Fueled by the increasing proliferation of citizen generated spatio-temporal data -- especially in participatory urban infrastructure monitoring -- municipal authorities are in need for ways to process and understand increasingly overwhelming amounts of data. However, duplicate issue reporting by citizens such as broken traffic lights, potholes or garbage can lead to bottlenecks in manual processing of such data. As contribution this paper examines which city issue report presentation methods are useful to support a human in analyzing and processing them. We compare presentation methods such as automatically clustered information, manual clustered information and mixes of both. Automatically clustering of information is performed by a data analytics algorithm which is also presented in this paper together with EstaVis, a prototype of an interactive visual urban analytics platform. Evaluation studies with 282 crowd-workers show how the platform can potentially help to speed-up report processing by detecting and aggregating duplicate reports by up to 3 orders of magnitude and discuss which lessons can be learned in terms of features and user experience pitfalls for this kind of system.
Crowdsensing has been widely investigated and its applications are being explored. We have used accelerometers embedded in the smartphone carried by bicycle users to detect road surface damages. However, detected damages locations are not aggregated and often, the same road surface damage can be detected several times at different locations due to measurement errors. In this study we propose a scheme to cluster damages locations to remove redundancy caused by measurement errors.
A number of local governments, businesses, and research institutions have discussed the usefulness and importance of "green networks" in recent years. However, most green sites are still perceived as isolated distributed patches, thereby potentially leading to their underuse and little awareness of their importance. We propose a participatory environment that can visualize connections of green sites based on the levels of visible green using celestial sphere images contributed by citizens. The system combines existing green coverage maps and smartphone-based image capture tools to show "green routes" in a city. We expect that the proposed approach can increase the networked uses of green resources, thereby increasing the awareness of green networks and potentially leading to collective efforts towards the development of richer connected green in cities.
This research investigates what characteristics foreign tourists expect of areas near tourist attractions in Japan, compared with what domestic tourists expect, in order to find out how to make the areas more attractive to foreign tourists. We develop a framework to use spatio-temporal data from Twitter and Foursquare to show the difference of preferences between foreign tourists and domestic tourists. First we extract the locations of tourist attractions from the data of domestic tourists. Then we characterize each location by using Foursquare's location information that has categories such as restaurants, shops, historic sites, etc. After characterizing each location, we make decision trees that explain what kind of combination of characteristics is important to attract foreign tourists and domestic tourists. Finally, from the results of the difference of preferences between foreign tourists and domestic tourists, we propose solutions to improve the environments of tourist attractions. In this paper, we apply this framework to the data gathered in August 2014 in Japan, and we conclude that foreign tourists expect nightlife spots (bars, nightclubs, etc) of the neighborhoods of tourist attractions whereas domestic tourists do not.
Geofencing mechanisms allow for timely message delivery to the visitors of predefined target areas. However, conventional geofencing approaches poorly support mobile data collection scenarios in which experts need in situ assistance. In this paper, we propose crowd geofencing environments, in which a large number of crowdworkers generate geofences to support mobile experts. As a first step to open up the possibilities of crowd geofencing, we have tested its feasibility by collecting more than one thousand geofences in an unfamiliar city prior to the visit to look into urban water and air quality issues. Our experience has revealed the strengths and weaknesses of crowd geofencing in terms of geofence quality and crowd-powered situated actions.
Previous works in architecture and social science found that aspects of the built environment such as density, connectivity, and house typologies are related to crime. However, these studies are qualitative, and thus hardly repeatable at larger scales. In this work, we overcome this limitation by offering a quantitative approach that explores the relationship between the configuration of the built environment and the activity of criminal groups in city areas. The method extracts a wide set of metrics related to aspects of urban form from openly accessible datasets. We then input these metrics in a step-wise logistic linear model, using presence of gang activity as dependent variable, and obtain a parsimonious model with an excellent fit when applied to the metropolitan area of London, UK. We then use values and slopes of model coefficients to build a narrative of the typical city area characterized by gang activity, re-connecting to previous theories. Outcomes of this research can help policy makers and architects in better understanding the relationship between neighborhood design and criminal activity.
While penetration of wireless information-centric spaces at university and industry campuses is steadily increasing, most are implemented as gateways to either wired (ADSL, etc.) or cellular (3G/LTE, etc.) Internet. Such infrastructure is expensive and requires intensive maintenance. This paper proposes an infrastructureless infocentric space built from a number of fully autonomous wireless hubs installed at density centers across the campus. Hubs themselves are not connected to the Internet (hence, low installation and maintenance cost) and instead depend on P2P syncs with wireless devices of passing by users. Hubs treat these syncs as virtual Internet connections and rely on them instead of a traditional infrastructure. This paper shows that such hubs would quickly fill up with content at which point the direction of syncs would reverse -- this is when a high proportion of users would download content from such hubs instead of relying on slow/congested/expensive 3G/LTE connections.
