Experimental evaluation of the spatial error distribution of indoor localization algorithms

As the research in indoor localization moves on, more and more algorithms and systems are proposed by academia as well as industry. However, most approaches are evaluated in simulation or small experiments. The results are often hard to compare as assumptions differ between all settings and the setups vary from lab conditions to real life scenarios. As one of the biggest problems of range based indoor localization are large errors due to signal reflection and other effects, several resilient lateration algorithms have been proposed. To evaluate a selection of algorithms we conducted a large experiment in a real life scenario. We used 25 anchor nodes and a mobile node installed on top of a robotic reference system to collect ranging values of an office floor. We show that there are significant differences between the results of our experiments and the results the authors showed in their original publications. We discuss what reasons cause algorithms to perform less optimally in a real world scenario and suggest important points to consider when modeling an indoor localization scenario for a simulation or designing a range based indoor localization algorithm. We further examine the influence of the anchor placement and density on ranging errors.