Many wheelchair users actively engage in wheelchair sports, for instance in the Paralympics. Trainers increasingly use the developments in sports technology for analysis. But wheelchair athletes do not benefit as much from these developments. A wheelchair can, for instance, disturb signal transmission. Recently, a set of studies describe the development of a wheelchair measurement system, the Wheelchair Mobility Performance Monitor (WMPM). It monitors wheelchair kinematics based on a wheelchair model and two Inertial Measurement Units (IMUs). The WMPM also gives relative location and orientation i.e. relative pose. But this does not allow for game analysis, which requires the absolute location of an athlete. A way to obtain the absolute location of an athlete is the use of Ultra-wideband (UWB) technology. It is often used due to its relatively high ranging accuracy and affordability. However, UWB signal transmission can be disturbed due to Non-Line-Of-Sight (NLoS) situations, where the body of an athlete, or the wheelchair, is located between two UWB nodes. This reduces the localization precision and accuracy. Game dynamics, i.e. quick changes in direction and velocity, also reduce the UWB localization precision and accuracy. Separately, UWB and the WMPM are not sufficient for proper absolute pose tracking. But a combination of these systems, i.e. sensor fusion, could be a good alternative. To provide trainers with real-time tracking data, the sensor data of these systems need to be fused on an embedded system. This requires an efficient UWB localization solution. A recent algorithm by Larsson formulates a localization problem with an easier-to-compute linear eigenvalue approach. In this work, this approach is validated and compared with a previous implementation via simulations and measurements. We show a execution time that is 5 times faster, without accuracy loss. An Unscented Kalman Filter (UKF) is implemented to fuse WMPM and UWB data and it is improved with a Rauch-Tung-Striebel smoother. An IR-camera system (Optitrack), providing the ground truth, is used to validate the UKF. The solution has a Root Mean Square Error (RMSE) of less than 20 cm in dynamic wheelchair situations. This work provides a basis for further development of a platform where WMPM and UWB are integrated for paralympic sports.