Abstract
In recent years, the mobile positioning has attracted many attentions since the proposition of E-911 purpose. This thesis adopt particle filter and convex optimization to complete the estimation. For the mobile positioning, particle filter based algorithm has become the common solution of solving the effect of non-linear channel. However, the solution of NLOS problem still does not come to an end. In this thesis, we adopt the mixed norm idea and the concepts of optimization problem of [1] to developing our positioning algorithm for solving NLOS problem. Due to the convex optimization problem of [1] is not general and too complicated, we introduce a ’virtual’ concept to describe the propagation condition. This concept reduces complexity of the optimization problem and enhances the performance simultaneously. Also, the idea of map factor introduced in [1] is discussed in detail and we determine how to assign the map factor. Another approach to improve the positioning performance by weighting the estimated distances come from particle filters is introduced. After constructing the algorithms, we analysed the architectures of particle filter and convex optimization processor. Because the resampling step of particle filter is too exhausted on timing, we proposed a modified resampling particle filter to improve the timing. The proposed modified resampling particle filter is verified without losing performance. Finally, we implement the architecture design on FPGA to verify our proposed algorithms. We also realize the distance estimation by RF modules and FPGA.