MinaQn: Web-based ParticipatorySensing Platform for Citizen-centricUrban Development

Mina SakamuraKeio University5322 EndohFujisawa, Kanagawa, Japanmina@ht.sfc.keio.ac.jp

Takuro YonezawaKeio University5322 EndohFujisawa, Kanagawa, Japantakuro@ht.sfc.keio.ac.jp

Tomotaka ItoKeio University5322 EndohFujisawa, Kanagawa, Japantomotaka@ht.sfc.keio.ac.jp

Jin NakazawaKeio University5322 EndohFujisawa, Kanagawa, Japanjin@ht.sfc.keio.ac.jp

Hideyuki TokudaKeio University5322 EndohFujisawa, Kanagawa, Japanhxt@ht.sfc.keio.ac.jp

Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies are notmade or distributed for profit or commercial advantage and that copies bearthis notice and the full citation on the first page. Copyrights for componentsof this work owned by others than ACM must be honored. Abstracting withcredit is permitted. To copy otherwise, or republish, to post on servers or toredistribute to lists, requires prior specific permission and/or a fee. Requestpermissions from Permissions@acm.org.Ubicomp/ISWC’15 Adjunct, September 07-11, 2015, Osaka, Japanc©2015 ACM. ISBN 978-1-4503-3575-1/15/09...$15.00

DOI: http://dx.doi.org/10.1145/2800835.2801632

AbstractThis paper presents MinaQn1, a web-based participatorysensing platform to enhance bidirectional communicationbetween municipality and citizens for citizen-centric urbandevelopment. To make cities more comfortable to live, itis necessary for municipality to collect and reflect citizens’demands in their administrative works. MinaQn enablescity officers to make various questionnaire to citizen easily,and citizen can answer the questionnaire with theirlocation information through any devices by using WEBinterface. The answers from citizen are collected as sensordata with unified format and protocol, so that the datacan be shared and used for various applications. Wedesigned and implemented MinaQn with web and XMPPtechnology, and carried out a two weeks experiment incorporation with 3 cities in Japan. Based on theexperiment, we measured the effectiveness of MinaQnwith qualitative and quantitative data. In addition, wealso evaluated how citizens consider about providing theirlocation information to the administrative works of thecities.

Author KeywordsSmart City; Participatory Sensing; City Applications;Citizen-centric Urban Development; Citizen Health and

1Mina means everybody in Japanese

Lifestyle Enhancement; Safety and Disaster Management;Experimentation; XMPP

ACM Classification KeywordsH.4.2 [Information Systems Applications]: Types ofSystems—Decision support,Logistics; J.0 [ComputerApplications]: General; H.5.2 [Information Interfaces AndPresentation (e.g., HCI) ]: User Interfaces—User-centereddesign

General TermsDesign,Experimentation,Human Factors

IntroductionThe increasing concentration of the world population intourban areas puts the cities in the center of thepreoccupations. On the other hand, developed countriesincluding Japan are facing the problems of rapidly agingsociety. Therefore, in order to leverage limited resourceeffectively and make city more comfortable to manycitizens, the importance of citizen-centric urbandevelopment will be more increasing in coming decades.

One of the key challenge towards citizen-centric urbandevelopment is to enhance communication amongmunicipality and citizens. Since citizens have specialand/or latest knowledge about their neighbourhoods,transferring those knowledge to municipality enables cityadministrative to work more effective. In addition, ifmunicipality can ask citizens about the latest informationof their area in disaster situation, it can help municipality’sassessment of the situation and rescue planning.

In this paper, we present MinaQn, a WEB-basedparticipatory sensing platform to enhance communicationbetween municipality and citizens for citizen-centric urbandevelopment. Participatory sensing[11] is the concept of

people contributing sensory information through themobile devices which has multiple sensors. The aim ofthis paper is to extend the concept of participatorysensing for the context of citizen-centric urbandevelopment. The information which municipality wantsto ask citizens dynamically changes day by day, especiallyin disaster situations. Therefore, simple and easyauthoring tool for creating sensing tasks (i.e., makingquestionnaire) is necessary. This intuitiveness is alsoneeded when citizens participate sensing campaign definedby municipality and other citizens. In addition, collectedinformation should be opened and reusable for variousapplications as much as possible since the information hasa great possibility to enhance city life.

