Implementation and analysis of GNSS software receiver on embedded CPU-GPU heterogeneous architecture

Abstract

In this study, a real-time GNSS software receiver is implemented and analyzed using CPU and GPU heterogeneous based embedded platform. A software and hardware configuration of GNSS receiver is designed by consideration of Jetson TK1 and specification of Tegra K1. To improve computation performance for Jetson TK1 based software receiver, three optimization techniques are applied; one technique is multi-core CPU based signal acquisition, another is optimum number of thread selection method for tracking loop, and the other is zero-copy based correlation. In order to verify performance of designed receiver, we measured and analyzed execution time of signal acquisition, signal tracking and totally receiver operation with various real GNSS signal. And we verified limitation of embedded based GNSS software receiver. As a results, The Jetson based software receiver can be operated in real-time with low sampling rate and ten channels.