An Empirical Study of Combining Participatory and Physical Sensing to Better Understand and Improve Urban Mobility Networks

The rapid rise of location-based services provides us an opportunity to achieve the information of human mobility, in the form of participatory sensing, where users can share their digital footprints (i.e., checkins) at different geo-locations (i.e., venues) with timestamps. These checkins provide a broad citywide coverage, but the instant number of checkins in urban areas is still limited. Smart traffic control systems can provide abundant traffic flow data by physical sensing, but each controlled region only covers a small area, and there is no user information in the data. Here we present a study combining participatory and physical sensing data, based on 3.4 million checkins collected in the Pittsburgh metropolitan area, and 125 million vehicle records collected in a subarea controlled by an adaptive urban traffic control system. Our aim is to disclose how we could utilize the combined data for a better understanding on urban mobility networks and activity patterns in urban environments, and how we may take advantage of such combined data to improve urban mobility applications such as anomaly traffic detection and reasoning, topic-based nontrivial traffic information extraction, and traffic demand analysis. Xie and Wang 2 INTRODUCTION Understanding human mobility (1, 2, 3, 4, 5) and activity patterns in urban environments is a significant and fundamental issue from various perspectives, e.g., understanding regional socioeconomics, improving traffic planning, providing local-based services, and promoting sustainable urban mobility. Traditionally, relevant information is however rarely obtained due to difficulties and costs in tracking the time-resolved locations of individuals over time. In recent years, an increasing attention has been placed on introducing smart traffic control systems (6, 7) into urban road networks. The primary objectives of such systems are to reduce travel time, resolve traffic congestion, and reduce vehicle emissions. Recent work in real-time, decentralized, schedule-driven control of traffic signals (7, 8) has demonstrated the strong potential of real-time adaptive signal control in urban environments. The smart and scalable urban traffic control system (9) achieved improvements of over 26% reductions in travel times, over 40% reductions in idle time, and a projected reduction in emissions of over 21%, in an initial urban deployment. To facilitate effective real-time control, vehicle flows can be monitored by different physical sensors, e.g., induction loops and video detectors, and pedestrian flows can be detected and inferred using push-buttons or other devices. Although traffic flow data in fine granularity can be logged in real time, usage of these physical sensing data is often limited to the region that is being controlled. For the broad uncontrolled regions instead, no information is available. The rise of location-based services provides us another way to achieve the information of human mobility, in the form of participatory sensing (10, 11, 12). With mobile devices, users can share their digital footprints at various geo-locations (i.e., venues) with timestamps through checkins, e.g., geo-enabled tweets and geo-tagged photos and videos. Using of these services grows fast worldwide, although the instant sampling rate of trajectories is still very limited, and some webbased services, e.g., Waze and Facebook, do not open their location-based data to public. Research work has been conducted to understand temporal, spatial, social patterns, and some combined patterns of human mobility (13, 14, 15, 16, 17, 18, 19, 20, 21, 22). In this study, we focus on exploring the potentials of combining participatory and physical sensing data in urban mobility networks. The participatory sensing data are collected in the Pittsburgh metropolitan area with the APIs provided by the location-based services including Twitter, Foursquare, Flickr, Picasa, and Panoramio. The physical traffic flow data are collected in an area controlled by the scalable urban traffic control system (9). Basic spatial and temporal characteristics are displayed for both the physical and participatory sensing data. We first study human mobility patterns for a better understand of the urban mobility network. User checkins are examined to disclose the distribution of user behaviors, a fundamental statistical property of mobility pattern. Geo-location based cluster analysis is performed to identify personal favorite places of users in the studied regions. User entropy is measured to reveal the degree of predictability of user activities. Time-dependent mobility patterns are analyzed to show the regularity of user behaviors, based on primary and secondary most visited places of users. We then further inform how we may benefit from combining physical and participatory sensing data to improve urban mobility applications with three