Stochastic Intensity Propagation via LFT Modelling

Abstract

AbstractRecently linear fractional transformation (LFT) modelling has demonstrated considerable potential for the filtering problem of highly nonlinear state space systems. A characteristic of the LFT model is the feedback loop which encapsulates the nonlinearity of the state-space model in a structure that is both simple and sparse. For a broad class of practical problems the LFT gives an equivalent representation of the nonlinear state space models which is more efficient due to its structure. With an approximation localized to a simple nonlinearity in the feedback path, it has been shown that this approach works reasonably well where conventional linearization techniques fail. In this paper, we propose nonlinear filtering via LFT modelling for propagating the stochastic conditional intensity function for problems in multi-target filtering. Simulation results are shown to demonstrate the performance of the proposed filter.