3Related WorkThis chapter gives an overview over related research that has been conducted oneither modelling the radio propagation of AP transmitters in section 3.1 and overlocalization algorithms exploiting this information in section 3.2. Since the largerfocus of this thesis was placed on the localization algorithms, especially the HMMand PF approach, the latter section is formed more broadly.An historically important project, especially for RSSI based indoor localization sys-tems, is given by the RADAR[2] framework. This framework combines the basicmethodology, the appliance of a radio propagation model(the WAF model) to alocalization algorithm(specifically the LMSE), that was also used in this thesis andother related works.3.1 Radio PropagationThe two major approaches for the radio propagation model are either manuallybuilding a database of RSSI readings or using an analytical model for simulatingthe radio propagation. Historically, the former approach has been used more promi-nently as it gives a simple procedure to build a map of RSSI values for an indoorscene. The latter, the modelling by reasoning about the physical nature of the radiopropagation, is a response to the expensive nature of the labour-intensive manualcollection and leads to ray tracing based models of different complexities. A selectionof localization systems that have employed analytical models based on 2D or 3D-raytracers are presented here. The following aspects of the systems are highlighted:1. How detailed is the simulation of the physical effects at the intersection pointsbetween rays and scene geometry?2. Is the transmitter of the signal a simple isotropic model or more complex?3. What is the source for the 2D/3D-scene geometry?