2012 Sixth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing
Participatory Sensing in Environmental Monitoring – Experiences
Ville Kotovirta, Timo Toivanen, Renne Tergujeff, Markku Huttunen VTT Technical Research Centre of Finland Espoo, Finland
[email protected]
research goal is to study the feasibility of participatory sensing for various environmental monitoring problems. We discuss our findings about privacy preservation, reliability of collected user observations and the problem of motivation, i.e. how to recruit and inspire people to produce useful information. For gathering observations, we have developed a participatory sensing tool called EnviObserver, which utilizes people as sensors by enabling them to report environmental observations with a mobile phone.
Abstract—In this paper we describe our experiences in applying the concept of participatory sensing to environmental monitoring. We have run pilot trials for air quality, water quality and plant disease monitoring. In these pilots, users have reported their personal observations or measurements of various environmental phenomena, using special locationbased applications in their mobile phones. We found a relevant correlation between algae observations by untrained citizens and by professionals, which supports the feasibility of participatory sensing as a complementary information source for algae monitoring. One key issue in collecting useful participatory datasets is managing to motivate people for acting as mobile environmental sensors. Other important issues discussed in the paper include privacy preservation and reliability of user observations.
II.
Lane & al. [9] compared theoretically a participatory model and an opportunistic model for using people as sensors. The participatory model requires people to take actively part in making observations, while the opportunistic model uses people’s mobile devices automatically for observation collection. Applications of participatory sensing have been developed for many areas, with various sensing methods. Eiman [4] presents a system based on opportunistic model that utilizes mobile phones’ GPS devices and microphones to collect data about noise levels in a city area, while Paxton et al. [12] describe a participatory study in which CO2 level measurements were observed with a handheld device and combined with written observations and video clips. Mednis et al. [10] developed a system for road irregularity detection by using data from accelerometers of mobile phones. Suggested architectures and platforms for participatory sensing include G-Sense [13], PRISM (Platform for Remote Sensing using Smartphones) [3], and the personal environmental impact report PEIR [11]. Participatory sensing tools have proved technically feasible, but motivating potential observers to collect data is still an open issue. Reddy et al. [15] suggest a recruitment framework for identifying potential participants for data collections, and Juong-Sik et al. [6] discuss an incentive mechanism for stimulating participatory sensing applications. Micro-payments as an incentive mechanism are explored by Reddy et al. [14]. Data privacy is one of the issues that must be considered in any participatory sensing system design, as usually data sent by users include information about user’s location and the time when the location was acquired. The privacy issue is discussed in a number of studies. Christin et al. [2] have made a survey on privacy in mobile participatory sensing applications. Their analysis shows that almost all applications capture location and time information. Methods for protecting privacy have been presented by Kazemi et al.
Keywords— participatory sensing; air quality; water quality; plant disease; mobile phone; architecture
I.
INTRODUCTION
Participatory sensing has recently become more popular method for collecting large environmental datasets, as people are more concerned about the global climate change and the state of the environment, and as mobile devices have become more capable and pervasive. Also, social media services enable people to share information more easily, including environmental data. Names for this emerging field are many, such as urban sensing, community sensing, people-centric sensing and citizen science, with each term having subtle nuances of meaning. By “participatory sensing” we refer to modern-day activity of citizens in collecting and reporting in-situ observations about their environment, and campaigns harnessing people for these activities, often performed using special location-based applications on mobile devices or via messaging in social media services such as Facebook and Twitter. Participatory sensing is anticipated to extend the sensor networks of environmental institutes and other organizations both spatially and temporally, and to produce useful information for situation awareness. Forecasting, remote sensing and scientific research can also benefit from wellprepared observation campaigns by providing ground-truth data for reference and model validation. In this paper we describe our experiences about applying participatory sensing to air quality, water quality and plant disease monitoring. Different pilots demonstrate different aspects of participatory sensing accounting for different target user groups, sensing methods and privacy needs. The 978-0-7695-4684-1/12 $26.00 © 2012 IEEE DOI 10.1109/IMIS.2012.70
RELATED WORK
155
[7], who propose a privacy-aware framework enabling participation of the users without compromising their privacy, and by Huang et al. [5], who propose a method for location privacy in participatory sensing. Determining the quality of user observations is important, especially when using the data for evaluating and validating predictive models or when it is used as groundtruth data for reference. Alabri et al. [1] describe a framework combining data quality control and trust metrics to enhance the reliability of citizen science data. Kuan et al. [8] and Yang et al. [17] propose reputation management to evaluate trustworthiness of gathered data. III.
