ION GNSS 2009
Session F1: Galileo Signal Structure, GPS/Galileo Interoperability

Title: Cellular-Aided Static Indoor GPS using a Vector Tracking Architecture: Implementation, Analysis, and Applications
Author(s): K.M. Pesyna Jr., J.A. Bhatti, and T.E. Humphreys, The University of Texas at Austin
Room: Room 102/103/104

A technique is presented that exploits vector tracking and ambient cellular CDMA signals to enable static indoor GPS-based positioning and timing. The technique is targeted to applications such as self-localizing and self-synchronizing femtocells and Wi-Fi access points, tracking of personal valuables, and even positioning of mobile devices when these can be assumed to remain stationary for several minutes at a time. In each of these applications, the GPS receiver is assumed to be referenced to a low-cost oscillator such as a temperature-compensated crystal oscillator (TCXO). Despite the unpredictable and relatively large frequency variations in such oscillators, coherent integration intervals of several seconds at the GPS frequencies are made possible because the vector tracking architecture fuses GPS signals with cellular CDMA signals, which are known to have excellent frequency stability [1,2]. The assumption of stationarity obviates the need to compensate for receiver motion during the coherent integration interval. Acquisition and tracking of severely attenuated GPS signals having carrier-to-noise ratios as low as 7 dB-Hz is demonstrated. It is further demonstrated that the other challenges of indoor use of GPS - multipath and near-far effects [3] - are substantially mitigated by the vector tracking architecture and by changing satellite geometry.

The applications for static indoor GPS are numerous. For example, consider cellular networks. To satisfy increasing demand for fast and reliable wireless communication, the trend in cellular networks is toward heterogeneous networks which combine traditional macro base stations with so-called pico and femtocell stations. Femtocells will be located inside of households and businesses, and, due to requirements put forth by the FCC and by the cellular standards, will need to self-locate and self-synchronize to meter- and microsecond-level accuracies, respectively [4]. The ability to acquire and track GPS signals indoors would enable femtocells to use GPS to meet these stringent requirements. Wi-Fi access points would also benefit from a static indoor GPS capability. As their ability to self-locate and self-synchronize improves, they effectively become positioning beacons, providing quick and accurate position information to connected mobile devices. Additionally, indoor GPS would enable tracking of personal valuables which become misplaced or stolen. The unique characteristic of all these applications is that they focus on the use of GPS in static devices. The assumption of stationarity allows the GPS receiver to perform long coherent integration without the added complexity of compensating for its own motion.

There are three primary challenges in indoor GPS: multipath, cross-correlation false alarms caused by the near-far effect, and signal attenuation. It has been demonstrated, in prior work by Thomas Pany, et al. [5], that these challenges can be overcome through the use of a vector tracking architecture employing long-coherent integration. However, to extend the coherent integration interval to several seconds requires knowledge of the user motion and receiver clock drift. The system presented in [5] met these requirements by employing an expensive tactical grade intertial measurement unit (IMU) and a highly stable oven-controlled crystal oscillator (OCXO). The solution presented in this paper eliminates both of these requirements. First, the need for an IMU is eliminated by the assumption that the receiver is stationary. Second, by exploiting the frequency stability of cellular CDMA signals, the OCXO can instead be replaced by a less-stable inexpensive temperature-controlled crystal oscillator (TCXO). To compensate for the significant frequency variations of the TCXO, the CDMA signals, which have been shown to have excellent frequency stability [2], are coupled into the receiver´s vector tracking architecture, allowing these errors to be estimated and automatically mitigated. This leads to a low-cost solution that can fit into a smaller package than that of [5], which required both expensive equipment (an IMU and an OCXO) and a large backpack unit to harness all of the equipment together. This low-cost smaller solution will be more appealing to manufacturers of the devices discussed earlier which exploit GPS signals for indoor positioning and timing.

