Multisensor data fusion in nonlinear Bayesian filtering

In this paper, an optimal multisensor data fusion method is proposed to estimate the state of a highly nonlinear dynamic model. Data fusion from spatially distributed sensors is expressed as a semi definite program (SDP) that aims at minimizing mean-squared error (MSE) of the state estimate under total transmit power constraints. Furthermore, a Bayesian filtering technique, based on unscented transformations and linear fractional transformations (LFT), is presented under multisensor framework to implement the SDP. Extensive simulations are performed to justify effectiveness of the proposed multisensor scheme over a single sensor supplied with the same power budget as that of the entire sensor network.