On the Utilization of GSM Cell Id for Mobile Device Tracking and Notification Yazan A. Alqudah 1, Sufyan Almajali 2 1 Department of Communications Engineering, 2 Department of Software Engineering Princess Sumaya University for Technology Amman, Jordan y.alqudah@psut.edu.jo Abstract Location based services utilize mobile location to provide enhanced services to users. Although Global Positioning System (GPS) provides an accurate determination of mobile location, it can drain the mobile device battery and its reliability might be compromised in urban environments with dense and tall buildings. This work proposes a battery-conscious solution for locating a mobile device by tracking serving base stations. For applications where mobiles travel specified routes, such as school buses and public transportation, the mobile location may be determined by collecting serving based stations information along the routes. Our drive tests indicate that collected cell Id data is repeatable and can be used to reliably determine mobile location. Keywords Location Based Services; GSM;Cellular Networks. I. INTRODUCTION With the proliferation of mobile devices, providers and developers are introducing new applications that run on the mobile phone. New class of applications that takes advantage of users locations is termed location based services (LBS) [1-6]. Location based services (LBS) rely on the knowledge of the mobile device location. Several techniques are used to obtain location information. These include GPS and Cell Tower Id. The introduction of mobile devices equipped with Global Positioning System (GPS) means it is possible to obtain accurate information about device location for navigation and other services. Although GPS provides device location, it can weigh heavily on battery usage due to high power consumption. Obtaining location based on Cell Tower Id relies on the unique Cell Id used to identify the base station and known cell tower location. In 1996, the US government passed a mandate that requires network operators to locate emergency callers [1]. This prompted operators to provide added services to balance investments in location services. The introduction of GPS on mobile and 3G broadband connectivity motivated further interest in location based services. A tool called OneBusAway was introduced by the University of Washington [2]. The tool provides real-time arrival information and bus schedule route browser for the Seattle area transit. Users access information by using an identifier at bus stops. Users can also view bus and stop information on a map through a web browser. The service is implemented on the iphone and it communicates with back end servers to request information. In GSM network, a mobile location estimation based on differences of downlink signal attenuations between serving and neighboring stations is proposed in [7]. Positioning algorithms called the focused time delay neural network (FTDNN) and the distributed time delay neural network (DTDNN) are proposed to efficiently learn the mobile location from sequentially received signal strength in GSM network [8]. The algorithms extract location information from temporal variation of radio signal strengths. The objective of this work is to validate the ability to develop a mobile service that is triggered based on a change in cell Id in the GSM network. The service can be used to provide notifications based on this change. The mobile is assumed to travel a known route that is consistent. Typical applications of this service: 1. Parents notifications of school bus arrival and delivery 2. Hospital and emergency room notification of emergency vehicle locations 3. Notifications of the arrival of public transportation buses and trains. The service is required to have low power consumption to enable the mobile to run without a need for charge during operation. The notification is not required to give exact time/location to a recipient but rather notification with certain tolerance. This work is organized as follows: In Section II we review the cellular concept in mobile communications that enables notifications based on cell Id changes, Section III describes the architecture of the notification system and its components, and Section IV introduces typical operation. The results of validation and field measurements are presented in Section V. Finally, the conclusion is provided in Section VI. ISBN: 978-1-941968-18-5 2015 SDIWC 46
II. BACKGROUND The widespread usage of mobile devices demands that service providers ensure uninterrupted coverage in usage area. This is made possible by dividing coverage area into cells and reutilizing resource in these cells. A mobile in service is connected to the base station that can offer the best quality of service. This is often based on signal strength. As the mobile travels, it stays connected to the base station as long as the signal strength meets a prescribed value. As the mobile reaches the border of a cell, the signal strength drops. The mobile searches for coverage in a neighboring cell. A handover is said to take place when the mobile connects to a new base station and leaves the serving base station. To ensure uninterrupted service, the mobile tries to connect to another base station before the signal with serving base station is lost. This process is referred to as handover. The new base station serves the mobile as long as it is within its coverage area. As the mobile moves farther, the signal strength drops. The mobile will then search for a new serving base station. This process continues as long as the mobile is moving in the coverage area. The knowledge of base station and cell Id can be used to determine the approximate location of the mobile. In this case, the mobile can be anywhere in the coverage area. Therefore, the accuracy of the location information is inversely proportional to the coverage area. To improve the location accuracy, the handover region is chosen to indicate mobile location. This region is defined by the intersection of the two base stations coverage areas. Fig. 1. Coverage of service area using cells. As the vehicle travels, it is connected to different cells. Figure 1 depicts a vehicle traveling in a coverage area. As the vehicle travels, it is connected to BS1, BS4, and BS5, consecutively. Our service takes