MinaQn was designed to solve such problems, andmotivating citizen to be in the loop of urban developmentby providing their knowledge to cities. With ourimplemented prototype of MinaQn, we carried out a twoweeks field trial of participatory sensing involvingmunicipality and citizens of 3 cities in Japan. Through theexperiment, we collected more than 1,000 of sensoryinformation from citizens. Of those collected data, wereceived more than 300 of answering data with GPS-basedlocation information of citizens.

In summary, our contributions in this paper are as follows:

• Designing and implementing MinaQn, a open andscalable web-based participatory sensing platformfor enhancing communication among municipalityand citizens

• Verifying the effectiveness of MinaQn bydemonstrating a two weeks experiment with bothquantitative and qualitative analytics

• Indicating guidelines to design future smart citydevelopment using participatory sensing

Design Considerations for Participatory Sens-ing towards Citizen-centric Urban Develop-mentUrban development/planning is a technical and politicalprocess concerned with the use of land and design of theurban environment2. In addition to the definition of urbandevelopment, we include city administrative works such ascity management and disaster control since these are veryimportant and key functionality of cities. Citizen-centricurban development (CUD), also known as participatoryplanning, means that including citizen in the process ofurban development. Many cities are currently focusing onCUD because CUD has great possibility to solve practicalcity problems with preventing conflict between variousstakeholders in cities. One of the key indicators in CUD isthe level of active communication among citystakeholders. Our aim is to leverage participatory sensingtechnology to enhance the communication. In urbandevelopment process, municipality wants to knowproblems and latest information about cities. If citizencan provide answers of such municipality’s question easilythrough their mobile devices, it helps city to be morecomfortable and reliable.

Figure 1: Participatory sensingmodel for citizen-centric urbandevelopment.

Figure 1 shows our proposal model of participatorysensing for CUD. In our model, there are two kinds ofsensing tasks - general tasks and specific tasks. Generaltasks include sensing tasks which useful in everyday life,such as reporting damage of city infrastructure, trafficcongestion, weather and so on. It also includes citizens’demands to municipality. Based on the sensing

2http://en.wikipedia.org/wiki/Urban planning

information from general tasks and also city events,sensing task authors (e.g., municipality) creates specialsensing tasks such as reporting damage of typhoon,finding beautiful cherry blossom or how many number ofkids get flu in your school etc. By repeating this dynamictask setting and answering, city officers can consider cityactions to get their city better.

To realize this CUD cycle, participatory sensing platformhas to fulfill following design requirements. Firstly, bothtask definition and task participation should be easily andintuitively achieved to involve many city officer andcitizens in the CUD. In addition, it’s better not to enforceany installation of the application in mobile devices ofmunicipality and citizens. Secondly, the sensing data,which has no privacy risks, should be opened and usablefor 3rd party application developers. Information fromcitizens are valuable to understand current status of cities,thus the data should be used by many developers ofindustry or NPOs. Thirdly, the platform should be secureand scalable because a lot number of citizens uses theplatform. In addition, data format and data deliveryprotocol should be unified and extensible. Thesedependability and flexibility also helps 3rd party developerto create their original applications, and this in turn leadscity eco-system among municipality and citizen.

There has been great efforts to improve and deployparticipatory sensing technology [15][12][10][13][14].Some of the works focused to sense special tasks such asgarbage volume[12], city noise[15] or congestionpattern[10]. In contrast to these works, our focusingplatform is necessary to define and distribute varioussensing tasks to support CUD. In this context, there aresome researches in tools or systems for task setting andfacilitation of providing tasks to build an application for

participatory sensing, such as OpenDataToolkit[3],sensr[13] and medusa [14]. However, these works haveseveral limitations to fulfill the requirements indicated -requiring application installation to the mobile devices ornot opening the data widely. Especially, we focus tocreate easy, open, flexible and scalable participatorysensing platform based on standard technology such asXMPP for CUD.

On the contrary, there are some research focusing onincentives for participants. As research on monetaryincentive, examining micro-payments for participatorysensing data collections[16] investigated the use ofmicro-payments as an incentive model. On the other handthere are researches on non-monetary incentive, forexample distributing an application that providesbeneficial information for users such as cophenhagenwheel[1]. Users can use the application as health care ornavigation system while they are sensing environmentalinformation for monitoring with their bicycles. Though wedo not focus incentive mechanism for participatory sensingin this paper, we will include these knowledge in futuredevelopment.