examples. First, we evaluate the attraction and limit of using sensing data for anomaly detection and reasoning of traffic flow. Second, we examine if nontrivial information could be extracted from participatory sensing data to effectively recognize traffic congestion in temporal and spatial dimensions. Third, by choosing two zones in the controlled region, we perform a close check on the correlation between physical and participatory sensing patterns. For the two zones, we also illustrate how we may investigate Xie and Wang 3 the origin and destination (O-D) patterns that are valuable for urban mobility from the transitions between user checkins. sensing data (a) Participatory Sensing Region (b) Physical Sensing Region (c) Checkin Locations (d) Heat Map of Checkins FIGURE 1 : Participatory and Physical Sensing Regions in Pittsburgh, PA. DATA DESCRIPTION Data Collection We implement our study in the Pittsburgh metropolitan area. The participatory sensing data contains a list of checkins. Each checkin can be represented as a tuple , where userID is associated with a unique user, venueID is associated with a venue at the geo-location of (latitude, longitude) with the precision of six decimal places. Our checkin data were collected (between March and July of 2014) from the geo-APIs of some location-sharing services, including geo-enabled tweets from Twitter and geo-tagged photos from Flickr, Picasa Xie and Wang 4 and Panoramio. We also included existing checkin data directly crawled from Foursquare (23). For studying urban mobility patterns, we only consider checkins at venues in the spatial latitude/longitude bounding box of (40.309640, -80.135014, 40.608740, -79.676678), as shown in Figure 1a. For studying up-to-date patterns, we only consider the recent data within the range of dates [1/1/2012, 7/1/2014]. The collected data contains 3,399,376 checkins of 74,658 users at 2,198,572 venues. The physical traffic flow data are collected in a road network which is currently controlled by the smart and scalable urban traffic control system (7, 8, 9) (see Figure 1a, and for more details see Figure 1b). The system was first installed on nine intersections (A to I) in the East Liberty neighborhood since June 2012 (see the pink region in Figures 1a and 1b), and then expanded to nine more intersections (J to R) in the Bakery Square neighborhood since October 2013 (see the green region in Figures 1a and 1b). The total area includes five major streets, Penn Ave, Centre Ave, Highland Ave, East Liberty Blvd, and Fifth Ave, with dynamic traffic flows throughout the day. For collecting real-time flow data, detectors were deployed on each entry/exit lane at the near end of each intersection and on entry lanes at the far end of boundary intersections. For each detector, a vehicle record is generated at the time when a vehicle is detected to pass the detector. Compared to that in a checkin, there is no userID and comment information available in a vehicle record. In total, the traffic flow data contains 125,369,318 vehicle records generated at 126 stop-bar detectors by the end of 7/1/2014. Basic Spatial and Temporal Characteristics Spatial Distribution of Checkins We first display the spatial distribution of all checkins in the whole region. Figure 1c shows the geo-distribution of checkins. It reflects the highly non-uniform dynamics of human mobility in the urban area. Figure 1d shows the heat map of checkins, where the high-density regions are clearly colored in red. Notice, one of the red regions in Figure 1d overlaps with the controlled region in Figure 1b. Temporal Patterns We are interested in the recurrent nature of human mobility over time. To investigate temporal mobility patterns, each week is segmented into 24×7 = 168 hourly bins (starting from Monday). For obtaining seasonal results, each year is divided into four seasons (A to D), and the binned results are averaged over 13 weeks in each season. We considered all four seasons in 2013 and the first two seasons in 2014. Figure 2a gives the seasonal checkin patterns in the participatory sensing region. It shows that the number of checkins has increased significantly over the seasons. Figure 2b shows the checkin frequency normalized by the total checkin size in each season. It shows that the checkin frequency has similar patterns for different seasons. The social day (11, 20) of Pittsburgh starts at around 4AM, and the checkin frequency peaks at around 8-9PM. A high checkin activity during Sunday is disclosed by Figure 2b. For vehicle flow in the controlled region in Figure 1b, we consider two pivotal intersections which service most vehicles in this road network: the intersection D of Centre Ave and Penn Ave in East Liberty, and the intersection P of Fifth Ave and Penn Ave in Bakery Square. For intersection P, the vehicle flow data is only available for the two seasons in 2014. The seasonal average vehicle flow patterns with physical sensing data of the intersections D and P are respectively shown in Fig-