A Spatio-Temporal Analysis of Mobile Internet Traffic in Public Transportation Systems

Abstract

Mobile devices are ubiquitous, and mobile-generated traffic is arguably a major component of today's web traffic. In particular, the use of smart-phones whilst commuting using public transport is a very popular and common practice in many countries. Mobile commuters however, often suffer from poor performance due to limited bandwidth and/or intermittent network coverage. This paper provides insights into the characteristics of web traffic generated by mobile commuters in these challenged conditions of public transportation systems.We use a dataset collected from 22 Inter-city Buses running on 6 different routes over 5 weeks in Sweden. By analyzing content similarity in time and across different routes, we discover a number of findings that reveal the existence of a spatio-temporal correlation of content popularity and that shed light on diurnal patterns of behavior of mobile commuters. We study popular content accessed by commuters and show that Social Networking and News content are predominant and are mutually exclusive. One of the salient findings is that mobile users' interest on buses is very concentrated, with 35% of the popular content solely accessed on a single day during the 5 weeks, and more than 70% of the popular content from a given day is accessed during one single hour of the day. We also observe high content similarity between specific routes which suggests that content caching within the bus can significantly improve user web experience. Our results indicate that based on the observed strong spatio-temporal correlation of content requests of mobile commuters, caching content inside the buses leads to a daily hit rate ranging from 10 to 20%, with a 20% savings of the daily bandwidth usage.

Spatiotemporal Applications of Big Data

“…Content similarity also found high between specific routes. Caching of content reduced the load of bandwidth usage in these cases [15].…”

uStash: A Novel Mobile Content Delivery System for Improving User QoE in Public Transport

“…The overall optimum system performance (H optimum ) is defined as the point with minimum distance to the origin (0,0,0), if we give equal significance to all three metrics i.e. t , b , u = 1 as per the Equation 9. Hence, all three performance metrics are normalized in a range between 0 and 1.…”

“…Despite the high spatial locality of user interest, preliminary results of caching content at the bus with DS1 suggests that a cache hit rate of over 20% on average can be achieved [9], which is more than twice as high as caching at the cellular base stations as shown in [10]. The difference in hit rate, we believe, lies in the fact that a particular bus is small enough to carry a set of users with similar interest, contradicting the common believe that caching performance is poor at mobile edge.…”

Effective Resource Sharing in Mobile-Cell Environments

“…It has been shown that the mobile-cells can effectively eliminate VPE, reduce the number of handovers, enhance quality of service (QoS) for commuting users, and increase network throughput. In [12], the authors have performed experiments to demonstrate that a cache mechanism (such as in mobile-cells) can greatly reduce bandwidth consumption for the public transport users. If the employed cache already contains the contents required by a vehicular user, the MC will not need to access the core network.…”