Towards an integrated approach to positioning

The quality and availability of positional information (location, orientation) is a key resource that strongly influences the way in which many systems incorporating spatial information can provide services to their user. In this extended abstract, we discuss a number of issues and ideas on how to cope with low-quality positional information, how to improve that quality and how integrating qualitative and quantitative approaches may be beneficial in this context. We present a number of examples from systems that we developed in the past and that we are currently developing, and we point out possible improvements. 1 Positioning techniques There is a large body of applications and systems that depend on knowledge about the location (and orientation) of their users and/or the position of the device used by them. These include (but are not limited to) mobile guides [3], car navigation systems [9], mobile phone based applications [8], geographic information systems [14] and ubiquitous computing environments [1]. There are various means for determining the current position of an entity. One of the most important ones is device based positioning, which is characterised by the fact that a (mobile) device itself determines its position albeit in conjunction with an existing positioning or communication infrastructure. Device based positioning methods can be divided into three major categories: • Measurement based positioning Measurement based positioning methods gather data from sensors and other positioning determination equipment and directly compute the device location. Perhaps the best known example of measurement based positioning is the Global Positioning System (GPS) (e. g. [11]). However, infrared beacons (e. g. [2]) and ultra-sound receivers (e. g. [1]) have also been used to the same effect. Furthermore, wireless communication networks such as GSM or 802.11 WLAN provide another means to determine the location of a mobile device (e. g. [3]). • Inference based positioning Inference based positioning methods perform reasoning to improve the quality and/or precision of location information derived with measurement based methods. The most important example of inference based positioning is dead reckoning. Most inference based methods combine direct measurements with knowledge about past device locations and current movement patterns to infer a device’s current location (cf. e. g. [10]). Inference based methods generate hypotheses: hypotheses are more or less reasonable but are not guaranteed to be correct. • Interactive positioning Interactive positioning is a method that uses an interactive dialogue between system and user to determine the position [7]. A confirmation dialogue is the most simple example for interactive positioning: the system asks the user to confirm whether they really are at the position that was computed. A more soph-