Cell Phone Location Data for Travel Behavior Analysis (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

133 The objective of this appendix is to document and explain in a stand-alone section some of the key new concepts related to the cell phone data and their analysis. The objective is to remove the jargon and use nontechnical terms to describe the essence of each term and its relevance to the study. active and passive CDR data. Active use of a cell phone device includes making a call, sending a message, or visiting a website; passive use includes receiving a call, receiving a message, or being pinged by an app running in the background. Both active and passive signals trigger the capture of CDR data from a single cell phone. advanced positioning. Advanced positioning techniques offer a finer spatial resolution than tower-based methods and include triangulation, transmission delay from multiple base sta- tions, and other techniques of identifying the location of phones anywhere in a cell. Addi- tional infrastructure such as monitoring devices along freeway segments and technology such as built-in GPS receivers in phones can provide accurate phone location. call detail record (CDR). Call detail records are automatically collected by service providers for billing purposes. Each record contains time-stamped coordinates of anonymized customers when they use their phone in the cellular network either actively or passively. CDR location data. The compiled data set of CDR records has locations for each cell phone device. Locations are inferred either by observing the cellular tower through which the phone is connected or by triangulation with nearby towers. cell phone tower. Network operators often record the location of mobile phones in terms of the cell tower to which the phones are currently connected, in part because of privacy consider- ations. In this case, users’ traces are represented by time-ordered sequences of cell tower IDs. If the geographic locations of cell towers are known, the latitude and longitude coordinates of the tower are used. clustering method. A second method used to aggregate spatial points. The clustering method recognizes the uncertainty in location estimates. Although a device remains at the same location, it may be assigned to multiple neighboring location estimates. Clustering-based approaches allow points to be aggregated by using zones with arbitrary shapes. These methods use distance or travel time thresholds as inputs. device oscillation. This is the phenomenon in which, although a cell phone is stationary, a radio signal from multiple nearby towers may reach the device. In such a case, a station- ary device may appear to move between multiple towers, which creates the appearance of movement. Glossary

134 Cell Phone Location Data for Travel Behavior Analysis grid method. One method of aggregating spatial points is by imposing a grid over the space and aggregating points that fall within each grid cell. This method depends on the layout of the grid, including the size and shape of the grid cell. grid-based stay extraction. This process identifies stay regions by using a grid-based clustering method to cluster stay points. A study area is divided into rectangular cells; all stay points are mapped to the appropriate cell; and an iterative method is used to merge each unlabeled cell with the maximum stay points and its unlabeled neighbors to create a stay region. inter-event time. The inter-event time is the time gap between successive uses of a cell phone for calls, text messages, and Internet data access. Frequent cell phone use results in shorter inter-event times and thus provides more location data points and a richer data set on travel patterns. Infrequent usage of a cell phone provides a more limited set of travel information with fewer available location data. location of cell phone device. The location of a device is recorded when a phone call, text mes- sage, or data request is registered by carriers for billing, network performance, and legal purposes. medoid. A medoid represents a cluster and is similar in concept to means or centroids. A medoid is a member of the data set and it minimizes the distance between points that belong in a cluster and a point designated as the center of that cluster. The term is used in data clustering algorithms in computer science. A set of medoids is first chosen at random and the distances to the other points are computed. Data are clustered according to the medoid to which they are most similar and the medoid set is optimized in an iterative process. point-based algorithm. This process identifies stay regions by using a point-based method that exploits the maximum spatial accuracy possible. This method is similar to those originally designed for processing GPS traces. It is tailored to process cell phone data, which have lower locational accuracy and gaps in space and time, to extract individuals’ whereabouts. positioning. The position of a device is recorded by network operators when a user com- municates with the network. Positioning data describe users’ locations only when an event occurs. Every time a user initiates a network connection event (voice call, text message, or data access), the cellular network operator needs to know the user’s location to determine the cell tower to channel this event. Cellular network operators do not maintain positions of users at all times in order to improve network performance, save bandwidth, and protect users’ privacy. spatial resolution. Unlike the accuracy of continuous GPS traces, the spatial resolution of cell phone data depends on the location, number, and density of cell towers; the land use and activity density; and the size of the zones used in the analysis. spatial resolution–tower-based CDR data. The spatial resolution of these data is determined by the density of cell towers, which varies from a few hundred meters in metropolitan areas to a few kilometers in rural regions. If a geographic area or zone that includes one or more cell towers is used, phone activity routed through a tower will result in a record with the zone location. The spatial resolution of location records greatly depends on the size of these zones. stay point. A stay or stay point corresponds to an activity location where the device and the individual user are engaged in an activity. stay region. Different stay points identified in a user’s several trajectories may refer to the same location, but the coordinates of these stay points are unlikely to be exactly the same. A grid- based clustering method is used to cluster stay points to obtain stay regions.

Glossary 135 stays and stay extraction. The identification of stays is important to identifying devices of users who engage in activities at a location instead of simply passing by the location while traveling along a trajectory. temporal resolution. Temporal resolution is the precision of a measurement with respect to time. The frequency of cell phone use for events such as calls, text messages, and Internet data access; the daily patterns of cell phone use; and the distribution of cell phone use over a typical day are important elements of temporal resolution. The quality of temporal resolu- tion depends on the mechanism that triggers what is recorded. Originally, each record cor- responded to calls made by cell phone users. In recent years, a record is generated each time an activity is performed on the cell phone, including calling, texting, and Internet browsing, resulting in a finer temporal resolution. tower-based CDR data. When a cell phone device connects to cellular networks for a call, message, or data transmission, a cellular tower ID is recorded with a time stamp. The tower ID identifies the cellular tower to which the device connected when its user made a call, sent a text message, or used data. triangulated CDR data. As technology advances and cell phones are used more frequently, the location of a device can be pinpointed more accurately while it connects to operators’ service networks. A time stamp records the connection of a cell phone to the network. Longitude- and latitude-pinpointed coordinate pairs are estimated with a reported accuracy of 200 to 300 meters. triangulation. The location of a cell phone device at any given time can be approximated by triangulating the signals sent from two or more cell towers to a phone device. In more densely populated urban areas, where towers are closely spaced, the location of a cell phone can be determined more accurately. uncertainty in location estimates. In areas with a single cell tower, the location of the cell phone is approximate, cannot be triangulated, and falls within a radius. This is often the case in rural areas. In dense urban areas with a higher density of cell towers, multiple signals are sent to the device and its location can be approximated. Triangulation can be more accurate but may also result in false signals that suggest movement even when the device is stationary.