ION GNSS 2012
Session D3: GNSS Algorithms & Methods 1: Signal Processing

Title: High Resolution Delay Estimation in Urban GNSS Vehicular Navigation
Author(s): N. Sokhandan, A. Broumandan, J.T. Curran, G. Lachapelle, University of Calgary, Canada
Date/Time: Thursday, September 20, 2012, 9:20 a.m.
Room: 204 (NCC)

Multipath propagation causes major impairments to satellite based navigation systems and still remains the dominant source of accuracy degradation and a major issue for high precision Global Navigation Satellite System (GNSS) applications. Multipath results in biased GNSS measurements, which leads to inaccurate position estimates or even loss of lock on the signal. An accurate Line-Of-Sight (LOS) delay estimation in multipath environments is one of the prerequisites of GNSS receivers. There are several existing solutions for reducing the effects of multipath proposed in literature; however this topic is still under active research focus. The most commonly encountered delay measurement algorithms implemented in today´s commercial GNSS receivers are the classical delay tracking techniques that make use of special correlators (e.g., the narrow correlator and the double-delta correlators) and are based on feedback code tracking loops. However, the performance of these techniques in severe multipath scenarios is still rather limited. The reason is that in dense multipath environments, the Pseudo Noise (PN) correlation peak might be shifted due to interferences of the adjacent paths. Since the stable lock point of these tracking techniques is, generally, near the maximum power of the autocorrelation function, the errors by adjacent paths cannot be completely compensated for unless the chip duration of the PN sequence is shorter than the arrival time intervals between the first path and the subsequent paths. However, in urban propagation channels, the delay between paths can be very short (of the order of tens of ns), and the available signal bandwidth is limited in practice by few tens of MHz. The question, therefore, is how to improve the accuracy of the delay estimation in severe multipath scenarios (e.g. urban environments where the number of signal reflection could be larger than two and/or some of the signal reflections might be stronger than LOS). In response, this paper analyzes high resolution subspace based TOA estimation techniques in an effort to achieve a higher TOA estimation accuracy. These techniques estimate the multipath delays in two steps. At the first step, a low-resolution channel profile, e.g., a PN correlation profile (Bouchereau et al 2001) or a frequency response (Li & Pahlavan 2004), is obtained and used to compute the signal covariance matrix. Next, the resolution of the channel profile is enhanced by a high resolution technique, e.g. eigenvalue decomposition via Multiple Signal Classification (MUSIC) technique. This technique is based on the orthogonality of the signal and noise subspaces. A more precise TOA is thus determined from the first peak detected on the enhanced channel profile. This methodology may provide an improvement in the estimation accuracy under certain kinds of multipath conditions.

Subspace based methods require a full-rank signal covariance matrix, which exists if the LOS and the multipath reflections are uncorrelated. However, in many cases in practice, the rank of this matrix reduces to unity due to the signal coherency (Bouchereau et al 2001). Therefore, different techniques such as diversity reception have been employed in practice to combat signal coherency. Common diversity techniques include antenna diversity, time diversity, frequency diversity and polarization diversity. Diversity techniques take advantage of the random nature of the radio propagation channel by combining uncorrelated signal versions. Spatial, temporal and polarization diversities have been reported in the literature to be non-effective for the purpose of signal decorrelation (Li & Pahlavan 2004). The reason is that in time diversity, the path gain coefficients remain unchanged together with path delays and in spatial and polarization diversities, radio channels from the transmitter to different diversity branches of the receiver are most likely not the same. Bouchereau et al (2001) have applied frequency diversity to de-correlate multipath signals. However, since only a few of GPS satellites transmit both L1 and L2 signals, frequency diversity cannot be an effective solution for this problem in the context of satellite navigation. Furthermore, the presence of atmospheric errors decreases the usefulness of this diversity. On the other hand, a fast fading wireless channel, where the receiver or the surrounding objects are in motion, intrinsically provides another opportunity to combat the problem of signal coherency. This means that, the received signal consists of a linear combination of independent frequency shifted copies of the transmitted signal (Sadowsky & KafedZiski 1998). These independent copies of signal produced by the wireless channel provide an inherent mechanism (Sayeed & Aazhang 1999) that can be exploited for the purpose of signal decorrelation via appropriate signal processing. The goal of this paper is to use a framework to fully take advantage of this opportunity in urban vehicular navigation. In this way, a trade-off can be held between the coherent integration time of the receiver and the number of available signal copies depending on the speed of the vehicle. Herein, the channel is assumed to follow the Rician fading model with a few strong multipath components (and numerous weak reflections) which fits most of the typical urban environments (Steingass & Lehner 2004). This implies that the LOS is assumed to be present but may be weaker than some of the signal reflections. In this paper, the Doppler spectrum broadening (Doppler spread) of the fast fading channel resulting from the motion of the receiver is employed for the purpose of signal decorrelation in high-resolution estimation of multipath delays through the subspace technique. In other words, delay-domain correlator outputs in different Doppler frequencies will be combined to enhance the rank of the covariance matrix. Theoretical, simulations and real-life data processing results (collected in downtown Calgary) will be presented to compare the performance of the proposed method with the state of the art algorithms such double delta correlators. The performance metrics are based upon pseudorange and positioning errors. In all of the real data tests, the computed pseudoranges and positions by the proposed techniques will be compared to the reference trajectory achieved by a high precision Inertial Measurement Unit (IMU) namely SPAN LCI.

References [1] Sayeed A. M., and B. Aazhang (1999) "Joint Multipath-Doppler Diversity in Mobile Wireless Communications," IEEE Transactions on Communications, vol. 47, no.1, January, pp. 123-132 [2] Sadowsky, J. S., and V. KafedZiski (1998) "On the Correlation and Scattering Functions of the WSSUS Channel for Mobile Communications," IEEE Transactions on Vehicular Technology, vol 47, no 1, January, pp. 270- 282 [3] Li, X. and K. Pahlavan (2004) "Super-Resolution TOA Estimation with Diversity for Indoor Geolocation," IEEE Transactions on Wireless Communications, vol 3, no 1, January, pp. 224 - 234 [4] Bouchereau, F., D. Brady and C. Lanzl (2001) "Multipath Delay Estimation Using a Super Resolution PN-Correlation Method," IEEE Transactions on Signal Processing, vol 49, no 5, May, pp. 938-949 [5] Steingass, A. and A. Lehner (2004), "Measuring the navigation multipath channel-A statistical analysis". ION GNSS 2004, 21- 24 September, Long Beach, CA.

Previous Abstract Return to Session D3 Next Abstract