Figure 1. EnviObserver’s conceptual architecture.
User’s location is determined automatically using either satellite positioning (GPS) or cell-based positioning data provided by the mobile network, depending on signal availability. GPS positioning is preferred, as positioning error of the cell-based method can be several hundred meters, even in a city environment. Usually, environmental observations are done outdoors where GPS availability is generally good. Information on which positioning method was actually used is available on some, but not all mobile platforms. Continuous usage of GPS for determining location can quickly drain the battery of a mobile phone, and therefore users are not expected to make observations continuously.
ENVIOBSERVER – A PARTICIPATORY SENSING TOOL
Human beings possess several qualities that make them good sensors. Firstly, humans themselves are pervasive – located practically everywhere without a need to install any sensors. Secondly, humans are sensitive to certain changes in the environment, and thirdly, humans often carry mobile devices capable of running software and accessing the network. Unlike many other participatory sensing applications that use integrated sensors of mobile phones or attached external sensors (e.g. [4][10][12]), EnviObserver is designed to gather subjective observations made by people, focusing on the types of observations that users make actively with their own senses, like vision and smell. With EnviObserver, users are able to report their estimate about the observed phenomenon, e.g. personal perception of air quality in the current location and time, or a symptom caused by environmental conditions. By correlating the observations with user profile information, e.g. by combining perceived breathing easiness with information about allergies, profound analyses can be performed. Entering readings from simple measurement devices such as a thermometer is also enabled. Due to a flexible data model design, automatic environmental sensors attached to mobile phones could be integrated as well. The architecture design of EnviObserver aims at enabling easy configuration of observational parameters, regardless of application area. The conceptual architecture (Fig. 1) encompasses the following parts: 1) mobile application for providing user observations, 2) interfaces for inputting observations and accessing the data, 3) visualization module for presenting data on a map, and 4) configurable alert services for notifying users about new interesting data.
B. Interfaces Interfaces are made available for receiving observations and for data retrieval, implemented as RESTful web services using HTTP POST and GET methods respectively. The interface for receiving observations enables development of various kinds of mobile applications for multiple platforms. The received observations, including textual and numeric data and images, are stored in a database, as are any profile data of registered users. A single observation consists of location (coordinates in WGS84 format), positioning method used (GPS or cell-based positioning), timestamp, the observed parameter, parameter value, and username, if one is provided. Observations are transmitted to the server over a secured channel. The data retrieval interface returns observations in KML (Keyhole Markup Language) format which can be easily used in visualization, for example in mapping services by Google. KML also supports the addition of arbitrary XML data, and with minor effort the data can be provided in other formats as well, such as proprietary XML or CSV formats.
A. Mobile Applications Users report their environmental observations with the help of mobile applications, designed to be simple so that reporting of observations is easy and fast. The user first selects the type of observation, defines the values for the parameters, optionally takes an accompanying picture and sends the observation to the server. If the number of parameters is small the user interface can be simplistic. So far, we have developed applications for Android and Java ME enabled devices.
C. Visualization and Services The visualization component (Fig. 2) allows users to browse near real-time and historical observations on a webbased map that can be freely zoomed and panned in space and time. The implementation is a mash-up built upon Google Maps and Google Earth components. The observations to be shown can be selected using a toolbar next to the map. Different types of observations are visualized with distinct symbols, which can be selected to view detailed information.
