Supercorrelation: Enhancing accuracy, sensitivity of commercial receivers

January 16, 2019  - By
Figure 1: Reflected signals in the local environment suffer different Doppler variations than the desired line­of­sight signal. This means that the supercorrelator that is created for a given satellite broadcast couples strongly to the desired line of sight version  of the signal, but attenuates any reflected  signals  arriving from different directions.  (Figure: Focal Point Positioning)

Figure 1: Reflected signals in the local environment suffer different Doppler variations than the desired line­of­sight signal. This means that the supercorrelator that is created for a given satellite broadcast couples strongly to the desired line of sight version of the signal, but attenuates any reflected signals arriving from different directions. (Figure: Focal Point Positioning)

The S­GPS/S­GNSS technology is a patent-protected suite of methods that provides software-based improvements to existing GNSS receivers. All methods within the software suite build upon a core technology called supercorrelation, which enables over a second of coherent integration while undergoing complex motions on low-cost platforms. The benefit is high sensitivity coupled with strong multipath mitigation capabilities, providing a high-accuracy and high-integrity positioning solution in traditionally difficult environments.

Many GNSS receivers perform a small amount of coherent integration, typically less than 20 milliseconds, and then optionally incoherently integrate over many hundreds of milliseconds to boost sensitivity if needed. The major problem with this approach is the resulting susceptibility to multipath interference. Incoherent integration destroys the phase information stored within the captured data before combining it, resulting in line-of-sight and non-line-of-sight signals accumulating within the same correlation peak, producing a distortion of the desired line-of-sight information. This distortion leads to erroneous codephase estimates, which in turn leads to erroneous position estimates.

Coherent integration can decorrelate signals arriving from different directions, but the degree of decorrelation depends on the user speed and the coherent integration time. Supercorrelator technology creates a clock-and-motion-compensated phasor correction sequence that provides over a second of coherent integration on low-cost consumer platforms. The outcome is signal tracking sensitivities down to nearly zero dBHz, combined with high multipath mitigation performance. Such long coherent integration times allow signals arriving from different directions to be separated out in the frequency domain, permitting new capabilities in anti-spoofing and 3D map-aiding methods by directly resolving GNSS angle-of-arrival using a single moving antenna.

Figure 2: The result of supercorrelation on positioning performance in the urban canyons of central San Francisco. The red line is a standard state­-of-­the-­art vector tracking GPS solution, and the green line is the same positioning engine with supercorrelation processing enabled. (Image: Focal Point Positioning)

Figure 2: The result of supercorrelation on positioning performance in the urban canyons of central San Francisco. The red line is a standard state­-of-­the-­art vector tracking GPS solution, and the green line is the same positioning engine with supercorrelation processing enabled. (Image: Focal Point Positioning)

Traditionally, very long coherent integration times were not practical on consumer devices due to limitations of data modulation bits, crystal oscillator stability, and unknown (often complicated) receiver motion. Supercorrelation overcomes these limitations with signal processing and sensor fusion. Data modulation bits are not an issue for modern pilot signals, and for legacy signals they can be removed with a variety of methods, ranging from prediction or provision of the bits over a datalink, to stripping them directly with signal-squaring methods. Receiver motion can be inferred from inertial sensors mounted alongside the GNSS receiver, as is the case for smartphones and smartwatches, or can be modeled using multi-hypothesis methods. Low-cost crystal oscillators cause phase instabilities which traditionally reduce coherent integration time, but can also be accounted for by multi-hypothesis testing and by joint estimation processes across multiple channels.

A decade ago, consumer GNSS receivers were typically an ASIC or similar hard-wired design. Modern designs incorporate a front-end correlator bank which may or may not be reprogrammable, feeding into a DSP stage which handles all tracking and navigation processing from the DLL, PLL, FLL stages onwards. The flexibility of reprogramming the code running on the DSP stage permits existing GNSS chipsets to be easily upgraded to support supercorrelation, without needing to design and fabricate a new receiver.

Focal Point aims to have S-GNSS enabled chips by early 2020, with licensing opportunities available from summer 2019 onwards.