Logo image
Low-Complexity Localization Algorithm and Architecture for Distributed Compressive Sensing Ultra-Wideband Radar System
Thesis

Low-Complexity Localization Algorithm and Architecture for Distributed Compressive Sensing Ultra-Wideband Radar System

Huang, Wei-Chin
Masters, 國立清華大學, 通訊工程研究所
2017

Abstract

定位演算法 超寬頻雷達系統 壓縮感知 Localization Algorithm Ultra-Wideband Radar System Compressive Sensing
In recent years, Compressive Sensing (CS) has been a hot research topic. The idea of compressive sensing is based on sparsity and incoherence, which is related to signal characteristic and measurement scheme respectively. Owing to signal characteristic of removing redundant information by reducing the sampling rate, CS technique has a wide range of applications. In the radar system, matched lter and the bandwidth requirement of analog to digital converter are very high, which can improve radar system by CS technique. All the CS problems pursue the high speed (low computational complexity) and high signal reconstruction performance, for example, orthogonal matching pursuit (OMP) is one of the popular algorithms. The signal reconstruction algorithm is the essentials of CS technique. Nevertheless, the complexity of reconstruction algorithms for CS radar increases with the resolution, which is the critical issue of CS radar applications. This study proposed Low-Complexity Localization Algorithm and Architecture for Distributed Compressive Sensing Ultra-Wideband Radar System. Proposed algorithm can reduce computation complexity efficiently, so it has a shorter operation time. Furthermore, this study also presents to combine with Two-Stage algorithm, which has lower complexity and better positioning performance than conventional OMP algorithm. The saving computation time and sparsity is linear relationship. We use the distributed radar system, in order to reduce the positioning blind spot, and improve the coverage ability in the simulation. The architecture of the proposed algorithm is implemented and veri ed by FPGA in the end.

Metrics

1 Record Views

Details

Logo image