Hardware Implementation Of Particle Filtering Algorithm

Firstly, this paper systematically introduces the basic principles and the key technologies of the particle filter which is properly used to solve the problems of the non-linear and non-Gaussian systems, and then introduces the digital signal processor and programmable logic devices in principle and internal structure. This thesis aims to overcome the particle degradation and reduce the complexity of the particle filter algorithm with better computation speed in order to design a particle filter which is good diversity of particle, fast, credible and using less hardware resource so as to be implemented in real-time system.To solve the above issues, the fission bootstrap particle filtering is introduced in this paper. The complexity and real-time problem of particle filter algorithm are analyzed and studied carefully. To solve the problem of more computational complexity, worse real-time and harder choosing condition of fission, the improved fission bootstrap particle filter is introduced. To clear the condition of fission, this paper has a detailed derivation on condition of fission. To reduce division operation and use less hardware resource, this paper puts the process of normalized weight into the processes of resulting output, testing weight's sharpening and resampling etc.Against more computing capability characteristic of particle filter, a high-performance digital signal processor and a field programmable gate array device are selected respectively to implement the improved fission bootstrap particle filter on hardware. Firstly, the Gauss rand number generator model and particle filter model are programmed from bottom to top by field programmable gate array. Then testing and verifying the correctness of models in waveform bench simulation. At last, these models are connected to implement the particle filter algorithm in field programmable gate array. The results of two implementations turn out that the improved fission bootstrap particle filter not only can improve diversity of particle, overcome the particle degradation and improve the filter accuracy, but also improve real-time and robustness?of?the?particle?filter?algorithm.?...