 Abstract— This paper introduces a new approach to perform very long coherent integration for HS-GNSS applications developed under ESA funded project “DINGPOS”. Very long coherent integration is an approach to increase a GNSS receiver’s post-correlation signal-to-noise ratio (SNR) and is especially helpful in the indoor scenario. For successful coherent integration very precise signal replicas have to be created. Taking into account an integration time in the second domain, the user motion has to be determined precisely. For that reason an Inertial Measurement Unit (IMU) has been chosen to track the user motion. Due to several constraints a MEMS-type IMU has to be used, hence the low performance of this sensor kind is denying the use of a classical strap-down calculation. The paper introduces an innovative approach using a Pedestrian Navigation System (PNS) as backbone for the system, taking comfort of the low drift and high macroscopic accuracy of that system. Using this position knowledge as a priori information, a corrected pseudo strap-down calculation can be used to produce a micro-trajectory of sufficient accuracy for replication of the GNSS signals. Finally an integration time of 2 seconds for a walking user shall be achieved. The paper discusses the need for that special procedure, the elements of the integrated system, the alternative processing of the micro-trajectory reproduction and discusses some initial results. Manuscript received February 2, 2009. This work has been conducted under the frame of the ESA contract DINGPOS (20834/07/NL/GLC). H. Niedermeier is a research associate at the Institute of Geodesy and Navigation, University of the Federal Armed Forces Munich, 85579 Neubiberg, Germany (phone: +49-(0)89-6004-4640, fax: +49-(0)89-6004- 3019, e-mail: herbert.niedermeier@unibw.de). G. Ameres is a research associate at the Institute of Geodesy and Navigation, University of the Federal Armed Forces Munich, 85579 Neubiberg, Germany (e-mail: gerald.ameres@unibw.de). Dr. T. Pany is a research associate at the Institute of Geodesy and Navigation, University of the Federal Armed Forces Munich, 85579 Neubiberg, Germany (e-mail: thomas.pany@unibw.de). Prof. Dr.-Ing. B. Eissfeller is full professor at the Institute of Geodesy and Navigation, University of the Federal Armed Forces Munich, 85579 Neubiberg, Germany (e-mail: bernd.eissfeller@unibw.de). Index Terms—High Sensitivity GNSS, MEMS INS, pedestrian navigation, sensor fusion I. INTRODUCTION LTHOUGH GNSS is on its way towards a “system of systems”, and it can be expected to have more than 100 useable GNSS satellites in orbit in several years, the indoor or high sensitivity application still is one of the major challenges in satellite navigation. Due to the high signal attenuation, the limited view to the sky inside buildings and the severe multipath indoors, signal acquisition and tracking often become extremely difficult, sometimes impossible. One of the major symptoms of the indoor domain is the extremely low SNR of the GNSS signals indoors, which limits the operation of standard GNSS receivers Longer signal integration is the most common way to improve the SNR of weak signals and has been used in GNSS receivers from the early days up to now. The integration period is nevertheless limited for standard processing techniques due to the CDMA signal’s navigation bit rate of 50 Hz to a maximum integration time of 20 ms (for the example of the GPS C/A code). Since the flip of the navigation data bit usually cannot be predicted without the knowledge of the navigation massage, integration over the data bit borderline could result in an instant loss of correlation gain. By aiding the receiver with the navigation data bits from an external reference (also known as assisted GNSS) this limitation can be overcome. Using advanced local oscillators or single-differences processing can also improve the quality of the reference signal for long coherent correlation. Nevertheless for mobile users these means are not sufficient due to the influence of the user motion on pseudo-ranges and signal Doppler. While the satellite motion can be predicted by almanac or precise ephemeris data, the determination of the user motion in a quality sufficient for trajectory replication without using GNSS is a non-trivial task. Obviously, a timely non-constant difference in signal replica and received signal carrier phase causes a significant correlation loss, which can also destroy the gain of long coherent integration if the difference reaches half of the signal wavelength. In the project DINGPOS an integrated replica error corresponding to 1.4 cm length or a loss of 1 dB has been defined as target. A crucial element of First results from supporting long coherent CDMA correlations by a MEMS INS and a Pedestrian Navigation System for HS-GNSS applications Herbert Niedermeier, Gerald Ameres, Thomas Pany, and Bernd Eissfeller A 978-1-4244-3293-6/09/$25.00 ©2009 IEEE PROCEEDINGS OF THE 6th WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION 2009 (WPNC’09) 5