Digital Signage (DS) is one of the popular IoT technologies deployed in the urban space. DS can provide wayfinding and urban information to city dwellers and convey targeted messaging and advertising to people approaching the DS. With the rise of the online-to-offline (O2O) mobile commerce, DS also become an important marketing tool in urban retailing. However, most digital signage systems today lack interactive feature and context-aware recommendation engine. Few interactive digital signage systems available today are also insufficient in engaging anonymous viewers and also not considering temporal interaction between viewer and DS system. To overcome the above challenges, this paper proposes a context-aware recommender system framework with novel temporal interaction scheme for IoT based interactive digital signage deployed in urban space to engage anonymous viewer. The results of experiments indicate that the proposed framework improves the advertising effectiveness for DS system deployed in public in urban space.
Non trivial part of IoT is sensor devices that are energy constrained. Assuming the devices are going to perform tasks together for a certain scenario, collaborating with each other and exchange semantics in an energy efficient way is a critical aspect. To address this issue, we propose a clustering algorithm for collaborative processing in IoT network. The algorithm is evaluated on actual IoT platform based simulator cooja. The parameters like network coverage, communication cost and power consumption analysis are evaluated by conducting experiments.
Recently, many researchers have been focusing on the detection and classification of urban events by information analysis on social networks. Previous works mainly use text analysis of users' posts on social networks for detecting urban events. However, this approach has a limitation that the users' posts must mention the event for the analysis to be conducted. We propose a new method for classifying urban events by extracting user interest from the location-based social network information without text analysis. The proposed method includes analyzing common friends of users in the vicinity of the event venue and extracting common the friends' attributes by referring to related Wikipedia information. We designed and implemented the proposed method, and conducted an experiment for evaluating our method. Our experimental result shows that our method can classify events well in cases where participants have similar interests.
Internet-enabled, location aware smart phones with sensor inputs have led to novel urban infra-structure monitoring applications exploiting unprecedented high levels of citizen participation in dense metropolitan areas. For policy makers, it is a key task to keep track of trends and developments of reported infra-structure issues for understanding and effectively reacting to problems around a city, specially in their early stages. In contrast to previous strategies which consider only limited information such as text and geographic locations, we analyze the urban dynamics of crowdsourced collected data using an existing approach that considers a novel modeling of heterogeneous attributes and relationships in the data. First, the underlying data is modeled into a heterogeneous network, in which it's measured for each node its current level of anomalousness for a desired time interval (e.g. a week) and then the most anomalous network's subgraph is extracted and described by means of problem category, geographical area, time and participants. First experiments illustrate the effectiveness and efficiency of leveraging this anomaly detection approach in our use-case (participatory infra-structure monitoring).
We present a pilot study that uses optical head-mounted displays (OHMD) as an augmented reality headset to view and define rules among various augmented objects in the environment. In the traditional augmented reality techniques on mobile devices, the users must hold the devices in their hands as a viewing window to the physical world. Instead, the OHMD are attached to the users' head, allowing them to view objects and related information by turning their heads and directly gazing at the objects. In a crowded urban environment surrounded by IoT devices, OHMD can reveal the presence and capability of the devices to users. Also, free movement of hands leave an opportunity to use them for interaction, e.g. via mid-air gestures. This paper describes an early prototype and a preliminary result from a pilot study to test feasibility of interacting with virtual objects that augment physical objects using OHMD. We found that basic interaction for building new rules among them were easy to learn and use, while fine-tuning of them using the conventional GUI components left rooms for improvement.
Ever-connected smart phones and advanced sensors have lead to new sensing paradigms that promise environmental monitoring in unprecedented spatio-temporal resolution. Especially in air quality sensing with low-cost sensors, regular in-situ device calibration is a helpful approach to ensure data quality. In participatory sensing scenarios, privacy implications arise, as personal sensor data, time and location need to be exchanged. We present a novel privacy-preserving multi-hop sensor calibration scheme that combines Private Proximity Testing and an anonymizing MIX network with cross-sensor calibration based on sensor rendezvous. Our evaluation with simulated ozone measurements and real-world taxicab mobility traces shows that our scheme provides privacy protection while maintaining competitive overall data quality in dense participatory sensing networks.