advantage of the change in base station to notify recipients of the approximate location of the vehicle and the approximate arrival time. III. BS1 BS2 BS3 BS4 BS5 SYSTEM DESIGN AND ARCHITECTURE In this section, we will describe the architecture and interaction of different system components. The application is proposed for school bus notifications. The route of the school bus is assumed fixed and does not change during the school year. Fig. 2: Components of the notifications system. The web browser is used to populate records in database. The mobile query database to get data a. Database DB Web browser The database contains information about the routes and their cell Id sequence, the notification messages, and the recipient s information. A table that contains previous and current base station Id, notification message, and recipient phone number are shown below in Table 1. Table I: Notifications table in database Route Current Id Previous Id Message Recipient 1 102 109 arrives in 10 min 07991112224 1 222 245 arrives in 5 min 07999339399 1 123 333 arrives in 20 min 07899999929 During operation, the mobile searches for a matching current and previous cell Id in the table. If a match is found, the mobile sends the notification message to the recipient. The message is written taking into account the proximity of cell change and the recipient s home location. This information is known and the table is populated during the route set up. b. Web Site The web site is used to create, update and delete records from the database via a web browser. When a bus route is created, records of messages that have to be sent to parents have to be added. For each message, previous and current cell Ids are added. The message should consider the approximate travel time between cell transition and recipient location. c. Mobile Service The mobile service runs on the mobile device of the driver or attendant and upon launch, it connects to the database to get route Id s and messages. It proceeds to monitor the cell Id of its serving base station. Upon change in base station Id, the service searches the database to see if any action is required. If cell Id transition is tied to a notification, the service obtains the notification message and recipient phone number. It sends a short message with the notification. The solution consists of a service that runs on a mobile device, a database that hosts routes information and notifications and web interface for updating the database. These components are illustrated in Fig. 2. Below we describe the components of the system. ISBN: 978-1-941968-18-5 2015 SDIWC 47
Along with collecting route information, the bus is required to stop and pick up students. Each time a student is picked up, a record is created that contains student Id, GPS information, and Step Id generating while picking up the student. Table 2 shows an example student data collected during the discovery stage. Table II: Student Data Collected During Discovery Stage Student Id GPS and Cell Info Step Id ( or range of Step Id s) S1 Lat1 and long 1, cell info 17-21 S2 Lat2 and long 2, cell info 50-54.. Sm. Fig. 3: System Architecture IV. OPERATION The solutions works using three stages: Registration, Discovery, and Usage. Below we describe each of these stages. Registration Stage: Before collecting route detailed information, the solution allows the school to establish the basic information of routes to be used by school s busses. Basic information includes identifying the names of the routes. In addition, it includes identifying the names of the students along with the routes they are going to be part of. This can happen via online form access or mobile application access. In addition, students parents can determine the timing specific for their kids, such as receiving a SMS notification 5 minutes before bus arrival. Discovery Stage: This stage happens at the beginning of the overall application use and after the registration stage. The purpose of this stage is to collect information about all students who are part of the bus route. Information includes students names, route GPS information, timing information, and cell towers information. The daily route of a bus starts at a specific point and continues through a set of streets where the bus stops to pick up students at certain points along the route. In this stage, the mobile application starts by connecting online, entering route Id, and downloading the initial route database which includes the names of the students located on this route. Once this happens, a bus can move to start picking up students. A specific point within the bus route will be designated as the start point (P1). Once this point is reached, the discovery mode is activated. The application starts recording information every X seconds. Collected information includes the GPS information and cell towers information available at the bus current location. Table 2 shows a sample of the information collected at this stage starting from point P1 to point Pn, where Pn is reached when the bus reaches the school. The discovery stage is designed to occur once in the beginning. If there is a change to the route at later times, there will be a need to run this stage again. Once the discovery stage is finished, the database of the discovered route is updated online. Usage Stage: The usage stage occurs once a day, each time the route is followed. In this stage, the application user is required to activate the usage stage once the bus reaches the starting point P1, which was defined during the discovery stage. Detecting P1 can happen automatically as well. Once P1 is reached, the system calculates the estimated arrival time (EAT) for each student using previously gathered discovery stage information. Along with estimated arrival time, the system calculates the notification trigger time (NTT) for each student based on EAT. Next, the application starts a continuous comparison process with the existing database and sends notifications to students if NTT of that student is reached. NTT is measured using step Ids. V. VALIDATION To validate the developed solution, a mobile application is built to collect information about the serving base stations. The application collects time, serving base station, and GPS locations. The number of samples collected is determined by the sampling rate (a configuration parameter that can be specified). Android platform was chosen for implementation due to its open nature and popularity in Jordan. The application is implemented as a service that runs in the background of the device allowing users to utilize the phone during operation. The data is collected in a log file that is stored on the mobile s storage memory. To validate the ability of the application to determine location based on changes to cell Id, real time data is needed. To collect the data, drive tests were conducted to collect cell Id and GPS locations. The path of the drive test path is illustrated in Fig. 4 below. The path is approximately 4 kilometers long ISBN: 978-1-941968-18-5 2015 SDIWC 48
