Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Predicting Loss of Communication Between Radio Enabled Devices Using Deep Recurrent Neural Networks

Rydh, Oscar LU and Klint, Joel LU (2019) EITM01 20182
Department of Electrical and Information Technology
Abstract
This thesis investigates the effectiveness of applying recurrent neural networks (RNN) to detect communication errors between radio devices, also known as supervision violations, on imbalanced data. The task is to classify whether a supervision violation is to occur within seven days. The available data is in the form of radio packets, which are being re-sampled and pre-processed such that they can be interpreted by RNNs. The RNNs are trained as both classifiers and generators to enable detection of supervision violations. Using RNNs, an extensive evaluation is made into different pre-processing methods, using multiple test sets and network architectures, applying average precision as metric and precision-recall curves as the main... (More)
This thesis investigates the effectiveness of applying recurrent neural networks (RNN) to detect communication errors between radio devices, also known as supervision violations, on imbalanced data. The task is to classify whether a supervision violation is to occur within seven days. The available data is in the form of radio packets, which are being re-sampled and pre-processed such that they can be interpreted by RNNs. The RNNs are trained as both classifiers and generators to enable detection of supervision violations. Using RNNs, an extensive evaluation is made into different pre-processing methods, using multiple test sets and network architectures, applying average precision as metric and precision-recall curves as the main evaluation technique. The results show that it is possible to achieve an average precision of 0.75, and that experimentation with pre-processing parameters along with multiple testsets are needed to ensure a generalised model. (Less)
Popular Abstract
The internet of things revolution is rapidly transforming the world we live in. As these gadgets rely on wireless communication, any distur- bance in the environment can cause the device to break down. Using AI technology, there is now a promising solution for predicting commu- nication errors.
Please use this url to cite or link to this publication:
author
Rydh, Oscar LU and Klint, Joel LU
supervisor
organization
course
EITM01 20182
year
type
H3 - Professional qualifications (4 Years - )
subject
keywords
Recurrent Neural Networks, Long Short-Term Memory, Predictive Maintenance, Imbalanced Data, Pre-processing
report number
LU/LTH-EIT 2019-683
language
English
additional info
Popular Science Summary is embedded in the report.
id
8971636
date added to LUP
2019-02-25 11:30:09
date last changed
2019-02-25 11:30:09
@misc{8971636,
  abstract     = {{This thesis investigates the effectiveness of applying recurrent neural networks (RNN) to detect communication errors between radio devices, also known as supervision violations, on imbalanced data. The task is to classify whether a supervision violation is to occur within seven days. The available data is in the form of radio packets, which are being re-sampled and pre-processed such that they can be interpreted by RNNs. The RNNs are trained as both classifiers and generators to enable detection of supervision violations. Using RNNs, an extensive evaluation is made into different pre-processing methods, using multiple test sets and network architectures, applying average precision as metric and precision-recall curves as the main evaluation technique. The results show that it is possible to achieve an average precision of 0.75, and that experimentation with pre-processing parameters along with multiple testsets are needed to ensure a generalised model.}},
  author       = {{Rydh, Oscar and Klint, Joel}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Predicting Loss of Communication Between Radio Enabled Devices Using Deep Recurrent Neural Networks}},
  year         = {{2019}},
}