CRAWDAD dataset kth/campus (v.2019-07-01)

Ljubica Pajevic; Gunnar Karlsson; Viktoria Fodor

doi:10.15783/c7-5r6x-4b46

Dataset of wireless network measurements at the KTH campuses, collected during 2014-2015. DOI: https://doi.org/10.15783/c7-5r6x-4b46

The dataset contains records of authenticated user associations to the wireless network of the KTH Royal Institute of Technology in Stockholm. The dataset also includes scan results and mapping information of Wi-Fi networks, collected by means of war-walking at the university's two largest campuses.

last modified

2019-07-01

nickname

campus

institution

KTH Royal Institute of Technology, Stockholm

reason for most recent change

the initial version

release date

2019-07-01

date/time of measurement start

2014-01-01

date/time of measurement end

2015-12-14

website

www.crawdad.org/kth/campus

network type

802.11 infrastructure

network type

GPS (Global Positioning System)

network type

DTN (Delay or Disruption Tolerant Network)

collection environment

The KTH wireless network provides coverage for buildings on one large and four small campuses located within metropolitan Stockholm area. The campus buildings are for non-residential use, housing classrooms, computer laboratories, libraries, offices, administrative premises, cafeterias and restaurants. Due to the high density of access points (APs) and proximity of campus buildings, most of the outdoor areas are covered as well. At the time of the trace collection (2014-2015) the university had around 18000 active students and employees, most of them accessing the wireless network via smartphones, laptops and other portable devices. The number of active wireless users that connected to the network for a weekday varied between 13000 and 15000.

network configuration

The number of deployed APs in the network varied from 934, active during the first month to 965 in the last month, with the maximum number of 985 APs in the middle of the experiment. All APs in the network are Cisco Aironet models. The APs are managed by Cisco 5508 Series Wireless Controllers. The wireless controllers are connected, through the university LAN, to an authentication server, in this case FreeRADIUS (version 2) server, which is responsible for handling wireless users access requests.

data collection methodology

The two tracesets included in the dataset were collected by different acquisition methodologies.

The first traceset consists of Eduroam associations, which were pulled from the authentication server deployed in the university network. These associations constitute 95 % of all associations in the wireless network. User association in an Eduroam network requires RADIUS authentication. For keeping record of all authentication events, FreeRADIUS uses a syslog module linelog which reports events in the F-Ticks format. In the case of Eduroam access, these records are logged and reported to the national operator for collecting statistics at national levels.

To obtain the second traceset, we developed a simple Android application to record scan results and coverage of all Wi-Fi networks detected while walking at the campus.

sanitization

In the eduroam trace, all user MAC addresses were, prior to logging, anonymized through hashing, which obfuscates user identities but keeps the hashed addresses preserved throughout the entire trace. The captured records contain both the MAC address (BSSID) and the actual name of each access point. The AP name consists of the name of the building, as well as the floor and the room or corridor number, where the AP is located. We removed AP MAC addresses and replaced the configured AP names with a more uniform labels, e.g., Bldg1AP1.The AP locations within the campus area can be looked up from the APlocations.csv file. Each AP location is given by Cartesian coordinates and the floor number (where available). The coordinates were inferred by map projection of the longitude and latitude of the building or, in case of large buildings, the section where the AP is situated. Thus, locations of multiple APs map to the same coordinates, and the locations of different APs on the same floor in the same building cannot be distinguished.

disruptions to data collection

None that we are familiar with.

error

Some APs were relocated during the course of the experiment, but their names were not modified to reflect these movements. This resulted in those relocated APs showing in the association logs with the name corresponding to the last location where the AP was deployed. We made our best efforts to detect all APs that belong in this group, and subsequently re-assigned names to correspond to the correct buildings (and floors where available) where the APs were deployed in the past.

limitation

The traceset contains only records of successful associations, lacking the information of session durations e.g., via de-authentication/disassociation events or accounting logs. The second limitation is the inability to infer any information about user devices from their MAC addresses---for instance, the device type from the OUI portion of the MAC address---since the entire addresses were hashed during logging.

url

/download/kth/campus/traceset2.zip
size="1.2MB" type="zip" md5="f9cf4b61ca7b867fded154502841633b"

url

/download/kth/campus/traceset1.zip
size="2.5GB" type="zip" md5="98e99419f8cb60c07b7e14af1dd35cfd"

This dataset contains the following 2 tracesets:

When writing a paper that uses CRAWDAD datasets, we would appreciate it if you could cite both the authors of the dataset and CRAWDAD itself, and identify the exact dataset using the appropriate version number. For this dataset, this citation would look like:

Ljubica Pajevic, Gunnar Karlsson, Viktoria Fodor, CRAWDAD dataset kth/campus (v. 2019‑07‑01), downloaded from https://crawdad.org/kth/campus/20190701, https://doi.org/10.15783/c7‑5r6x‑4b46, Jul 2019.

We also provide bibliographic information in common citation formats below:

If you do not use the provided citation formats, please include a reference with the same information, as described in the CRAWDAD FAQ.

The kth/campus dataset (v. 2019-07-01)

This dataset contains the following 2 tracesets: