Multiple-PE particle filter processor IC for mobile positioning systems

Abstract

This paper presents a particle filter processor IC for fingerprinting-based positioning in mobile communication systems. The proposed particle filter improves the positioning accuracy from algorithmic and architectural perspectives. For the algorithmic perspective, the proposed particle filter solves the bottleneck of large information exchange of particles by proposing a threshold IMH resampling method to raise the hardware utilization rate and reduce processing latency. For the architectural perspective, the proposed multiple-processing-element (multiple-PE) architecture based on central unit (CU) for threshold IMH resampling can further reduce the processing latency because no particle routing is needed in the PE-CU architecture. The proposed particle filter processor IC was designed and implemented by TSMC 0.18 μ m technology. The processor chip achieves 118MHz clock frequency with 1.48mm2 core area and 3.64mm2 chip area. The IC measurement results show that the proposed multiple-PE particle filter chip has two times hardware efficiency of one-PE particle filter.