An alternative approach for the resampling in the bootstrap filter

The resampling step in the bootstrap filter aims at reducing the well-known particle degeneracy phenomenon. In practical applications of particle filters, a large number of particles is used for computing the estimates of desired states. However, the parallelizability of the filter is affected by the resampling that increases the hardware complexity. This paper proposes an alternative method to replace this step by one that, in accordance to the so-called attractor function, moves the particles in the likelihood estimate to regions where the weights are higher. The implementation of this new method allows the parallelization of the algorithm and the particle filter performance is similar to that obtained by the traditional bootstrap method