Cellular-aided static indoor GPS acquisition and tracking will be accomplished using the Generalized Radionavigation Interfusion Device (GRID) software receiver [6]. The receiver will be modified from its original scalar tracking loop architecture in which signals are tracked independently, to a vector tracking loop architecture in which tracking of signals is performed jointly and information is shared across channels through the use of a Kalman filter. The Kalman filter ingests all signal observables from a long-coherent integration interval, in particular pseudoranges and carrier phases, and uses this information to compute position and time corrections. These corrections are then used to dictate to each correlator exactly what parameters to use for the next integration interval. Within this architecture, CDMA signals are treated just like GPS signals. CDMA observables are ingested into the Kalman filter, and, because they are derived from a stable frequency and time source [2], are used by the filter to estimate more accurately the receiver´s position and clock error than could be provided on the basis of GPS acting alone. The improved position and receiver clock error estimates are then used to help track weaker GPS signals. In addition to their tracking benefits, the accurate time and frequency information provided by the CDMA signals will also assist the receiver in narrowing its time-frequency search space during GPS signal acquisition. A reduced search space allows the receiver to perform faster GPS signal acquisition with a higher probability of signal detection under a desired probability of false alarm.

The Kalman filter implemented in the GRID receiver as part of the vector tracking architecture will have two unique characteristics. First, the filter will be of square-root information form, also known as the square-root information filter (SRIF) [7]. The SRIF is more numerically stable and thus is better suited for a finite-precision software-based implementation than the conventional Kalman filter. Second, the filter will also have a backward smoother [8]. Backward smoothing is important because it removes abrupt dynamics introduced by innovations during the forward-pass filtering. These abrupt dynamics, if left unsmoothed, would violate the filter´s state dynamics model, even when this model is an accurate reflection of reality [9]. The receiver clock offset is one state element of the filter for which smoothing is particularly critical. An accurate estimate of the clock offset´s time history is needed to perform long coherent integrations.

References:

[1] Lashley, M. and Bevly, D., "What are vector tracking loops, and what are their benefits and drawbacks?" GNSS Solutions Column, Inside GNSS , Vol. 4, No. 3, 2009, pp. 16-21.

[2] Wesson, K., Pesyna, K., Bhatti, J., and Humphreys, T. E., "Opportunistic Frequency Stability Transfer for Extending the Coherence Time of GNSS Receiver Clocks," Proceedings of the ION GNSS Meeting, Institute of Navigation, Portland, Oregon, 2010.

[3] Seco-Granados, G., Lopez-Salcedo, J., Jimenez-Banos, D., and Lopez-Risueno, G., "Challenges in Indoor Global Navigation Satellite Systems: Unveiling its core features in signal processing," Signal Processing Magazine, IEEE , Vol. 29, No. 2, march 2012, pp. 108 -131.

[4] Pesyna, Jr., K. M., Wesson, K., Heath, Jr., R. W., and Humphreys, T. E., "Extending the Reach of GPS-assisted Femtocell Synchronization and Localization Through Tightly-Coupled Opportunistic Navigation," GLOBECOM Workshops (GC Wkshps), 2011 IEEE , 2011.

[5] Thomas Pany, Bernhard Riedl, J. W. T. W. R. S., "Coherent Integration Time: The Longer, the Better," Inside GNSS , Vol. 4, No. 6, November/December 2009, pp. 52-61.

[6] Humphreys, T. E., Ledvina, B. M., Psiaki, M. L., and Kintner, Jr., P. M., "GNSS Receiver Implementation on a DSP: Status, Challenges, and Prospects," Proceedings of the ION GNSS Meeting, Institute of Navigation, Fort Worth, TX, 2006.

[7] Bierman, G., Belzer, M., Vandergraft, J., and Porter, D., "Maximum likelihood estimation using square root information filters," Automatic Control, IEEE Transactions on, Vol. 35, No. 12, 1990, pp. 1293-1298.

[8] Psiaki, M., "Backward-smoothing extended Kalman filter," Journal of guidance, control, and dynamics, Vol. 28, No. 5, 2005, pp. 885-894.

[9] Pesyna, K., Kassas, Z., Bhatti, J., and Humphreys, T. E., "Tightly-Coupled Opportunistic Navigation for Deep Urban and Indoor Positioning," Proceedings of the ION GNSS 2011 , 2011.

Return to Session F1