4DesignThis chapter introduces the different concepts that were needed for the realization ofthe localization framework. The core components, their use cases and their individ-ual relationships are presented. The various communication channels that connectthese components are described and the motivations for the individual architecturaldecisions are explained.4.1 General OverviewThe localization framework is build around a central server instance that communi-cates with different consumers for various purposes over HTTP. The different con-sumers are composed of mobile devices, browsers, GPU-nodes and fat clients. Thedifferent purposes for these communications are localization, measuring, evaluat-ing of results and distributing computational hard problems. The Server combinesfive major components for approaching the topics of this thesis. These componentsconsists of:1. The Optimizer, the component responsible for training the radio propagationmodel through searching for optimal material parameter solutions by compar-ing raytracer results with a given set of initial measurements.2. The Simulator distributes raytracing jobs to the available GPU-nodes andcollects the processing results. Either in the context of an Optimizer run orfor manual inspection of the radio propagation results for the different APs anddifferent material parameters in a debugging environment like the Fat Client.3. The Localizer component represents the HMM and PF driven algorithms withthe associated environment needed for positioning and tracking of mobile de-vices. Therefore, its also used by the Evaluator for evaluating offline recordedmeasurements annotated with tracked positions.