The nexus of changes in personal technology and human behavior has created new opportunities to understand cities by mapping the large-scale movements of goods and people through the use of GPS and GIS. Examples now abound making use of these technologies at the large-scale; we have chosen to look more carefully at individuals in cities. Our research has produced initial detailed studies from which more general urban behavioral and space syntactical patterns begin to emerge. We present two case studies that use food as a proxy resource relative to behavior. Our data collection methodology included both digital and traditional techniques, and our evolving analytical methodology draws upon visualization, interview-based observations and statistical analysis to offer qualitative and quantitative observations.
Over the past years, the impact of spatial characteristics on subjective well-being has started to receive attention but mostly on the macro granularity of sub-national level. In addition the studies that focused on the spatial scale of urban cities and their neighbourhood have mainly examined the influence of environmental perspectives, land use and urban morphological features. The influence of geographical contexts such as city attributes however have been studied sporadically with somewhat contradicting observations. In this work we focus on the theoretical foundation of subjective well-being and through the discrepancy theory we examine the impact of what is offered and desired by citizens on their subjective well-being. We model functionalities a neighbourhood offers in terms of density, diversity and rarity of its human-made amenities. To infer whether these functionalities fit the desire of residents we model their propensity to travel in and out of an area through large scale analysis of urban mobility flows. Our analysis supports discrepancy theory by showing that the the gap between what is offered and desired is a good predictor for subjective well-being.
Detailed information about the flow of potential customers in a city is extremely relevant for strategic decisions of various service providers such as taxi companies or advertising agencies. The knowledge about highly frequented regions as well as peak times in specific areas provides a crucial business advantage to competitors. Today, business relevant decisions about the positioning of service providers and advertising spaces or the balancing of capacity are primarily based on experience only. In this paper, we present a novel approach to gain knowledge about the distribution of potential customers over time and space based on the data of taxi rides, which have been recorded for documentation purposes. By leveraging the performance of in-memory databases, we build an application, which allows the user to analyze about 700 million taxi rides in real-time. The application allows companies to get an impression in which areas and in what timeframes they can reach a large audience of potential customers. Additionally, we demonstrate that the developed visualization concept enables the comparison of different regions and allows to analyze trends in the customer flow over time.
Choosing the right technologies to build an urban-scale IoT system can be challenging. There is often a focus on low-level architectural details such as the scalability of message handling. In our experience building an IoT information system requires a high-level holistic approach that mixes traditional data collection from vendor-specific cloud backends, together with data collected directly from embedded hardware and mobile devices. Supporting this heterogeneous environment can prove challenging and lead to complex systems that are difficult to develop and deploy in a timely fashion. In this paper we describe how we address these challenges by proposing a three-tiered DevOps model which we used to build an information system that is capable of providing real-time analytics of Electric Vehicle (EV) mobility usage and management within a smart city project.
In recent years, as GPS-enabled mobile phones have made spreading location data much more accessible than before. Various applications especially concerning urban IoT service collect users' historical location data and consider how to exploit this data to improve the application or service. This paper describes that location log data is useful to estimate user preference and it is verified whether our hypothesis holds true. Two methods to recommend news articles using location log data are proposed. These methods are evaluated by actual application and then counting the number of articles that prove interesting to users compared with using and not using location log data. It is found that the best method for news article recommendation is the method, that labels location log data by Bayesian model "location hierarchical Dirichlet process" (LocHDP) and classifies users, thus demonstrating the usefulness of location log data in terms of news recommendation.
How can we estimate the location of user-generated content using textual data without location-specific information to understand urban space? Understanding urban space is important to tackle the issues that cities face, e.g. disasters, traffic congestion. Although event information reported with location data on microblog are very informational, many users post them without their locations because of the privacy concerns. To address this issue, some studies have attempted to estimate the location where the users post their tweets by analyzing the text. While those works have introduced various techniques for effective estimation, they have taken a lot of effort to do so. In this paper, we propose an approach that can estimate the location without those efforts. To achieve this goal, we adopt bidirectional Long-Short Term Memory (BLSTM). In our experiment, we use the geotagged tweets that were posted in Japan and treat location estimation as a multi-class classification problem where the location of tweets should be classified into administrative districts. As a result, we show that our proposed method can classify the location of tweets with higher accuracy than baseline methods.
Collaborative data processing is needed for sensing, analyzing and visualization of heterogeneous data across IoT platforms. Localization is required to maintain up-to-date position information, both for indoor and outdoor network infrastructures. This helps in real time context aware processing and service delivery in smart city applications. In this paper, a two layer localization scheme is proposed for a distributed network using the Internet of Things.
