Proceedings Chapter
Reference
Deriving Connectivity and Application Usage Patterns From Longitudinal Mobile Phone Usage Data
BARMPOUNAKIS, Sokratis, WAC, Katarzyna
Abstract
In this paper, based on the Mobile Data Challenge data obtained from the Lausanne data collection campaign, our research aims first to derive network connectivity (e.g., WLAN, 3G) and its Quality of Service (QoS) and mobility patterns of the mobile users, as this connectivity and QoS relate to the user's application activity. Second, we aim to understand how these patterns relate to the overall Quality of Experience (QoE) of the user. Concerning the mobility patterns, we define indoor and outdoor activity for each mobile user. Moreover, we attempt to define semantic places using time filters and GIS techniques, which could also be correlated to the application activity of the users. By correlating the above with the application activity of the users, as well as the hour and weekday patterns, certain inferences can be extracted, concerning the users' spatial and temporal behaviour. These inferences could be used further in developing methods for assurance of the mobile users' QoE.
BARMPOUNAKIS, Sokratis, WAC, Katarzyna. Deriving Connectivity and Application Usage Patterns From Longitudinal Mobile Phone Usage Data. In: Laurila, J. & Gatica - Perez, D.
Mobile Data Challenge MDC 2012 (by Nokia) Workshop organized in conjunction with the 20th International Conference on Pervasive Computing 2012 . 2012. p. 6
Available at:
http://archive-ouverte.unige.ch/unige:55845
Disclaimer: layout of this document may differ from the published version.
1 / 1
Locations (indoors/outdoors) and Semantic Places
S002: GPS Speed Visualization S002: CellIDs mapped on GPS
S002: GPS points density clusters
Applications Usage in Time
There are no significant differences per application type and location status.
Users seem to be more active during work hours, with an exception of early afternoon.
Different types of applications show different temporal activity pattern., e.g. E-mail vs. Maps.
Connectivity vs. Applications
Wireless access network types:
• Cellular 2G (+): GPRS, EDGE 3G (+): UMTS, CDMA, HSPA
• WLAN
Connectivity vs. Location Overview
The MDC data analysis carried out focuses mainly on the mobile users’ Application Usage in context of their wireless access network Connectivity and Mobility patterns, as these patterns relate to the Quality of Service (QoS) provided to the user, and the overall Quality of Experience (QoE).
The analysis was focused only on applications, which are network-dependent and make use of online application data exchange, with distinctive traffic models, like:
email web maps VoIP search MMS sharing
Deriving Connectivity and Application Usage Patterns From Longitudinal Mobile Phone Usage Data
The application usage was correlated with the user connectivity, as well as with the current location status.
• When outdoors, mainly the 3G(+) network type is used
• When indoors, mainly the WLAN network type is used
Application Usage Per Connectivity Type
Connectivity Type For Overall Application Usage
S009: Location Type per Application
Connectivity Type vs. Location Type Connectivity Type vs. Location Type, indoors omitted
Sokratis Barmpounakis & Katarzyna Wac, {Sokratis.Barmpounakis, Katarzyna.Wac}@unige.ch
The main algorithm used GPS, GSM, WLAN and Application MDC logs’ data.
Location status types: indoors, outdoors, non–indoors
Users are indoors on average 96% for the overall application usage.
• 70% of the time using cellular network: 2G(+) - 19 %, 3G(+) - 51%
• When on cellular network, 80%+ applications used are Maps, Web, Search and MMS
• Certain application types’ activity is highly influenced by the user’s connectivity type, like:
• Maps over 2G(+)/3G(+)
• VoIP over WLAN (but used by only 9/38 users)
FACULTY OF ECONOMIC AND SOCIAL SCIENCES Department of Management Studies
Institute of Services Science Center for Informatics Quality of Life group
Location Type For Overall Application Usage
Indoors 2G(+) 17%
Outdoors 2G(+) 1%
Non-indoors 2G(+) 1%
Indoors 3G(+) 42%
Outdoors 3G(+) 4%
Non-indoors 3G(+) 5%
Indoors WLAN 30%
Outdoors WLAN
0% Non-indoors WLAN
0% Outdoors 2G(+)
5%
Non-indoors 2G(+) 9%
Outdoors 3G(+) Non-indoors 34%
3G(+) 46%
Outdoors WLAN 1%
Non-indoors WLAN
0% 5%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Maps Web Sharing Search MMS E-mail VoIP
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2 5 7 9 10 17 23 26 34 42 50 51 56 60 63 68 75 77 82 83 89 94 109 111 117 120 123 126 127 139 141 160 165 169 172 179 185 186 avg 2G(+) 3G(+) WLAN
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2 5 7 9 10 17 23 26 34 42 50 51 56 60 63 68 75 77 77 83 89 94 109 111 117 120 123 126 127 139 141 160 165 169 172 179 185 186 avg
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2 5 7 9 10 17 23 26 34 42 50 51 56 60 63 68 75 77 82 83 89 94 109 111 117 120 123 126 127 139 141 160 165 169 172 179 185 186 avg
Connectivity Type For Maps and VoIP
85%
86%
87%
88%
89%
90%
91%
92%
93%
94%
95%
96%
97%
98%
99%
100%
2 5 7 9 10 17 23 26 34 42 50 51 56 60 63 68 75 77 82 83 89 94 109 111 117 120 123 126 127 139 141 160 165 169 172 179 185 186 avg Additional analysis based on the GPS speed information, contributed in extracting useful semantic locations’ information.
17%
31%
4%
18%
12%
18%
30%
25%
5%
17%
10%
7%
6%
34%
3% 26%
14%
11%
5% 7%
2G(+) 3G(+) WLAN
tsx - wlan_ts >
120s yes, cell net
no
cellID >
65535 yes
no WLAN
2G(+)
3G(+) prev_row
cell net indoors
gps_count ≥ (tsx-tsx-1)/10 outdoors yes
non-indoors
GPS WLAN GSM
logs
!"#$%&#
!"#$'%#
yes
Hour of a Day For Overall Application Usage Email and Maps Applications Usage in a Week
0 50 100 150 200 250
Mo Tu We Th Fr Sa Su
0 200 400 600 800 1000 1200
Mo Tu We Th Fr Sa Su
0 200 400 600 800 1000 1200 1400 1600 1800
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24