System ArchitectureFigure2 shows system architecture of MinaQn with thewhole flow of participatory sensing process. The flow canbe devided into 5 main parts. At first, city officer createsa new participatory sensing task as questionnaire by usingauthoring web interface. After setting a questionnaire, hecan also choose which questionnaire should be active. Ifhe changed the questionnaire, the redirect destinationURL would be changed dynamically. Therefore, once astatic questionnaire URL is integrated with existingwebsites, it is not necessary to change a questionnaireURL every time even if city officer changes a providing

questionnaire. A new sensing task will be created as apartcipatory sensing virtual sensor node. We designed andelivering data architecture for various applicationscapable of ensuring security and scalability. Citizens canbrowse active sensing tasks on the website and answer itanytime. After citizens answer the tasks, the sensing datawill be delivered through virtual sensor nodes tosubscribers of the nodes. In addition to real-time datadelivery via PubSub functionality of MinaQn, collecteddata from citizens is also stored in a database as historydata. Finally, the city officer can see all sensing results,and citizens can also see simple results without privacydata.

Authoring  Interface  Publish Sensing Task

Task Definition

Get Sensing Task

Generating Answer Interface

Publish Sensor Data

Answering    Interface  

Answer the task Other subscribers

if subscribing is allowed (e.g., App. Developers)

City officer

Citizens

Database  Visualizer  

Real-time Data Delivery

Get Historical Data

Put Historical Data

Par7cipatory    Sensing  System  (PubSub-­‐based)  

Real-time Data Delivery

Results include all information

Results without privacy information

Meta node

Data node

Figure 2: System architecture of MinaQn with the whole flowof participatory sensing process.

ImplementationMinaQn was implemented to fulfill the requirements ofparticipatory sensing for CUD.

XMPP-based Sensor Network for Flexibility and ScalabilityTo treat citizens’ opinions as sensor data, we used XMPPtechnology[9], which is the general term for protocol,client and server of open-source instant messenger basedon XML. XMPP has a strong security, scalability and alsoflexiblity because XEP(XMPP Extension Protocol) isoffered. In our architecture, as same as SensorAndrew[17], we used publish-subscribe extension[7] ofXEP. Using publish-subscribe extension, we can publishand subscribe sensor data from participatory sensing to avirtual event node which is defined as a participatorysensing task. When an application uses sensor data, it isnot so important for the application where sensor data isgathered from. For any application, it would be enough ifit could get only the necessary information. Therefore, weused publish-subscribe extension that any application canget sensor data by the unified API.

Q:  Do  you  know  your  evacua0on  place?   Yes  (17)  

No  (52)    

Q:  How  many  days  do  you  usually  have  your  breakfast  in  a  week?  

Every  day(9)  

4-­‐6  days(5)  3-­‐5  

days(24)  

0-­‐2  days  (10)  

Figure 3: Different types ofsensing data visualization:map-based (above) andchart-based (Bottom).

Unified sensor data format for participatory sensing by ex-tending Sensor-Over-XmppOur architecture is based on Sensor Andrew[17], which isdeveloped in Carnegie Mellon University. Sensor Andrew isa middleware that can comprehensively handle sensor dataof different sensor nodes and implemented on XMPPframework. Sensor Andrew also offers theSensor-Over-XMPP(SOX) XEP[6] to cope with variousphysical sensors and actuators with unified format. Whenwe use sensor data for applications, we need to know it’sdata format. In SOX, we treat a virtual event node as apair, meta node and data node. Meta node has aninformation format of sensor data of participatory sensing.However, units about participatory sensing such as freeform text, free form number, single choice(radio button)and moltiple choice(checkbox) were not defined in originalSOX. Therefore, we extended Sensor-Over-XMPP[4] todeal with various participatory sensing information. By

giving a description of definition about participatorysensing in meta node, we can get information about auser interface dynamically and generate suitable userinterface for each questionnaire automatically when a userparticipates in sensing. On the other hand, data node hasinformation about a citizen’s position and an answer.

Intuitive Authoring and Answering WEB Interface with col-lecting location sensor dataTo realize web based participatory sensing using XMPP,we used the Bidirectional-streams Over SynchronousHTTP (BOSH) technology[8]. BOSH is a mechanism toexchange XMPP protocol between a Web Browser and aBOSH server over HTTP. So, a Web XMPP client willrun in the browser and use BOSH to communicate withan XMPP server. As shown in Figure4, on the top page ofquestion management website city officers can create anew questionnaire and manage the existing questions.About management existing questions, you can see actualanswer screen, check answer data, change which questionto carry out, and delete a question. The question which isaccepting answers is shown in red-colored characters. Inanswer part, we used an anonymous user to answer thequestion for ensuring privacy. The answer image isdescribed in the next section.

