Ranking and optimization of target tracking algorithms

摘要: In this thesis the performance of target tracking algorithms, and in particular sequential multi-sensor joint probabilistic data association algorithm is analyzed. The main contribution development a tool for ranking selecting designs where metric standard deviation (e.g., tracking, root mean square position error), not network throughput), assuming that error has Gaussian distribution. Further improvement variance selection achieved when modeled as mixture distribution (a weighted sum distributions, defined by set weights, means, variances). Parameter estimation k-component obtained applying modified version well known expectation maximization algorithm. order to reduce long simulations necessary achieve good confidence observed fusion an optimized computing budget allocation approach applied, using improved models track error. developed leads more than magnitude reduction computational effort produces results with high levels comparing different orders processing sensor information Application new technique various types particle filters also demonstrates can be accomplished efficiently levels. The proposed applied other algorithms provided or