Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 42884 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Behnia, Fereidoon
- Abstract:
- In this thesis, Particle Filter Methods are investigated in the context of moving targets tracking and the associated performance analysis. The application scope of this algorithm which is a particular case of Sequential Monte Carlo Method is far broader than tracking of moving targets. This algorithm can be used for mathematical calculations such as estimation of mathematical expectations, integrals, surface area of curves and many other mathematical calculations. In addition, it has applications in other branches of science like genetics. This algorithm is based on random sampling of a probability density function and resampling from the extracted samples. We change this algorithm in three stages and eventually obtain an algorithm which compared to Monte Calro methods has better performance in terms of decision making and error probability. To change Monte Carlo algorithm, we focus on the entropy of samples’ weight and try to obtain some boundaries for the entropy and based on them to decide on the necessity or non-necessity of a resampling step. Finally, we compare the recommended algorithm and the Monte Carlo algorithm both using simulations and theoretically and demonstrate superiority of the recommended algorithm. In addition, we focus on the threshold in variance and entropy criteria and demonstrate that the two thresholds are not independent and satisfy the same inequality. This inequality causes the required space for search of the threshold’s surfaces smaller and ease the problem of finding the desirable surfaces.
Keywords: Particle Filter Method, Sequential Monte Carlo Method, sampling, resampling - Keywords:
- Monte Carlo Method ; Particle Filter ; Resampling ; Sequential Monte Carlo Method ; Random Sampling
Digital Object List-
محتواي پايان نامه - view
Bookmark