Quality of Location: Estimation, system integration and application

Accurate location measurement is an important research topic in pervasive computing systems and applications. To achieve high performance measurements, the knowledge of the quality of a measurement, a sensor cue, or an inferred location value is required. This paper presents a novel approach in the deliverance of an independent, unified quality of location (QoL) value for location systems. The proposed approach is highly flexible, independent of technology and location inference mechanisms and approaches, integrable into any existing location system, and does neither require knowledge of sensor, nor of application characteristics. The paper proposes both a method to retrieve QoL for a given system, and shows its application in a setting using a simple ultrasound location system. Retrieving QoL requires a multi step process including a unsupervised subtractive clustering method for initial learning, and a supervised network based fuzzy inference systems (ANFIS) for refinement of the parameters. The approach described can be used in settings using heterogeneous systems, devices, and sensors. It is also usable at any abstraction layer and is able to run on small sensor node devices. Technical foundations of the algorithms are an adaptive network based fuzzy inference systems (ANFIS). In this paper we will show the technical principles, its application and evaluate the performance of the system.