Research Of Particle Filter Algorithm Based On FPGA Implementation

Developing in 90's last century, Particle filter, a new filter method based on Monte Carlo and recursive Bayesian estimation, has special advantages in dealing with the state and the parameter estimation in the nonlinear and non-Gaussian system. However, the disadvantages of complex algorithm architecture, enormous computations and low speed have restricted its implementation in real-time system. This thesis aims to reduce the complexity of the particle filter algorithm with better computation speed in order to design a particle filter which is fast, credible and using less hardware resource so as to be implemented in real-time system, i.e. object tracking etc..Field Programmable Gate Array is a hardware logic device, which implements the real parallel computation. In theory, the FPGA can output one result in one clock period. Particle filter can reach the highest computation speed in FPGA system. So, this thesis chooses the FPGA as the target device to implement particle filter.This thesis, at first, introduces the status in quo of particle filter and the meaning of research. Secondly, it introduces the basics theory of the particle filter and the characteristics of FPGA. Thirdly, the complexity of the algorithm has been reduced and the computation speed has been advanced by improving methods on the three steps in sampling, weighting computation, and resampling based on analyzing thoroughly the processes, complexity, parallel of SIRF algorithm. And it simulates in matlab using bearing-only system object tracking for example to check the correctness of the advanced algorithms. And then, it designs the models, such as the Gauss rand number generator, the resampling model, sample model and weight computation model in FPGA in the basis of the advanced SIRF algorithm, as well as uses the waveform bench simulation to test the correctness of the models. At last, these models are connected to implement the particle filter algorithm in FPGA.This thesis has designed and implemented the advanced particle filter algorithm method in FPGA, and the waveform bench simulation and the report of resource using indicate that this advanced algorithm, with its characteristics of high speed, using small resource, can be used in real-time system, i.e. object tracking, navigation, image processing, parameter estimation etc..