A New SIMD Correlator Algorithm for GNSS Software Receivers to Process Complex IF Data

Abstract

GNSS software receivers implement digital signal processing algorithms on programmable software platform (such as PC, DSP), which traditional ones execute on dedicated hardware. It is highly flexible, convenient for debug and could be adapted to a platform for GNSS algorithm research. However, the computation of down shifting of the GNSS signal to baseband and correlation with the locally generated ranging code is too expensive for normal GNSS software receivers to work in real time. The computational cost could be reduced by utilizing Single Instruction Multiple Data (SIMD) operations. The article proposes a new SIMD correlator algorithm for complex GNSS IF signal processing on x86 processors. It firstly demonstrates why normal GNSS software receivers couldn’t achieve real-time processing by using Single Instruction Single Data (SISD) operations and the improvements achieved by using existing SIMD algorithms; then proposes a new SIMD correlator algorithm, outlines its implementation principle and compares it with the SISD and the existing SIMD algorithms. Performance gains achieved via the new SIMD algorithms are then demonstrated in an analysis. Finally, the implementation and experimental results of the new algorithm are presented. The experimental results show that compared to the SISD and existing SIMD algorithms the new SIMD algorithm can effectively reduce the computation of down shifting of the GNSS signal to baseband and correlation with the locally generated ranging code.