension of the Real-Time-Kinematic-Library (RTK-LIB). RTK-LIB is also used for RTK and Precise point Positioning (PPP) processing. The GNSS data flow can be inverted to use the receiver as a GNSS signal generator (transceiver concept). For ease of data analysis, a dedicated data visualization software named MuSNAT-Analyzer is developed which interfaces to the software receiver MuSNAT-Core via a Structured Query Language (SQL)-database. The SQL-database serves also as basis for further MATLAB based analysis. For complex receiver configurations, the demands on the SQL-performance are significant. An automated test framework ensures the stability of the software and is based on the Jenkins-environment.