156
able to make observations and view them on a map. During the pilot, 101 air quality observations were received from 13 different users. Motivating the users to make observations appeared more difficult than anticipated, and extensive dataset could not be collected. The user pilot succeeded in demonstrating the technical feasibility of the concept, but it also demonstrated the problem of recruiting and motivating people to make observations. This problem is discussed in more detail in the discussion section. B. Water Quality The water quality pilot was initiated in 2010 with monitoring of water temperature, turbidity and depth of visibility. Here, observers did not rely solely on their own senses, but took advantage of two items of equipment: a thermometer for temperature readings, and a special device and method for observing water turbidity and depth of visibility, developed by the Finnish Environment Institute. Turbidity and visibility were post-processed on the server from the images taken and sent by users. Feasibility of the solution was verified on a field trial in Vesijärvi lake in Southern Finland by a group of students from University of Helsinki. In addition, an alert service was set up for water temperature in the swimming beaches of the Helsinki area. SMS alerts were sent automatically whenever user-defined temperature threshold values for selected beaches were exceeded. The alerts relied solely on temperature measurements by users themselves, as no automatic sensors with open data interfaces were available. The second water quality pilot in the summer of 2011 concentrated on algae observations at the Baltic Sea and inner lakes of Finland. Occurrences of algae were evaluated using a four-level scale (none / some / plenty / extreme amount of algae), and users could attach images from their mobile phone cameras to show their algae observations to others. This pilot was open to everyone without registration, requiring only a download of a mobile application, which was made available for Nokia (Java ME) and Android-based mobile phones. The applications were designed to be as easy to use as possible, and they included image-supported instructions for making the observations. During this second pilot, the Android application was installed 173 times and the Nokia application was downloaded 307 times. We received 374 algae observations in total. The number of distinct users is not known, as making observations was possible without registration and no mechanism was implemented to identify sources of observations – an issue that needs to be taken into account in the future work. In this pilot, no feedback was provided to the users about their observations’ relation to other observations or the current algae situation; this is an identified target for future development. More effort was devoted to marketing, which seems to have positively affected the observation activity. The observations are analyzed in more detail in the evaluation chapter, and the reliability issue is discussed in the discussion section.
Figure 2. EnviObserver’s web-based visualization. A user-submitted algae observation is opened for a closer look, with a description and an image.
Mechanism for alert services has also been implemented. The service alerts users automatically with SMS messages whenever the observed value of a certain parameter exceeds the user-defined limits. Thus users do not need to visit the visualization page regularly for the latest observations. The alert service has been tested in the water quality pilot for water temperature observations. IV.
PILOT APPLICATIONS
We have applied participatory sensing in pilot applications for monitoring air quality, water quality and plant diseases. These pilots demonstrate various aspects of participatory sensing accounting for different target user groups, sensing methods and privacy needs. A. Air Quality The air quality pilot was implemented during the summer of 2009, demonstrating the technical feasibility of the concept and combining participatory sensing observations with other data sources and services. The pilot relied on human senses exclusively; no additional sensors were used. The collected parameters were a subjective estimate about the overall air quality (good / average / bad) and any personally experienced symptoms related to air quality. In the air quality pilot, EnviObserver formed a part of a larger service prototype, which included a map presentation of official air quality measurements and forecasts within the Helsinki region, an SMS alert service for allergic people about changes in the pollen levels in the air, and a presentation of participatory sensing observations. The role of EnviObserver was to provide user observations to supplement the spatial density of the measurements from sparsely deployed official air quality measurement stations. The official data sources included air quality measurements collected by the Helsinki Region Environmental Services Authority (HSY), air quality model forecasts by the Finnish Meteorological Institute (FMI), and pollen data analysis from the Aerobiology Unit of University of Turku. During the pilot, 143 users registered to the SMS pollen alert service. Around 93% of the 90 users who answered the questionnaire about the service found it useful, and 62% claimed that it helped them control their allergies. 95 persons registered as EnviObserver users; only registered users were
157
official observations always include also the observations where no algae are present, and such bias is therefore not expected for the barometer values. We estimated the correlation of participatory sensing and official observations by using the Spearman’s rank correlation coefficient, which takes into account also nonlinear correlation between two datasets compared to normal correlation coefficient. The Spearman’s rank correlation between the participatory sensing observations and the algae barometer 2011 is 0.79 for sea areas and 0.55 for inner lakes. This indicates that participatory sensing and official algae observations have a relevant correlation, suggesting that participatory sensing can produce useful additional information for algae monitoring. To estimate how the amount of algae affects the observation activity, we computed the Spearman’s rank correlation also between the number of participatory observations and the barometer 2011 for sea areas (0.73) and inner lakes (-0.45). The negative correlation coefficient for inner lakes indicates that during the summer the activity of making observations has decreased even though the amount of algae has increased. This is assumed to be related to the motivation issue discussed below.
