Many location-aware applications rely on data from fine-grained location systems. During deployment such systems require a survey, specifying the locations of their environment-based components. Most current surveying methods are time-consuming, and require costly and bulky equipment. This paper presents the concept of self-surveying, i.e. methods by which a location system can survey itself. Such methods are user-friendly, fast, and require little or no extra equipment. Experimental results show self-survey accuracies comparable to the accuracy of the underlying location system.
A close collaboration with Intel Research Cambridge, this is a study of methods for auto-calibration in fine-grained location systems. Such systems require accurate, labour-intensive surveys of fixed infrastructure, a prohibitive barrier to their deployment. Using the renowned ""Bat"" location system at Cambridge, exhaustive experiments were performed in five rooms, characterising three auto-calibration methods. Survey accuracy (ranging 3-25 cm) was shown to be directly related to the obtrusiveness/sophistication of the data-gathering. This is one of the leading papers on auto-calibration for ubiquitous localisation. It has resulted in ongoing Lancaster PhD work, jointly funded by Intel Research. (acceptance rate = 17.6%) RAE_import_type : Conference contribution RAE_uoa_type : Computer Science and Informatics
