Signals from the Crowd: Uncovering Social Relationshipsthrough Smartphone ProbesMarco V. Barbera, Alessandro Epasto, Alessandro Mei, Vasile C. Perta, and Julinda StefaDepartment of Computer Science, Sapienza University of Rome, Italy,{barbera, epasto, mei, perta, stefa}@di.uniroma1.it.ABSTRACTThe ever increasing ubiquitousness of WiFi access points, cou-pled with the diffusion of smartphones, suggest that Internet everytime and everywhere will soon (if not already has) become a re-ality. Even in presence of 3G connectivity, our devices are builtto switch automatically to WiFi networks so to improve user ex-perience. Most of the times, this is achieved by recurrently broad-casting automatic connectivity requests (known as Probe Requests)to known access points (APs), like, e.g., “Home WiFi”, “CampusWiFi”, and so on. In a large gathering of people, the number ofthese probes can be very high. This scenario rises a natural ques-tion: “Can signif i cant information on the social structure of a largecrowd and on its socioeconomic status be inferred by looking atsmartphone probes?”.In this work we give a positive answer to this question. We or-ganized a 3-months long campaign, through which we collectedaround 11M probes sent by more than 160K different devices. Dur-ing the campaign we targeted national and international events thatattracted large crowds as well as other gatherings of people. Then,we present a simple and automatic methodology to build the un-derlying social graph of the smartphone users, starting from theirprobes. We do so for each of our target events, and f i nd that theyall feature social-network properties. In addition, we show that, bylooking at the probes in an event, we can learn important sociolog-ical aspects of its participants—language, vendor adoption, and soon.Categories and Subject DescriptorsC.2 [Computer-communication networks]: Network Architec-ture and Design—Wireless communicationKeywordsSmartphones; Wi-Fi probe requests; social networks.1. MOTIVATION AND GOALSWiFi access points (APs) are becoming increasingly ubiquitousin our homes, off i ces and public places. Initially, the APs werePermission to make digital or hard copies of all or part of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor prof i t or commercial advantage and that copies bear this notice and the full cita-tion on the f i rst page. Copyrights for components of this work owned by others thanACM must be honored. Abstracting with credit is permitted. To copy otherwise, or re-publish, to post on servers or to redistribute to lists, requires prior specif i c permissionand/or a fee. Request permissions from permissions@acm.org.IMC’13, October 23–25, 2013, Barcelona, Spain.Copyright 2013 ACM 978-1-4503-1953-9/13/10 ...$15.00.http://dx.doi.org/10.1145/2504730.2504742.used to free our portable computers (laptops) from the ADSL/LANcable. Nowadays, APs also represent a viable option for mobiledevices to get fast and cheap connectivity. So, it is becoming moreand more common for mobile devices to automatically switch toWiFi connectivity whenever possible. To facilitate this automaticprocess, currentsmartphoneOSesstorethelistofthenames(SSID)of the networks the user typically connects to. Periodically, manyof our smartphones broadcast these SSIDs in the form of ProbeRequest to search for available networks [20, 11, 4]. This is doneevery few seconds, even when we are far from the WiFi accesspoints we usually connect to. In a large crowd of people, a veryhigh number of probe requests are sent every minute. In this paperweconsiderthefollowingquestions: “Whatinformationcanwegeton a large crowd from their probe requests?”; “Is it possible to inferimportantinformationonthecrowdlikeitssocialstructureoritsso-cioeconomic status?”; and lastly: “If yes, can this analysis be donein a simple and automatic way?”. To answer all these questions,we organized a campaign of probe collection: We targeted largegatherings of people at city-wide, national, and international eventsas well as a university campus. Our campaign lasted three months,and we managed to collect, using commodity hardware only, a totalof 11,136,711 probes sent by 164,740 different devices.Our main contribution and f i ndings in this work are the follow-ing:• We develop a simple and automated methodology that al-lows to extract, starting from our datasets, the existing socialconnections among the smartphone owners, and use it to un-cover, for the f i rst time, the underlying social network of theparticipants in each event;• we analyze the properties of these social graphs and showthat they all feature social-network attributes in many aspectssuch as diameter, clustering coeff i cient and degree distribu-tion;• we show that important information on the nature of largeevents can be learnt from the probes:– the distribution of the languages of SSIDs, as detectedbyourmethodology, showsaclearrelationshipbetweenthe international nature of the event and the density offoreign participants;– the distribution of the smartphone vendors varies acrossthe events and matches the expected socioecomomiccharacteristics of the participants.• by using our datasets, we validate the well-known sociologi-cal theories of homophily and social inf l uence in the contextof smartphone vendor adoption;265
