Identifying social groups and movement patterns in a city-wide wireless network

From Master Projects
Jump to: navigation, search

Identifying social groups and movement patterns in a city-wide wireless network
status: ongoing
Master: project within::Parallel and Distributed Computer Systems
Student name: student name::Daniel Urda
Start start date:=2013/01/01
End end date:=2013/08/01
Supervisor: Maarten van Steen
Second supervisor: Matthew Dobson
Second reader: has second reader::Spyros Voulgaris
Thesis: has thesis::Media:Thesis.pdf
Poster: has poster::Media:Posternaam.pdf

Signature supervisor



Wireless Arnhem (WA) is an initiative to provide people ease of Internet access in the city of Arnhem. The basic infrastructure consists of a collection of hotspots spread across the city. It aims to eventually offer a few tens of wireless hotspots. Each hotspot will allow mobile devices in their range, primarily smartphones, to connect to the Internet. At the same time, the hotspots will be used to collect some data about the WiFi traffic of the devices in their range, without infringing on the privacy of the device owners.

It is assumed that such data can be collected from all the nodes that use the WA system in order to connect to the Internet. Furthermore, it is assumed that the hotspots are also able to "sniff" WiFi packets not directed (or originating) at the hotspot, by using the Monitor mode of the 802.11 protocol. Another assumption is that the hotspots are able to collect at lest the MAC address of the source and destination of each packet, as well as a timestamp for each transmission event. The properties of the hotspots must also be known in advance(power, range), the positions should generally fixed (with movements known beforehand) and the equipment should always able to capture WiFi traffic (any period of downtime is clearly identified). No other smartphone traffic will be recorded (GSM, Bluetooth, NFC).

We postulate that data collected by the hotspot can be used in order to detect social interaction between device owner. Such relations can be inferred by studying patterns of communication such as timing of WiFi traffic, recurring events of simultaneous communication between devices and hotspot, communication between devices (detected by capturing inter-device packets of 802.11 ad-hioc networks), events of simultaneous communication between certain devices and a hotspot observable in several locations (by different hotspots not covering the same perimeter). We further postulate that movement patterns can be detected by observing the presence of the certain devices across hotspots in a fashion consistent with directed movement (timestamps and locations consistent with a person moving on a route), as well as identifying the timeframe when such movement happens.