Dissemination in Opportunistic Mobile Ad-hoc Networks: the Power of the Crowd

Download Dissemination in Opportunistic Mobile Ad-hoc Networks:  the Power of the Crowd

Post on 21-Jan-2016

29 views

Category:

Documents

0 download

DESCRIPTION

Dissemination in Opportunistic Mobile Ad-hoc Networks: the Power of the Crowd. Gjergji Zyba and Geoffrey M. Voelker University of California, San Diego. Stratis Ioannidis and Christophe Diot Technicolor. Reported by Wentao Li. In IEEE INFOCOM 2011. Abstract. - PowerPoint PPT Presentation

TRANSCRIPT

  • Dissemination in Opportunistic Mobile Ad-hoc Networks: the Power of the CrowdGjergji Zyba and Geoffrey M. VoelkerUniversity of California, San Diego

    Stratis Ioannidis and Christophe DiotTechnicolorReported by Wentao LiIn IEEE INFOCOM 2011*/47

  • AbstractOpportunistic mobile ad-hoc networks The relationship between properties of human interactions and information disseminationUse three different experimental traces to studyUser classification: Socials and Vagabonds

    the effectiveness of dissemination predominantly depends on the number of users rather than their social behavior*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • IntroductionMobile ad-hoc communication and social networking applications supported entirely through opportunistic contacts in the physical world

    Communication in opportunistic mobile ad-hoc networks is challenging the volatility of contactscommunication technologiesresource limitat

    Communication is also strongly impacted by human mobility, which is driven by user social behavior*/47

  • IntroductionProgress in understanding opportunistic mobile ad-hoc networks is mainly limiteddifculty to collect complete tracesdifculty to model large systems with realistic assumptions (absence of large experimental data sets)

    Difculty in the experimentalcollect traces that contain enough information about each device (mobility, social prole, contact opportunities, duration of contacts, communication technology)not biased by constraints due to experimental conditions*/47

  • IntroductionData collectiona need to collect and consider data that encompasses the behavior of all devices in a population not just experimental devices

    process traces by subdividing each trace based on a specic social or professional geographical area of interest

    dene two classes of populations with different presence characteristics, namely Socials and Vagabonds*/47

  • IntroductionContributionStudy data dissemination spanning a large range of Social and Vagabond compositions

    Observe that the efciency of content propagation is not only a consequence of the devices social status, but also a consequence of the number and density of devices

    Study both experimentally and analytically the tipping point beyond which the population size becomes more signicant than the social status

    */47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • Related workSocial-based routing protocolsUse social-based metrics to make opportunistic forwarding decisionsSimBet, Bubble Rap and PeopleRankOur work exploring the role and potential of non-social, vagabond devices for communication and data disseminationSocial networking concepts have been used in mobile opportunistic applicationsOur analysis focuses mostly on epidemic message dissemination*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • Data setsThree data sets - represent distinct and considerably different mobile environmentsDartmouthSan Francisco (SF) Second Life (SL)

    We further subdivide Dartmouth and SF into smaller geographical areas which have different social behavior characteristics*/47

  • DartmouthData set comprises logs of association and disassociation events between wireless devices and access points at Dartmouth CollegeThe logs span 60 days and include events from 4920 devices(4248 avaiable)In contact - when associated with the same access point

    Subdivide areasEngineering (300m200m)Medical (300m300m) Dining (150m150m) */47

  • San FranciscoData set consists of GPS coordinates of 483 cabs operating in the San Francisco area collected over a period of three consecutive weeksIn contact - whenever their distance is less than 100 meters(realistic range for WiFi transmissions)

    Subdivide areasSunset (2km6km)Airport (0.7km1km)Downtown (2km2km)*/47

  • Second LifeData set captures avatar mobility in the Second Life (SL) virtual worldconsists of the virtual coordinates of all 3126(2713 avaiable) avatars that visit a virtual region during 10 daysIn contact - when they are within a vicinity of 10 meters (a reasonable range for Bluetooth)

    We do not dene sub-areas in this data set as the virtual region is small (300m300m)*/47

  • Data sets*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • Socals and VagabondsWe rst classify users according to their social mobility Behavior

    Socials devices that appear regularly, predictably in a given area

    Vagabonds devices that visit an area rarely and unpredictably

    */47

  • Identifying Vagabonds and SocialsMethods based onhow long they stay in a given areathe regularity of their appearance in an area

    A ve-day consecutive weekday period

    Identifying methodsLeast Total AppearanceFourierBin

    */47

  • Least Total AppearanceTotal appearance the total time spent by each device within an area during the ve-day period

    */47

  • Least Total Appearance75% of the population appears less than 20% of the timeFew devices stay within an area for longer periods SocialInection pointssignicant changes in curvaturelinear regression in the range [0, t]error in approximation when the correlation r is such that r2 < 0.9Inection point: above is social */47

  • Fourierdetects periodicityrelies upon the Fourier transformation and the autocorrelation of the appearance of a user in an areaDartmouth MedicalThreshold: above is social */47

  • Binis motivated by the observation that peoples mobility patterns exhibit a diurnal behaviordetects if a user appears every day in an area, and consistently during the same time period

    Binsdivide our measurement period into bins of equal size bBinary stirngrepresent the appearance of each device over timeag a time bin with 1 if the device appears in the area during the period corresponding to this bin, 0 otherwise

