Improving the Gaussian Sum Particle filtering by MMSE constraint

摘要

In the paper,we proposed a new modification to the Gaussian Sum Particle filtering (GSPF) by incorporating a Minimum Mean Square Error constraint (MMSE-GSPF) in tracking parameter variation of Dynamic State Space model,the algorithm complexity is reduced yet.In addition,the MMSE-GSPF also shows better performance in tracking of time-varying parameters than the original Gaussian Sum Particle filtering.