Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Multi-Factor Mobile based Access Control Solution

Larsson, August LU and Ohlsson, Alvin LU (2020) EITM01 20201
Department of Electrical and Information Technology
Abstract
In this thesis, alternatives for replacing an RFID card + PIN based access controlsolution for office buildings with a mobile applications are evaluated. The proposedsolutions are based on a probabilistic fingerprinting approach where distributionsof RSSI are approximated and stored in a fingerprint database during an initialoffline phase. Unlike traditional fingerprinting applications, the solution utilize thefact that only position relative to the different doors is required by solely creatingfingerprints at each doors position as opposed to in a grid of the building. In thefollowing online phase, Maximum Likelihood Classification is applied to find thebest match between a users RSSI and objects in the database, granting access tothe... (More)
In this thesis, alternatives for replacing an RFID card + PIN based access controlsolution for office buildings with a mobile applications are evaluated. The proposedsolutions are based on a probabilistic fingerprinting approach where distributionsof RSSI are approximated and stored in a fingerprint database during an initialoffline phase. Unlike traditional fingerprinting applications, the solution utilize thefact that only position relative to the different doors is required by solely creatingfingerprints at each doors position as opposed to in a grid of the building. In thefollowing online phase, Maximum Likelihood Classification is applied to find thebest match between a users RSSI and objects in the database, granting access tothe respective door.Alterations are made to the original solution to counteract a) restrictions imposedon the Android API which slow down Wi-Fi RSSI scanning and b) decreased ac-curacy with too many doors densely placed. The alternate Wi-Fi solution usesthe previously received RSSI sample to create a coarser estimation of location,allowing the user to choose between the three closest doors for instant authenti-cation. For improved accuracy with tightly grouped doors, a geomagnetism basedsolution is used that use fingerprints similar to the Wi-Fi solution, but that arenot subjects to spacial variations in the same manner.Additionally, behaviour of Wi-Fi signals in an indoor office environment is mea-sured in terms of spatial variations and line of sight obstruction. In turn, differentAP setups and how they affect localization performance are evaluated, along withimprovements made to the RSSI sampling process to account for human obstruc-tion of AP line of sight during use.Finally, results from testing in a dedicated test environment show that all of thesolutions can be suitable for real use in different scenarios. The thesis providesconcrete conclusions of how each solution can be applicable for different use cases,or improved upon further with other hardware or technologies. (Less)
Please use this url to cite or link to this publication:
author
Larsson, August LU and Ohlsson, Alvin LU
supervisor
organization
course
EITM01 20201
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Indoor positioning Wi-Fi Fingerprinting Magnetic Fingerprinting
report number
LU/LTH-EIT 2020-769
language
English
id
9024653
date added to LUP
2020-07-08 15:55:59
date last changed
2020-07-08 15:55:59
@misc{9024653,
  abstract     = {{In this thesis, alternatives for replacing an RFID card + PIN based access controlsolution for office buildings with a mobile applications are evaluated. The proposedsolutions are based on a probabilistic fingerprinting approach where distributionsof RSSI are approximated and stored in a fingerprint database during an initialoffline phase. Unlike traditional fingerprinting applications, the solution utilize thefact that only position relative to the different doors is required by solely creatingfingerprints at each doors position as opposed to in a grid of the building. In thefollowing online phase, Maximum Likelihood Classification is applied to find thebest match between a users RSSI and objects in the database, granting access tothe respective door.Alterations are made to the original solution to counteract a) restrictions imposedon the Android API which slow down Wi-Fi RSSI scanning and b) decreased ac-curacy with too many doors densely placed. The alternate Wi-Fi solution usesthe previously received RSSI sample to create a coarser estimation of location,allowing the user to choose between the three closest doors for instant authenti-cation. For improved accuracy with tightly grouped doors, a geomagnetism basedsolution is used that use fingerprints similar to the Wi-Fi solution, but that arenot subjects to spacial variations in the same manner.Additionally, behaviour of Wi-Fi signals in an indoor office environment is mea-sured in terms of spatial variations and line of sight obstruction. In turn, differentAP setups and how they affect localization performance are evaluated, along withimprovements made to the RSSI sampling process to account for human obstruc-tion of AP line of sight during use.Finally, results from testing in a dedicated test environment show that all of thesolutions can be suitable for real use in different scenarios. The thesis providesconcrete conclusions of how each solution can be applicable for different use cases,or improved upon further with other hardware or technologies.}},
  author       = {{Larsson, August and Ohlsson, Alvin}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Multi-Factor Mobile based Access Control Solution}},
  year         = {{2020}},
}