C. Plant Diseases The plant disease pilot is an example of a participatory sensing application aimed at a specialized group of users with special data needs. The pilot was executed during the summer of 2011, when the users made observations about the nine most common diseases affecting barley, wheat, rye and oat at nine farms in southern and central parts of Finland. The pilot was planned together with researchers from MTT Agrifood Research Finland, and the observers included one farming advisor and four plant disease researchers. The main motivation was to get additional information on the disease situation in order to estimate and decide when, where and how to spray pesticides. The observational data is utilized together with weather data and disease risk prognosis, which is anticipated to provide added value in terms of lower pesticide use and cost, and higher amount and quality of yield and environmental benefits. In the pilot a small but dedicated group of five users recorded 27 observations during the pilot. It appears that in a specific case where the benefits can be demonstrated, it is easier to find motivated users to utilize participatory sensing. This pilot also introduced the need for privacy preservation, as the farmers were not willing to make their data public in detail. The privacy issue is treated in more detail in the discussion section. V.
EVALUATION
To evaluate the usefulness of received participatory observations, we compared the collected data with reference data from other sources. In preparing the evaluation, a preprocessing was required to remove apparent test uses of the application, duplicate data records, and other susceptible observations that would distort the evaluation results. As algae is observed regularly by the Finnish Environment Institute, we chose to compare participatory cyanobacteria observations collected during the 2011 algae observation campaign with the statistics provided by the institute in the so called “algae barometer”. The barometer values are based on professional observations in total of 339 fixed locations in coastal sea areas and inner lakes around Finland. The values represent weekly averages of the amount of algae using the same four-level scale as in the user observations (none / some / plenty / extreme amount of algae, values 0-3). Fig. 3 and Fig. 5 present weekly counts of received user observations in sea areas and in inner lakes, respectively. Fig. 4 and Fig. 6 present the weekly averages of the participatory observations, together with the official algae barometer values and a long-term average of the barometer. Visual interpretation of the charts indicates some correlation between the algae barometer and user observations, even though the average values for amount of algae are clearly much higher for participatory observations than for the barometer. One probable reason for higher values for participatory observations might be people’s tendency to report algae occurrences more frequently than situations where no algae is observed. This causes a bias to the participatory sensing average values. On the contrary,
Figure 3. Number of received weekly user observations in sea areas.
Figure 4. Sea areas of Finland, 2011: comparison of observed amount of algae based on participatory observations and the professionally produced algae barometer. Also shown is a long-term average of the algae barometer.
158
interface, simplified the system by reducing the amount of observable parameters, and connected the visualization of observations to a popular lake information service run by the Finnish Environment Institute (www.jarviwiki.fi – currently only in Finnish). Water quality as the pilot subject probably also contributed to the improvement in observation activity, as normally the air quality is fairly good in Finland and concerns only allergic and sensitive people, while occurrences of algae are quite common in the summer and visible to everyone moving close to water. The plant disease pilot serves as an example of a case where a small group of motivated users can provide an interesting set of observations. Experiences from this pilot support the conclusion derived from the air quality pilot: to create a successful participatory sensing application, observers have to get clear benefits from providing data. Plant disease researchers were eager to utilize participatory sensing for plant disease monitoring and the feedback received was very encouraging. In this case, plant disease researchers will use the observation data in plant disease prognosis. As this pilot was designed together with plant disease researchers, our experiences in this pilot support the arguments presented by Sunyoung et al. [16] that creating a successful participatory sensing application requires designing the application together with various stakeholders and ensuring that the gathered data can be put to use. To summarize, participatory sensing applications need to be designed in a way that motivates users to make observations. An incentive or feedback mechanism should be included, and the incentive does not necessarily have to be monetary. We have recognized some ways to motivate the observers:
Figure 5. Number of received weekly user observations in inner lakes.
