KOTOHIRAGU NAVIGATOR: An Open Experiment of Location-Aware Service for Popular Mobile Phones

Hiroyuki Tarumi, Yuko Tsurumi, Kazuya Matsubara, Yusuke Hayashi, Yuki Mizukubo, Makoto Yoshida, Fusako Kusunoki: KOTOHIRAGU NAVIGATOR: An Open Experiment of Location-Aware Service for Popular Mobile Phones. LoCA 2006: 48-63

A fuzzy paper on the design and deployment of a mobile location-based sightseeing system that mixed the real environment with virtual creatures. The outcome that this system is better accepted by young people and that there is a generation gap towards it is not really convincing (I see a strong bias on the interface). Anyway, the authors mention at several occasions the impact of location accuracy on their system and the users:

Another problem was the GPS inaccuracies. As we just used the raw location data obtained from the gpsOne system, we sometimes had GPS errors of more than 10m, which confused users to find agents or buildings in the virtual world.

Therefore they design a GPS error compensation (unfortunately not described in details)
After the January’s experiment, we conducted another experiment recruiting ten student subjects. We developed five location compensation algorithms and input real location data obtained by popular GPS-phones to each algorithm. Using modified location data output from each algorithm, simulated virtual scenes were computed and shown to the subjects. The algorithms were map-matching, moving-average, avoiding big jumps, etc. We have found that we need the strongest algorithm, map-matching, for our purpose of virtual world navigation.
[...]
Despite the imperfectness of compensation, we can still say that the service quality has been much improved compared to the January’s system.

On of their outcome is to suggest that “Subjects are always requiring more accurate location-based system”. However:

In order to understand the bad effects of GPS inaccuracy on the service, we have calculated correlation coefficients of the evaluation data. All absolute values of coefficients are less than 0.4. This shows that subjects recognized the GPS inaccuracy as an independent problem from the system’s value.

On a light note, the authors provide a picture of the “worst place for GPS”

Relation to my thesis: GPS inaccuracy remained a problem even after the integration of a GPS error compensation mechanism. The user recognized it as a different problem, not part of the information service quality.