Previous Abstract Return to Session C6 Next Abstract

Session C6: Harsh Urban and Indoor GNSS

Expanding High Accuracy Positioning Through the Mitigation of Influence Factors in Highly Urbanized Areas
A. Chamorro, J.C. Lopez, J. Mate, E. Carbonell, A. Gonzalez, D. Calle, GMV
Alternate Number 2

It is well known that GNSS high accuracy solutions find some of the most challenging environmental conditions for optimal accuracy and reliability in highly urbanized areas. Multiple factors can have a negative impact in GNSS/INS integrated navigation, adding errors to the typical behavior of RF signals and complexity into the input sensor real time processing. Thorough knowledge of the influence factors taking part on these challenging environments is key to plan and implement proper mitigations enabling to further expand the areas of applicability of High Accuracy solutions.
The presence of buildings or other bulk infrastructures typically mask satellites with a low elevation, leaving the user with a reduced input raw data stream. Data processing of different constellations and frequencies help diversify the source of different observations to ultimately increase the number of satellites in view for a given user. Additionally, report trends observed in the receiver markets are showing the increased interest into GPS L5 and Galileo E5a (transmitting in a similar frequency band) complementing the tracking of E1/L1 frequencies. One of the main important reasons for this selection is the higher power level and more robust signal modulations which should provide the user with less noisy and more available measurements.
On the other hand, sensor fusion algorithms are considered one of the strategies with a higher potential to improve dead reckoning and other poor satellite visibility situations. By correctly weighting external measurements, it is expected to achieve better accuracies and higher availability. On the other hand, the usage of different sensors for an absolute positioning solution provides a great advantage as well in terms of integrity, as the different failure modes of the sensors can in general be combined into smaller error bounds for the same target integrity risks.
Typical scenarios recorded in urban and suburban scenarios typically present a much higher level of cycle slips (due to natural trajectories cutting satellite lines of sight, traffic lights, signals, road and city topology), strongly affecting high accuracy carrier-based positioning algorithms. This aspect is even more challenging when it is combined with the feature of performing single point positioning or positioning not based in dual frequency measurements. One of the main strategies to deal with the erratic behaviour of the signal tracking in challenging scenarios is to reduce dependencies on the presence of a minimum quality or number frequencies, by allowing the processing of whatever signal is available (SF, one constellation) at a given moment in time.
Last, but not least, GNSS signals may be reflected by buildings, walls, vehicles and the ground. Glass, metal, and wet surfaces can be particularly strong reflectors which can lead to high positioning errors, as the onboard receivers can track the same signal via multiple paths (Line Of Sight Multipath) or even only the reflected signal (Non-Line Of Sight Multipath). Detection of this phenomenon is not evident and is one of the biggest issues currently preventing reliable processing of GNSS signals into these challenging environments. Several strategies to improve current multipath detection are presented. One of them is to better classify different correlated effects present in Multipath signals with the support of AI-based strategies. On the other hand, a well-known factor present in MP scenarios is a higher perceived noise in the pseudorange measurements. Fine tuning of the existing mechanisms monitoring these effects will also be described, as well as the implementation of additional monitors based in the comparison of concurrent PVT. This philosophy inspired in ARAIM algorithms for EGNOS in aviation domains is not typically a widespread option in the automotive domain due to the required computational resources. However, similar lightweight checks are presented, which can be useful in the detection and rejection of harmful input data.

Previous Abstract Return to Session C6 Next Abstract