Figure 6. Inner lakes of Finland, 2011: comparison of observed amount of algae based on participatory observations and the professionally produced algae barometer. Also shown is a long-term average of the algae barometer.
• VI.
DISCUSSION •
A. Motivation Based on our findings, efforts are needed to recruit suitable observers and to motivate them to continue making observations, especially when observations are wanted from among the general public. In the air quality pilot and the first water quality pilot, we utilized advertisements on various web sites, read by thousands of people. This attracted about one hundred registered users, of whom, however, only 13 provided observations. Marketing of the services was deemed unsuccessful, and the users probably expected to receive information as well, not solely to produce it. This pilot showed that users generally won’t provide observations without getting any benefits from it. In the second water quality pilot, we took a more thorough approach to recruiting and relied on the help of the media. The recruitment invitation was published in local and national news services at the beginning of the algae season in Finland. Also, the possibility to share observations on Facebook was implemented in order to utilize social networks in the marketing. The marketing strategy was more successful than in the previous pilots, and over 300 algae observations were received. In addition to improving marketing, we enhanced the mobile application user
• •
• •
159
Relevance: observation campaigns should be meaningful to the user and relate to his or her everyday life and personal interests. Recognition: users should receive immediate feedback about the contributed observation, ranging from a simple “thank you” message to real-time map visualization of observations. Of additional interest are statistics about the accumulation of observations, both personally and in relation to the user’s reference groups, and user’s current position in various rankings based on observation activity. Status: users should gain the status of a recognized observer through continued observation activity. Reward: in compensation for their efforts, users should in return receive reward, either monetary or service value, or exclusive information that is not easily or freely available elsewhere. For example, SMS notifications can be offered about risen pollen levels in the user’s current location. Social linking: users should be able to connect and share experiences with peer observers. Scientific contribution: users should be given access to information about the observation campaign and the effect that their observations have. Users can even be invited to participate in the design and
•
implementation of the campaign and the analysis of the results. Reminding: observation activity can be revitalized via kindly reminding the users to continue making observations and by inviting them to new campaigns.
according to individual sensitivity and mood. Observation tasks should be designed to cope with these error sources. In the described pilots, no error handling for positioning was performed for the observations whose location was inaccurate due to cell-based positioning. Depending on the programming interfaces of various mobile device platforms, information about the used positioning method is not always available. This issue needs to be addressed in cases where exact positioning is crucial. During the performed pilot trials, no service misuse was detected, and no actions to remedy that kind of activity were necessary. Also, due to the experimental nature of the pilots, obvious test uses of the system were detected manually and no automatic identification of faulty or accidental observations was implemented. All observations were stored and made available as such, without corrective actions; i.e., interpretation and assessment of the observations were left to the user of the data. In production use, however, automatic identification of faulty observations is important, and a more advanced assessment of the reliability of observations is needed. Especially when the information is used in critical decision making, for example in avoiding areas of bad air quality or pollen levels, or deciding whether to take medicine or not. Based on our pilot experiences, we have identified a number of methods for managing the issue of reliability:
B. Privacy In our work, we discovered a contradiction between privacy preservation and the usability of the data. On the one hand, details about the observers are needed for evaluating the observational dataset, but on the other hand, requiring user profile data probably limits the amount of interested observers and thus useful observations. The first version of the air quality pilot required registration, but in later versions the mandatory registration was removed, e.g. the algae monitoring pilot was implemented completely without registration. Based on our experiences, requiring registration reduces the amount of observers as all users are not willing to fill in the registration forms with profile data. Our experiences from the pilots seem to support this assumption, although a more systematic analysis would be interesting. Even though registration would enable recognition of users and more accurate correlation between users and observations, each extra step required in the installation phase reduces the amount of interested users. As not all users are willing to share their observations with others, the possibility to submit anonymous observations would probably attract new users to the system. On the other hand, since the identity of the participants is not known by the application, it will not be able to identify and filter observers that report invalid or faulty data. Not requiring registration leaves also a need for identifying the number of distinct observers, for utilization in data analysis. One way to accomplish this would be using identification code in the client application. In the data analysis, anonymous user ID could be then used to identify how many observations a single user has sent. In addition, user reputation systems usually require identifying users to evaluate the past performance of the observers [1] [17]. In the plant disease pilot, we faced another issue concerning privacy. The farmers were happy to help researchers in making plant disease observations, but they were not willing to tell everybody about the plant disease findings concerning their farm. As we wanted to show every observation in the public map presentation, privacy of the farmers was preserved by distorting the exact shown location of disease observations and by limiting the zoom-in level of the map, so that viewers could not recognize individual farms. The exact coordinates of the observations were stored for analysis purposes.
