LAUSR

Many positioning and tracking applications use spatially distributed sensor stations, each equipped with coherent measurement channels. The coherent data set of each node is incoherently measured to the data sets of all other nodes, avoiding expensive synchronization procedures between the stations. Usually, the measurements are evaluated by mapping the incoherently measured complex valued data on real valued data like angle of arrival or received signal strength. After this preprocessing step, recursive filters fuse the real valued data to estimate a system state. Unfortunately, even though the original measurements are performed in additive white Gaussian noise environments, the preprocessing step can result in correlated, noise shaped errors, whose variance is system state dependent. Hence, the additive white Gaussian noise assumption, which is commonly drawn in Kalman filters, is violated. Therefore, this paper proposes an iterative extended Kalman filter, which estimates the state of a nonlinear real valued system via a complex valued measurement model that consists of incoherent sensor stations, each with several coherent measurement channels. Since the proposed iterative extended Kalman filter is well suited for distributed positioning systems with several stations, a transmitter is localized in a simulation via two sensor stations, each measuring the received signal’s amplitude and phase at four channels. To illustrate the advantages of the proposed algorithm, the direct measurement evaluation using the proposed algorithm is compared to a Kalman filter that evaluates the received signal strengths and angle of arrivals at each sensor station. Finally, a reflecting wall is incorporated into the simulation scenario to demonstrate the flexibility of the proposed Kalman filter.