Analysis of crowdsourced data for estimating data speeds across service areas of India
- PDF / 1,539,079 Bytes
- 16 Pages / 595.276 x 790.866 pts Page_size
- 76 Downloads / 184 Views
Analysis of crowdsourced data for estimating data speeds across service areas of India V. Sridhar1
· K. Girish1 · M. Badrinarayan1
Accepted: 20 October 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract The intense adoption of Information Technology by businesses and government have increased data consumption across the world. While some countries have augmented their telecom infrastructure, data speeds are still very low in countries such as India. In this paper, we collected about 25 million records of crowdsourced data obtained through the mobile app deployed by the regulator in India. We have built a panel data regression model and analyzed the effect of supply-side variables such as radio spectrum holding of the operator, the mobile access infrastructure deployed by the operators, the technology deployed (3G/4G), and the demand side variable such as the mobile subscriber base. Our analysis indicates that a lower amount of spectrum holding, poor receive signal strength at mobile handsets, and the technology deployed (3G/4G) negatively affect the users’ download data speeds. The subscriber base also has a moderate effect on the data speeds. We conclude by prescribing policy recommendations on spectrum allocation and improvements in mobile access technologies to augment users’ quality of experience. Keywords Quality of service · Mobile broadband · Radio spectrum · Panel data analysis · 4G technology · Indian telecom, 3G services
1 Introduction Worldwide, Quality of Service (QoS) measurement and telecom network monitoring have been a tedious task for telecom regulators. One of the biggest challenges faced when dealing with QoS over the Internet is how to deliver QoS over an unregulated, connectionless network that was designed, deployed, operated, managed, and commercialized without any QoS perspectives [5, 21]. It was indicated by Vicente et al. [55] that Japan was found to be the only country out of 42 countries studied that was prepared to deliver the quality required for next-generation web applications. Further, with many autonomous systems managed by hundreds of network service providers, it is not easy to measure and benchmark end-to-end QoS for data services. The diversity of applications such as Internet Telephony, email, and video streaming adds to the complexity of QoS measurement. Though QoS parameters to check the end-to-end quality of
B 1
V. Sridhar [email protected]
calls and “user-perceived speech quality” using techniques such as Mean Opinion Score (MOS)/Perceptual Evaluation of Speech Quality (PESQ) exist for voice, comparative measurement techniques for video services are still evolving. The drawbacks of 3G mobile networks in providing adequate service quality for data and Internet services have been addressed through various network and application architectures [20]. QoS is also measured across different customer life cycle stages, including customer acquisition, retention, and exit. While QoS emphasizes measurable dimensions with more emphasis on tech
Data Loading...