•
•
•
•
•
C. Reliability In any data gathering task relying on volunteer user contributions, there is a risk of faulty input by human errors or even by service misuse. The risk for misuse is obviously higher when no registration to the service is required. In addition, when the observations are based on human senses the measurements are not of uniform quality, but vary
Do not rely on single observations. Single observations should not be made available to endusers without appropriate notice. Before presenting the data, several observations of the same parameter should be collected from different persons, from the same geographical area and the same time period. Many similar observations can be assumed to reflect a real phenomenon in the environment. Discard outliers. In cases where a small portion of many observations of the same parameter (in the same area and period of time) is clearly in contrast with majority of observations, it may be reasonable to judge the outliers as erroneous and discard them from the rest of the analysis. Build trust in individual observers. Observations made by individual users can be monitored and their credibility validated over an extended period of time. More weight can then be put on reports of trustworthy observers. Train observers. Observations made by users who have been educated on the subject, can be assigned elevated trust. However, in many cases training is impractical to implement. Utilize peer review. Peer review can be feasible in some but probably not in all applications.
Additionally, in the evaluation performed we noted that while simple averaging works fine as an indicator value for the professional observations (algae barometer) in fixed locations and fixed time intervals, it is not best suited for the user observations, because the number of consecutive observations is not controlled. An enthusiastic observer can
160
REFERENCES
provide half a dozen reports of the same phenomenon in a short timeframe and at the same approximate spot. These observations have a considerable effect on average values calculated from that area, even though they represent essentially only a single observation. This shows that simple averaging of observations without a proper preprocessing is not enough.
[1]
[2]
VII. CONCLUSION AND FUTURE WORK
[3]
In this paper we have presented our experiences about using participatory sensing in environmental monitoring. We have utilized participatory sensing for monitoring air quality, water quality and plant diseases. For gathering observations from people we have developed a tool for participatory sensing called EnviObserver. The tool consists of a mobile application for providing user observations, a data model and database for storing data, interfaces for accessing and for inputting data, a map-based visualization module and an alert service that is used to deliver information to the users. Based on our findings, participatory sensing and official algae observations have relevant correlation, and participatory sensing can therefore produce useful additional information for algae monitoring. This is an encouraging preliminary result that supports plans for more thorough evaluations in future campaigns. To carry out successful participatory sensing campaigns, the users need to be motivated to report their observations. A feedback mechanism should be implemented to not only remind the user to make observations, but also to provide feedback about the user’s observations in relation to other observations in the area. It is important to design the participatory sensing applications together with relevant stakeholders of the campaign, and to ensure that the collected data will be valuable and have concrete use. Privacy and reliability of observations are also important issues to be solved when creating participatory sensing applications. In our future work, we are aiming toward a flexible design that would enable more straightforward configuration of new observable parameters. The user groups, observable parameters and other settings are defined in the server side, and the mobile application user interface with environmentrelated questions is formed dynamically when the user starts the application. Dynamism of the system provides also a possibility to recruit observers from one application area to facilitate making observations for other applications.
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
ACKNOWLEDGMENT [13]
The authors would like to thank the following people for their contributions and cooperation in the various participatory sensing pilots: Timo Pyhälahti, Sampsa Koponen, Matti Lindholm, Jari Silander and Maria Kämäri from the Finnish Environment Institute and Ville Peltola from IBM (water quality pilot); Sirpa Thessler and Marja Jalli from MTT Agrifood Research Finland (plant disease pilot); Ari Karppinen and Virpi Tarvainen from the Finnish Meteorological Institute, Pontus Lindman from Medixine Ltd. and Kostas Karatzas and his group from Aristotle University of Thessaloniki, Greece (air quality pilot).
