Massive Parallel Algorithms for Software GNSS Signal Simulation using GPU

Iva Bartunkova and Bernd Eissfeller

Abstract:	GNSS signal simulators are important tools for adjustable and repeatable testing and verification of GNSS receivers. Simulator offer many advantages in contrast to costly field testing and enable development and verification of receivers of future GNSS systems and services not yet being in service. Therefore GNSS signal simulators became widely used in testing procedures of all components of GNSS receivers. The topic of this paper is a software-based GNSS IF signal simulator that uses for signal generation a Graphics Processing Unit (GPU) to reach real-time performance. This solution addresses the disadvantages of the typical solutions with FPGAs and DSPs that have very high costs per unit. Moreover the evolution of these components is slow, as they are deployed in low numbers in very specific non-mass market products. The programming of these units needs special programming tools and deep insight into the hardware functionality. In contrast to it, the GPUs are sold as mass-market products in form of motherboard components and graphics cards for gaming and video streaming purposes. The price is very low and swift evolution of processing capacity follows the Moore law similarly to evolution of general microprocessors (CPU). Also miniaturization and energy consumption per operation progresses accordingly. The programming interface development keeps pace with evolution and has improved dramatically from graphics-specific programming towards general programming interface during the last years. This paper introduces a GPU based design and implementation of a signal samples generation module of the simulator. The GPS L1 C/A signal service was implemented. The objective was to reach real-time performance for one signal service with all possibly received satellite channels and maximize the sampling rate generated together with minimal GPU size settings. In the paper, the key issues for real-time performance and acceleration of the signal samples generation are identified. Algorithmic solutions exploiting the massive parallel architecture of the GPU are explained and discussed. The results of the signal simulator verification are presented. The performance of the implemented simulator was benchmarked with respect to number of generated channels and the size of used GPU and results are given. Additionally, comparison to the only another similar solution, to the best author’s knowledge, is given. This is the CUDA based GNSS signal simulator solution that was published after the work on results presented in this paper had already been done [1].
Published in:	Proceedings of the 25th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2012) September 17 - 21, 2012 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	118 - 126
Cite this article:	Bartunkova, Iva, Eissfeller, Bernd, "Massive Parallel Algorithms for Software GNSS Signal Simulation using GPU," Proceedings of the 25th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2012), Nashville, TN, September 2012, pp. 118-126.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In