The Inadequacy of the Spectral Separation Coefficient and Aggregate Gain Factor for Quantifying the Effects of GPS C/A Code Self Interference

A.J. Van Dierendonck, R.J. Erlandson, K. Shallberg, S. Ericson

Abstract:	Some would call C/A Code self interference by tne name cross-correlation. Indeed, that is a more-correct characterization. Unlike most other GNSS codes, the C/A code is a short code (1023 chips) designed in the early 1970s as a short acquisition code for aiding acquisition of the much longer GPS P(Y)-Code. Since the P(Y)-Code is not available for civil applications (except in a semi-codeless mode), the C/A code has also became the standard code for civil applications. It comes with the short code disadvantages, but disadvantages that receiver designers have overcome to the point where the use of the GPS C/A code has been certified for aviation applications. One of these short code disadvantages, multipath errors, has been mitigated to the point that the C/A code now provides accuracy nearly equal to that of most longer modernized codes in receivers with similar bandwidths. However, C/A cross-correlation effects are still worse than those for more recently designed RNSS signals (GPS L1C and L5, Galileo, etc.) An analytical method has been derived to quantify the effects of self-interference through what is called a Spectral Separation Coefficient - (SSC) that works well when applying it to long codes, basically using spectral analysis. This analysis works well for these longer codes primarily because they have a near-continuous noise-like spectral density. Since the C/A codes have a modified "line-spectrum" and more severe cross-correlation properties, the spectral SSC method does not work so well, so a technique using time-domain derivation of SSC has been used instead has been proposed. Unfortunately, C/A code self-noise is miss-named "noise", even though under certain conditions, it can be noise-like if receiver pre-detection intervals are short. Cross-correlation is a better description because it is mostly not noise-like much of the time. It somewhat has the characteristics of multipath resulting from code-correlation peak distortion, and is modeled more appropriately as signal degradation (like multipath fading) than as noise. In fact, if multipath mitigation is used in the receiver, the mitigation tends to reduce the effects of cross-correlation in the receiver as well, and most aviation receivers have some form of multipath mitigation that reduces fading as well as improving code-tracking performance. This was shown in our ION-GNSS-2002 paper. Also shown in that paper is that, at low C/N0 conditions associated with other (specified) external interference, the effects of cross-correlation are minimized because the C/A code self interference is dominated by the specified external interference. Aviation receivers are designed to operate and meet performance requirements in that external interference environment. However, the cross-corellation effects do show up at higher C/N0, but that is a "don't-care" situation. Even with fading at those higher C/N0 conditions, the receiver easily meets performance requirements. The problem is that when the time-domain SSCs are computed without the noise the results appear to be much worse than they would be at lower C/N0 when it really matters. In our 1999 ION-GNSS paper, we presented derivations of expected receiver performance when subjected to C/A-code self interference. In our 2002 ION-GNSS paper, we presented both software simulations and RF hardware test results describing the effects of GPS C/A Code self interence. We did not compute Spectral Separation Coefficients (SSCs) associated with those results. Our intent is to do that in this paper. Since we presented the 1999 and 2002 papers, another concept was developed for evaluating the impact of C/A-code self interference by RTCA, Inc. for aviation. This concept, documented in RTCA DO-235B, only evaluates the impact of the C/A-code interference on critcal C/A signals required to complete the aviation mission at hand with the required integrity -- such as a precision landing operation using WAAS or LAAS. Essentially, the interference is not deemed critical if the mission can be completed (with the required integriy) without the availability of the satellite C/A signal that may be lost (loss of lock) due to interference. Of course, this can potentially happen only to signals from low-elevation satellites, Usually, there are enough visible satellites providing signal power that this happens with an acceptable very low probability, or, in the case of expanded GPS constellations, not at all, even accounting for one or two failed satellites. This concept works quite well, except that a lot of computation is required because possible losses of signal have to be evaluated. In other more recent papers or presentations on the effects of C/A on C/A interference, this critical satellite concept has been recognized, and the fact that with the expanded 30 to 36 satellite constellations, there are no "critical" satellites. Thus, the impact of the interference has been evaluated without low-elevation satellites, taking advantage of increased received power. We call this a "poor man's" critical satellite evaluation, but it is acceptable under expanded constellation scenarios, and, of course, doesn't require a lot of computation. There a problems with all the above approaches in evaluating the implact of C/A code self-interference -- either they require an unreasonable amount of computation (and, and in the case of the time-domain SSC approach, an associated amount of understanding), they are overly conservative, or they require an expanded GPS constellation, which, of course, cannot be guaranteed. The only past publications that included test results and detailed receiver simulation results were our ION-GNSS 1999 and 2002 papers. Thus, the purpose of this paper is to use the results of those receiver simulations to modify the application of C/A code SSCs to represent the results of those tests and simulations in environments where there are other external interferences, such as those imposed upon aviation certified receivers. In the end, there should be no reason why the same application of modified SSCs cannot be applied to other non-aviation GPS applications using the C/A code.
Published in:	Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013) September 16 - 20, 2013 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	1435 - 1444
Cite this article:	Van Dierendonck, A.J., Erlandson, R.J., Shallberg, K., Ericson, S., "The Inadequacy of the Spectral Separation Coefficient and Aggregate Gain Factor for Quantifying the Effects of GPS C/A Code Self Interference," Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013), Nashville, TN, September 2013, pp. 1435-1444.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In