Top  Page    

Click

Ques1on  se3ng  screen  

Figure 4: WEB interface for authoring participatory sensingtasks.

We used the HTML Geolocation API[2] to get thegeographical position of a user.Since this can compromiseuser privacy, the position is not available unless the userapproves it. As shown in Figure3, we visualized the resultsby both 3D mapping in realtime and a pie chart withhistory data.

Experiments and Results

A  :  Pinpoint  version

 B  :  Mesh  version

Detected  loca6on  

 (not  to  post)  

The  number  of  the  square  (  to  post  )  

Figure 5: 2 type of ways howcitizens privide locationinformation.

In order to validate our system and overall approach, wecarried out our experimentation in 3 cities, Fujisawa City,Chigasaki City and Samukawa Town in KanagawaPrefecture, which is next to Tokyo. The population is40.9 million, 23.5 million and 4.7 million for each. Theexperiments were carried out from Feb. 10 2015 to Feb.24 2015. WEB interface for participating sensing tasks isshown as Figure6. By clicking the question link on theexisiting website, the question page is shown as in the leftimage. We used 5 different websites including municipalwebsite and the city information website by NPO. As forthe question assumed disasters, we prepared 2 types ofways how citizens privide local information shown asFigure5. One is the pinpoint version, which provides theexact location where you are. Another one is the meshversion, which provides the number of the mesh devidedinto 500 meters square according to one’s location. Eitherone of those were shown in a random order to the citizens.

ex)Fujisawa  city  web  site  

Click

Answer  scren  

Figure 6: WEB interface for participating sensing tasks.

Results in case of normal conditionsTable1 presents participatory sensing task results. Thetotal number of answers were 939. Each question has onecommon question which asks a place to live and the restwas related to each question category. As a remarkableresult, there was a differenece between quantitative dataand qualitative data sensed from people. For example, thequestion about symptoms of hay fever was carried out 3times and the distribution of answers differed clearly. As aresult of questions, the probability of complaining of thesymptom increased in the order of Feb. 13, 17, 23.However, according to the pollen count observationsystem by the Ministry of Environment[5] ,the totaldetected pollen count of each day increased in the orderof Feb. 17, 13, 23. Therefore, we know that it is differentbetween the place where citizens answered and theobservation place, however, there is a possibility thatparticipatory sensing will be useful for discovering asituation of a city that can not be predicted byquantitative data.

Results in case of abnormal conditions(with location infor-mation)The experiments were carried out in Feb. 18 and Feb. 21.The total number of answers were 339 and the questionswere about the place they live, their gender, age andphysical condition. According to an access log, a ratio ofusers who did not shift from the explaination page to theactual question page is 44.33% in the pinpoint versionand 41.43% in the mesh version. There is little differencein 2 types of providing location information, thereforewhen natual disasters occur, citizens would not hesitateproviding their location information because of theprivacy.As a result, nearly 90% of the total answered thatwhen natural disasters occur, they would provide theirlocation information or profile to a government willingly.

It is very meaningful for both governance and citizens tocarry out participatory sensing with location informationin the event of a natural disaster. What is more, over 80%of the total answered that they feel a need to shareinformation with a local govenment and citizens.

Feedbacks from a municipality staffWe asked the munitipality staff who conductedquestionnaires for 2 weeks to get feedbacks about oursystem.

Table 1: Participatory sensing task results.

Date Question categoryThe numberof questions

The numberof answers

Feb. 10 Disaster prevention 3 173Feb. 11 Health 3 82Feb. 12 Eco-life 3 101

Feb. 13Symptoms of

hay fever 2 77Feb. 14 Disaster prevention 3 72Feb. 15 Sleep degree 2 81

Feb. 16Locally-grown andlocally-consumed 2 51

Feb. 17Symptoms of

hay fever 2 62

Feb. 18Location

and condition 5 166Feb. 19 In Time of Disaster 2 72Feb. 20 Health 3 41

Feb. 21Location

and condition 5 173Feb. 22 Crime Prevention 3 42

Feb. 23Symptoms of

hay fever 2 44Feb. 24 Disaster prevention 3 41

He thinks that the system was trial production so that theuser interface should be updated, and he would use thesystem accordingly depend on contents of questionnaires.Also he answered that the system is very effective andneeded for communication among municipality andcitizens, because it can help municipality to get real timeinformation including in time of disasters. Furthermore, heanswered that it would be good to conduct questionnairesabout not only municipal administration but also otherthings in a wide category.

