University of Twente Student Theses


Efficient proximity detection among mobile clients using the GSM network

Witteman, M.T. (2007) Efficient proximity detection among mobile clients using the GSM network.

[img] PDF
Abstract:Personalized services, such as customized ringtone recommendations and location based services increase the popularity of mobile phones. A location based service (LBS) is a type of personalized service that uses the location of the mobile user. Proximity detection is defined as the capability of a location based service to automatically detect when a pair of mobile users is closer to each other than a certain proximity distance. To realize this function the location of the mobile users have to be permanently tracked. The main objective of this research is to find to what extent it is possible to use GSM cell-ids for proximity detections among mobile entities. We studied the typical GSM characteristics and the state of the art proximity detection algorithms found in literature. All the proximity algorithms we found use X, Y coordinates as location identifier. Dynamic Circles by George True et. al. was stated in literature to perform the best. This algorithm is used as inspiration for four new algorithms that use the GSM network to implement proximity detection. The proposed algorithms are divided into two groups, the basic group and the graph group. The basic group contains two algorithms that use the information available in the GSM network, called cell-id algorithm and LAC algorithm. The graph group uses a cell-id graph representing all the cell switches, to get more efficient algorithms. This group contains two algorithms, named Graph1 algorithm and Graph2 algorithm. For the comparison of the algorithms a simulator was developed. The simulator can simulate the amount of messages, the traffic over TCP/IP, missed proximities and the accuracy of proximity distance using GSM. The input data of the simulator are user movement paths. These paths can be generated or can be real logged data gathered by a mobile device. For a realistic cell topology it was found that the median of differences between real distance and cell-id distance is around 1100 meters. Using GSM cells as location identifier introduces missed proximities, i.e. false negatives. We found that for a realistic cell layer and a proximity distance of 1100 meters the number of missed proximities is 3.5% of the total proximity detections. A simulation of a distributed scenario, representing rush hour, stated that for low density of users the Graph algorithms outperform the Basic algorithms in terms of generated messages and traffic. From this simulation it is found that for 30 buddies a user consumes a total of 30MB of data traffic per month.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page