Jani Käppi

Bringing RTK to Cellular Terminals Using a LowCost Single-Frequency AGPS Receiver and Inertial Sensors

Today an ever-increasing number of handsets come equipped with a GPS receiver and some even with inertial sensors. Moreover, an even higher number of terminals are already capable of connecting to an add-on device with such capabilities. However, the full potential of these devices is not yet exploited. This paper introduces the mobile RTK (mRTK) solution, which can be included in the wireless standards to enable high-precision double-difference carrier phase positioning in handsets at no extra hardware cost. mRTK differs from the current OTF/RTK solutions in that it is a software-only solution using the hardware and wireless connections already existing in handsets. Moreover, the mRTK solution can utilize information from on-board inertial sensors. These are the key differentiating factors compared to the previous solutions. The paper shows that the sensors supplying information on baseline changes during the ambiguity initialization significantly assist the ambiguity resolution. A new communication protocol and messaging was defined in order to be able to exchange information between mRTK-capable handsets. The protocol includes reservations for additional GPS frequencies as well as for other Global Navigation Satellite Systems (GNSSs), such as Galileo. This protocol can be directly included in the wireless standards. Challenges in the current implementation include using only the L1 frequency for ambiguity resolution. Utilizing an L1-only receiver necessarily leads to penalties in the baseline accuracy due to inherent problems in the ambiguity resolution and validation. However, this paper shows that the baseline obtained is still better than the plain difference of positions. This paper shows that the mRTK solution significantly improves A-GPS performance. The mRTK solution also brings near-professional-quality positioning performance to the mass market. It would, therefore, be beneficial to include mRTK in wireless standards in order to expand A-GPS use cases in the short term and A-GNSS use cases in the long term.

Robust Misalignment Handling in Pedestrian Dead Reckoning

Modern mobile devices consists various sensors such as accelerometers and gyroscopes that can be used to aid and complement Global Navigation Satellite System position and velocity updates. Especially these sensors are now capable of providing good navigation solutions for pedestrians with competitive battery power consumption when compared to radio navigation systems. In order to achieve this energy efficiency, the navigation solutions could primarily be based on onboard inertial sensors alone, and additional aiding systems such as GNSS or indoor map used only when requested, such as to initialize the starting position. One of the most critical problems in such mobile device navigation solutions is the inability to accurately estimate the misalignment between the pedestrian walking direction and the mobile device orientation, eventually causing the navigation algorithm to output wrong location solution. In this paper we propose methods to detect such situations and means to reduce the errors due to the misalignment estimation error.

Semantic Labeling of User Location Context Based on Phone Usage Features

In mobile phones, the awareness of the user’s context allows services better tailored to the user’s needs. We propose a machine learning based method for semantic labeling that utilizes phone usage features to detect the user’s home, work, and other visited places. For place detection, we compare seven different classification methods. We organize the phone usage data based on periods of uninterrupted time that the user has been in a certain place. We consider three approaches to represent this data: visits, places, and cumulative samples. Our main contribution is semantic place labeling using a small set of privacy-preserving features and novel data representations suitable for resource constrained mobile devices. The contributions include (1) introduction of novel data representations including accumulation and averaging of the usage, (2) analysis of the effect of the data accumulation time on the accuracy of the place classification, (3) analysis of the confidence on the classification outcome, and (4) identification of the most relevant features obtained through feature selection methods. With a small set of privacy-preserving features and our data representations, we detect the user’s home and work with probability of 90% or better, and in 3-class problem the overall classification accuracy was 89% or better.