Open AccessSimulation and analysis of 60 GHz MMW for localisation based on machine learning
Author(s) -
Dou Mingwu,
Zhang Wenwu
Publication year - 2017
Publication title -
the journal of engineering
Language(s) - English
Resource type - Journals
ISSN - 2051-3305
DOI - 10.1049/joe.2017.0264
Subject(s) - robustness (evolution) , computer science , time of arrival , artificial neural network , energy (signal processing) , noise (video) , signal (programming language) , channel (broadcasting) , artificial intelligence , electronic engineering , real time computing , telecommunications , statistics , engineering , mathematics , biochemistry , chemistry , gene , image (mathematics) , programming language
The 60 GHz pulse is more practical for the indoor localisation system due to its better time resolution. Usually, time‐of‐arrival (TOA) estimation with higher accuracy is critical to the indoor localisation. To acquire precise TOA estimations, a novel threshold crossing technique using neural network (NN) is discussed via analysing the characteristics of the received pulses based on the energy detection receiver. The relationship between the optimum thresholds and the signal‐to‐noise ratios (SNRs) are researched. Meanwhile, the influences caused by changing integration periods and channel models are examined. Results show that the proposed NN method can provide better accuracy and robustness to the lower SNR in the models designed by TG 3c.