IRJET- Comparative Study on Embedded Feature Selection Techniques for Interne...
REU2016_XavidRamireez_Poster
1. Developing a System to Monitor Railway
Induced Congestion via Wi-Fi Probe Requests
Xavid Ramirez
Department of Computer Engineering, The University of Texas Rio Grande Valley
• The current status of the system needs
on site testing to determine the viability
of counting vehicles based on Wi-Fi
devices within the vehicle.
• Triangulation: attempt to pinpoint and
only record devices that appear on
streets based on multiple node point
triangulation.
• Implement a machine learning
algorithm so they system can self train
itself on determining railway induced
congestion based on multisource data.
Introduction
Methodology
Understanding 802.11 Packet
Frames
Designing a packet capturing
device.
Results
Research Question
•Can a system to monitor railway induced
congestion via Wi-Fi tracking be devised?
Conclusions
• The University of Texas Rio Grande
Valley.
• The University of Nebraska- Lincoln.
• The University Transportation Center for
Railway Safety.
• The Nebraska Transportation Center.
References
[1] 802.11 Association process explained.
(2015). Retrieved August 03, 2016, from
https://documentation.meraki.com/MR/WiFi
_Basics_and_Best_Practices/802.11_Asso
ciation_process_explained
[2] 802.11 Sniffer Capture Analysis -
Management Frames and Open Auth.
(n.d.). Retrieved August 03, 2016, from
https://supportforums.cisco.com/document/
101431/80211-sniffer-capture-analysis-
management-frames-and-open-auth
[3] Keeble, E. (2014, February 26). Passive
WiFi Tracking. Retrieved August 03, 2016,
from
http://edwardkeeble.com/2014/02/passive-
wifi-tracking/
[4] Documentation. (n.d.). Retrieved July
08, 2016, from https://www.aircrack-ng.org/
[5] Chaffey, D. (2016). Mobile marketing
statistics 2016. Retrieved August 03, 2016,
from http://www.smartinsights.com/mobile-
marketing/mobile-marketing-
analytics/mobile-marketing-statistics/
Acknowledgment
Airodump-ng sample data from in
lab testing
System Diagram including
Network Layout
Congestion can be visually seen by a
person driving down a road, the problem
persists that computerized systems must
be taught how to determine congestion.
The ability to monitor congestion and even
predict future congestion can help future
studies on improving roads and how traffic
is handled during high traffic hours.
With the ever growing use of technologies
and smartphones becoming and every day
utility, a driver carrying a smart phone has a
high statistical probability.
The purpose of this study is to develop a
system to count vehicles as they travel
across an intersection by detecting the Wi-Fi
device within the vehicle. Based on the
gathered data, the system should be able to
detect congestion levels by taking into
account the amount of devices crossing at a
given time.
Wi-Fi connectivity uses the 802.11 Association
Process.
[1] The process of sending packets of
information in the 802.11 packet standard
between mobile devices and Access Points.
[2] 802.11 Packet frames are separated
into 3 types and 10 subtypes.
Mobile Devices send out Probe Requests
which are Type 1 subtype 04 as show
below:
Setting up a Raspberry Pi along with a Wi-Fi
USB dongle that support Monitor mode and
running Airodump-ng generates a csv file with
a list of Access Points and Station Clients
[4] Airodump-ng will sniff out all the Access
Points and Mobile Devices in the area, record
all packet information for each device and log
it into the csv file once Airodump-ng has been
terminated.
Portability is always ideal. Technology has
allowed computers the size of factory rooms
to fit into the palm of the hand.
Using the Raspberry PI, along with a
portable battery to keep it running, a system
that is portable and can be placed anywhere
to capture data is shown below.
Modeling congestion from
monitored devices.
Traffic congestion is characterized by the
longer travel times, slower vehicular speeds,
and increased vehicle queueing.
The Raspberry PI nodes will need to bet set
up at points of interest through the
intersection where congestion will
presumably occur, and collect data based
on that assumption.
The system has not been tested in a real
site setting. In lab development testing
proved promising but no data available to
validation the ability of the system to
detect vehicles based on Wi-Fi devices.
The Network between the test site location
and the Central Server prevents the
Raspberry Pi’s from sending data back to
the Central Server, nor allows the Central
Server network to connect to the Pi’s.
In order to get the system up and running,
the Network Routing Tables and Firewall
have to be properly configured to allow
Port 80 and Port 22 for the subnet the
Raspberry Pi’s are on to send data to and
from the subnet the Central Server is on.