For the real time problem of particle filter in GPS/INS integrated navigation system, an improved particle filter
algorithm based on linear & nonlinear structural decomposition is proposed. In the improved algorithm, state equation is
decomposed into linear & nonlinear structures, individually predicted and educed by importance sampling and linear Kalman
filter. By bringing the advantages of particle filter and Kalman filter into the prediction process, the computation workload
of particle filter is reduced and the real time capability of integrated filtering is improved with the precision of GPS/INS
integrated navigation guaranteed.