Changing patterns of mobile network (WLAN) Usage: Smart-Phones vs. Laptops

Authors:

Abstract
The proliferation of smart-phones has been simply unprecedented. Unlike laptops, smart-phones have a much smaller form factor, are ultra mobile, include a rich set of sensors (e.g., location, motion), and are tightly coupled with the users. These factors pose a fundamental question regarding the characteristics of mobile users' on-line behavior for these devices, and how their differences (vis-a-vis laptops) impact networking protocols and services. Over the past decade the networking community has used extensive data sets, mainly capturing usage of laptops in WLANs, to develop and validate mobility (and usage) models, used in turn to analyze, simulate and design network services and protocols. In this study, using recent WLAN traces, we aim to compare and contrast the spatio-temporal on-line behavioral patterns of mobile users when using smart-phones vs. laptops. The main challenges in our study include the collection of extensive recent traces, the classification of mobile devices, and the analysis of various behavioral metrics on a large-scale. We collect over 46M records of 120K devices during 4 months in 2011. We introduce a classification of smart-phones based on MAC address filters, then use several spatio-temporal and predictability metrics to compare the traces. The result of our analysis shows significant differences between usage of smart-phones and laptops. The differences include shorter session durations for smart-phones (by 67% as compared to laptops), higher number of sessions per day (2 times that of laptops) and visiting new location (3 times that of laptops). We find that these differences are significant and will allow researchers to create better models and future services.