Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Simple RSRP Fingerprint Collection Setup and Indoor Positioning in 5G

Flink, Sofi LU and Tallund, Annie LU (2022) EITM01 20212
Department of Electrical and Information Technology
Abstract (Swedish)
Solving for a way to accurately predict the position of a user equipment is
crucial in an array of applications within 5G new radio. One approach is to
form unique identifiers, also known as fingerprints, and map them to positional
data in an area. Previous research suggests that radio signal received strength
is a promising indicator to use for the fingerprint, in order to achieve accurate
indoor positioning.
The thesis proposes an automated data collection setup. It is capable of col-
lecting radio signal information through modem logs. Additionally, it uses a
navigation system which offers a way to map modem log data to a ground-truth
coordinate. A data preprocessing pipeline is presented as to turn the raw modem
logs into... (More)
Solving for a way to accurately predict the position of a user equipment is
crucial in an array of applications within 5G new radio. One approach is to
form unique identifiers, also known as fingerprints, and map them to positional
data in an area. Previous research suggests that radio signal received strength
is a promising indicator to use for the fingerprint, in order to achieve accurate
indoor positioning.
The thesis proposes an automated data collection setup. It is capable of col-
lecting radio signal information through modem logs. Additionally, it uses a
navigation system which offers a way to map modem log data to a ground-truth
coordinate. A data preprocessing pipeline is presented as to turn the raw modem
logs into fingerprints.
The outcome is a data set for supervised learning. As a proof-of-concept
the data is classified using two machine learning algorithms: naive bayes and k
nearest neighbor. In the latter, a F1 (macro) score of 98% is obtained in predicting
the user equipment position. (Less)
Please use this url to cite or link to this publication:
author
Flink, Sofi LU and Tallund, Annie LU
supervisor
organization
course
EITM01 20212
year
type
H2 - Master's Degree (Two Years)
subject
keywords
5G, indoor positioning, radio fingerprint, k nearest neighbor, naive bayes, automated data collection
report number
LU/LTH-EIT 2022-880
language
English
id
9092127
date added to LUP
2022-06-22 10:57:37
date last changed
2022-06-22 10:57:37
@misc{9092127,
  abstract     = {{Solving for a way to accurately predict the position of a user equipment is
crucial in an array of applications within 5G new radio. One approach is to
form unique identifiers, also known as fingerprints, and map them to positional
data in an area. Previous research suggests that radio signal received strength
is a promising indicator to use for the fingerprint, in order to achieve accurate
indoor positioning.
The thesis proposes an automated data collection setup. It is capable of col-
lecting radio signal information through modem logs. Additionally, it uses a
navigation system which offers a way to map modem log data to a ground-truth
coordinate. A data preprocessing pipeline is presented as to turn the raw modem
logs into fingerprints.
The outcome is a data set for supervised learning. As a proof-of-concept
the data is classified using two machine learning algorithms: naive bayes and k
nearest neighbor. In the latter, a F1 (macro) score of 98% is obtained in predicting
the user equipment position.}},
  author       = {{Flink, Sofi and Tallund, Annie}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Simple RSRP Fingerprint Collection Setup and Indoor Positioning in 5G}},
  year         = {{2022}},
}