Optimizing route prior knowledge for map-aided fingerprint-based positioning systems

This paper investigates how positioning accuracy is affected in map-aided positioning systems, when a user's typical route is described by different probability distribution types. Probability distributions are introduced in an effort to better explain any reasonable route deviations from the user's center line of movement. The user route is assumed to be a-priori knowledge. Such knowledge can be extracted by utilizing information from environment maps and user mobility behaviour within the area of interest. In our research work, several probability distributions are tested along the center line of a user's route. The effect of the distribution width, radius ¿, on positioning accuracy is also investigated, by varying the value of ¿ for both sides of the route. In this way, the allocated weight probability for locations at the proximity of the user's center line route can be controlled. Analysis suggests that the introduction of statistical distributions to describe the users movement around a straight path, leads to better positioning results, compared with the simplified approach where no map constraints are imposed. Best results have been observed when the distance ratio distribution was used.