A Target Recognition-Based NLOS Identification Algorithm

In this paper, we develop a target recognition-based algorithm for line-of-sight (LOS) and non-line-of-sight (NLOS) identification, which can be used for channel modeling and parameter analysis. The algorithm first extracts angular information by using target recognition algorithms and then builds a classifier by using support vector machines. Based on the realistic channel measurements, the proposed algorithm achieves fairly high accuracy on LOS/NLOS identification.

I. Introduction

Multi-path propagation plays an important role in wireless communication systems, since line-of-sight (LOS) and non-line-of-sight (NLOS) conditions between transmitter and receiver have a great influence on wireless communications [1]. In the past, NLOS identification in wireless communication systems usually distinguished LOS and NLOS in time domain by using the features like maximum amplitude of the received signal, RMS delay spread, Kurtosis, etc [2], [3], whereas angular features like clusters in multi-path components are used only for visual inspection [4]. In general, the following limitations exit: i) the accuracy of visual inspection is strongly affected by inspector; ii) visual inspection is time-consuming for a large number of measurement data.