NLOS Identification for UWB Body Communications

Abstract

In the last few years, a great attention has been paid to wireless communications for body area networks especially since the IEEE 802.15.6 standard. The main objective of this work is to present a good technique for identifying between both Line-Of-Sight (LOS) and Non-Line-Of-Sight (NLOS) propagation schemes for UWB both of both on-body and offbody communication. Our work is focalize in the first to extract the information using traditional features compared with our proposed methods and secondly to classify it using Support Vector Machine for objective to given a good recognition rate of identification between LOS and NLOS phenomena. This characterized was applied for UWB measurement by the antenna Electromagnetics Group (Body WiSeR).

Deep Learning-Based LOS and NLOS Identification in Wireless Body Area Networks

“…This results in a greater or similar effectiveness of the THM relative to SVM for several learning datasets, which in these cases calls into question the legitimacy of using the more complex SVM method. The obtained values of the THM classification efficiency and SVM methods are similar to these presented in the literature [5,6,10,11]—described in more detail in Section 2.…”

“…During the measurements, over 40,000 pairs of CIR parameter values were obtained for all measurement scenarios. It was decided to carry out classification effectiveness η measurements of the proposed DFNN, THM, and SVM methods, which were widely used in the studies presented in the literature [5,6,10,11]. It should be mentioned that the SVM method was also tested with the use of only two input parameters, i.e., FPP and TP.…”

“…Currently, various methods for identifying LOS and NLOS conditions in UWB networks are known in the literature, including solutions based on ML, i.e., SVM (Support Vector Machine). In general, these methods can be divided into those that uniquely determine LOS and NLOS conditions based on current parameters in real time [5,6,10,11] or those that have a certain time inertia, e.g., due to the additional data filtration usage and data processing with memory [2]. In this article, only real-time working methods will be analyzed.…”

“…Similar NLOS conditions detection research is also conducted in WBANs, where the human body may obstruct the direct visibility of nodes. In [11], the authors obtained 87.5% classification efficiency for on-body and off-body communication by reducing the full CIR by PCA (Principal Component Analysis) used as an input for the SVM method.…”