资源描述:
《Book Optimization and Modeling of Content Delivery Algorithms for Dense Mobile Social Networks》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、OptimizationandModelingofContentDeliveryAlgorithmsforDenseMobileSocialNetworksFawadNazirAdissertationsubmittedinfulfilmentoftherequirementsforthedegreeofDoctorofPhilosophyTheSchoolofElectricalEngineeringandTelecommunicationsTheUniversityofNewSouthWalesMay2010iiAb
2、stractDuetotheincreaseinmobilephones,Ad-hocMobileSocialNetworks(MSNs)arereceivingresearchandindustryattention,i.e.informationexchangeinanoppor-tunisticfashion:whenevertwopeoplemeet,theonewhosecontentismostrecentpushes(usingBluetoothorAdhocwireless)totheonewhosec
3、ontentisoutdated.Mobileuserscandelivercontentviaexistingcellularnetworks.However,thereareseveralbenefitsofusingsuchAd-hocMSNs.Theseopportunisticcontentexchangescanbeusedtoextendthenetwork’scoverageandimprovetheserviceprovidedtoanyuserwhoseaccesstothewirelessnetwo
4、rkisintermittent.Mostimportantly,suchexchangescanincreasenetwork’scapacity.Inparticular,byutilizingthebandwidthofconnectionsbetweenusers,theserviceproviderscansupportmoresubscribersatalowercost.Moreover,MSNscanhelpincontentfilteringasusuallypeopleonlywanttoexchan
5、geprofile,exchangedata,getopinions,trustinformation,takerecommendations,geteventnotification,andreceivenewsetcfromthepeopletheytrust.Thecontributionsofthisthesisaretwofolds.First,weproposeatime-constrainedcontentdeliveryalgorithmforAd-hocMSNs.Thecurrentcontentdeli
6、veryal-gorithmsforMSNsdonotsupporttime-constraineddeliveryofcontent.AsintheseapproachesanimportantbehaviourofMSNsisignoredi.e.predictabilityofthetimeofupcomingindividualencounters/re-encounters.Consideringpredictableencounter-timeintheprotocolsgiveshighprobabili
7、tyofmessagedeliveryfortimeconstrainedapplication(suchas:emergencysituationsnotification,trafficconges-tioninformation,shorteventnoticeetc).Thereforeinthisthesisweaddresstheresearchquestion,HowtoexploitthepredictablebehaviourofpeopleinMSNstodisseminatetimeconstraine
8、ddynamiccontent?iiiSecondly,whilevalidatingourcontentdeliveryalgorithmsweidentifiedtwomajorshortcomingsofthecurrentsimulationmodels,thateithertheyarebasedonrandommodel