The Bartlett Centre for Advanced Spatial Analysis


Crowd-Sourced Global Navigation Satellite Systems (GNSS) Data for Indoor Positioning

Summary: The Global Navigation Satellite Systems (GNSS), e.g. GPS, are main positioning technology for many Location-Based Service (LBS) applications, such as navigation and emergency services. While existing indoor positioning services, e.g. those based on Wi-Fi and Bluetooth, do not provide a free, accurate, continuous, reliable and privacy-preserving or “GPS-like” indoor positioning service, inside the buildings, where people spend most of their time, GNSS signals can be blocked, attenuated or reflected, making indoor positioning unreliable or impossible.

This research proposes two novel techniques to provide a seamless (indoor/outdoor) GNSS-only positioning service. The first technique combines measurements over a short period of time, or at close-enough locations, to compute the user’s position. It will be used when fewer than four satellites (absolute minimum requirement of GNSS-based positioning), are available at a particular time and location. The second technique, GNSS fingerprinting, extracts the spatio-temporal patterns from the signals, e.g. the obscuration/unavailability patterns of the satellites or the signal attenuation patterns, from historical GNSS signals stored over time (potentially contributed by public and volunteers), orbital data, 3D model of city (i.e. obstructions and barriers). In the positioning mode, the newly received signals are matched with the stored database and the most likely location is found.

These step-changing techniques will enable: -researchers in many areas, e.g. intelligent mobility and smart city, to apply/extend the project’s concepts and techniques, -wide public participation in research, -LBS (e.g. navigation, tracking, emergency, security, and special assistance services) to provide continuous and reliable services, enhancing quality of, and potentially saving, lives.

Workflow/Deliverables:This project proposes two novel techniques to provide seamless (indoor-outdoor), continuous, free-to use, privacy preserving GNSS-based positioning services, requiring no further infrastructure or mobile device modification. This can be used by many LBS applications, including life-saving emergency and security services. GNSS positioning in difficult environments, i.e. urban canyons and indoors, suffers from several types of error, including multipath, non-line-of-sight (NLOS) signals, signal attenuation and signal blockage. GNSS signals can be reflected by  urfaces of objects (NLOS) or blocked by objects e.g. buildings and trees, or attenuated with respect to distance travelled through an object or medium, e.g. windows, (signal attenuation). The reflected GNSS signals can interfere with reception of the signals received directly from the satellites (multipath). Since GNSS positioning is based on ranging measurements (i.e. time taken for the signal to get to the receiver from the satellite), NLOS, multipath and signal attenuation can all cause positioning errors. Blockage of GNSS signals may result in a lack of availability of the minimum of four satellites in-view ((absolute minimum requirement of GNSS-based positioning) and consequently lead to a failure in the continuity of the positioning service.

This project proposes two techniques, based on the concepts of a virtual spatial diversity antenna and GNSS signal fingerprinting. The first technique combines the raw GNSS observations (which have been made accessible recently on mobile devices running Android 7 and higher operating systems) over a short period of time, or at close-enough locations, to compute the user’s position. It will be used when fewer than four satellites, are available at a particular time and location. This enables any currently available mobile devices to calculate position, by having more observations over a longer period of time or from another location that is close enough (depend on applications and scenarios) to be considered as one single point. Each epoch (set of observations) adds one more unknown, its clock offset; therefore in total at least n+3 observations are required, where "n" is number of epochs. As the measurements do not have to be made at the same time, it can be applied in a collaborative or crowd-sourced scheme, where spatially close users can share measurements and localise themselves. Crowd-sourced-gathered datasets and collaborative positioning schemes can improve the quality and availability of input datasets.

Secondly, GNSS fingerprinting, extracts the spatio-temporal patterns from the raw measurements, e.g. the obscuration patterns of the satellites, and the signal attenuation. These patterns are extracted from historical GNSS signals stored over time or contributed by the crowd, orbital data, 3D model of city, e.g. obstructions and barriers at a high level of detail (e.g. LoD4). In the positioning mode, the newly received signals are matched with the stored database and the most likely location is found.

Project Workflow Diagram

Dr Ana Basiri

View Ana's profile
Send Ana an email