[14]
[15]
[16]
161
Alabri, A., Hunter, J. Enhancing the quality and trust of citizen science data. (2010) Proceedings - 2010 6th IEEE International Conference on e-Science, eScience 2010, art. no. 5693902, pp. 81-88. Christin, D., Reinhardt, A., Kanhere, S.S., Hollick, M. A survey on privacy in mobile participatory sensing applications. (2011) Journal of Systems and Software, 84 (11), pp. 1928-1946. Das, T., Mohan, P., Padmanabhan, V., Ramjee, R. and Sharma, A. PRISM: Platform for remote sensing using smartphones. In Proceedings of the 8th International Conference on Mobile Systems, Applications, and Services (MobiSys '10). ACM, New York, NY, USA, 63-76. Eiman, K. 2010. NoiseSPY: A Real-Time Mobile Phone Platform for Urban Noise Monitoring and Mapping. Mob. Netw. Appl. 15, 4 (August 2010), 562-574. Huang, K. L., Kanhere, S.S. and Hu, W. Preserving privacy in participatory sensing systems. Computer Communications, 33, 11 (2010), 1266-1280. Juong-Sik, L. and Hoh, B. Dynamic pricing incentive for participatory sensing. Pervasive and Mobile Computing, 6, 6 (2010), 693-708. Kazemi, L. and Shahabi, C. Towards preserving privacy in participatory sensing. 2011 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops) (March 2011), IEEE Computer Society, Seattle, WA, United States Kuan L. H., Salil S. K., and Wen H. Are you contributing trustworthy data?: the case for a reputation system in participatory sensing. In Proceedings of the 13th ACM international conference on Modeling, analysis, and simulation of wireless and mobile systems (MSWIM '10). ACM, New York, NY, USA, 14-22. Lane, N., D., Eisenman S., B., Musolesi M., Miluzzo, E. and Campbell, A., T. Urban sensing systems: opportunistic or participatory? Proceedings of the 9th workshop on Mobile computing systems and applications (2008). Mednis, A., Strazdins, G., Zviedris, R. and Kanonirs, G. Real time pothole detection using Android smartphones with accelerometers. In Proceedings of the 7th IEEE International Conference on Distributed Computing in Sensor Systems and Workshops (DCOSS’11), pp. 1-6. Mun, M., Reddy, S., Shilton K., Yau, N., Burke, J., Estrin D., Hansen, M., Howard, E., West, R. and Boda, P. PEIR, the personal environmental impact report, as a platform for participatory sensing systems research. MobiSys '09 Proceedings of the 7th international conference on Mobile systems, applications, and services. 2009. Paxton, M. and Benford, S. Experiences of participatory sensing in the wild. In Proceedings of the 11th International Conference on Ubiquitous Computing (UbiComp '09). ACM, New York, NY, USA, 265-274. Perez, A., Labrador, M. and Barbeau, J. G-Sense: A Scalable Architecture for Global Sensing and Monitoring. IEEE Networks. 24, 4 (2010), 57-64. Reddy, S., Estrin D., Hansen M. and Srivastava, M. Examining Micro-Payments for Participatory Sensing Data Collections. In Proceedings of the 12th ACM International Conference on Ubiquitous Computing (Ubicomp '10). ACM, New York, NY, USA, 33-36. Reddy, S., Estrin, D. and Srivastava, M. Recruitment framework for participatory sensing data collections. In Proceedings of the 8th International Conference on Pervasive Computing (May 2010), pp. 138-155. Sunyoung, K., Robson, C., Zimmerman, T., Pierce, J. and Haber, E. Creek Watch: Pairing Usefulness and Usability for
Successful Citizen Science. In Proceedings of the 2011 Annual Conference on Human Factors in Computing Systems (CHI '11). ACM, New York, NY, USA, 2125-2134.
[17] Yang H., Zhang J., Roe P. Using Reputation Management in
Participatory Sensing for Data Classification, Procedia Computer Science, Volume 5, 2011, Pages 190-197, ISSN 1877-0509, 10.1016/j.procs.2011.07.026.
162