QoE Control of Network using Collective Intelligence of SNS in Large-Scale Disasters Chihiro Maru Ochanomizu University Bunkyo, Tokyo, Japan Email: chihiro@ogl.is.ocha.ac.jp Miki Enoki IBM Research - Tokyo Chuo, Tokyo, Japan Email: enomiki@jp.ibm.com Akihiro Nakao University of Tokyo Bunkyo, Tokyo, Japan Email: nakao@iii.u-tokyo.ac.jp Shu Yamamoto University of Tokyo Bunkyo, Tokyo, Japan Email: shu@iii.u-tokyo.ac.jp Saneyasu Yamaguchi Kogakuin Univirsity Shinjuku, Tokyo, Japan Email: sane@cc.kogakuin.ac.jp Masato Oguchi Ochanomizu University Bunkyo, Tokyo, Japan Email: oguchi@is.ocha.ac.jp Abstract— When the Great East Japan Earthquake occurred in 2011，it was difficult to immediately grasp all telecommuni- cations network conditions using only information from network monitoring devices because the damage was considerably heavy and a severe congestion control state occurred．Moreover，at the time of the earthquake，telephone and e-mail services could not be used in many cases，although social networking services (SNSs) were still available．In an emergency，such as an earth- quake，users proactively convey information on telecommunica- tions network conditions through SNSs． Therefore， the collective intelligence of SNSs is suitable as a means of information detec- tion complementary to conventional observation through network monitoring devices． In this paper， we propose a network failure detection system that detects telephony failures with a high degree of accuracy by using the collective intelligence of Twitter，one of the most widely used SNSs．We also show that network control can be performed automatically and autonomically using infor- mation on telecommunications network conditions detected with our system．We developed a network control system on a deeply programmable network (DPN) environment and implemented it on a wide-area network testbed． I. I NTRODUCTION Large-scale disasters，such as earthquakes，often cause telephony failures because base stations and network facilities become damaged and many users try to access the telecommu- nications network at the same time． In such emergencies， it is important that communications via telephone and e-mail ser- vices be available．Usually，network conditions can only be grasped using network monitoring devices．However，when the Great East Japan Earthquake [1] occurred in 2011，it was difficult to immediately grasp all telecommunications network conditions using only information from network monitoring devices because the damage was considerably heavy and a severe congestion control state occurred [2]． Conventionally，telecommunications network conditions are monitored using information from inside a network， using only network monitoring devices [3]. To solve the above- mentioned problem，we propose a network failure detection system using information from outside a network that is complementary to network monitoring devices． The proposed system was developed on a deeply programmable network (DPN) environment called FLARE [28], [29] and implemented on a wide-area network testbed. In subsequent research on the Great East Japan Earthquake [4]，survey participants responded that they were able to use social networking services (SNSs)．Such services are also advantageous in that they can obtain information from users in real time．In an emergency，such as an earthquake，users proactively convey information about telecommunications net- work conditions through SNSs. For example，Twitter can be used to obtain information on the locations and causes of telephony failures and on the degree of impact to users， which cannot be obtained using only network monitoring devices．Therefore，the collective intelligence of SNSs is suitable as a means of information detection complementary to conventional observation using network monitoring devices． The objective of this study was to achieve automatic and autonomic network control by using collective intelligence analyzed from Twitter [5]， one of the most widely used SNSs． This system is targeted to network managers who need to automatically detect telephony failures during emergencies． Twitter accessibility is an issue when Internet services are down．However，if wireless LAN access is not available， other services such as 3G networks and LTE networks may be used．Moreover，people in areas where failures have not occurred can provide information on telephony failures． The contributions of this work are summarized as follows． 1) By designing and prototyping an SNS-based network failure detection system，we can detect information on telephony failures even for a finely-divided part of cities． 2) By integrating our SNS-based network failure detection system into a network control system，we can automat-