Presented at ION GNSS 2012, Session D3, Nashville, TN, USA, September 17-21, 2012. 1 Improving the Performance of the FFT-based Parallel Code-phase Search Acquisition of GNSS Signals by Decomposition of the Circular Correlation Jérôme Leclère, Cyril Botteron, Pierre-André Farine, Electronics and Signal Processing Laboratory (ESPLAB), École Polytechnique Fédérale de Lausanne (EPFL), Switzerland BIOGRAPHIES Jérôme Leclère received the Master Degree in Electronics and Signal Processing from the ENSEEIHT, Toulouse, France, in 2008. He is currently performing his Ph.D. thesis in the GNSS field at EPFL, focusing his researches in the acquisition and high sensitivity areas, with application to hardware receivers, especially using FPGAs. Dr. Cyril Botteron leads the GNSS and UWB groups in the electronics and signal processing laboratory at EPFL. He received his PhD degree from the University of Calgary, Canada, in 2003. His current research interests comprise the development of low power radio frequency (RF) integrated circuits and advanced signal processing techniques for ultra-low power communications and global and local positioning applications. Prof. Pierre-André Farine is professor in electronics and signal processing at EPFL, and is head of the electronics and signal processing laboratory. He received the M.Sc. and Ph.D. degrees in Microtechnology from the University of Neuchâtel, Switzerland, in 1978 and 1984, respectively. He is active in the study and implementation of low-power solutions for applications covering wireless telecommunications, ultra-wideband, global navigation satellite systems, and video and audio processing. He is the author or coauthor of more than 100 publications in conference and technical journals and 50 patent families (more than 270 patents). ABSTRACT This paper proposes alternative architectures to perform a circular correlation using the Fast Fourier Transform (FFT) by decomposing the initial circular correlation into several smaller circular correlations. The approach used is similar to the Fast Finite Impulse Response (FIR) Algorithms (FFAs). These architectures improve the performance in terms of reduced processing time or resource usage, and consequently lower the energy consumption. The results can be applied to any system that performs circular convolution or correlation. In this paper, the application is the acquisition of Global Navigation Satellite System (GNSS) signals with the FFT-based Parallel Code-phase Search (PCS), and more precisely on the GPS L1 C/A signal, when the target considered is a Field Programmable Gate Array (FPGA). In this context, it is for example shown that it is possible with one of the proposed architectures to reduce the logic usage by 11 %, the memory usage by 41 %, and the Digital Signal Processing (DSP) block usage by 32 %, while keeping the same processing time. With another architecture, it is shown that the processing time can be halved by increasing the logic usage by only 35 %, while reducing the memory usage and keeping the same DSP usage. Note that the proposed approach is not based on an approximation of the traditional method, but a modified implementation providing the same result. Thus, there is no loss of sensitivity. 1. INTRODUCTION The acquisition of GNSS signals consists mainly in three steps : 1) multiplication of the input signal with a local carrier replica to remove the offset in frequency due to the Doppler effect, 2) multiplication with a local pseudo- random noise (PRN) code replica; 3) Integration. The process has to be repeated for different carrier frequencies and code phases of the replicas until they are both aligned with the received ones. This is thus a two-dimension search (for each satellite). Together, the last two steps are equivalent to a circular correlation (due to the code repetition). As a consequence, the FFT can be used to compute it. This enables getting the correlation result for all the code-phases simultaneously. This is known as Parallel Code-phase Search acquisition. However, the processing time can still be relatively long, because of : 1) the different carrier frequencies to test; 2) the results over dozens or hundreds of code periods that are usually accumulated to increase the sensitivity; 3) the process has to be repeated for several satellites (up to a few dozens, if there is no a priori information and several constellations are considered). Therefore, finding new methods to perform the search faster is still topical. Within this context, we use an approach that consists in decomposing the initial circular correlation into several