ION GNSS 2012
Session E4: Software Receivers

Title: Assisted-GPS Based Snap-shot GPS Receiver with FFT-accelerated Collective Detection: Time Synchronisation and Search Space Analysis
Author(s): J.W. Cheong, J. Wu, A.G. Dempster, C. Rizos, The University of New South Wales, Australia
Date/Time: Thursday, September 20, 2012, 4:00 p.m.
Room: Grand Ballroom West (Renaissance)

In this paper, the architecture of the Assisted-GPS based Snap-shot GPS receiver is introduced. Such a system would be suitable for emergency services, freeing up the processing burden of the client handset receiver while a novel ´search engine´ sub-system handles a "snap-shot" of transferred raw radio-frequency data at an A-GPS server. This system can be treated as an advanced hybrid technology that comprises the concepts of Assisted-GPS, Snap-shot GPS receiver and Collective Detection.

1.1 Assisted-GPS

Assisted-GPS (A-GPS) improves upon a conventional GPS receiver by significantly reducing the Time to First Fix (TTFF), which is mainly comprised of satellite acquisition time and satellite data download - which can potentially take 30 seconds (one frame) to acquire the Time Of Week (TOW) and ephemeris data. Typically, the receiver has an A-GPS chipset integrated within a network-connected device (e.g. a cellular phone, WiFi device, etc.) The A-GPS assistance data contains the satellite ephemeris provided by a nearby cellular network. In addition, the Mobile Station (MS) is time-synchronised to the cellular network (or the Base Station, i.e. BS). Hence, a MS-based A-GPS receiver can compute its position once it has its code phase measurements, obtaining other relevant information from the A-GPS assistance data.

In the acquisition stage, A-GPS improves its TTFF over conventional GPS by reducing its search space by accounting for A-GPS assistance data provided via a network connection. Typical assistance data for MS-based A-GPS include the satellite transmit time (accuracy of ñ2 seconds), reference frequency correction (accuracy of 50ppm), nearby cell-tower location (typically within 3km radius) and up-to-date almanac and/or ephemeris data. By knowing its approximate location, the number of satellites and Doppler frequency bins to search can be significantly reduced. Hence, the overall reduction in search space can be used either to increase the receiver´s dwell time such that its sensitivity can be increased or to reduce its TTFF [1].

1.2 Snapshot Receiver

In a Snap-shot receiver a short period of signal (a few milliseconds (ms) of data) is collected at the MS. The raw Intermediate Frequency (IF) data is then processed either on-chip by the MS or off-chip by a server. Conventional signal processing methods (i.e. serial acquisition and then tracking) are not suitable for dealing with such a short signal sequence. Solving for 1ms ambiguity (coarse time) for position estimation is also not addressed in conventional methods [2]. With the help of A-GPS assistance data - which provides a priori information such as timing, satellite ephemeris, and approximate geographical position - the 1ms ambiguity can be resolved as discussed in [3, 4].

Briefly, the positioning technique for A-GPS via a snapshot receiver can be broken down into four steps. The first step computes the a priori position-clock vector (i.e. position vector augmented with Common Clock Bias) from the assistance data. Secondly, the full pseudoranges are predicted based on the a priori state. Then the fractional pseudoranges (derived from code delay measurements) are used to reconstruct the measured full pseudoranges. The a priori state is finally updated via a Least Squares (LS) procedure to reflect the difference between the measured full pseudorange and the a priori full pseudorange. This is iterated multiple times from the second step onwards, essentially performing iterative Least Squares (ILS) process to compute the correct position.

1.3 Collective Detection

Collective Detection is a GPS receiver signal processing technique that combines the correlator output of all satellite channels and projects them onto the position-clock space to enhance the overall signal detection probability, given that a priori knowledge of satellite ephemeris, and approximate user location is known to the receiver. This is an enhancement to pre-existing GPS signal processing techniques, where the gains acquired from Collective Detection can be leveraged on longer coherent/non-coherent integration periods, making it viable even at C/No of 20dB-Hz. On top of achieving higher sensitivity and higher detection rates for weak signals, Collective Detection is also shown to be able estimate the receiver position prior to entering tracking loops. This is a useful unique feature of collective detection that makes feasible the implementation of the Snap-shot signal processing algorithms on the search engine for position estimation despite having relatively short RF data captured by the Snap-shot receiver front-end.

Aims and Contribution:

Perfect a priori knowledge of the ephemeris reference time is usually assumed for conceptual investigation of Collective Detection. This is especially fallacious for the common clock bias domain, which has yet to be solved realistically using fine step sizes and complete uncertainty. It is the aim of this research to investigate a practical solution to improve the system functionality particularly by taking into account this random clock bias. An efficient FFT-accelerated matched filter is proposed to solve for the common clock bias. The use of FFT-based pre-processing improves the efficiency but degrades the accuracy due to the quantisation of floating-point code phases.

Methodologies:

An A-GPS-based CD SDR platform has been developed in MATLAB by the authors to process the IF signals from both BS and MS. One of the contributions of this paper is the proposed acceleration algorithm for the processing of CD using a FFT-accelerated matched filter to especially solve for the vast Common Clock Bias (CCB) search range. The practical implementation is then used to produce investigation results of CD in real-life non-ideal scenarios. The use of an FFT-accelerated matched filter requires quantisation of the floating-point code phase, which results in a PDPC containing undesired artefacts. This is unavoidable but discussed in detail in this paper. Specifically, the resulting position accuracy and precision of non-ideal time synchronisation at the MS is shown. Also, the effects of the usage of various spatial step sizes are shown. Finally a simple method to improve upon a computationally efficient implementation of CD is proposed to produce results with better precision. Unlike many previous research contributions, in all investigations, the entire CCB search range (i.e. 0-300km) is considered in the CD search space. Findings:

There is a trade-off between precision and computation load regarding step size selection, which is characterised in this paper. Live signals are used in real-life non-ideal scenarios for performance evaluations. The associated accuracy and precision are empirically characterised as a function of coarse time error and position-clock domain step size. A simple averaging approach is proven to improve precision using limited computational resources, regardless of its large uncertainty region.

Keyword: Assisted-GPS, Snap-shot Receiver, Collective Detection, Time Synchronization, Clock Bias

[1]. Van Diggelen, F.S.T., A-GPS: assisted GPS, GNSS, and SBAS. 2009: Artech House Publishers [2]. Kaplan, E. and Hegarty, C., Understanding GPS: Principles and Applications. 2006: Artech House [3]. Hein, G., et al., Platforms for a future GNSS Receiver. Inside GNSS, 2006, 1(2), 56-62. [4]. Bisiules, P.J. and Yang, C., Dualband base station antenna using ring antenna elements. 2009, EP Patent 2,051,331.

Previous Abstract Return to Session E4 Next Abstract