Return to Session B3

Session B3: Future Trends in GNSS Augmentation Systems

Light Integrity Algorithm Considering Local Environmental Feared Events
Yuki Sato, Rui Hirokawa, Masao Higuchi, Akinori Taira, Mitsubishi Electric Corporation

We propose a novel integrity algorithm to compute protection level primarily for urban navigation. A nonlinear optimization problem is derived from conventional RAIM and solved just once per epoch. The nonlinear optimization problem is solved for a bias vector, where lower and upper bounds of each component of the vector are given as constraints. The magnitude of each feared event is simply expressed by the bounds of corresponding component. Therefore, this algorithm does not require distinction of faulty/unfaulty states and measurements, as is usually required in other algorithms such as Solution-Separation. Also, this algorithm does not limit the number of assumed fault modes. It avoids combinatorial increase of computational load, as it is usually problematic in general RAIM algorithm with increasing number of fault modes. Also, this algorithm does not require dedicated selection of measurement for rejection.
The upper and lower bounds of bias vector can be modeled or be given externally. This algorithm can be applied not only to snapshot positioning using least-squares method, but also to positioning using Kalman filter. Application to Kalman filter is the most practical and only requires few changes in matrices used in the nonlinear optimization problem. In this case, the bias vector expresses not only bias error in the measurement at a certain epoch, but also bias error of prior states caused by faulty measurement in the previous epochs. We suppose this light integrity algorithm is suitable for urban navigation where local environmental feared events frequently occur, and their effects propagate in the filter states. For evaluation, we processed dataset we took with low cost antenna and receiver. The course includes highway and urban areas. Dual frequency pseudo-range, carrier-phase and Doppler measurements of GPS, Galileo and QZSS were used together with correction data provided by QZSS-L6 CLAS. Horizontal positioning error and horizontal protection level were computed for each epoch. Epochs when test statistics exceed the threshold or when the number of measurements were fewer than five were excluded. In the integrity algorithm, we set the bounds of bias error for ambiguity states and pseudo-range measurement. These bounds were provided based on experience. In this case, fault fixing of the integer ambiguity and environmental bias error in the measurement after correction were considered. Probability of false alarm and missed detection were set to 10-4 and 10-5, respectively. Stanford plot showed that this algorithm can be applied to high-integrity application at required risk of 10-5.
One important remark is that this algorithm is also suitable for navigation aided by local location server. The bounds of bias error in the measurement cause by local environmental feared events can be disseminated by the server as a part of integrity assistance data being standardized in 3GPP and RTCM. Given an approximate position of user, the accurate bounds are disseminated by the server, which usually has better knowledge about the environment. The server may utilize big data obtained in its local service area or compute the bounds by using a three-dimensional building map or placement information of signal interference source. Those assistance data address the local environmental feared events. Together with other integrity assistance data for satellite and atmosphere feared events, the user can compute appropriate protection level in various kind of environments.



Return to Session B3