and it loops to return to its starting point. In real application, the full path (or part) of the bus route can be used. The objective of the drive test is to see whether changes in cell Id are captured by the mobile device and whether the changes are consistent and repeatable. The simplest and least accurate method to determine location is for the mobile device to maintain current and up to N-1 previous cell Id s. The mobile compares the N values of stored values and if the majority of the cell Ids match, the mobile can be certain that it is in a specified location. In the above example, 7 previous cell Ids are maintained, when the majority of the cells match, the mobile is said to have traveled the path above. A more accurate localization method can be obtained by observing Fig. 5. The figure shows that all drive tests will have cell Id 2501. However, this cell Id is obtained at different distance traveled. The distance varies between approximately 1600 and 3000 meters. Since this cell Id can occur more than once during a single drive, the previous cell Id can be used to uniquely identify this location. Assuming a 30km/h vehicle speed, this distance will correspond to approximately a 3 minutes variation in the arrival of a notification based on the serving cell Id being equal to 2501. Fig. 4: Path selected for collecting cell Id data Several drive tests were conducted to collect information about the serving cell Id. Table 3 below shows the results of 14 drive tests conducted over several days. The table shows the changes in cell Id as the mobile travels along the test path. The last digit in the cell Id refers to the sector Id (i.e. same cell Id can have different least significant digits indicating different sectors). Fig. 5: Changes in Cell Id vs. distance for drive tests Cell Id Table III: Cell Id s changes during drive tests. Drive 1 2 3 4 5 6 7 8 9 10 11 12 13 14 25162 25162 25213 25213 25162 25213 25162 25162 25404 25213 25212 25213 25162 25213 25011 25441 25441 25441 25011 25162 25404 25404 25441 25404 25404 25162 25441 25162 25401 25011 25403 25381 25401 25401 25441 25441 25381 25441 25011 25404 25381 25441 25441 25332 25381 25011 25441 25011 25381 25381 25011 25381 25403 25441 25011 25381 25381 25441 25012 25404 25381 25404 25011 25011 25403 25011 25011 25403 56243 25011 25011 25403 25011 25441 25011 25441 25332 25381 25441 25381 25381 25011 25403 25381 25403 25011 25404 25212 25403 25381 25401 25332 25213 25162 25332 25441 25441 25162 25441 25404 25441 25441 25011 25441 25441 25441 25441 25213 25404 25441 25213 25441 25213 25213 25332 25212 25404 25404 25404 25213 25404 25213 25441 25212 25211 25213 25212 25213 The cell Id vs. distance and time are plotted in Fig 5 and 6, respectively. The table and the graphs indicate that during the 14 drive tests, a total of 22 unique cell Id s are encountered. These cells belong to 13 different base stations (the last digit in the cell Id indicates the sector). The maximum number of cell Id changes is 11 and on average 8 cell Id s changes occur in each drive test. There are 3 common base station Id s in all drive tests. These are 2501, 2540, and 2544. Future work will consider the most efficient and accurate ways to utilize cell information to determine location. We briefly present some ideas based on observed results. Fig. 6: Changes in Cell Id vs. time for drive tests ISBN: 978-1-941968-18-5 2015 SDIWC 49
VI. CONCLUSION This work proposed a solution that enables location based notification by utilizing cellular communication infrastructure. By capturing the serving base stations along a travel path, a mobile device is able to determine its locations. This enables a notification to be sent of arrival time. Although GPS can be used for accurate location determination, GPS suffers from high energy consumption that weighs heavily on the mobile device s battery. Another disadvange of GPS is the unrealiablity in urban enviroment where tall buildings can block GPS signal. This work presented data of serving cell Id collected during drive tests. The results show that it will be possible to develop a technique that utilize cell Id s to determine the location of the mobile device a long known path. REFERENCES [1] Vaughan-Nichols, Steven J. "Will mobile computing's future be location, location, location?." Computer 42, no. 2 (2009): 14-17. [2] Ferris, Brian, Kari Watkins, and Alan Borning. "Location-aware tools for improving public transit usability." IEEE Pervasive Computing 1 (2009): 13-19. [3] Bozzon, Alessandro, Marco Brambilla, Stefano Ceri, and Silvia Quarteroni. "A framework for integrating, exploring, and searching location-based web data."internet Computing, IEEE 15, no. 6 (2011): 24-31. [4] Barkhuus, Louise, and Anind K. Dey. "Location-Based Services for Mobile Telephony: a Study of Users' Privacy Concerns." In INTERACT, vol. 3, pp. 702-712. 2003. [5] Kjærgaard, Mikkel Baun. "Location-based services on mobile phones: minimizing power consumption." IEEE Pervasive Computing 1 (2010): 67-73. [6] Bellavista, Paolo, Axel Kupper, and Sumi Helal. "Location-based services: Back to the future." Pervasive Computing, IEEE 7, no. 2 (2008): 85-89. [7] Lin, Ding-Bing, and Rong-Terng Juang. "Mobile location estimation based on differences of signal attenuations for GSM systems." Vehicular Technology, IEEE Transactions on 54, no. 4 (2005): 1447-1454. [8] Fang, Shih-Hau, Bo-Cheng Lu, and Ying-Tso Hsu. "Learning location from sequential signal strength based on GSM experimental data." Vehicular Technology, IEEE Transactions on 61, no. 2 (2012): 726-736. ISBN: 978-1-941968-18-5 2015 SDIWC 50