Fusion of multiple positioning algorithms

With the proliferation of location based services (LBS), various indoor positioning techniques have been explored based on received signal strength (RSS). To improve performance, many hybrid or fusion approaches have been proposed in the literature. In this paper, a new fusion approach is proposed to achieve better positioning performance, with a focus on the optimal utilization of RSS measurements in wireless local area network (WLAN). First, a fusion architecture is developed to make use of multiple observations from the different positioning algorithms and by employing this architecture, more than 20 percent reduction in the mean distance error is achieved. Additionally, a novel online training method is employed to estimate the covariance of the observations to achieve further improvement.