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

Building up knowledge through passive WiFi probes

“…Therefore such analysis must be repeated from time to time with the objective of identifying potential important changes in how the two classes of devices differ [27][28][29][30]. Clearly, this operation calls for automatic methods able to classify traffic flows as belonging to one or the other class.…”

“…Different approaches to perform device classification are available in the literature. A trivial and inexpensive method is to perform device classification by relying solely on the first 3 octets of a MAC address, known as the Organizationally Unique Identifier (OUI), which identifies uniquely the manufacturer of a device [28,29]. The method works for those manufacturers who make only smartphones (e.g., HTC, Rim, Nokia) or laptops (e.g., Asus, Dell).…”

“…• Number of probe requests with broadcast/known SSID: For each source MAC address s we store the number of probe request frames with a "Broadcast" destination SSID N b, s and the number of probe request frames with a textual SSID (that is, the SSID of a Wi-Fi network to • Device manufacturer: Several works in the past have exploited the vendor information contained in the MAC address to infer the class of a device ( [28,29]). Given that some vendors produce only mobile or laptop devices, it is reasonable to include the vendor as a feature for classification.…”

“…This approach applies machine learning techniques to the same information available to other approaches available in the literature [28,29].…”

Passive Classification of Wi-Fi Enabled Devices

“…In this context, the knowledge (or prediction) of what type of traffic (and devices) is actually using the network can be leveraged to optimize the network configuration and/or implement and support several services (e.g., management of wide WiFi networks, smart content caching approaches, etc.) [11,8,3,15,9,10,19,18].…”