    */47

  • BinPeriodicappears every day, at a specic period of the daya device whose corresponding string has a 1 every 24/b bits is periodicFor exibility, we identify a device as periodic even when an exact bin is not agged but a neighboring bin isPeriodic Social, otherwise VagabondBin sizes of 3 hoursaround this time granularity the results are not very sensitive to the bin size*/47

  • Classifying Vagabonds and Socialsunder all methods, in most of the areas Vagabonds represent the majority of the population*/47

  • Classifying Vagabonds and Socialsoverlaps are similar for LTA and Bin yet surprisingly different for Fourier. For balance, we choose Bin.*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • Contact propertiesContact characteristics are key in the effectiveness of opportunistic ad-hoc communicationExamine three different contact metricsContact rateInter-contact timeContact durationStudy these metrics for four different contact scenariosSocial-meets-Socials (SS)Vagabond-meets-Socials (VS)Social-meets-Vagabonds (SV) Vagabond-meets-Vagabonds (VV)*/47

  • Contact rateContact rateFor each device, we compute the number of contacts per hour with other devices in the social or vagabond groupNormalize - to remove the bias introduced by the size of the target populationSocials dominateBalanceVagabonds dominate*/47

  • Contact ratethe SS contact rate is an order of magnitude higher than the VV contact ratethe VS and SV contact rates somewhere in betweenVS and VV - long tails, SS and SV - short tails*/47

  • Inter-contact timeInter-contact timeThe time interval of a device that starts with the end of a contact and ends with the beginning of the next contact It characterizes the periods during which a device cannot forward any content to other devices

    */47

  • Inter-contact timetwo different parts in each curvethe main body (roughly below 12 hours) the tail of the distribution (above 12 hours)it characterizes the mobility patterns that are specic to each arealonger tails when the device met is a Vagabond, which explains later why vagabonds are not individually as effective at contentdissemination*/47

  • Contact durationThe amount of data that can be transmitted between two devices depends both on contact durations and on the communication technology

    It is difcult to interpret and does not characterize the performanceMostly a characteristic of the mobility in the areaWe nd that Socials and Vagabonds experience comparable contact characteristics and their distributions are very similar*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • Data disseminationTrace driven simulationsreplay each trace multiple times using only Socials, only Vagabonds, or any device to propagate messagesObservationin areas in which Vagabonds outnumber Socials signicantly, dissemination using Vagabonds outperforms dissemination using Socials, despite the lower contact rate experienced by VagabondsPredictpopulation is going to be more effective at propagating information*/47

  • MethodologyFloodingNomorize - Repeat the simulation by uniformly sampling many start times between the beginning of the selected week (Sunday midnight) and the middle of that week (Wednesday noon)Simulations last 2.5 days to ensure they all complete within the week-long traceAssume that message transfers are instantaneousMetric contaminationthe number of devices that receive a given message as a function of time

    */47

  • EvaluationSocials outperformVagabonds in areas where they are the majority (SF Downtown) or of comparable population size (Dartmouth Engineering)in areas where Vagabonds largely dominate, they exhibit better contamination characteristics than Socials (Second Life)large populations of Vagabonds can achieve the same contamination performance as Socials*/47

  • EvaluationWe decrease the number of Socials(or Vagabonds) when the social(Vagabonds) group performs better in an area, until we observe a similar contamination ratio for disseminationTo have comparable contamination ratios, Vagabonds need to number two to six times more than Socials*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • AnalysisGoalformally characterize the relationship between the population size and the social behavior of users under which such phenomena occur

    Approacha so-called mean eld limit applied to epidemic dissemination*/47

  • Model DescriptionNotations*/47

  • Model DescriptionVagabonds and SocialsN mobile users visiting an area APartitioned into the two classes: Vagabonds(Nv) and Socials(Ns)Time is slotted, and at each timeslot a Vagabond(Socials) enters A with probability v(s), independently - occupancy rateAssume: v svNv (sNs) - the density of each class*/47

  • Model DescriptionContacts between users and data disseminationAt each timeslot, we select two users uniformly at random among all (unordered) pairs of the N users in the systemIf both of these users are within the area A then a contact takes place between themdenote by vv, vs, sv, and ss the probabilities that transmissions succeed across and within classes

    Main Result(Theorem 1)For large enough N, the epidemic dissemination using Vagabonds eventually dominates dissemination using Socials if and only if Nvv2 > Nss2*/47

  • Proof of Theorem 1A uid limitFractions of the total population: rv = Nv/N, rs = Ns/NNumber of infected(susceptible): Iv, Is(Sv, Ss)Fractions of infected(susceptible): iv = Iv/N, is = Is/N (sv = Sv/N, ss = Ss/N )Number of infected users in each class, is a stochastic process: i(t) can be approximated with arbitrary accuracy through the solution of the following ordinary differential equation (ODE)

    */47

  • Proof of Theorem 1*/47

  • Numerical Validation*/47

  • OutlineIntroductionRelated workData setsSocals and VagabondsContact propertiesData disseminationAnalysisConclusions*/47

  • ConclusionsWe separating users into two behavioral classesAlthough Socials form an active population subset, most areas are dominated by Vagabonds in terms of population sizeVagabonds, often excluded as unimportant, can often play a central role in opportunistic networksThis work is just a rst step in studying the impact of social behavior of users on information disseminationInteresting directionsthe characteristics of inter-area message propagationthe dynamics of user social behavior (e.g., Vagabonds becoming Socials in other areas)the interactions between Vagabonds and Socials in supporting information dissemination*/47

Recommended

View more >