دانلود رایگان مقاله خوب، بد و پیامدهای پروفایل پوشش پهن باند تلفن همراه

Abstract

Pervasive coverage and continuous connectivity of Mobile Broadband (MBB) networks are common goals for regulators and operators. Given the increasing heterogeneity of technologies in the last mile of MBB networks, further support for seamless connectivity across multiple network types relies on understanding the prevalent network coverage profiles that capture different available technologies in an area. Correlating these coverage profiles with network performance metrics is of great importance in order to forestall disturbances for applications running on top of MBB networks. In this paper, we aim to profile MBB coverage and its performance implications from the end-user’s perspective along critical transport infrastructures (i.e., railways in Norway). For this, we deploy custom measurement nodes on-board five Norwegian inter-city trains and we collect a unique geo-tagged dataset along the train routes. We then build a coverage mosaic, where we divide the routes into segments and analyze the coverage of individual operators in each segment. We propose and evaluate the use of hierarchical clustering to describe prevalent coverage profiles of MBB networks along the train routes and classify each segment accordingly. We further analyze the areas we classify with each profile and assess the packet-loss and HTTP download performance of the networks in those areas.

8. Conclusions and future work

MBB networks are the key infrastructure for people to stay connected, especially in high mobility scenarios (e.g., when using public transport). MBB coverage profiling from the end-user experience while on critical public transport routes are of great importance to many stakeholders. At the same time, this is a challenging problem, since even a straight-forward classification of coverage into “good” or “bad” is very difficult to grasp in quantitative thresholds. In this paper, we evaluate the use of hierarchical clustering to build a coverage mosaic of MBB technologies in an area and analyze its implications in terms of network performance and application performance. By piggy-backing network measurements onto public transportation vehicles via the NNE platform, we first obtained a unique dataset that (i) captures the coverage and performance from user’s perspective and (ii) provides repetitive measurement runs on the same route, in similar conditions. Moreover, an important perk of such measurement platforms is allowing other parties, including public transport companies, to assess and compare the MBB coverage along their infrastructure to verify their service level agreement. We then leveraged hierarchical clustering in order to identify and characterize prevalent coverage profiles. Though in this study we look at the case of railways in Norway, the methodology can easily be generalized for running a similar study in other regions or applying it to a different datasets, (e.g. crowd-sourced data). A copy of the dataset we used in this paper is available for open access in Zenodo6, as well as the code for the clustering approach.