Conclusion and Future WorkTo make cities more comfortable to live, it is necessary formunicipality to collect and reflect citizens’ demands intheir administrative works. MinaQn enables city officers tomake various questionnaire to citizen easily, and citizencan answer the questionnaire with their locationinformation through any devices by using WEB interface.The answers from citizen are collected as sensor data withunified format and protocol, so that the data can beshared and used for various applications. We implementedMinaQn with web and XMPP technology and carried outa two weeks experiment in corporation with 3 cities inJapan. As a result, we found that MinaQn enables tohave a conversation with city officers and citizens.However, we still need to get more answers from citizensfor practical use. For example, to make a clear aim foreach questionnaire and show them to citizens would beone of the effective ways for getting more people toparticipate. Also, we will design a better interface andintegrate our system with not only existing websites butalso existing smartphone applications which can receive apush notification in appropriate timing. At the same timewe will enhance system availability and implement thesystem which is easy to maintenance, to carry outlarge-scale and long-term experiments by increasing the

number of participants, question linked webpages, types ofquestions and conducting at other cities. We will also tryto use not only participatory sensing but also othersensing data, so that we can realize multi-level sensingenvironment using our system. By sharing sensor datawith other cities we believe that we can form a communityexceeding a framework of an individual city.

AcknowledgementsThis research was partly supported by National Instituteof Information and Communications Technology (NICT)and MIC G-space project.

References[1] copenhagen wheel project - mit senseable city lab.

http://senseable.mit.edu/copenhagenwheel/index.html.[2] Geolocation api.

http://www.w3.org/TR/geolocation-API/.[3] Open data toolkit. http://opendatakit.org.[4] Our extended sox schema.

http://sox.ht.sfc.keio.ac.jp/schema/schema.html.[5] Pollen count observation system by the ministry of

environment. http://kafun.taiki.go.jp/index.aspx.[6] Sensor-over-xmpp.

http://xmpp.org/extensions/inbox/sensors.html.[7] Xep-0060: Publish-subscribe.

http://xmpp.org/extensions/xep-0060.html.[8] Xep-0124:bidirectional-streams over synchronous http

(bosh). http://xmpp.org/extensions/xep-0124.html.[9] Xmpp standard foundation. http://xmpp.org/.

[10] Berkeley/Nokia/NAVTEQ, U. Mobile millennium.http://traffic.berkeley.edu.

[11] Burke, J. A., Estrin, D., Hansen, M., Parker, A.,Ramanathan, N., Reddy, S., and Srivastava, M. B.Participatory sensing.

[12] CENS/UCLA. Participatory sensing / urban sensingprojects. http://research.cens.ucla.edu/.

[13] Kim, S., Mankoff, J., and Paulos, E. Sensr:evaluating a flexible framework for authoring mobiledata-collection tools for citizen science. InProceedings of the 2013 conference on Computersupported cooperative work, ACM (2013),1453–1462.

[14] Ra, M.-R., Liu, B., La Porta, T. F., and Govindan,R. Medusa: A programming framework forcrowd-sensing applications. In Proceedings of the10th international conference on Mobile systems,applications, and services, ACM (2012), 337–350.

[15] Rana, R. K., Chou, C. T., Kanhere, S. S., Bulusu, N.,and Hu, W. Ear-phone: An end-to-end participatoryurban noise mapping system. In Proceedings of the9th ACM/IEEE International Conference onInformation Processing in Sensor Networks, IPSN’10, ACM (New York, NY, USA, 2010), 105–116.

[16] Reddy, S., Estrin, D., Hansen, M., and Srivastava,M. Examining micro-payments for participatorysensing data collections. In Proceedings of the 12thACM international conference on Ubiquitouscomputing, ACM (2010), 33–36.

[17] Rowe, A., Berges, M., Bhatia, G., Goldman, E.,Rajkumar, R., Jr., J. H. G., Moura, J. M. F., andSoibelman, L. Sensor andrew: Large-scalecampus-wide sensing and actuation. IBM Journal ofResearch and Development 55, 1 (2011), 6.