Publication detail

Comparison of Neural Models of UWB and 60GHz In-car Transmission Channels

KOTOL, M. RAIDA, Z.

Original Title

Comparison of Neural Models of UWB and 60GHz In-car Transmission Channels

English Title

Comparison of Neural Models of UWB and 60GHz In-car Transmission Channels

Type

conference paper

Language

en

Original Abstract

Knowledge of characteristics of the transmission channel is advantageous for the selection of a suitable location of transmitting and receiving antennas, choice of the carrier frequency and the transmission parameters such as bit rate, modulation type, coding, etc. However, the description of properties of the transmission channel can be computationally time consuming, and the computational complexity increases with the increasing frequency. The transmission channel can be modeled by an artificial neural network to reduce the computational complexity compared to the analysis using full-wave simulation programs (CST, HFSS, etc.). Two neural network architectures were selected (a feed-forward one and a radial basis function one) to model an in-car transmission channel. For each neural model, a study of the model error, the speed of training and the network complexity is given.

English abstract

Knowledge of characteristics of the transmission channel is advantageous for the selection of a suitable location of transmitting and receiving antennas, choice of the carrier frequency and the transmission parameters such as bit rate, modulation type, coding, etc. However, the description of properties of the transmission channel can be computationally time consuming, and the computational complexity increases with the increasing frequency. The transmission channel can be modeled by an artificial neural network to reduce the computational complexity compared to the analysis using full-wave simulation programs (CST, HFSS, etc.). Two neural network architectures were selected (a feed-forward one and a radial basis function one) to model an in-car transmission channel. For each neural model, a study of the model error, the speed of training and the network complexity is given.

Keywords

Artificial neural network; feed-forward network; radial basis function network; transmission channel measurement; estimation of in-car channel parameters; transfer function

Released

14.09.2016

Publisher

IEEE Xplore

ISBN

978-1-5090-2269-4

Book

CoBCom 2016

Pages from

64

Pages to

68

Pages count

150

URL

BibTex


@inproceedings{BUT129209,
  author="Martin {Kotol} and Zbyněk {Raida}",
  title="Comparison of Neural Models of UWB and 60GHz
In-car Transmission Channels
",
  annote="Knowledge of characteristics of the transmission channel is advantageous for the selection of a suitable location of transmitting and receiving antennas, choice of the carrier frequency and the transmission parameters such as bit rate, modulation type, coding, etc. However, the description of properties of the transmission channel can be computationally time consuming, and the computational complexity increases with the increasing frequency. The transmission channel can be modeled by an artificial neural network to reduce the computational complexity compared to the analysis using full-wave simulation programs (CST, HFSS, etc.). Two neural network architectures were selected (a feed-forward one and a radial basis function one) to model an in-car transmission channel. For each neural model, a study of the model error, the speed of training and the network complexity is given.",
  address="IEEE Xplore",
  booktitle="CoBCom 2016",
  chapter="129209",
  doi="10.1109/COBCOM.2016.7593493",
  howpublished="online",
  institution="IEEE Xplore",
  year="2016",
  month="september",
  pages="64--68",
  publisher="IEEE Xplore",
  type="conference paper"
}